diff --git a/.clang-tidy b/.clang-tidy
index a30f9e592..c02b7152f 100644
--- a/.clang-tidy
+++ b/.clang-tidy
@@ -64,8 +64,8 @@ Checks: '*,
-readability-identifier-length,
-misc-no-recursion,
-concurrency-mt-unsafe,
- -bugprone-easily-swappable-parameters'
-
+ -bugprone-easily-swappable-parameters,
+ -bugprone-unchecked-optional-access'
WarningsAsErrors: ''
HeaderFilterRegex: 'src/.*'
AnalyzeTemporaryDtors: false
diff --git a/.github/workflows/diff.yaml b/.github/workflows/diff.yaml
index a2dc0aef2..49b7d4273 100644
--- a/.github/workflows/diff.yaml
+++ b/.github/workflows/diff.yaml
@@ -19,11 +19,16 @@ on:
jobs:
community_build:
name: "Community build"
- runs-on: [self-hosted, Linux, X64, Diff]
+ runs-on: [self-hosted, Linux, X64, DockerMgBuild]
+ timeout-minutes: 60
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
+ OS: debian-11
+ TOOLCHAIN: v5
+ ARCH: amd
+ BUILD_TYPE: RelWithDebInfo
steps:
- name: Set up repository
@@ -33,35 +38,56 @@ jobs:
# branches and tags. (default: 1)
fetch-depth: 0
- - name: Build community binaries
+ - name: Spin up mgbuild container
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ run
- # Initialize dependencies.
- ./init
-
- # Build community binaries.
- cd build
- cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo -DMG_ENTERPRISE=OFF ..
- make -j$THREADS
+ - name: Build release binaries
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph --community
- name: Run unit tests
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph unit
- # Run unit tests.
- cd build
- ctest -R memgraph__unit --output-on-failure -j$THREADS
+ - name: Stop mgbuild container
+ if: always()
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ stop --remove
code_analysis:
name: "Code analysis"
- runs-on: [self-hosted, Linux, X64, Diff]
+ runs-on: [self-hosted, Linux, X64, DockerMgBuild]
+ timeout-minutes: 60
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
+ OS: debian-11
+ TOOLCHAIN: v5
+ ARCH: amd
+ BUILD_TYPE: Debug
steps:
- name: Set up repository
@@ -71,6 +97,14 @@ jobs:
# branches and tags. (default: 1)
fetch-depth: 0
+ - name: Spin up mgbuild container
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ run
+
# This is also needed if we want do to comparison against other branches
# See https://github.community/t/checkout-code-fails-when-it-runs-lerna-run-test-since-master/17920
- name: Fetch all history for all tags and branches
@@ -78,11 +112,13 @@ jobs:
- name: Initialize deps
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Initialize dependencies.
- ./init
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph --init-only
- name: Set base branch
if: ${{ github.event_name == 'pull_request' }}
@@ -96,45 +132,43 @@ jobs:
- name: Python code analysis
run: |
- CHANGED_FILES=$(git diff -U0 ${{ env.BASE_BRANCH }}... --name-only --diff-filter=d)
- for file in ${CHANGED_FILES}; do
- echo ${file}
- if [[ ${file} == *.py ]]; then
- python3 -m black --check --diff ${file}
- python3 -m isort --profile black --check-only --diff ${file}
- fi
- done
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph code-analysis --base-branch "${{ env.BASE_BRANCH }}"
- name: Build combined ASAN, UBSAN and coverage binaries
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- cd build
- cmake -DTEST_COVERAGE=ON -DASAN=ON -DUBSAN=ON ..
- make -j$THREADS memgraph__unit
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph --coverage --asan --ubsan
- name: Run unit tests
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Run unit tests. It is restricted to 2 threads intentionally, because higher concurrency makes the timing related tests unstable.
- cd build
- LSAN_OPTIONS=suppressions=$PWD/../tools/lsan.supp UBSAN_OPTIONS=halt_on_error=1 ctest -R memgraph__unit --output-on-failure -j2
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph unit-coverage
- name: Compute code coverage
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Compute code coverage.
- cd tools/github
- ./coverage_convert
-
- # Package code coverage.
- cd generated
- tar -czf code_coverage.tar.gz coverage.json html report.json summary.rmu
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph code-coverage
- name: Save code coverage
uses: actions/upload-artifact@v4
@@ -144,21 +178,36 @@ jobs:
- name: Run clang-tidy
run: |
- source /opt/toolchain-v4/activate
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph clang-tidy --base-branch "${{ env.BASE_BRANCH }}"
- # Restrict clang-tidy results only to the modified parts
- git diff -U0 ${{ env.BASE_BRANCH }}... -- src | ./tools/github/clang-tidy/clang-tidy-diff.py -p 1 -j $THREADS -path build -regex ".+\.cpp" | tee ./build/clang_tidy_output.txt
-
- # Fail if any warning is reported
- ! cat ./build/clang_tidy_output.txt | ./tools/github/clang-tidy/grep_error_lines.sh > /dev/null
+ - name: Stop mgbuild container
+ if: always()
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ stop --remove
debug_build:
name: "Debug build"
- runs-on: [self-hosted, Linux, X64, Diff]
+ runs-on: [self-hosted, Linux, X64, DockerMgBuild]
+ timeout-minutes: 100
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
+ OS: debian-11
+ TOOLCHAIN: v5
+ ARCH: amd
+ BUILD_TYPE: Debug
steps:
- name: Set up repository
@@ -168,58 +217,95 @@ jobs:
# branches and tags. (default: 1)
fetch-depth: 0
- - name: Build debug binaries
+ - name: Spin up mgbuild container
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ run
- # Initialize dependencies.
- ./init
-
- # Build debug binaries.
- cd build
- cmake ..
- make -j$THREADS
+ - name: Build release binaries
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph
- name: Run leftover CTest tests
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Run leftover CTest tests (all except unit and benchmark tests).
- cd build
- ctest -E "(memgraph__unit|memgraph__benchmark)" --output-on-failure
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph leftover-CTest
- name: Run drivers tests
run: |
- ./tests/drivers/run.sh
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph drivers
- name: Run integration tests
run: |
- tests/integration/run.sh
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph integration
- name: Run cppcheck and clang-format
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Run cppcheck and clang-format.
- cd tools/github
- ./cppcheck_and_clang_format diff
-
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph cppcheck-and-clang-format
+
- name: Save cppcheck and clang-format errors
uses: actions/upload-artifact@v4
with:
name: "Code coverage(Debug build)"
path: tools/github/cppcheck_and_clang_format.txt
+ - name: Stop mgbuild container
+ if: always()
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ stop --remove
+
release_build:
name: "Release build"
- runs-on: [self-hosted, Linux, X64, Diff]
+ runs-on: [self-hosted, Linux, X64, DockerMgBuild]
+ timeout-minutes: 100
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
+ OS: debian-11
+ TOOLCHAIN: v5
+ ARCH: amd
+ BUILD_TYPE: Release
steps:
- name: Set up repository
@@ -229,26 +315,33 @@ jobs:
# branches and tags. (default: 1)
fetch-depth: 0
+ - name: Spin up mgbuild container
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ run
- name: Build release binaries
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Initialize dependencies.
- ./init
-
- # Build release binaries.
- cd build
- cmake -DCMAKE_BUILD_TYPE=Release ..
- make -j$THREADS
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph
- name: Run GQL Behave tests
run: |
- cd tests
- ./setup.sh /opt/toolchain-v4/activate
- cd gql_behave
- ./continuous_integration
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph gql-behave
- name: Save quality assurance status
uses: actions/upload-artifact@v4
@@ -260,14 +353,19 @@ jobs:
- name: Run unit tests
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Run unit tests.
- cd build
- ctest -R memgraph__unit --output-on-failure -j$THREADS
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --threads $THREADS \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph unit
+ # This step will be skipped because the e2e stream tests have been disabled
+ # We need to fix this as soon as possible
- name: Ensure Kafka and Pulsar are up
+ if: false
run: |
cd tests/e2e/streams/kafka
docker-compose up -d
@@ -276,13 +374,17 @@ jobs:
- name: Run e2e tests
run: |
- cd tests
- ./setup.sh /opt/toolchain-v4/activate
- source ve3/bin/activate_e2e
- cd e2e
- ./run.sh
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph e2e
+ # Same as two steps prior
- name: Ensure Kafka and Pulsar are down
+ if: false
run: |
cd tests/e2e/streams/kafka
docker-compose down
@@ -291,59 +393,92 @@ jobs:
- name: Run stress test (plain)
run: |
- cd tests/stress
- source ve3/bin/activate
- ./continuous_integration
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph stress-plain
- name: Run stress test (SSL)
run: |
- cd tests/stress
- source ve3/bin/activate
- ./continuous_integration --use-ssl
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph stress-ssl
- name: Run durability test
run: |
- cd tests/stress
- source ve3/bin/activate
- python3 durability --num-steps 5
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph durability
- name: Create enterprise DEB package
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
- cd build
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ package-memgraph
- # create mgconsole
- # we use the -B to force the build
- make -j$THREADS -B mgconsole
-
- # Create enterprise DEB package.
- mkdir output && cd output
- cpack -G DEB --config ../CPackConfig.cmake
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ copy --package
- name: Save enterprise DEB package
uses: actions/upload-artifact@v4
with:
name: "Enterprise DEB package"
- path: build/output/memgraph*.deb
+ path: build/output/${{ env.OS }}/memgraph*.deb
+
+ - name: Copy build logs
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ copy --build-logs
- name: Save test data
uses: actions/upload-artifact@v4
if: always()
with:
name: "Test data(Release build)"
- path: |
- # multiple paths could be defined
- build/logs
+ path: build/logs
+
+ - name: Stop mgbuild container
+ if: always()
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ stop --remove
release_jepsen_test:
name: "Release Jepsen Test"
- runs-on: [self-hosted, Linux, X64, Debian10, JepsenControl]
- #continue-on-error: true
+ runs-on: [self-hosted, Linux, X64, DockerMgBuild]
+ timeout-minutes: 80
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
+ OS: debian-12
+ TOOLCHAIN: v5
+ ARCH: amd
+ BUILD_TYPE: RelWithDebInfo
steps:
- name: Set up repository
@@ -353,16 +488,31 @@ jobs:
# branches and tags. (default: 1)
fetch-depth: 0
+ - name: Spin up mgbuild container
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ run
+
- name: Build release binaries
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
- # Initialize dependencies.
- ./init
- # Build only memgraph release binarie.
- cd build
- cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo ..
- make -j$THREADS memgraph
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph
+
+ - name: Copy memgraph binary
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ copy --binary
- name: Refresh Jepsen Cluster
run: |
@@ -381,13 +531,27 @@ jobs:
name: "Jepsen Report"
path: tests/jepsen/Jepsen.tar.gz
+ - name: Stop mgbuild container
+ if: always()
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ stop --remove
+
release_benchmarks:
name: "Release benchmarks"
- runs-on: [self-hosted, Linux, X64, Diff, Gen7]
+ runs-on: [self-hosted, Linux, X64, DockerMgBuild, Gen7]
+ timeout-minutes: 60
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
+ OS: debian-11
+ TOOLCHAIN: v5
+ ARCH: amd
+ BUILD_TYPE: Release
steps:
- name: Set up repository
@@ -397,25 +561,33 @@ jobs:
# branches and tags. (default: 1)
fetch-depth: 0
+ - name: Spin up mgbuild container
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ run
+
- name: Build release binaries
run: |
- # Activate toolchain.
- source /opt/toolchain-v4/activate
-
- # Initialize dependencies.
- ./init
-
- # Build only memgraph release binaries.
- cd build
- cmake -DCMAKE_BUILD_TYPE=release ..
- make -j$THREADS
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --build-type $BUILD_TYPE \
+ --threads $THREADS \
+ build-memgraph
- name: Run macro benchmarks
run: |
- cd tests/macro_benchmark
- ./harness QuerySuite MemgraphRunner \
- --groups aggregation 1000_create unwind_create dense_expand match \
- --no-strict
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph macro-benchmark
- name: Get branch name (merge)
if: github.event_name != 'pull_request'
@@ -429,30 +601,49 @@ jobs:
- name: Upload macro benchmark results
run: |
- cd tools/bench-graph-client
- virtualenv -p python3 ve3
- source ve3/bin/activate
- pip install -r requirements.txt
- ./main.py --benchmark-name "macro_benchmark" \
- --benchmark-results "../../tests/macro_benchmark/.harness_summary" \
- --github-run-id "${{ github.run_id }}" \
- --github-run-number "${{ github.run_number }}" \
- --head-branch-name "${{ env.BRANCH_NAME }}"
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph upload-to-bench-graph \
+ --benchmark-name "macro_benchmark" \
+ --benchmark-results "../../tests/macro_benchmark/.harness_summary" \
+ --github-run-id ${{ github.run_id }} \
+ --github-run-number ${{ github.run_number }} \
+ --head-branch-name ${{ env.BRANCH_NAME }}
- # TODO (andi) No need for path flags and for --disk-storage and --in-memory-analytical
- name: Run mgbench
run: |
- cd tests/mgbench
- ./benchmark.py vendor-native --num-workers-for-benchmark 12 --export-results benchmark_result.json pokec/medium/*/*
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph mgbench
- name: Upload mgbench results
run: |
- cd tools/bench-graph-client
- virtualenv -p python3 ve3
- source ve3/bin/activate
- pip install -r requirements.txt
- ./main.py --benchmark-name "mgbench" \
- --benchmark-results "../../tests/mgbench/benchmark_result.json" \
- --github-run-id "${{ github.run_id }}" \
- --github-run-number "${{ github.run_number }}" \
- --head-branch-name "${{ env.BRANCH_NAME }}"
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ --enterprise-license $MEMGRAPH_ENTERPRISE_LICENSE \
+ --organization-name $MEMGRAPH_ORGANIZATION_NAME \
+ test-memgraph upload-to-bench-graph \
+ --benchmark-name "mgbench" \
+ --benchmark-results "../../tests/mgbench/benchmark_result.json" \
+ --github-run-id "${{ github.run_id }}" \
+ --github-run-number "${{ github.run_number }}" \
+ --head-branch-name "${{ env.BRANCH_NAME }}"
+
+ - name: Stop mgbuild container
+ if: always()
+ run: |
+ ./release/package/mgbuild.sh \
+ --toolchain $TOOLCHAIN \
+ --os $OS \
+ --arch $ARCH \
+ stop --remove
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 028406447..c02039497 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -300,6 +300,19 @@ endif()
option(ENABLE_JEMALLOC "Use jemalloc" ON)
+option(MG_MEMORY_PROFILE "If build should be setup for memory profiling" OFF)
+if (MG_MEMORY_PROFILE AND ENABLE_JEMALLOC)
+ message(STATUS "Jemalloc has been disabled because MG_MEMORY_PROFILE is enabled")
+ set(ENABLE_JEMALLOC OFF)
+endif ()
+if (MG_MEMORY_PROFILE AND ASAN)
+ message(STATUS "ASAN has been disabled because MG_MEMORY_PROFILE is enabled")
+ set(ASAN OFF)
+endif ()
+if (MG_MEMORY_PROFILE)
+ add_compile_definitions(MG_MEMORY_PROFILE)
+endif ()
+
if (ASAN)
message(WARNING "Disabling jemalloc as it doesn't work well with ASAN")
set(ENABLE_JEMALLOC OFF)
diff --git a/environment/os/rocky-9.3.sh b/environment/os/rocky-9.3.sh
index 571278654..6ec045b42 100755
--- a/environment/os/rocky-9.3.sh
+++ b/environment/os/rocky-9.3.sh
@@ -59,7 +59,7 @@ MEMGRAPH_BUILD_DEPS=(
doxygen graphviz # source documentation generators
which nodejs golang custom-golang1.18.9 # for driver tests
zip unzip java-11-openjdk-devel java-17-openjdk java-17-openjdk-devel custom-maven3.9.3 # for driver tests
- sbcl # for custom Lisp C++ preprocessing
+ cl-asdf common-lisp-controller sbcl # for custom Lisp C++ preprocessing
autoconf # for jemalloc code generation
libtool # for protobuf code generation
cyrus-sasl-devel
@@ -162,6 +162,30 @@ install() {
fi
continue
fi
+ if [ "$pkg" == doxygen ]; then
+ if ! dnf list installed doxygen >/dev/null 2>/dev/null; then
+ dnf install -y https://dl.rockylinux.org/pub/rocky/9/CRB/x86_64/os/Packages/d/doxygen-1.9.1-11.el9.x86_64.rpm
+ fi
+ continue
+ fi
+ if [ "$pkg" == cl-asdf ]; then
+ if ! dnf list installed cl-asdf >/dev/null 2>/dev/null; then
+ dnf install -y https://pkgs.sysadmins.ws/el8/base/x86_64/cl-asdf-20101028-18.el8.noarch.rpm
+ fi
+ continue
+ fi
+ if [ "$pkg" == common-lisp-controller ]; then
+ if ! dnf list installed common-lisp-controller >/dev/null 2>/dev/null; then
+ dnf install -y https://pkgs.sysadmins.ws/el8/base/x86_64/common-lisp-controller-7.4-20.el8.noarch.rpm
+ fi
+ continue
+ fi
+ if [ "$pkg" == sbcl ]; then
+ if ! dnf list installed sbcl >/dev/null 2>/dev/null; then
+ dnf install -y https://pkgs.sysadmins.ws/el8/base/x86_64/sbcl-2.0.1-4.el8.x86_64.rpm
+ fi
+ continue
+ fi
if [ "$pkg" == PyYAML ]; then
if [ -z ${SUDO_USER+x} ]; then # Running as root (e.g. Docker).
pip3 install --user PyYAML
diff --git a/environment/os/run.sh b/environment/os/run.sh
index e7c370f62..654636a5f 100755
--- a/environment/os/run.sh
+++ b/environment/os/run.sh
@@ -5,17 +5,20 @@ IFS=' '
# NOTE: docker_image_name could be local image build based on release/package images.
# NOTE: each line has to be under quotes, docker_container_type, script_name and docker_image_name separate with a space.
# "docker_container_type script_name docker_image_name"
+# docker_container_type OPTIONS:
+# * mgrun -> running plain/empty operating system for the purposes of testing native memgraph package
+# * mgbuild -> running the builder container to build memgraph inside it -> it's possible create builder images using release/package/run.sh
OPERATING_SYSTEMS=(
- "mgrun amzn-2 amazonlinux:2"
- "mgrun centos-7 centos:7"
- "mgrun centos-9 dokken/centos-stream-9"
- "mgrun debian-10 debian:10"
- "mgrun debian-11 debian:11"
- "mgrun fedora-36 fedora:36"
- "mgrun ubuntu-18.04 ubuntu:18.04"
- "mgrun ubuntu-20.04 ubuntu:20.04"
- "mgrun ubuntu-22.04 ubuntu:22.04"
- # "mgbuild centos-7 package-mgbuild_centos-7"
+ # "mgrun amzn-2 amazonlinux:2"
+ # "mgrun centos-7 centos:7"
+ # "mgrun centos-9 dokken/centos-stream-9"
+ # "mgrun debian-10 debian:10"
+ # "mgrun debian-11 debian:11"
+ # "mgrun fedora-36 fedora:36"
+ # "mgrun ubuntu-18.04 ubuntu:18.04"
+ # "mgrun ubuntu-20.04 ubuntu:20.04"
+ # "mgrun ubuntu-22.04 ubuntu:22.04"
+ # "mgbuild debian-12 memgraph/memgraph-builder:v5_debian-12"
)
if [ ! "$(docker info)" ]; then
@@ -33,14 +36,24 @@ print_help () {
# NOTE: This is an idempotent operation!
# TODO(gitbuda): Consider making docker_run always delete + start a new container or add a new function.
docker_run () {
- cnt_name="$1"
- cnt_image="$2"
+ cnt_type="$1"
+ if [[ "$cnt_type" != "mgbuild" && "$cnt_type" != "mgrun" ]]; then
+ echo "ERROR: Wrong docker_container_type -> valid options are mgbuild, mgrun"
+ exit 1
+ fi
+ cnt_name="$2"
+ cnt_image="$3"
if [ ! "$(docker ps -q -f name=$cnt_name)" ]; then
if [ "$(docker ps -aq -f status=exited -f name=$cnt_name)" ]; then
echo "Cleanup of the old exited container..."
docker rm $cnt_name
fi
- docker run -d --volume "$SCRIPT_DIR/../../:/memgraph" --network host --name "$cnt_name" "$cnt_image" sleep infinity
+ if [[ "$cnt_type" == "mgbuild" ]]; then
+ docker run -d --volume "$SCRIPT_DIR/../../:/memgraph" --network host --name "$cnt_name" "$cnt_image"
+ fi
+ if [[ "$cnt_type" == "mgrun" ]]; then
+ docker run -d --volume "$SCRIPT_DIR/../../:/memgraph" --network host --name "$cnt_name" "$cnt_image" sleep infinity
+ fi
fi
echo "The $cnt_image container is active under $cnt_name name!"
}
@@ -55,9 +68,9 @@ docker_stop_and_rm () {
cnt_name="$1"
if [ "$(docker ps -q -f name=$cnt_name)" ]; then
docker stop "$1"
- if [ "$(docker ps -aq -f status=exited -f name=$cnt_name)" ]; then
- docker rm "$1"
- fi
+ fi
+ if [ "$(docker ps -aq -f status=exited -f name=$cnt_name)" ]; then
+ docker rm "$1"
fi
}
@@ -71,7 +84,7 @@ start_all () {
docker_name="${docker_container_type}_$script_name"
echo ""
echo "~~~~ OPERATING ON $docker_image as $docker_name..."
- docker_run "$docker_name" "$docker_image"
+ docker_run "$docker_container_type" "$docker_name" "$docker_image"
docker_exec "$docker_name" "/memgraph/environment/os/$script_name.sh install NEW_DEPS"
echo "---- DONE EVERYHING FOR $docker_image as $docker_name..."
echo ""
diff --git a/include/_mgp.hpp b/include/_mgp.hpp
index 8b67bc36a..b1d9e26d5 100644
--- a/include/_mgp.hpp
+++ b/include/_mgp.hpp
@@ -326,6 +326,21 @@ inline mgp_vertex *graph_get_vertex_by_id(mgp_graph *g, mgp_vertex_id id, mgp_me
return MgInvoke<mgp_vertex *>(mgp_graph_get_vertex_by_id, g, id, memory);
}
+inline bool graph_has_text_index(mgp_graph *graph, const char *index_name) {
+ return MgInvoke<int>(mgp_graph_has_text_index, graph, index_name);
+}
+
+inline mgp_map *graph_search_text_index(mgp_graph *graph, const char *index_name, const char *search_query,
+ text_search_mode search_mode, mgp_memory *memory) {
+ return MgInvoke<mgp_map *>(mgp_graph_search_text_index, graph, index_name, search_query, search_mode, memory);
+}
+
+inline mgp_map *graph_aggregate_over_text_index(mgp_graph *graph, const char *index_name, const char *search_query,
+ const char *aggregation_query, mgp_memory *memory) {
+ return MgInvoke<mgp_map *>(mgp_graph_aggregate_over_text_index, graph, index_name, search_query, aggregation_query,
+ memory);
+}
+
inline mgp_vertices_iterator *graph_iter_vertices(mgp_graph *g, mgp_memory *memory) {
return MgInvoke<mgp_vertices_iterator *>(mgp_graph_iter_vertices, g, memory);
}
diff --git a/include/mg_procedure.h b/include/mg_procedure.h
index 93ef241d8..117dc66ab 100644
--- a/include/mg_procedure.h
+++ b/include/mg_procedure.h
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -891,6 +891,36 @@ enum mgp_error mgp_edge_iter_properties(struct mgp_edge *e, struct mgp_memory *m
enum mgp_error mgp_graph_get_vertex_by_id(struct mgp_graph *g, struct mgp_vertex_id id, struct mgp_memory *memory,
struct mgp_vertex **result);
+/// Result is non-zero if the index with the given name exists.
+/// The current implementation always returns without errors.
+enum mgp_error mgp_graph_has_text_index(struct mgp_graph *graph, const char *index_name, int *result);
+
+/// Available modes of searching text indices.
+MGP_ENUM_CLASS text_search_mode{
+ SPECIFIED_PROPERTIES,
+ REGEX,
+ ALL_PROPERTIES,
+};
+
+/// Search the named text index for the given query. The result is a map with the "search_results" and "error_msg" keys.
+/// The "search_results" key contains the vertices whose text-indexed properties match the given query.
+/// In case of a Tantivy error, the "search_results" key is absent, and "error_msg" contains the error message.
+/// Return mgp_error::MGP_ERROR_UNABLE_TO_ALLOCATE if there’s an allocation error while constructing the results map.
+/// Return mgp_error::MGP_ERROR_KEY_ALREADY_EXISTS if the same key is being created in the results map more than once.
+enum mgp_error mgp_graph_search_text_index(struct mgp_graph *graph, const char *index_name, const char *search_query,
+ enum text_search_mode search_mode, struct mgp_memory *memory,
+ struct mgp_map **result);
+
+/// Aggregate over the results of a search over the named text index. The result is a map with the "aggregation_results"
+/// and "error_msg" keys.
+/// The "aggregation_results" key contains the vertices whose text-indexed properties match the given query.
+/// In case of a Tantivy error, the "aggregation_results" key is absent, and "error_msg" contains the error message.
+/// Return mgp_error::MGP_ERROR_UNABLE_TO_ALLOCATE if there’s an allocation error while constructing the results map.
+/// Return mgp_error::MGP_ERROR_KEY_ALREADY_EXISTS if the same key is being created in the results map more than once.
+enum mgp_error mgp_graph_aggregate_over_text_index(struct mgp_graph *graph, const char *index_name,
+ const char *search_query, const char *aggregation_query,
+ struct mgp_memory *memory, struct mgp_map **result);
+
/// Creates label index for given label.
/// mgp_error::MGP_ERROR_NO_ERROR is always returned.
/// if label index already exists, result will be 0, otherwise 1.
diff --git a/include/mgp.hpp b/include/mgp.hpp
index 3f7ed591e..f35231062 100644
--- a/include/mgp.hpp
+++ b/include/mgp.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -32,6 +32,15 @@
namespace mgp {
+class TextSearchException : public std::exception {
+ public:
+ explicit TextSearchException(std::string message) : message_(std::move(message)) {}
+ const char *what() const noexcept override { return message_.c_str(); }
+
+ private:
+ std::string message_;
+};
+
class IndexException : public std::exception {
public:
explicit IndexException(std::string message) : message_(std::move(message)) {}
@@ -4306,12 +4315,12 @@ inline void AddParamsReturnsToProc(mgp_proc *proc, std::vector<Parameter> ¶m
}
} // namespace detail
-inline bool CreateLabelIndex(mgp_graph *memgaph_graph, const std::string_view label) {
- return create_label_index(memgaph_graph, label.data());
+inline bool CreateLabelIndex(mgp_graph *memgraph_graph, const std::string_view label) {
+ return create_label_index(memgraph_graph, label.data());
}
-inline bool DropLabelIndex(mgp_graph *memgaph_graph, const std::string_view label) {
- return drop_label_index(memgaph_graph, label.data());
+inline bool DropLabelIndex(mgp_graph *memgraph_graph, const std::string_view label) {
+ return drop_label_index(memgraph_graph, label.data());
}
inline List ListAllLabelIndices(mgp_graph *memgraph_graph) {
@@ -4322,14 +4331,14 @@ inline List ListAllLabelIndices(mgp_graph *memgraph_graph) {
return List(label_indices);
}
-inline bool CreateLabelPropertyIndex(mgp_graph *memgaph_graph, const std::string_view label,
+inline bool CreateLabelPropertyIndex(mgp_graph *memgraph_graph, const std::string_view label,
const std::string_view property) {
- return create_label_property_index(memgaph_graph, label.data(), property.data());
+ return create_label_property_index(memgraph_graph, label.data(), property.data());
}
-inline bool DropLabelPropertyIndex(mgp_graph *memgaph_graph, const std::string_view label,
+inline bool DropLabelPropertyIndex(mgp_graph *memgraph_graph, const std::string_view label,
const std::string_view property) {
- return drop_label_property_index(memgaph_graph, label.data(), property.data());
+ return drop_label_property_index(memgraph_graph, label.data(), property.data());
}
inline List ListAllLabelPropertyIndices(mgp_graph *memgraph_graph) {
@@ -4340,6 +4349,58 @@ inline List ListAllLabelPropertyIndices(mgp_graph *memgraph_graph) {
return List(label_property_indices);
}
+namespace {
+constexpr std::string_view kErrorMsgKey = "error_msg";
+constexpr std::string_view kSearchResultsKey = "search_results";
+constexpr std::string_view kAggregationResultsKey = "aggregation_results";
+} // namespace
+
+inline List SearchTextIndex(mgp_graph *memgraph_graph, std::string_view index_name, std::string_view search_query,
+ text_search_mode search_mode) {
+ auto results_or_error = Map(mgp::MemHandlerCallback(graph_search_text_index, memgraph_graph, index_name.data(),
+ search_query.data(), search_mode));
+ if (results_or_error.KeyExists(kErrorMsgKey)) {
+ if (!results_or_error.At(kErrorMsgKey).IsString()) {
+ throw TextSearchException{"The error message is not a string!"};
+ }
+ throw TextSearchException(results_or_error.At(kErrorMsgKey).ValueString().data());
+ }
+
+ if (!results_or_error.KeyExists(kSearchResultsKey)) {
+ throw TextSearchException{"Incomplete text index search results!"};
+ }
+
+ if (!results_or_error.At(kSearchResultsKey).IsList()) {
+ throw TextSearchException{"Text index search results have wrong type!"};
+ }
+
+ return results_or_error.At(kSearchResultsKey).ValueList();
+}
+
+inline std::string_view AggregateOverTextIndex(mgp_graph *memgraph_graph, std::string_view index_name,
+ std::string_view search_query, std::string_view aggregation_query) {
+ auto results_or_error =
+ Map(mgp::MemHandlerCallback(graph_aggregate_over_text_index, memgraph_graph, index_name.data(),
+ search_query.data(), aggregation_query.data()));
+
+ if (results_or_error.KeyExists(kErrorMsgKey)) {
+ if (!results_or_error.At(kErrorMsgKey).IsString()) {
+ throw TextSearchException{"The error message is not a string!"};
+ }
+ throw TextSearchException(results_or_error.At(kErrorMsgKey).ValueString().data());
+ }
+
+ if (!results_or_error.KeyExists(kAggregationResultsKey)) {
+ throw TextSearchException{"Incomplete text index aggregation results!"};
+ }
+
+ if (!results_or_error.At(kAggregationResultsKey).IsString()) {
+ throw TextSearchException{"Text index aggregation results have wrong type!"};
+ }
+
+ return results_or_error.At(kAggregationResultsKey).ValueString();
+}
+
inline bool CreateExistenceConstraint(mgp_graph *memgraph_graph, const std::string_view label,
const std::string_view property) {
return create_existence_constraint(memgraph_graph, label.data(), property.data());
diff --git a/init b/init
index 9187ee5aa..5d1799237 100755
--- a/init
+++ b/init
@@ -14,6 +14,7 @@ function print_help () {
echo "Optional arguments:"
echo -e " -h\tdisplay this help and exit"
echo -e " --without-libs-setup\tskip the step for setting up libs"
+ echo -e " --ci\tscript is being run inside ci"
}
function setup_virtualenv () {
@@ -35,6 +36,7 @@ function setup_virtualenv () {
}
setup_libs=true
+ci=false
if [[ $# -eq 1 && "$1" == "-h" ]]; then
print_help
exit 0
@@ -45,6 +47,10 @@ else
shift
setup_libs=false
;;
+ --ci)
+ shift
+ ci=true
+ ;;
*)
# unknown option
echo "Invalid argument provided: $1"
@@ -76,11 +82,13 @@ if [[ "$setup_libs" == "true" ]]; then
fi
# Fix for centos 7 during release
-if [ "${DISTRO}" = "centos-7" ] || [ "${DISTRO}" = "debian-11" ] || [ "${DISTRO}" = "amzn-2" ]; then
- if python3 -m pip show virtualenv >/dev/null 2>/dev/null; then
- python3 -m pip uninstall -y virtualenv
+if [[ "$ci" == "false" ]]; then
+ if [ "${DISTRO}" = "centos-7" ] || [ "${DISTRO}" = "debian-11" ] || [ "${DISTRO}" = "amzn-2" ]; then
+ if python3 -m pip show virtualenv >/dev/null 2>/dev/null; then
+ python3 -m pip uninstall -y virtualenv
+ fi
+ python3 -m pip install virtualenv
fi
- python3 -m pip install virtualenv
fi
# setup gql_behave dependencies
@@ -119,14 +127,16 @@ fi
# Install precommit hook except on old operating systems because we don't
# develop on them -> pre-commit hook not required -> we can use latest
# packages.
-if [ "${DISTRO}" != "centos-7" ] && [ "$DISTRO" != "debian-10" ] && [ "${DISTRO}" != "ubuntu-18.04" ] && [ "${DISTRO}" != "amzn-2" ]; then
- python3 -m pip install pre-commit
- python3 -m pre_commit install
- # Install py format tools for usage during the development.
- echo "Install black formatter"
- python3 -m pip install black==23.1.*
- echo "Install isort"
- python3 -m pip install isort==5.12.*
+if [[ "$ci" == "false" ]]; then
+ if [ "${DISTRO}" != "centos-7" ] && [ "$DISTRO" != "debian-10" ] && [ "${DISTRO}" != "ubuntu-18.04" ] && [ "${DISTRO}" != "amzn-2" ]; then
+ python3 -m pip install pre-commit
+ python3 -m pre_commit install
+ # Install py format tools for usage during the development.
+ echo "Install black formatter"
+ python3 -m pip install black==23.1.*
+ echo "Install isort"
+ python3 -m pip install isort==5.12.*
+ fi
fi
# Link `include/mgp.py` with `release/mgp/mgp.py`
diff --git a/libs/CMakeLists.txt b/libs/CMakeLists.txt
index 7d568d548..ab6a313f1 100644
--- a/libs/CMakeLists.txt
+++ b/libs/CMakeLists.txt
@@ -295,6 +295,32 @@ set_path_external_library(jemalloc STATIC
import_header_library(rangev3 ${CMAKE_CURRENT_SOURCE_DIR}/rangev3/include)
+ExternalProject_Add(mgcxx-proj
+ PREFIX mgcxx-proj
+ GIT_REPOSITORY https://github.com/memgraph/mgcxx
+ GIT_TAG "v0.0.4"
+ CMAKE_ARGS
+ "-DCMAKE_INSTALL_PREFIX=<INSTALL_DIR>"
+ "-DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}"
+ "-DENABLE_TESTS=OFF"
+ INSTALL_DIR "${PROJECT_BINARY_DIR}/mgcxx"
+)
+ExternalProject_Get_Property(mgcxx-proj install_dir)
+set(MGCXX_ROOT ${install_dir})
+
+add_library(tantivy_text_search STATIC IMPORTED GLOBAL)
+add_dependencies(tantivy_text_search mgcxx-proj)
+set_property(TARGET tantivy_text_search PROPERTY IMPORTED_LOCATION ${MGCXX_ROOT}/lib/libtantivy_text_search.a)
+
+add_library(mgcxx_text_search STATIC IMPORTED GLOBAL)
+add_dependencies(mgcxx_text_search mgcxx-proj)
+set_property(TARGET mgcxx_text_search PROPERTY IMPORTED_LOCATION ${MGCXX_ROOT}/lib/libmgcxx_text_search.a)
+# We need to create the include directory first in order to be able to add it
+# as an include directory. The header files in the include directory will be
+# generated later during the build process.
+file(MAKE_DIRECTORY ${MGCXX_ROOT}/include)
+set_property(TARGET mgcxx_text_search PROPERTY INTERFACE_INCLUDE_DIRECTORIES ${MGCXX_ROOT}/include)
+
# Setup NuRaft
import_external_library(nuraft STATIC
${CMAKE_CURRENT_SOURCE_DIR}/nuraft/lib/libnuraft.a
diff --git a/libs/setup.sh b/libs/setup.sh
index 9c2a38c47..e23c5efef 100755
--- a/libs/setup.sh
+++ b/libs/setup.sh
@@ -127,6 +127,7 @@ declare -A primary_urls=(
["jemalloc"]="http://$local_cache_host/git/jemalloc.git"
["range-v3"]="http://$local_cache_host/git/range-v3.git"
["nuraft"]="http://$local_cache_host/git/NuRaft.git"
+ ["asio"]="http://$local_cache_host/git/asio.git"
)
# The goal of secondary urls is to have links to the "source of truth" of
@@ -157,6 +158,7 @@ declare -A secondary_urls=(
["jemalloc"]="https://github.com/jemalloc/jemalloc.git"
["range-v3"]="https://github.com/ericniebler/range-v3.git"
["nuraft"]="https://github.com/eBay/NuRaft.git"
+ ["asio"]="https://github.com/chriskohlhoff/asio.git"
)
# antlr
@@ -266,13 +268,13 @@ repo_clone_try_double "${primary_urls[jemalloc]}" "${secondary_urls[jemalloc]}"
pushd jemalloc
./autogen.sh
-MALLOC_CONF="retain:false,percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:5000,dirty_decay_ms:5000" \
+MALLOC_CONF="background_thread:true,retain:false,percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:5000,dirty_decay_ms:5000" \
./configure \
--disable-cxx \
--with-lg-page=12 \
--with-lg-hugepage=21 \
--enable-shared=no --prefix=$working_dir \
- --with-malloc-conf="retain:false,percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:5000,dirty_decay_ms:5000"
+ --with-malloc-conf="background_thread:true,retain:false,percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:5000,dirty_decay_ms:5000"
make -j$CPUS install
popd
@@ -286,5 +288,7 @@ nuraft_tag="v2.1.0"
repo_clone_try_double "${primary_urls[nuraft]}" "${secondary_urls[nuraft]}" "nuraft" "$nuraft_tag" true
pushd nuraft
git apply ../nuraft2.1.0.patch
+asio_tag="asio-1-29-0"
+repo_clone_try_double "${primary_urls[asio]}" "${secondary_urls[asio]}" "asio" "$asio_tag" true
./prepare.sh
popd
diff --git a/query_modules/CMakeLists.txt b/query_modules/CMakeLists.txt
index 41dbb495c..1336f3eb0 100644
--- a/query_modules/CMakeLists.txt
+++ b/query_modules/CMakeLists.txt
@@ -6,6 +6,8 @@ project(memgraph_query_modules)
disallow_in_source_build()
+find_package(fmt REQUIRED)
+
# Everything that is installed here, should be under the "query_modules" component.
set(CMAKE_INSTALL_DEFAULT_COMPONENT_NAME "query_modules")
string(TOLOWER ${CMAKE_BUILD_TYPE} lower_build_type)
@@ -58,6 +60,22 @@ install(PROGRAMS $<TARGET_FILE:schema>
# Also install the source of the example, so user can read it.
install(FILES schema.cpp DESTINATION lib/memgraph/query_modules/src)
+add_library(text SHARED text_search_module.cpp)
+target_include_directories(text PRIVATE ${CMAKE_SOURCE_DIR}/include)
+target_compile_options(text PRIVATE -Wall)
+target_link_libraries(text PRIVATE -static-libgcc -static-libstdc++ fmt::fmt)
+# Strip C++ example in release build.
+if (lower_build_type STREQUAL "release")
+ add_custom_command(TARGET text POST_BUILD
+ COMMAND strip -s $<TARGET_FILE:text>
+ COMMENT "Stripping symbols and sections from the C++ text_search module")
+endif()
+install(PROGRAMS $<TARGET_FILE:text>
+ DESTINATION lib/memgraph/query_modules
+ RENAME text.so)
+# Also install the source of the example, so user can read it.
+install(FILES text_search_module.cpp DESTINATION lib/memgraph/query_modules/src)
+
# Install the Python example and modules
install(FILES example.py DESTINATION lib/memgraph/query_modules RENAME py_example.py)
install(FILES graph_analyzer.py DESTINATION lib/memgraph/query_modules)
diff --git a/query_modules/text_search_module.cpp b/query_modules/text_search_module.cpp
new file mode 100644
index 000000000..8e4405058
--- /dev/null
+++ b/query_modules/text_search_module.cpp
@@ -0,0 +1,149 @@
+// Copyright 2024 Memgraph Ltd.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+// License, and you may not use this file except in compliance with the Business Source License.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+#include <string>
+#include <string_view>
+
+#include <fmt/format.h>
+
+#include <mgp.hpp>
+
+namespace TextSearch {
+constexpr std::string_view kProcedureSearch = "search";
+constexpr std::string_view kProcedureRegexSearch = "regex_search";
+constexpr std::string_view kProcedureSearchAllProperties = "search_all";
+constexpr std::string_view kProcedureAggregate = "aggregate";
+constexpr std::string_view kParameterIndexName = "index_name";
+constexpr std::string_view kParameterSearchQuery = "search_query";
+constexpr std::string_view kParameterAggregationQuery = "aggregation_query";
+constexpr std::string_view kReturnNode = "node";
+constexpr std::string_view kReturnAggregation = "aggregation";
+const std::string kSearchAllPrefix = "all";
+
+void Search(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory);
+void RegexSearch(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory);
+void SearchAllProperties(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory);
+void Aggregate(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory);
+} // namespace TextSearch
+
+void TextSearch::Search(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory) {
+ mgp::MemoryDispatcherGuard guard{memory};
+ const auto record_factory = mgp::RecordFactory(result);
+ auto arguments = mgp::List(args);
+
+ try {
+ const auto *index_name = arguments[0].ValueString().data();
+ const auto *search_query = arguments[1].ValueString().data();
+ for (const auto &node :
+ mgp::SearchTextIndex(memgraph_graph, index_name, search_query, text_search_mode::SPECIFIED_PROPERTIES)) {
+ auto record = record_factory.NewRecord();
+ record.Insert(TextSearch::kReturnNode.data(), node.ValueNode());
+ }
+ } catch (const std::exception &e) {
+ record_factory.SetErrorMessage(e.what());
+ }
+}
+
+void TextSearch::RegexSearch(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory) {
+ mgp::MemoryDispatcherGuard guard{memory};
+ const auto record_factory = mgp::RecordFactory(result);
+ auto arguments = mgp::List(args);
+
+ try {
+ const auto *index_name = arguments[0].ValueString().data();
+ const auto *search_query = arguments[1].ValueString().data();
+ for (const auto &node : mgp::SearchTextIndex(memgraph_graph, index_name, search_query, text_search_mode::REGEX)) {
+ auto record = record_factory.NewRecord();
+ record.Insert(TextSearch::kReturnNode.data(), node.ValueNode());
+ }
+ } catch (const std::exception &e) {
+ record_factory.SetErrorMessage(e.what());
+ }
+}
+
+void TextSearch::SearchAllProperties(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result,
+ mgp_memory *memory) {
+ mgp::MemoryDispatcherGuard guard{memory};
+ const auto record_factory = mgp::RecordFactory(result);
+ auto arguments = mgp::List(args);
+
+ try {
+ const auto *index_name = arguments[0].ValueString().data();
+ const auto *search_query = fmt::format("{}:{}", kSearchAllPrefix, arguments[1].ValueString()).data();
+ for (const auto &node :
+ mgp::SearchTextIndex(memgraph_graph, index_name, search_query, text_search_mode::ALL_PROPERTIES)) {
+ auto record = record_factory.NewRecord();
+ record.Insert(TextSearch::kReturnNode.data(), node.ValueNode());
+ }
+ } catch (const std::exception &e) {
+ record_factory.SetErrorMessage(e.what());
+ }
+}
+
+void TextSearch::Aggregate(mgp_list *args, mgp_graph *memgraph_graph, mgp_result *result, mgp_memory *memory) {
+ mgp::MemoryDispatcherGuard guard{memory};
+ const auto record_factory = mgp::RecordFactory(result);
+ auto arguments = mgp::List(args);
+
+ try {
+ const auto *index_name = arguments[0].ValueString().data();
+ const auto *search_query = arguments[1].ValueString().data();
+ const auto *aggregation_query = arguments[2].ValueString().data();
+ const auto aggregation_result =
+ mgp::AggregateOverTextIndex(memgraph_graph, index_name, search_query, aggregation_query);
+ auto record = record_factory.NewRecord();
+ record.Insert(TextSearch::kReturnAggregation.data(), aggregation_result.data());
+ } catch (const std::exception &e) {
+ record_factory.SetErrorMessage(e.what());
+ }
+}
+
+extern "C" int mgp_init_module(struct mgp_module *module, struct mgp_memory *memory) {
+ try {
+ mgp::MemoryDispatcherGuard guard{memory};
+
+ AddProcedure(TextSearch::Search, TextSearch::kProcedureSearch, mgp::ProcedureType::Read,
+ {
+ mgp::Parameter(TextSearch::kParameterIndexName, mgp::Type::String),
+ mgp::Parameter(TextSearch::kParameterSearchQuery, mgp::Type::String),
+ },
+ {mgp::Return(TextSearch::kReturnNode, mgp::Type::Node)}, module, memory);
+
+ AddProcedure(TextSearch::RegexSearch, TextSearch::kProcedureRegexSearch, mgp::ProcedureType::Read,
+ {
+ mgp::Parameter(TextSearch::kParameterIndexName, mgp::Type::String),
+ mgp::Parameter(TextSearch::kParameterSearchQuery, mgp::Type::String),
+ },
+ {mgp::Return(TextSearch::kReturnNode, mgp::Type::Node)}, module, memory);
+
+ AddProcedure(TextSearch::SearchAllProperties, TextSearch::kProcedureSearchAllProperties, mgp::ProcedureType::Read,
+ {
+ mgp::Parameter(TextSearch::kParameterIndexName, mgp::Type::String),
+ mgp::Parameter(TextSearch::kParameterSearchQuery, mgp::Type::String),
+ },
+ {mgp::Return(TextSearch::kReturnNode, mgp::Type::Node)}, module, memory);
+
+ AddProcedure(TextSearch::Aggregate, TextSearch::kProcedureAggregate, mgp::ProcedureType::Read,
+ {
+ mgp::Parameter(TextSearch::kParameterIndexName, mgp::Type::String),
+ mgp::Parameter(TextSearch::kParameterSearchQuery, mgp::Type::String),
+ mgp::Parameter(TextSearch::kParameterAggregationQuery, mgp::Type::String),
+ },
+ {mgp::Return(TextSearch::kReturnAggregation, mgp::Type::String)}, module, memory);
+ } catch (const std::exception &e) {
+ std::cerr << "Error while initializing query module: " << e.what() << std::endl;
+ return 1;
+ }
+
+ return 0;
+}
+
+extern "C" int mgp_shutdown_module() { return 0; }
diff --git a/release/package/amd-builders-v4.yml b/release/package/amd-builders-v4.yml
new file mode 100644
index 000000000..2e3d4c5d2
--- /dev/null
+++ b/release/package/amd-builders-v4.yml
@@ -0,0 +1,73 @@
+version: "3"
+services:
+ mgbuild_v4_amzn-2:
+ image: "memgraph/mgbuild:v4_amzn-2"
+ build:
+ context: amzn-2
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_amzn-2"
+
+ mgbuild_v4_centos-7:
+ image: "memgraph/mgbuild:v4_centos-7"
+ build:
+ context: centos-7
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_centos-7"
+
+ mgbuild_v4_centos-9:
+ image: "memgraph/mgbuild:v4_centos-9"
+ build:
+ context: centos-9
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_centos-9"
+
+ mgbuild_v4_debian-10:
+ image: "memgraph/mgbuild:v4_debian-10"
+ build:
+ context: debian-10
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_debian-10"
+
+ mgbuild_v4_debian-11:
+ image: "memgraph/mgbuild:v4_debian-11"
+ build:
+ context: debian-11
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_debian-11"
+
+ mgbuild_v4_fedora-36:
+ image: "memgraph/mgbuild:v4_fedora-36"
+ build:
+ context: fedora-36
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_fedora-36"
+
+ mgbuild_v4_ubuntu-18.04:
+ image: "memgraph/mgbuild:v4_ubuntu-18.04"
+ build:
+ context: ubuntu-18.04
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_ubuntu-18.04"
+
+ mgbuild_v4_ubuntu-20.04:
+ image: "memgraph/mgbuild:v4_ubuntu-20.04"
+ build:
+ context: ubuntu-20.04
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_ubuntu-20.04"
+
+ mgbuild_v4_ubuntu-22.04:
+ image: "memgraph/mgbuild:v4_ubuntu-22.04"
+ build:
+ context: ubuntu-22.04
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_ubuntu-22.04"
diff --git a/release/package/amd-builders-v5.yml b/release/package/amd-builders-v5.yml
new file mode 100644
index 000000000..d37bc6092
--- /dev/null
+++ b/release/package/amd-builders-v5.yml
@@ -0,0 +1,81 @@
+version: "3"
+services:
+ mgbuild_v5_amzn-2:
+ image: "memgraph/mgbuild:v5_amzn-2"
+ build:
+ context: amzn-2
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_amzn-2"
+
+ mgbuild_v5_centos-7:
+ image: "memgraph/mgbuild:v5_centos-7"
+ build:
+ context: centos-7
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_centos-7"
+
+ mgbuild_v5_centos-9:
+ image: "memgraph/mgbuild:v5_centos-9"
+ build:
+ context: centos-9
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_centos-9"
+
+ mgbuild_v5_debian-11:
+ image: "memgraph/mgbuild:v5_debian-11"
+ build:
+ context: debian-11
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_debian-11"
+
+ mgbuild_v5_debian-12:
+ image: "memgraph/mgbuild:v5_debian-12"
+ build:
+ context: debian-12
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_debian-12"
+
+ mgbuild_v5_fedora-38:
+ image: "memgraph/mgbuild:v5_fedora-38"
+ build:
+ context: fedora-38
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_fedora-38"
+
+ mgbuild_v5_fedora-39:
+ image: "memgraph/mgbuild:v5_fedora-39"
+ build:
+ context: fedora-39
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_fedora-39"
+
+ mgbuild_v5_rocky-9.3:
+ image: "memgraph/mgbuild:v5_rocky-9.3"
+ build:
+ context: rocky-9.3
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_rocky-9.3"
+
+ mgbuild_v5_ubuntu-20.04:
+ image: "memgraph/mgbuild:v5_ubuntu-20.04"
+ build:
+ context: ubuntu-20.04
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_ubuntu-20.04"
+
+ mgbuild_v5_ubuntu-22.04:
+ image: "memgraph/mgbuild:v5_ubuntu-22.04"
+ build:
+ context: ubuntu-22.04
+ args:
+ TOOLCHAIN_VERSION: "v5"
+ container_name: "mgbuild_v5_ubuntu-22.04"
diff --git a/release/package/amzn-2/Dockerfile b/release/package/amzn-2/Dockerfile
index ba3ad34b9..80fdb9893 100644
--- a/release/package/amzn-2/Dockerfile
+++ b/release/package/amzn-2/Dockerfile
@@ -7,9 +7,34 @@ RUN yum -y update \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz
+# Download and install toolchain
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-amzn-2-x86_64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/amzn-2.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/amzn-2.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+# Install PyYAML (only for amzn-2, centos-7, cento-9 and rocky-9)
+RUN pip3 install --user PyYAML
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/arm-builders-v4.yml b/release/package/arm-builders-v4.yml
new file mode 100644
index 000000000..18cd5c4e9
--- /dev/null
+++ b/release/package/arm-builders-v4.yml
@@ -0,0 +1,18 @@
+version: "3"
+
+services:
+ mgbuild_v4_debian-11-arm:
+ image: "memgraph/mgbuild:v4_debian-11-arm"
+ build:
+ context: debian-11-arm
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_debian-11-arm"
+
+ mgbuild_v4_ubuntu_v4_22.04-arm:
+ image: "memgraph/mgbuild:v4_ubuntu-22.04-arm"
+ build:
+ context: ubuntu-22.04-arm
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_v4_ubuntu-22.04-arm"
diff --git a/release/package/arm-builders-v5.yml b/release/package/arm-builders-v5.yml
new file mode 100644
index 000000000..c276a49a8
--- /dev/null
+++ b/release/package/arm-builders-v5.yml
@@ -0,0 +1,18 @@
+version: "3"
+
+services:
+ debian-12-arm:
+ image: "memgraph/mgbuild:v5_debian-12-arm"
+ build:
+ context: debian-12-arm
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_debian-12-arm"
+
+ ubuntu-22.04-arm:
+ image: "memgraph/mgbuild:v5_ubuntu-22.04-arm"
+ build:
+ context: ubuntu-22.04-arm
+ args:
+ TOOLCHAIN_VERSION: "v4"
+ container_name: "mgbuild_ubuntu-22.04-arm"
diff --git a/release/package/arm-builders.yml b/release/package/arm-builders.yml
deleted file mode 100644
index d52f3bb26..000000000
--- a/release/package/arm-builders.yml
+++ /dev/null
@@ -1,11 +0,0 @@
-version: "3"
-
-services:
- debian-11-arm:
- build:
- context: debian-11-arm
- container_name: "mgbuild_debian-11-arm"
- ubuntu-2204-arm:
- build:
- context: ubuntu-22.04-arm
- container_name: "mgbuild_ubuntu-22.04-arm"
diff --git a/release/package/centos-7/Dockerfile b/release/package/centos-7/Dockerfile
index ff386c551..d34dfee77 100644
--- a/release/package/centos-7/Dockerfile
+++ b/release/package/centos-7/Dockerfile
@@ -7,9 +7,33 @@ RUN yum -y update \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-centos-7-x86_64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/centos-7.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/centos-7.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+# Install PyYAML (only for amzn-2, centos-7, cento-9 and rocky-9)
+RUN pip3 install --user PyYAML
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/centos-9/Dockerfile b/release/package/centos-9/Dockerfile
index 1c4c9541e..7c559a932 100644
--- a/release/package/centos-9/Dockerfile
+++ b/release/package/centos-9/Dockerfile
@@ -7,9 +7,33 @@ RUN yum -y update \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-centos-9-x86_64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/centos-9.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/centos-9.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+# Install PyYAML (only for amzn-2, centos-7, cento-9 and rocky-9)
+RUN pip3 install --user PyYAML
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/debian-10/Dockerfile b/release/package/debian-10/Dockerfile
index abd4a7f0e..a3ef12cc6 100644
--- a/release/package/debian-10/Dockerfile
+++ b/release/package/debian-10/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-debian-10-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/debian-10.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/debian-10.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/debian-11-arm/Dockerfile b/release/package/debian-11-arm/Dockerfile
index 5040d8041..9b3aa3c8c 100644
--- a/release/package/debian-11-arm/Dockerfile
+++ b/release/package/debian-11-arm/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-arm64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/debian-11-arm.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/debian-11-arm.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/debian-11/Dockerfile b/release/package/debian-11/Dockerfile
index cf47f253e..62070eccf 100644
--- a/release/package/debian-11/Dockerfile
+++ b/release/package/debian-11/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-debian-11-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/debian-11.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/debian-11.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/debian-12-arm/Dockerfile b/release/package/debian-12-arm/Dockerfile
new file mode 100644
index 000000000..6d4200391
--- /dev/null
+++ b/release/package/debian-12-arm/Dockerfile
@@ -0,0 +1,39 @@
+FROM debian:12
+
+ARG TOOLCHAIN_VERSION
+
+# Stops tzdata interactive configuration.
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt update && apt install -y \
+ ca-certificates wget git
+# Do NOT be smart here and clean the cache because the container is used in the
+# stateful context.
+
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-arm64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-arm64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-arm64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-arm64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/debian-12-arm.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/debian-12-arm.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/debian-12/Dockerfile b/release/package/debian-12/Dockerfile
new file mode 100644
index 000000000..c35640eb3
--- /dev/null
+++ b/release/package/debian-12/Dockerfile
@@ -0,0 +1,39 @@
+FROM debian:12
+
+ARG TOOLCHAIN_VERSION
+
+# Stops tzdata interactive configuration.
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt update && apt install -y \
+ ca-certificates wget git
+# Do NOT be smart here and clean the cache because the container is used in the
+# stateful context.
+
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-debian-12-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/debian-12.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/debian-12.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/docker-compose.yml b/release/package/docker-compose.yml
deleted file mode 100644
index 4da0526ba..000000000
--- a/release/package/docker-compose.yml
+++ /dev/null
@@ -1,38 +0,0 @@
-version: "3"
-services:
- mgbuild_centos-7:
- build:
- context: centos-7
- container_name: "mgbuild_centos-7"
- mgbuild_centos-9:
- build:
- context: centos-9
- container_name: "mgbuild_centos-9"
- mgbuild_debian-10:
- build:
- context: debian-10
- container_name: "mgbuild_debian-10"
- mgbuild_debian-11:
- build:
- context: debian-11
- container_name: "mgbuild_debian-11"
- mgbuild_ubuntu-18.04:
- build:
- context: ubuntu-18.04
- container_name: "mgbuild_ubuntu-18.04"
- mgbuild_ubuntu-20.04:
- build:
- context: ubuntu-20.04
- container_name: "mgbuild_ubuntu-20.04"
- mgbuild_ubuntu-22.04:
- build:
- context: ubuntu-22.04
- container_name: "mgbuild_ubuntu-22.04"
- mgbuild_fedora-36:
- build:
- context: fedora-36
- container_name: "mgbuild_fedora-36"
- mgbuild_amzn-2:
- build:
- context: amzn-2
- container_name: "mgbuild_amzn-2"
diff --git a/release/package/fedora-36/Dockerfile b/release/package/fedora-36/Dockerfile
index f84af132f..954a5b2bf 100644
--- a/release/package/fedora-36/Dockerfile
+++ b/release/package/fedora-36/Dockerfile
@@ -8,9 +8,30 @@ RUN yum -y update \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-36-x86_64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/fedora-36.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/fedora-36.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/fedora-38/Dockerfile b/release/package/fedora-38/Dockerfile
new file mode 100644
index 000000000..cf7454ae4
--- /dev/null
+++ b/release/package/fedora-38/Dockerfile
@@ -0,0 +1,37 @@
+FROM fedora:38
+
+ARG TOOLCHAIN_VERSION
+
+# Stops tzdata interactive configuration.
+RUN yum -y update \
+ && yum install -y wget git
+# Do NOT be smart here and clean the cache because the container is used in the
+# stateful context.
+
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-38-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-38-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-38-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-38-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/fedora-38.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/fedora-38.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/fedora-39/Dockerfile b/release/package/fedora-39/Dockerfile
new file mode 100644
index 000000000..263017a3d
--- /dev/null
+++ b/release/package/fedora-39/Dockerfile
@@ -0,0 +1,37 @@
+FROM fedora:39
+
+ARG TOOLCHAIN_VERSION
+
+# Stops tzdata interactive configuration.
+RUN yum -y update \
+ && yum install -y wget git
+# Do NOT be smart here and clean the cache because the container is used in the
+# stateful context.
+
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-39-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-39-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-39-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-fedora-39-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/fedora-39.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/fedora-39.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/mgbuild.sh b/release/package/mgbuild.sh
new file mode 100755
index 000000000..e24776f60
--- /dev/null
+++ b/release/package/mgbuild.sh
@@ -0,0 +1,667 @@
+#!/bin/bash
+set -Eeuo pipefail
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+SCRIPT_NAME=${0##*/}
+PROJECT_ROOT="$SCRIPT_DIR/../.."
+MGBUILD_HOME_DIR="/home/mg"
+MGBUILD_ROOT_DIR="$MGBUILD_HOME_DIR/memgraph"
+
+DEFAULT_TOOLCHAIN="v5"
+SUPPORTED_TOOLCHAINS=(
+ v4 v5
+)
+DEFAULT_OS="all"
+SUPPORTED_OS=(
+ all
+ amzn-2
+ centos-7 centos-9
+ debian-10 debian-11 debian-11-arm debian-12 debian-12-arm
+ fedora-36 fedora-38 fedora-39
+ rocky-9.3
+ ubuntu-18.04 ubuntu-20.04 ubuntu-22.04 ubuntu-22.04-arm
+)
+SUPPORTED_OS_V4=(
+ amzn-2
+ centos-7 centos-9
+ debian-10 debian-11 debian-11-arm
+ fedora-36
+ ubuntu-18.04 ubuntu-20.04 ubuntu-22.04 ubuntu-22.04-arm
+)
+SUPPORTED_OS_V5=(
+ amzn-2
+ centos-7 centos-9
+ debian-11 debian-11-arm debian-12 debian-12-arm
+ fedora-38 fedora-39
+ rocky-9.3
+ ubuntu-20.04 ubuntu-22.04 ubuntu-22.04-arm
+)
+DEFAULT_BUILD_TYPE="Release"
+SUPPORTED_BUILD_TYPES=(
+ Debug
+ Release
+ RelWithDebInfo
+)
+DEFAULT_ARCH="amd"
+SUPPORTED_ARCHS=(
+ amd
+ arm
+)
+SUPPORTED_TESTS=(
+ clang-tidy cppcheck-and-clang-format code-analysis
+ code-coverage drivers durability e2e gql-behave
+ integration leftover-CTest macro-benchmark
+ mgbench stress-plain stress-ssl
+ unit unit-coverage upload-to-bench-graph
+
+)
+DEFAULT_THREADS=0
+DEFAULT_ENTERPRISE_LICENSE=""
+DEFAULT_ORGANIZATION_NAME="memgraph"
+
+print_help () {
+ echo -e "\nUsage: $SCRIPT_NAME [GLOBAL OPTIONS] COMMAND [COMMAND OPTIONS]"
+ echo -e "\nInteract with mgbuild containers"
+
+ echo -e "\nCommands:"
+ echo -e " build Build mgbuild image"
+ echo -e " build-memgraph [OPTIONS] Build memgraph binary inside mgbuild container"
+ echo -e " copy OPTIONS Copy an artifact from mgbuild container to host"
+ echo -e " package-memgraph Create memgraph package from built binary inside mgbuild container"
+ echo -e " pull Pull mgbuild image from dockerhub"
+ echo -e " push [OPTIONS] Push mgbuild image to dockerhub"
+ echo -e " run [OPTIONS] Run mgbuild container"
+ echo -e " stop [OPTIONS] Stop mgbuild container"
+ echo -e " test-memgraph TEST Run a selected test TEST (see supported tests below) inside mgbuild container"
+
+ echo -e "\nSupported tests:"
+ echo -e " \"${SUPPORTED_TESTS[*]}\""
+
+ echo -e "\nGlobal options:"
+ echo -e " --arch string Specify target architecture (\"${SUPPORTED_ARCHS[*]}\") (default \"$DEFAULT_ARCH\")"
+ echo -e " --build-type string Specify build type (\"${SUPPORTED_BUILD_TYPES[*]}\") (default \"$DEFAULT_BUILD_TYPE\")"
+ echo -e " --enterprise-license string Specify the enterprise license (default \"\")"
+ echo -e " --organization-name string Specify the organization name (default \"memgraph\")"
+ echo -e " --os string Specify operating system (\"${SUPPORTED_OS[*]}\") (default \"$DEFAULT_OS\")"
+ echo -e " --threads int Specify the number of threads a command will use (default \"\$(nproc)\" for container)"
+ echo -e " --toolchain string Specify toolchain version (\"${SUPPORTED_TOOLCHAINS[*]}\") (default \"$DEFAULT_TOOLCHAIN\")"
+
+ echo -e "\nbuild-memgraph options:"
+ echo -e " --asan Build with ASAN"
+ echo -e " --community Build community version"
+ echo -e " --coverage Build with code coverage"
+ echo -e " --for-docker Add flag -DMG_TELEMETRY_ID_OVERRIDE=DOCKER to cmake"
+ echo -e " --for-platform Add flag -DMG_TELEMETRY_ID_OVERRIDE=DOCKER-PLATFORM to cmake"
+ echo -e " --init-only Only run init script"
+ echo -e " --no-copy Don't copy the memgraph repo from host."
+ echo -e " Use this option with caution, be sure that memgraph source code is in correct location inside mgbuild container"
+ echo -e " --ubsan Build with UBSAN"
+
+ echo -e "\ncopy options:"
+ echo -e " --binary Copy memgraph binary from mgbuild container to host"
+ echo -e " --build-logs Copy build logs from mgbuild container to host"
+ echo -e " --package Copy memgraph package from mgbuild container to host"
+
+ echo -e "\npush options:"
+ echo -e " -p, --password string Specify password for docker login"
+ echo -e " -u, --username string Specify username for docker login"
+
+ echo -e "\nrun options:"
+ echo -e " --pull Pull the mgbuild image before running"
+
+ echo -e "\nstop options:"
+ echo -e " --remove Remove the stopped mgbuild container"
+
+ echo -e "\nToolchain v4 supported OSs:"
+ echo -e " \"${SUPPORTED_OS_V4[*]}\""
+
+ echo -e "\nToolchain v5 supported OSs:"
+ echo -e " \"${SUPPORTED_OS_V5[*]}\""
+
+ echo -e "\nExample usage:"
+ echo -e " $SCRIPT_NAME --os debian-12 --toolchain v5 --arch amd run"
+ echo -e " $SCRIPT_NAME --os debian-12 --toolchain v5 --arch amd --build-type RelWithDebInfo build-memgraph --community"
+ echo -e " $SCRIPT_NAME --os debian-12 --toolchain v5 --arch amd --build-type RelWithDebInfo test-memgraph unit"
+ echo -e " $SCRIPT_NAME --os debian-12 --toolchain v5 --arch amd package"
+ echo -e " $SCRIPT_NAME --os debian-12 --toolchain v5 --arch amd copy --package"
+ echo -e " $SCRIPT_NAME --os debian-12 --toolchain v5 --arch amd stop --remove"
+}
+
+check_support() {
+ local is_supported=false
+ case "$1" in
+ arch)
+ for e in "${SUPPORTED_ARCHS[@]}"; do
+ if [[ "$e" == "$2" ]]; then
+ is_supported=true
+ break
+ fi
+ done
+ if [[ "$is_supported" == false ]]; then
+ echo -e "Error: Architecture $2 isn't supported!\nChoose from ${SUPPORTED_ARCHS[*]}"
+ exit 1
+ fi
+ ;;
+ build_type)
+ for e in "${SUPPORTED_BUILD_TYPES[@]}"; do
+ if [[ "$e" == "$2" ]]; then
+ is_supported=true
+ break
+ fi
+ done
+ if [[ "$is_supported" == false ]]; then
+ echo -e "Error: Build type $2 isn't supported!\nChoose from ${SUPPORTED_BUILD_TYPES[*]}"
+ exit 1
+ fi
+ ;;
+ os)
+ for e in "${SUPPORTED_OS[@]}"; do
+ if [[ "$e" == "$2" ]]; then
+ is_supported=true
+ break
+ fi
+ done
+ if [[ "$is_supported" == false ]]; then
+ echo -e "Error: OS $2 isn't supported!\nChoose from ${SUPPORTED_OS[*]}"
+ exit 1
+ fi
+ ;;
+ toolchain)
+ for e in "${SUPPORTED_TOOLCHAINS[@]}"; do
+ if [[ "$e" == "$2" ]]; then
+ is_supported=true
+ break
+ fi
+ done
+ if [[ "$is_supported" == false ]]; then
+ echo -e "TError: oolchain version $2 isn't supported!\nChoose from ${SUPPORTED_TOOLCHAINS[*]}"
+ exit 1
+ fi
+ ;;
+ os_toolchain_combo)
+ if [[ "$3" == "v4" ]]; then
+ local SUPPORTED_OS_TOOLCHAIN=("${SUPPORTED_OS_V4[@]}")
+ elif [[ "$3" == "v5" ]]; then
+ local SUPPORTED_OS_TOOLCHAIN=("${SUPPORTED_OS_V5[@]}")
+ else
+ echo -e "Error: $3 isn't a supported toolchain_version!\nChoose from ${SUPPORTED_TOOLCHAINS[*]}"
+ exit 1
+ fi
+ for e in "${SUPPORTED_OS_TOOLCHAIN[@]}"; do
+ if [[ "$e" == "$2" ]]; then
+ is_supported=true
+ break
+ fi
+ done
+ if [[ "$is_supported" == false ]]; then
+ echo -e "Error: Toolchain version $3 doesn't support OS $2!\nChoose from ${SUPPORTED_OS_TOOLCHAIN[*]}"
+ exit 1
+ fi
+ ;;
+ *)
+ echo -e "Error: This function can only check arch, build_type, os, toolchain version and os toolchain combination"
+ exit 1
+ ;;
+ esac
+}
+
+
+##################################################
+######## BUILD, COPY AND PACKAGE MEMGRAPH ########
+##################################################
+build_memgraph () {
+ local build_container="mgbuild_${toolchain_version}_${os}"
+ local ACTIVATE_TOOLCHAIN="source /opt/toolchain-${toolchain_version}/activate"
+ local ACTIVATE_CARGO="source $MGBUILD_HOME_DIR/.cargo/env"
+ local container_build_dir="$MGBUILD_ROOT_DIR/build"
+ local container_output_dir="$container_build_dir/output"
+ local arm_flag=""
+ if [[ "$arch" == "arm" ]] || [[ "$os" =~ "-arm" ]]; then
+ arm_flag="-DMG_ARCH="ARM64""
+ fi
+ local build_type_flag="-DCMAKE_BUILD_TYPE=$build_type"
+ local telemetry_id_override_flag=""
+ local community_flag=""
+ local coverage_flag=""
+ local asan_flag=""
+ local ubsan_flag=""
+ local init_only=false
+ local for_docker=false
+ local for_platform=false
+ local copy_from_host=true
+ while [[ "$#" -gt 0 ]]; do
+ case "$1" in
+ --community)
+ community_flag="-DMG_ENTERPRISE=OFF"
+ shift 1
+ ;;
+ --init-only)
+ init_only=true
+ shift 1
+ ;;
+ --for-docker)
+ for_docker=true
+ if [[ "$for_platform" == "true" ]]; then
+ echo "Error: Cannot combine --for-docker and --for-platform flags"
+ exit 1
+ fi
+ telemetry_id_override_flag=" -DMG_TELEMETRY_ID_OVERRIDE=DOCKER "
+ shift 1
+ ;;
+ --for-platform)
+ for_platform=true
+ if [[ "$for_docker" == "true" ]]; then
+ echo "Error: Cannot combine --for-docker and --for-platform flags"
+ exit 1
+ fi
+ telemetry_id_override_flag=" -DMG_TELEMETRY_ID_OVERRIDE=DOCKER-PLATFORM "
+ shift 1
+ ;;
+ --coverage)
+ coverage_flag="-DTEST_COVERAGE=ON"
+ shift 1
+ ;;
+ --asan)
+ asan_flag="-DASAN=ON"
+ shift 1
+ ;;
+ --ubsan)
+ ubsan_flag="-DUBSAN=ON"
+ shift 1
+ ;;
+ --no-copy)
+ copy_from_host=false
+ shift 1
+ ;;
+ *)
+ echo "Error: Unknown flag '$1'"
+ exit 1
+ ;;
+ esac
+ done
+
+ echo "Initializing deps ..."
+ # If master is not the current branch, fetch it, because the get_version
+ # script depends on it. If we are on master, the fetch command is going to
+ # fail so that's why there is the explicit check.
+ # Required here because Docker build container can't access remote.
+ cd "$PROJECT_ROOT"
+ if [[ "$(git rev-parse --abbrev-ref HEAD)" != "master" ]]; then
+ git fetch origin master:master
+ fi
+
+ if [[ "$copy_from_host" == "true" ]]; then
+ # Ensure we have a clean build directory
+ docker exec -u mg "$build_container" bash -c "rm -rf $MGBUILD_ROOT_DIR && mkdir -p $MGBUILD_ROOT_DIR"
+ echo "Copying project files..."
+ docker cp "$PROJECT_ROOT/." "$build_container:$MGBUILD_ROOT_DIR/"
+ fi
+ # Change ownership of copied files so the mg user inside container can access them
+ docker exec -u root $build_container bash -c "chown -R mg:mg $MGBUILD_ROOT_DIR"
+
+ echo "Installing dependencies using '/memgraph/environment/os/$os.sh' script..."
+ docker exec -u root "$build_container" bash -c "$MGBUILD_ROOT_DIR/environment/os/$os.sh check TOOLCHAIN_RUN_DEPS || /environment/os/$os.sh install TOOLCHAIN_RUN_DEPS"
+ docker exec -u root "$build_container" bash -c "$MGBUILD_ROOT_DIR/environment/os/$os.sh check MEMGRAPH_BUILD_DEPS || /environment/os/$os.sh install MEMGRAPH_BUILD_DEPS"
+
+ echo "Building targeted package..."
+ # Fix issue with git marking directory as not safe
+ docker exec -u mg "$build_container" bash -c "cd $MGBUILD_ROOT_DIR && git config --global --add safe.directory '*'"
+ docker exec -u mg "$build_container" bash -c "cd $MGBUILD_ROOT_DIR && $ACTIVATE_TOOLCHAIN && ./init --ci"
+ if [[ "$init_only" == "true" ]]; then
+ return
+ fi
+
+ echo "Building Memgraph for $os on $build_container..."
+ docker exec -u mg "$build_container" bash -c "cd $container_build_dir && rm -rf ./*"
+ # Fix cmake failing locally if remote is clone via ssh
+ docker exec -u mg "$build_container" bash -c "cd $MGBUILD_ROOT_DIR && git remote set-url origin https://github.com/memgraph/memgraph.git"
+
+ # Define cmake command
+ local cmake_cmd="cmake $build_type_flag $arm_flag $community_flag $telemetry_id_override_flag $coverage_flag $asan_flag $ubsan_flag .."
+ docker exec -u mg "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && $ACTIVATE_CARGO && $cmake_cmd"
+
+ # ' is used instead of " because we need to run make within the allowed
+ # container resources.
+ # Default value for $threads is 0 instead of $(nproc) because macos
+ # doesn't support the nproc command.
+ # 0 is set for default value and checked here because mgbuild containers
+ # support nproc
+ # shellcheck disable=SC2016
+ if [[ "$threads" == 0 ]]; then
+ docker exec -u mg "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && $ACTIVATE_CARGO "'&& make -j$(nproc)'
+ docker exec -u mg "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && $ACTIVATE_CARGO "'&& make -j$(nproc) -B mgconsole'
+ else
+ docker exec -u mg "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && $ACTIVATE_CARGO "'&& make -j$threads'
+ docker exec -u mg "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && $ACTIVATE_CARGO "'&& make -j$threads -B mgconsole'
+ fi
+}
+
+package_memgraph() {
+ local ACTIVATE_TOOLCHAIN="source /opt/toolchain-${toolchain_version}/activate"
+ local build_container="mgbuild_${toolchain_version}_${os}"
+ local container_output_dir="$MGBUILD_ROOT_DIR/build/output"
+ local package_command=""
+ if [[ "$os" =~ ^"centos".* ]] || [[ "$os" =~ ^"fedora".* ]] || [[ "$os" =~ ^"amzn".* ]] || [[ "$os" =~ ^"rocky".* ]]; then
+ docker exec -u root "$build_container" bash -c "yum -y update"
+ package_command=" cpack -G RPM --config ../CPackConfig.cmake && rpmlint --file='../../release/rpm/rpmlintrc' memgraph*.rpm "
+ fi
+ if [[ "$os" =~ ^"debian".* ]]; then
+ docker exec -u root "$build_container" bash -c "apt --allow-releaseinfo-change -y update"
+ package_command=" cpack -G DEB --config ../CPackConfig.cmake "
+ fi
+ if [[ "$os" =~ ^"ubuntu".* ]]; then
+ docker exec -u root "$build_container" bash -c "apt update"
+ package_command=" cpack -G DEB --config ../CPackConfig.cmake "
+ fi
+ docker exec -u mg "$build_container" bash -c "mkdir -p $container_output_dir && cd $container_output_dir && $ACTIVATE_TOOLCHAIN && $package_command"
+}
+
+copy_memgraph() {
+ local build_container="mgbuild_${toolchain_version}_${os}"
+ case "$1" in
+ --binary)
+ echo "Copying memgraph binary to host..."
+ local container_output_path="$MGBUILD_ROOT_DIR/build/memgraph"
+ local host_output_path="$PROJECT_ROOT/build/memgraph"
+ mkdir -p "$PROJECT_ROOT/build"
+ docker cp -L $build_container:$container_output_path $host_output_path
+ echo "Binary saved to $host_output_path"
+ ;;
+ --build-logs)
+ echo "Copying memgraph build logs to host..."
+ local container_output_path="$MGBUILD_ROOT_DIR/build/logs"
+ local host_output_path="$PROJECT_ROOT/build/logs"
+ mkdir -p "$PROJECT_ROOT/build"
+ docker cp -L $build_container:$container_output_path $host_output_path
+ echo "Build logs saved to $host_output_path"
+ ;;
+ --package)
+ echo "Copying memgraph package to host..."
+ local container_output_dir="$MGBUILD_ROOT_DIR/build/output"
+ local host_output_dir="$PROJECT_ROOT/build/output/$os"
+ local last_package_name=$(docker exec -u mg "$build_container" bash -c "cd $container_output_dir && ls -t memgraph* | head -1")
+ mkdir -p "$host_output_dir"
+ docker cp "$build_container:$container_output_dir/$last_package_name" "$host_output_dir/$last_package_name"
+ echo "Package saved to $host_output_dir/$last_package_name"
+ ;;
+ *)
+ echo "Error: Unknown flag '$1'"
+ exit 1
+ ;;
+ esac
+}
+
+
+##################################################
+##################### TESTS ######################
+##################################################
+test_memgraph() {
+ local ACTIVATE_TOOLCHAIN="source /opt/toolchain-${toolchain_version}/activate"
+ local ACTIVATE_VENV="./setup.sh /opt/toolchain-${toolchain_version}/activate"
+ local ACTIVATE_CARGO="source $MGBUILD_HOME_DIR/.cargo/env"
+ local EXPORT_LICENSE="export MEMGRAPH_ENTERPRISE_LICENSE=$enterprise_license"
+ local EXPORT_ORG_NAME="export MEMGRAPH_ORGANIZATION_NAME=$organization_name"
+ local BUILD_DIR="$MGBUILD_ROOT_DIR/build"
+ local build_container="mgbuild_${toolchain_version}_${os}"
+ echo "Running $1 test on $build_container..."
+
+ case "$1" in
+ unit)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $BUILD_DIR && $ACTIVATE_TOOLCHAIN "'&& ctest -R memgraph__unit --output-on-failure -j$threads'
+ ;;
+ unit-coverage)
+ local setup_lsan_ubsan="export LSAN_OPTIONS=suppressions=$BUILD_DIR/../tools/lsan.supp && export UBSAN_OPTIONS=halt_on_error=1"
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $BUILD_DIR && $ACTIVATE_TOOLCHAIN && $setup_lsan_ubsan "'&& ctest -R memgraph__unit --output-on-failure -j2'
+ ;;
+ leftover-CTest)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $BUILD_DIR && $ACTIVATE_TOOLCHAIN "'&& ctest -E "(memgraph__unit|memgraph__benchmark)" --output-on-failure'
+ ;;
+ drivers)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR "'&& ./tests/drivers/run.sh'
+ ;;
+ integration)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR "'&& tests/integration/run.sh'
+ ;;
+ cppcheck-and-clang-format)
+ local test_output_path="$MGBUILD_ROOT_DIR/tools/github/cppcheck_and_clang_format.txt"
+ local test_output_host_dest="$PROJECT_ROOT/tools/github/cppcheck_and_clang_format.txt"
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tools/github && $ACTIVATE_TOOLCHAIN "'&& ./cppcheck_and_clang_format diff'
+ docker cp $build_container:$test_output_path $test_output_host_dest
+ ;;
+ stress-plain)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tests/stress && source ve3/bin/activate "'&& ./continuous_integration'
+ ;;
+ stress-ssl)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tests/stress && source ve3/bin/activate "'&& ./continuous_integration --use-ssl'
+ ;;
+ durability)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tests/stress && source ve3/bin/activate "'&& python3 durability --num-steps 5'
+ ;;
+ gql-behave)
+ local test_output_dir="$MGBUILD_ROOT_DIR/tests/gql_behave"
+ local test_output_host_dest="$PROJECT_ROOT/tests/gql_behave"
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tests && $ACTIVATE_VENV && cd $MGBUILD_ROOT_DIR/tests/gql_behave "'&& ./continuous_integration'
+ docker cp $build_container:$test_output_dir/gql_behave_status.csv $test_output_host_dest/gql_behave_status.csv
+ docker cp $build_container:$test_output_dir/gql_behave_status.html $test_output_host_dest/gql_behave_status.html
+ ;;
+ macro-benchmark)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && export USER=mg && export LANG=$(echo $LANG) && cd $MGBUILD_ROOT_DIR/tests/macro_benchmark "'&& ./harness QuerySuite MemgraphRunner --groups aggregation 1000_create unwind_create dense_expand match --no-strict'
+ ;;
+ mgbench)
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tests/mgbench "'&& ./benchmark.py vendor-native --num-workers-for-benchmark 12 --export-results benchmark_result.json pokec/medium/*/*'
+ ;;
+ upload-to-bench-graph)
+ shift 1
+ local SETUP_PASSED_ARGS="export PASSED_ARGS=\"$@\""
+ local SETUP_VE3_ENV="virtualenv -p python3 ve3 && source ve3/bin/activate && pip install -r requirements.txt"
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tools/bench-graph-client && $SETUP_VE3_ENV && $SETUP_PASSED_ARGS "'&& ./main.py $PASSED_ARGS'
+ ;;
+ code-analysis)
+ shift 1
+ local SETUP_PASSED_ARGS="export PASSED_ARGS=\"$@\""
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && cd $MGBUILD_ROOT_DIR/tests/code_analysis && $SETUP_PASSED_ARGS "'&& ./python_code_analysis.sh $PASSED_ARGS'
+ ;;
+ code-coverage)
+ local test_output_path="$MGBUILD_ROOT_DIR/tools/github/generated/code_coverage.tar.gz"
+ local test_output_host_dest="$PROJECT_ROOT/tools/github/generated/code_coverage.tar.gz"
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && $ACTIVATE_TOOLCHAIN && cd $MGBUILD_ROOT_DIR/tools/github "'&& ./coverage_convert'
+ docker exec -u mg $build_container bash -c "cd $MGBUILD_ROOT_DIR/tools/github/generated && tar -czf code_coverage.tar.gz coverage.json html report.json summary.rmu"
+ mkdir -p $PROJECT_ROOT/tools/github/generated
+ docker cp $build_container:$test_output_path $test_output_host_dest
+ ;;
+ clang-tidy)
+ shift 1
+ local SETUP_PASSED_ARGS="export PASSED_ARGS=\"$@\""
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && export THREADS=$threads && $ACTIVATE_TOOLCHAIN && cd $MGBUILD_ROOT_DIR/tests/code_analysis && $SETUP_PASSED_ARGS "'&& ./clang_tidy.sh $PASSED_ARGS'
+ ;;
+ e2e)
+ # local kafka_container="kafka_kafka_1"
+ # local kafka_hostname="kafka"
+ # local pulsar_container="pulsar_pulsar_1"
+ # local pulsar_hostname="pulsar"
+ # local setup_hostnames="export KAFKA_HOSTNAME=$kafka_hostname && PULSAR_HOSTNAME=$pulsar_hostname"
+ # local build_container_network=$(docker inspect $build_container --format='{{ .HostConfig.NetworkMode }}')
+ # docker network connect --alias $kafka_hostname $build_container_network $kafka_container > /dev/null 2>&1 || echo "Kafka container already inside correct network or something went wrong ..."
+ # docker network connect --alias $pulsar_hostname $build_container_network $pulsar_container > /dev/null 2>&1 || echo "Kafka container already inside correct network or something went wrong ..."
+ docker exec -u mg $build_container bash -c "pip install --user networkx && pip3 install --user networkx"
+ docker exec -u mg $build_container bash -c "$EXPORT_LICENSE && $EXPORT_ORG_NAME && $ACTIVATE_CARGO && cd $MGBUILD_ROOT_DIR/tests && $ACTIVATE_VENV && source ve3/bin/activate_e2e && cd $MGBUILD_ROOT_DIR/tests/e2e "'&& ./run.sh'
+ ;;
+ *)
+ echo "Error: Unknown test '$1'"
+ exit 1
+ ;;
+ esac
+}
+
+
+##################################################
+################### PARSE ARGS ###################
+##################################################
+if [ "$#" -eq 0 ] || [ "$1" == "-h" ] || [ "$1" == "--help" ]; then
+ print_help
+ exit 0
+fi
+arch=$DEFAULT_ARCH
+build_type=$DEFAULT_BUILD_TYPE
+enterprise_license=$DEFAULT_ENTERPRISE_LICENSE
+organization_name=$DEFAULT_ORGANIZATION_NAME
+os=$DEFAULT_OS
+threads=$DEFAULT_THREADS
+toolchain_version=$DEFAULT_TOOLCHAIN
+command=""
+while [[ $# -gt 0 ]]; do
+ case "$1" in
+ --arch)
+ arch=$2
+ check_support arch $arch
+ shift 2
+ ;;
+ --build-type)
+ build_type=$2
+ check_support build_type $build_type
+ shift 2
+ ;;
+ --enterprise-license)
+ enterprise_license=$2
+ shift 2
+ ;;
+ --organization-name)
+ organization_name=$2
+ shift 2
+ ;;
+ --os)
+ os=$2
+ check_support os $os
+ shift 2
+ ;;
+ --threads)
+ threads=$2
+ shift 2
+ ;;
+ --toolchain)
+ toolchain_version=$2
+ check_support toolchain $toolchain_version
+ shift 2
+ ;;
+ *)
+ if [[ "$1" =~ ^--.* ]]; then
+ echo -e "Error: Unknown option '$1'"
+ exit 1
+ else
+ command=$1
+ shift 1
+ break
+ fi
+ ;;
+ esac
+done
+check_support os_toolchain_combo $os $toolchain_version
+
+if [[ "$command" == "" ]]; then
+ echo -e "Error: Command not provided, please provide command"
+ print_help
+ exit 1
+fi
+
+if docker compose version > /dev/null 2>&1; then
+ docker_compose_cmd="docker compose"
+elif which docker-compose > /dev/null 2>&1; then
+ docker_compose_cmd="docker-compose"
+else
+ echo -e "Missing command: There has to be installed either 'docker-compose' or 'docker compose'"
+ exit 1
+fi
+echo "Using $docker_compose_cmd"
+
+##################################################
+################# PARSE COMMAND ##################
+##################################################
+case $command in
+ build)
+ cd $SCRIPT_DIR
+ if [[ "$os" == "all" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml build
+ else
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml build mgbuild_${toolchain_version}_${os}
+ fi
+ ;;
+ run)
+ cd $SCRIPT_DIR
+ pull=false
+ if [[ "$#" -gt 0 ]]; then
+ if [[ "$1" == "--pull" ]]; then
+ pull=true
+ else
+ echo "Error: Unknown flag '$1'"
+ exit 1
+ fi
+ fi
+ if [[ "$os" == "all" ]]; then
+ if [[ "$pull" == "true" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml pull --ignore-pull-failures
+ elif [[ "$docker_compose_cmd" == "docker compose" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml pull --ignore-pull-failures --policy missing
+ fi
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml up -d
+ else
+ if [[ "$pull" == "true" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml pull mgbuild_${toolchain_version}_${os}
+ elif ! docker image inspect memgraph/mgbuild:${toolchain_version}_${os} > /dev/null 2>&1; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml pull --ignore-pull-failures mgbuild_${toolchain_version}_${os}
+ fi
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml up -d mgbuild_${toolchain_version}_${os}
+ fi
+ ;;
+ stop)
+ cd $SCRIPT_DIR
+ remove=false
+ if [[ "$#" -gt 0 ]]; then
+ if [[ "$1" == "--remove" ]]; then
+ remove=true
+ else
+ echo "Error: Unknown flag '$1'"
+ exit 1
+ fi
+ fi
+ if [[ "$os" == "all" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml down
+ else
+ docker stop mgbuild_${toolchain_version}_${os}
+ if [[ "$remove" == "true" ]]; then
+ docker rm mgbuild_${toolchain_version}_${os}
+ fi
+ fi
+ ;;
+ pull)
+ cd $SCRIPT_DIR
+ if [[ "$os" == "all" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml pull --ignore-pull-failures
+ else
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml pull mgbuild_${toolchain_version}_${os}
+ fi
+ ;;
+ push)
+ docker login $@
+ cd $SCRIPT_DIR
+ if [[ "$os" == "all" ]]; then
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml push --ignore-push-failures
+ else
+ $docker_compose_cmd -f ${arch}-builders-${toolchain_version}.yml push mgbuild_${toolchain_version}_${os}
+ fi
+ ;;
+ build-memgraph)
+ build_memgraph $@
+ ;;
+ package-memgraph)
+ package_memgraph
+ ;;
+ test-memgraph)
+ test_memgraph $@
+ ;;
+ copy)
+ copy_memgraph $@
+ ;;
+ *)
+ echo "Error: Unknown command '$command'"
+ exit 1
+ ;;
+esac
diff --git a/release/package/rocky-9.3/Dockerfile b/release/package/rocky-9.3/Dockerfile
new file mode 100644
index 000000000..4ee3a0d78
--- /dev/null
+++ b/release/package/rocky-9.3/Dockerfile
@@ -0,0 +1,40 @@
+FROM rockylinux:9.3
+
+ARG TOOLCHAIN_VERSION
+
+# Stops tzdata interactive configuration.
+RUN yum -y update \
+ && yum install -y wget git
+# Do NOT be smart here and clean the cache because the container is used in the
+# stateful context.
+
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-rocky-9.3-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-rocky-9.3-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-rocky-9.3-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-rocky-9.3-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/rocky-9.3.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/rocky-9.3.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
+
+# Install PyYAML (only for amzn-2, centos-7, cento-9 and rocky-9.3)
+RUN pip3 install --user PyYAML
+
+ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/run.sh b/release/package/run.sh
deleted file mode 100755
index bbd5ff48a..000000000
--- a/release/package/run.sh
+++ /dev/null
@@ -1,208 +0,0 @@
-#!/bin/bash
-
-set -Eeuo pipefail
-
-SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
-SUPPORTED_OS=(
- centos-7 centos-9
- debian-10 debian-11 debian-11-arm
- ubuntu-18.04 ubuntu-20.04 ubuntu-22.04 ubuntu-22.04-arm
- fedora-36
- amzn-2
-)
-
-SUPPORTED_BUILD_TYPES=(
- Debug
- Release
- RelWithDebInfo
-)
-
-PROJECT_ROOT="$SCRIPT_DIR/../.."
-TOOLCHAIN_VERSION="toolchain-v4"
-ACTIVATE_TOOLCHAIN="source /opt/${TOOLCHAIN_VERSION}/activate"
-HOST_OUTPUT_DIR="$PROJECT_ROOT/build/output"
-
-print_help () {
- # TODO(gitbuda): Update the release/package/run.sh help
- echo "$0 init|package|docker|test {os} {build_type} [--for-docker|--for-platform]"
- echo ""
- echo " OSs: ${SUPPORTED_OS[*]}"
- echo " Build types: ${SUPPORTED_BUILD_TYPES[*]}"
- exit 1
-}
-
-make_package () {
- os="$1"
- build_type="$2"
-
- build_container="mgbuild_$os"
- echo "Building Memgraph for $os on $build_container..."
-
- package_command=""
- if [[ "$os" =~ ^"centos".* ]] || [[ "$os" =~ ^"fedora".* ]] || [[ "$os" =~ ^"amzn".* ]]; then
- docker exec "$build_container" bash -c "yum -y update"
- package_command=" cpack -G RPM --config ../CPackConfig.cmake && rpmlint --file='../../release/rpm/rpmlintrc' memgraph*.rpm "
- fi
- if [[ "$os" =~ ^"debian".* ]]; then
- docker exec "$build_container" bash -c "apt --allow-releaseinfo-change -y update"
- package_command=" cpack -G DEB --config ../CPackConfig.cmake "
- fi
- if [[ "$os" =~ ^"ubuntu".* ]]; then
- docker exec "$build_container" bash -c "apt update"
- package_command=" cpack -G DEB --config ../CPackConfig.cmake "
- fi
- telemetry_id_override_flag=""
- if [[ "$#" -gt 2 ]]; then
- if [[ "$3" == "--for-docker" ]]; then
- telemetry_id_override_flag=" -DMG_TELEMETRY_ID_OVERRIDE=DOCKER "
- elif [[ "$3" == "--for-platform" ]]; then
- telemetry_id_override_flag=" -DMG_TELEMETRY_ID_OVERRIDE=DOCKER-PLATFORM"
- else
- print_help
- exit
- fi
- fi
-
- echo "Copying project files..."
- # If master is not the current branch, fetch it, because the get_version
- # script depends on it. If we are on master, the fetch command is going to
- # fail so that's why there is the explicit check.
- # Required here because Docker build container can't access remote.
- cd "$PROJECT_ROOT"
- if [[ "$(git rev-parse --abbrev-ref HEAD)" != "master" ]]; then
- git fetch origin master:master
- fi
-
- # Ensure we have a clean build directory
- docker exec "$build_container" rm -rf /memgraph
-
- docker exec "$build_container" mkdir -p /memgraph
- # TODO(gitbuda): Revisit copying the whole repo -> makese sense under CI.
- docker cp "$PROJECT_ROOT/." "$build_container:/memgraph/"
-
- container_build_dir="/memgraph/build"
- container_output_dir="$container_build_dir/output"
-
- # TODO(gitbuda): TOOLCHAIN_RUN_DEPS should be installed during the Docker
- # image build phase, but that is not easy at this point because the
- # environment/os/{os}.sh does not come within the toolchain package. When
- # migrating to the next version of toolchain do that, and remove the
- # TOOLCHAIN_RUN_DEPS installation from here.
- # TODO(gitbuda): On the other side, having this here allows updating deps
- # wihout reruning the build containers.
- echo "Installing dependencies using '/memgraph/environment/os/$os.sh' script..."
- docker exec "$build_container" bash -c "/memgraph/environment/os/$os.sh install TOOLCHAIN_RUN_DEPS"
- docker exec "$build_container" bash -c "/memgraph/environment/os/$os.sh install MEMGRAPH_BUILD_DEPS"
-
- echo "Building targeted package..."
- # Fix issue with git marking directory as not safe
- docker exec "$build_container" bash -c "cd /memgraph && git config --global --add safe.directory '*'"
- docker exec "$build_container" bash -c "cd /memgraph && $ACTIVATE_TOOLCHAIN && ./init"
- docker exec "$build_container" bash -c "cd $container_build_dir && rm -rf ./*"
- # TODO(gitbuda): cmake fails locally if remote is clone via ssh because of the key -> FIX
- if [[ "$os" =~ "-arm" ]]; then
- docker exec "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && cmake -DCMAKE_BUILD_TYPE=$build_type -DMG_ARCH="ARM64" $telemetry_id_override_flag .."
- else
- docker exec "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN && cmake -DCMAKE_BUILD_TYPE=$build_type $telemetry_id_override_flag .."
- fi
- # ' is used instead of " because we need to run make within the allowed
- # container resources.
- # shellcheck disable=SC2016
- docker exec "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN "'&& make -j$(nproc)'
- docker exec "$build_container" bash -c "cd $container_build_dir && $ACTIVATE_TOOLCHAIN "'&& make -j$(nproc) -B mgconsole'
- docker exec "$build_container" bash -c "mkdir -p $container_output_dir && cd $container_output_dir && $ACTIVATE_TOOLCHAIN && $package_command"
-
- echo "Copying targeted package to host..."
- last_package_name=$(docker exec "$build_container" bash -c "cd $container_output_dir && ls -t memgraph* | head -1")
- # The operating system folder is introduced because multiple different
- # packages could be preserved during the same build "session".
- mkdir -p "$HOST_OUTPUT_DIR/$os"
- package_host_destination="$HOST_OUTPUT_DIR/$os/$last_package_name"
- docker cp "$build_container:$container_output_dir/$last_package_name" "$package_host_destination"
- echo "Package saved to $package_host_destination."
-}
-
-case "$1" in
- init)
- cd "$SCRIPT_DIR"
- if ! which "docker-compose" >/dev/null; then
- docker_compose_cmd="docker compose"
- else
- docker_compose_cmd="docker-compose"
- fi
- $docker_compose_cmd build --build-arg TOOLCHAIN_VERSION="${TOOLCHAIN_VERSION}"
- $docker_compose_cmd up -d
- ;;
-
- docker)
- # NOTE: Docker is build on top of Debian 11 package.
- based_on_os="debian-11"
- # shellcheck disable=SC2012
- last_package_name=$(cd "$HOST_OUTPUT_DIR/$based_on_os" && ls -t memgraph* | head -1)
- docker_build_folder="$PROJECT_ROOT/release/docker"
- cd "$docker_build_folder"
- ./package_docker --latest "$HOST_OUTPUT_DIR/$based_on_os/$last_package_name"
- # shellcheck disable=SC2012
- docker_image_name=$(cd "$docker_build_folder" && ls -t memgraph* | head -1)
- docker_host_folder="$HOST_OUTPUT_DIR/docker"
- docker_host_image_path="$docker_host_folder/$docker_image_name"
- mkdir -p "$docker_host_folder"
- cp "$docker_build_folder/$docker_image_name" "$docker_host_image_path"
- echo "Docker images saved to $docker_host_image_path."
- ;;
-
- package)
- shift 1
- if [[ "$#" -lt 2 ]]; then
- print_help
- fi
- os="$1"
- build_type="$2"
- shift 2
- is_os_ok=false
- for supported_os in "${SUPPORTED_OS[@]}"; do
- if [[ "$supported_os" == "${os}" ]]; then
- is_os_ok=true
- break
- fi
- done
- is_build_type_ok=false
- for supported_build_type in "${SUPPORTED_BUILD_TYPES[@]}"; do
- if [[ "$supported_build_type" == "${build_type}" ]]; then
- is_build_type_ok=true
- break
- fi
- done
- if [[ "$is_os_ok" == true && "$is_build_type_ok" == true ]]; then
- make_package "$os" "$build_type" "$@"
- else
- if [[ "$is_os_ok" == false ]]; then
- echo "Unsupported OS: $os"
- elif [[ "$is_build_type_ok" == false ]]; then
- echo "Unsupported build type: $build_type"
- fi
- print_help
- fi
- ;;
-
- build)
- shift 1
- if [[ "$#" -ne 2 ]]; then
- print_help
- fi
- # in the vX format, e.g. v5
- toolchain_version="$1"
- # a name of the os folder, e.g. ubuntu-22.04-arm
- os="$2"
- cd "$SCRIPT_DIR/$os"
- docker build -f Dockerfile --build-arg TOOLCHAIN_VERSION="toolchain-$toolchain_version" -t "memgraph/memgraph-builder:${toolchain_version}_$os" .
- ;;
-
- test)
- echo "TODO(gitbuda): Test all packages on mgtest containers."
- ;;
-
- *)
- print_help
- ;;
-esac
diff --git a/release/package/ubuntu-18.04/Dockerfile b/release/package/ubuntu-18.04/Dockerfile
index 97eceead4..b4b090984 100644
--- a/release/package/ubuntu-18.04/Dockerfile
+++ b/release/package/ubuntu-18.04/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-18.04-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/ubuntu-18.04.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/ubuntu-18.04.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/ubuntu-20.04/Dockerfile b/release/package/ubuntu-20.04/Dockerfile
index 9fea7fd79..30d6cda8e 100644
--- a/release/package/ubuntu-20.04/Dockerfile
+++ b/release/package/ubuntu-20.04/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-20.04-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/ubuntu-20.04.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/ubuntu-20.04.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/ubuntu-22.04-arm/Dockerfile b/release/package/ubuntu-22.04-arm/Dockerfile
index 56cac8f76..aa7b5b63f 100644
--- a/release/package/ubuntu-22.04-arm/Dockerfile
+++ b/release/package/ubuntu-22.04-arm/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-arm64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/ubuntu-22.04-arm.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/ubuntu-22.04-arm.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/release/package/ubuntu-22.04/Dockerfile b/release/package/ubuntu-22.04/Dockerfile
index 6bd22589b..652de0f5c 100644
--- a/release/package/ubuntu-22.04/Dockerfile
+++ b/release/package/ubuntu-22.04/Dockerfile
@@ -10,9 +10,30 @@ RUN apt update && apt install -y \
# Do NOT be smart here and clean the cache because the container is used in the
# stateful context.
-RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/${TOOLCHAIN_VERSION}/${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz \
- -O ${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz \
- && tar xzvf ${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz -C /opt \
- && rm ${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz
+RUN wget -q https://s3-eu-west-1.amazonaws.com/deps.memgraph.io/toolchain-${TOOLCHAIN_VERSION}/toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz \
+ -O toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz \
+ && tar xzvf toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz -C /opt \
+ && rm toolchain-${TOOLCHAIN_VERSION}-binaries-ubuntu-22.04-amd64.tar.gz
+
+# Install toolchain run deps and memgraph build deps
+SHELL ["/bin/bash", "-c"]
+RUN git clone https://github.com/memgraph/memgraph.git \
+ && cd memgraph \
+ && ./environment/os/ubuntu-22.04.sh install TOOLCHAIN_RUN_DEPS \
+ && ./environment/os/ubuntu-22.04.sh install MEMGRAPH_BUILD_DEPS \
+ && cd .. && rm -rf memgraph
+
+# Add mgdeps-cache and bench-graph-api hostnames
+RUN echo -e "10.42.16.10 mgdeps-cache\n10.42.16.10 bench-graph-api" >> /etc/hosts
+
+# Create mg user and set as default
+RUN useradd -m -s /bin/bash mg
+USER mg
+
+# Install rust
+RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+# Fix node
+RUN curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
ENTRYPOINT ["sleep", "infinity"]
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 4d5d523c6..af88e624a 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -45,7 +45,7 @@ set(mg_single_node_v2_sources
add_executable(memgraph ${mg_single_node_v2_sources})
target_include_directories(memgraph PUBLIC ${CMAKE_SOURCE_DIR}/include)
target_link_libraries(memgraph stdc++fs Threads::Threads
- mg-telemetry mg-communication mg-communication-metrics mg-memory mg-utils mg-license mg-settings mg-glue mg-flags mg::system mg::replication_handler)
+ mg-telemetry mgcxx_text_search tantivy_text_search mg-communication mg-communication-metrics mg-memory mg-utils mg-license mg-settings mg-glue mg-flags mg::system mg::replication_handler)
# NOTE: `include/mg_procedure.syms` describes a pattern match for symbols which
# should be dynamically exported, so that `dlopen` can correctly link th
diff --git a/src/coordination/raft_state.cpp b/src/coordination/raft_state.cpp
index 38acfd85e..fd93160b6 100644
--- a/src/coordination/raft_state.cpp
+++ b/src/coordination/raft_state.cpp
@@ -12,6 +12,7 @@
#ifdef MG_ENTERPRISE
#include <chrono>
+#include <spdlog/spdlog.h>
#include "coordination/coordinator_config.hpp"
#include "coordination/coordinator_exceptions.hpp"
#include "coordination/raft_state.hpp"
@@ -123,7 +124,7 @@ auto RaftState::AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_po
spdlog::info("Request to add server {} to the cluster accepted", endpoint);
} else {
throw RaftAddServerException("Failed to accept request to add server {} to the cluster with error code {}",
- endpoint, cmd_result->get_result_code());
+ endpoint, int(cmd_result->get_result_code()));
}
// Waiting for server to join
@@ -173,7 +174,8 @@ auto RaftState::AppendRegisterReplicationInstanceLog(CoordinatorClientConfig con
spdlog::info("Request for registering instance {} accepted", config.instance_name);
if (res->get_result_code() != nuraft::cmd_result_code::OK) {
- spdlog::error("Failed to register instance {} with error code {}", config.instance_name, res->get_result_code());
+ spdlog::error("Failed to register instance {} with error code {}", config.instance_name,
+ int(res->get_result_code()));
return false;
}
@@ -194,7 +196,7 @@ auto RaftState::AppendUnregisterReplicationInstanceLog(std::string_view instance
spdlog::info("Request for unregistering instance {} accepted", instance_name);
if (res->get_result_code() != nuraft::cmd_result_code::OK) {
- spdlog::error("Failed to unregister instance {} with error code {}", instance_name, res->get_result_code());
+ spdlog::error("Failed to unregister instance {} with error code {}", instance_name, int(res->get_result_code()));
return false;
}
return true;
@@ -214,7 +216,7 @@ auto RaftState::AppendSetInstanceAsMainLog(std::string_view instance_name) -> bo
spdlog::info("Request for promoting instance {} accepted", instance_name);
if (res->get_result_code() != nuraft::cmd_result_code::OK) {
- spdlog::error("Failed to promote instance {} with error code {}", instance_name, res->get_result_code());
+ spdlog::error("Failed to promote instance {} with error code {}", instance_name, int(res->get_result_code()));
return false;
}
return true;
@@ -233,7 +235,7 @@ auto RaftState::AppendSetInstanceAsReplicaLog(std::string_view instance_name) ->
spdlog::info("Request for demoting instance {} accepted", instance_name);
if (res->get_result_code() != nuraft::cmd_result_code::OK) {
- spdlog::error("Failed to promote instance {} with error code {}", instance_name, res->get_result_code());
+ spdlog::error("Failed to promote instance {} with error code {}", instance_name, int(res->get_result_code()));
return false;
}
@@ -252,7 +254,7 @@ auto RaftState::AppendUpdateUUIDLog(utils::UUID const &uuid) -> bool {
spdlog::info("Request for updating UUID accepted");
if (res->get_result_code() != nuraft::cmd_result_code::OK) {
- spdlog::error("Failed to update UUID with error code {}", res->get_result_code());
+ spdlog::error("Failed to update UUID with error code {}", int(res->get_result_code()));
return false;
}
diff --git a/src/csv/include/csv/parsing.hpp b/src/csv/include/csv/parsing.hpp
index 66f2913c8..0accc616d 100644
--- a/src/csv/include/csv/parsing.hpp
+++ b/src/csv/include/csv/parsing.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -119,6 +119,8 @@ class Reader {
auto GetHeader() const -> Header const &;
auto GetNextRow(utils::MemoryResource *mem) -> std::optional<Row>;
+ void Reset();
+
private:
// Some implementation issues that need clearing up, but this is mainly because
// I don't want `boost/iostreams/filtering_stream.hpp` included in this header file
diff --git a/src/csv/parsing.cpp b/src/csv/parsing.cpp
index 6d03dc7fd..6961a42e4 100644
--- a/src/csv/parsing.cpp
+++ b/src/csv/parsing.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -34,6 +34,10 @@ struct Reader::impl {
[[nodiscard]] bool HasHeader() const { return read_config_.with_header; }
[[nodiscard]] auto Header() const -> Header const & { return header_; }
+ void Reset() {
+ line_buffer_.clear();
+ line_buffer_.shrink_to_fit();
+ }
auto GetNextRow(utils::MemoryResource *mem) -> std::optional<Reader::Row>;
@@ -42,7 +46,7 @@ struct Reader::impl {
void TryInitializeHeader();
- std::optional<utils::pmr::string> GetNextLine(utils::MemoryResource *mem);
+ bool GetNextLine();
ParsingResult ParseHeader();
@@ -55,6 +59,8 @@ struct Reader::impl {
Config read_config_;
uint64_t line_count_{1};
uint16_t number_of_columns_{0};
+ uint64_t estimated_number_of_columns_{0};
+ utils::pmr::string line_buffer_{memory_};
Reader::Header header_{memory_};
};
@@ -129,17 +135,16 @@ void Reader::impl::InitializeStream() {
MG_ASSERT(csv_stream_.is_complete(), "Should be 'complete' for correct operation");
}
-std::optional<utils::pmr::string> Reader::impl::GetNextLine(utils::MemoryResource *mem) {
- utils::pmr::string line(mem);
- if (!std::getline(csv_stream_, line)) {
+bool Reader::impl::GetNextLine() {
+ if (!std::getline(csv_stream_, line_buffer_)) {
// reached end of file or an I/0 error occurred
if (!csv_stream_.good()) {
csv_stream_.reset(); // this will close the file_stream_ and clear the chain
}
- return std::nullopt;
+ return false;
}
++line_count_;
- return std::move(line);
+ return true;
}
Reader::ParsingResult Reader::impl::ParseHeader() {
@@ -170,6 +175,8 @@ void Reader::impl::TryInitializeHeader() {
const Reader::Header &Reader::GetHeader() const { return pimpl->Header(); }
+void Reader::Reset() { pimpl->Reset(); }
+
namespace {
enum class CsvParserState : uint8_t { INITIAL_FIELD, NEXT_FIELD, QUOTING, EXPECT_DELIMITER, DONE };
@@ -179,6 +186,8 @@ Reader::ParsingResult Reader::impl::ParseRow(utils::MemoryResource *mem) {
utils::pmr::vector<utils::pmr::string> row(mem);
if (number_of_columns_ != 0) {
row.reserve(number_of_columns_);
+ } else if (estimated_number_of_columns_ != 0) {
+ row.reserve(estimated_number_of_columns_);
}
utils::pmr::string column(memory_);
@@ -186,13 +195,12 @@ Reader::ParsingResult Reader::impl::ParseRow(utils::MemoryResource *mem) {
auto state = CsvParserState::INITIAL_FIELD;
do {
- const auto maybe_line = GetNextLine(mem);
- if (!maybe_line) {
+ if (!GetNextLine()) {
// The whole file was processed.
break;
}
- std::string_view line_string_view = *maybe_line;
+ std::string_view line_string_view = line_buffer_;
// remove '\r' from the end in case we have dos file format
if (line_string_view.back() == '\r') {
@@ -312,6 +320,11 @@ Reader::ParsingResult Reader::impl::ParseRow(utils::MemoryResource *mem) {
fmt::format("Expected {:d} columns in row {:d}, but got {:d}", number_of_columns_,
line_count_ - 1, row.size()));
}
+ // To avoid unessisary dynamic growth of the row, remember the number of
+ // columns for future calls
+ if (number_of_columns_ == 0 && estimated_number_of_columns_ == 0) {
+ estimated_number_of_columns_ = row.size();
+ }
return std::move(row);
}
@@ -319,7 +332,7 @@ Reader::ParsingResult Reader::impl::ParseRow(utils::MemoryResource *mem) {
std::optional<Reader::Row> Reader::impl::GetNextRow(utils::MemoryResource *mem) {
auto row = ParseRow(mem);
- if (row.HasError()) {
+ if (row.HasError()) [[unlikely]] {
if (!read_config_.ignore_bad) {
throw CsvReadException("CSV Reader: Bad row at line {:d}: {}", line_count_ - 1, row.GetError().message);
}
@@ -333,7 +346,7 @@ std::optional<Reader::Row> Reader::impl::GetNextRow(utils::MemoryResource *mem)
} while (row.HasError());
}
- if (row->empty()) {
+ if (row->empty()) [[unlikely]] {
// reached end of file
return std::nullopt;
}
diff --git a/src/dbms/dbms_handler.hpp b/src/dbms/dbms_handler.hpp
index b0bbd5758..482423ebf 100644
--- a/src/dbms/dbms_handler.hpp
+++ b/src/dbms/dbms_handler.hpp
@@ -311,7 +311,7 @@ class DbmsHandler {
stats.triggers += info.triggers;
stats.streams += info.streams;
++stats.num_databases;
- stats.indices += storage_info.label_indices + storage_info.label_property_indices;
+ stats.indices += storage_info.label_indices + storage_info.label_property_indices + storage_info.text_indices;
stats.constraints += storage_info.existence_constraints + storage_info.unique_constraints;
++stats.storage_modes[(int)storage_info.storage_mode];
++stats.isolation_levels[(int)storage_info.isolation_level];
diff --git a/src/dbms/inmemory/replication_handlers.cpp b/src/dbms/inmemory/replication_handlers.cpp
index 3e4a31884..f9ce7a9d8 100644
--- a/src/dbms/inmemory/replication_handlers.cpp
+++ b/src/dbms/inmemory/replication_handlers.cpp
@@ -589,7 +589,6 @@ uint64_t InMemoryReplicationHandlers::ReadAndApplyDelta(storage::InMemoryStorage
if (timestamp < storage->timestamp_) {
continue;
}
-
SPDLOG_INFO(" Delta {}", applied_deltas);
switch (delta.type) {
case WalDeltaData::Type::VERTEX_CREATE: {
@@ -616,6 +615,7 @@ uint64_t InMemoryReplicationHandlers::ReadAndApplyDelta(storage::InMemoryStorage
auto vertex = transaction->FindVertex(delta.vertex_add_remove_label.gid, View::NEW);
if (!vertex)
throw utils::BasicException("Invalid transaction! Please raise an issue, {}:{}", __FILE__, __LINE__);
+ // NOTE: Text search doesn’t have replication in scope yet (Phases 1 and 2)
auto ret = vertex->AddLabel(transaction->NameToLabel(delta.vertex_add_remove_label.label));
if (ret.HasError() || !ret.GetValue())
throw utils::BasicException("Invalid transaction! Please raise an issue, {}:{}", __FILE__, __LINE__);
@@ -628,18 +628,21 @@ uint64_t InMemoryReplicationHandlers::ReadAndApplyDelta(storage::InMemoryStorage
auto vertex = transaction->FindVertex(delta.vertex_add_remove_label.gid, View::NEW);
if (!vertex)
throw utils::BasicException("Invalid transaction! Please raise an issue, {}:{}", __FILE__, __LINE__);
+ // NOTE: Text search doesn’t have replication in scope yet (Phases 1 and 2)
auto ret = vertex->RemoveLabel(transaction->NameToLabel(delta.vertex_add_remove_label.label));
if (ret.HasError() || !ret.GetValue())
throw utils::BasicException("Invalid transaction! Please raise an issue, {}:{}", __FILE__, __LINE__);
break;
}
case WalDeltaData::Type::VERTEX_SET_PROPERTY: {
- spdlog::trace(" Vertex {} set property {} to {}", delta.vertex_edge_set_property.gid.AsUint(),
- delta.vertex_edge_set_property.property, delta.vertex_edge_set_property.value);
+ spdlog::trace(" Vertex {} set property", delta.vertex_edge_set_property.gid.AsUint());
+ // NOLINTNEXTLINE
auto *transaction = get_transaction(timestamp);
+ // NOLINTNEXTLINE
auto vertex = transaction->FindVertex(delta.vertex_edge_set_property.gid, View::NEW);
if (!vertex)
throw utils::BasicException("Invalid transaction! Please raise an issue, {}:{}", __FILE__, __LINE__);
+ // NOTE: Phase 1 of the text search feature doesn't have replication in scope
auto ret = vertex->SetProperty(transaction->NameToProperty(delta.vertex_edge_set_property.property),
delta.vertex_edge_set_property.value);
if (ret.HasError())
@@ -684,8 +687,7 @@ uint64_t InMemoryReplicationHandlers::ReadAndApplyDelta(storage::InMemoryStorage
break;
}
case WalDeltaData::Type::EDGE_SET_PROPERTY: {
- spdlog::trace(" Edge {} set property {} to {}", delta.vertex_edge_set_property.gid.AsUint(),
- delta.vertex_edge_set_property.property, delta.vertex_edge_set_property.value);
+ spdlog::trace(" Edge {} set property", delta.vertex_edge_set_property.gid.AsUint());
if (!storage->config_.salient.items.properties_on_edges)
throw utils::BasicException(
"Can't set properties on edges because properties on edges "
@@ -854,6 +856,14 @@ uint64_t InMemoryReplicationHandlers::ReadAndApplyDelta(storage::InMemoryStorage
throw utils::BasicException("Invalid transaction! Please raise an issue, {}:{}", __FILE__, __LINE__);
break;
}
+ case WalDeltaData::Type::TEXT_INDEX_CREATE: {
+ // NOTE: Text search doesn’t have replication in scope yet (Phases 1 and 2)
+ break;
+ }
+ case WalDeltaData::Type::TEXT_INDEX_DROP: {
+ // NOTE: Text search doesn’t have replication in scope yet (Phases 1 and 2)
+ break;
+ }
case WalDeltaData::Type::EXISTENCE_CONSTRAINT_CREATE: {
spdlog::trace(" Create existence constraint on :{} ({})", delta.operation_label_property.label,
delta.operation_label_property.property);
@@ -917,5 +927,4 @@ uint64_t InMemoryReplicationHandlers::ReadAndApplyDelta(storage::InMemoryStorage
spdlog::debug("Applied {} deltas", applied_deltas);
return applied_deltas;
}
-
} // namespace memgraph::dbms
diff --git a/src/flags/experimental.cpp b/src/flags/experimental.cpp
index 123903c96..8c29142a1 100644
--- a/src/flags/experimental.cpp
+++ b/src/flags/experimental.cpp
@@ -18,14 +18,15 @@
// Bolt server flags.
// NOLINTNEXTLINE (cppcoreguidelines-avoid-non-const-global-variables)
-DEFINE_string(experimental_enabled, "",
- "Experimental features to be used, comma seperated. Options [system-replication, high-availability]");
-
+DEFINE_string(
+ experimental_enabled, "",
+ "Experimental features to be used, comma-separated. Options [system-replication, text-search, high-availability]");
using namespace std::string_view_literals;
namespace memgraph::flags {
auto const mapping = std::map{std::pair{"system-replication"sv, Experiments::SYSTEM_REPLICATION},
+ std::pair{"text-search"sv, Experiments::TEXT_SEARCH},
std::pair{"high-availability"sv, Experiments::HIGH_AVAILABILITY}};
auto ExperimentsInstance() -> Experiments & {
@@ -45,7 +46,7 @@ bool AreExperimentsEnabled(Experiments experiments) {
void InitializeExperimental() {
namespace rv = ranges::views;
- auto const connonicalize_string = [](auto &&rng) {
+ auto const canonicalize_string = [](auto &&rng) {
auto const is_space = [](auto c) { return c == ' '; };
auto const to_lower = [](unsigned char c) { return std::tolower(c); };
@@ -56,7 +57,7 @@ void InitializeExperimental() {
auto const mapping_end = mapping.cend();
using underlying_type = std::underlying_type_t<Experiments>;
auto to_set = underlying_type{};
- for (auto &&experiment : FLAGS_experimental_enabled | rv::split(',') | rv::transform(connonicalize_string)) {
+ for (auto &&experiment : FLAGS_experimental_enabled | rv::split(',') | rv::transform(canonicalize_string)) {
if (auto it = mapping.find(experiment); it != mapping_end) {
to_set |= static_cast<underlying_type>(it->second);
}
diff --git a/src/flags/experimental.hpp b/src/flags/experimental.hpp
index 5a19889fe..0b209a4e8 100644
--- a/src/flags/experimental.hpp
+++ b/src/flags/experimental.hpp
@@ -23,7 +23,8 @@ namespace memgraph::flags {
// old experiments can be reused once code cleanup has happened
enum class Experiments : uint8_t {
SYSTEM_REPLICATION = 1 << 0,
- HIGH_AVAILABILITY = 1 << 1,
+ TEXT_SEARCH = 1 << 1,
+ HIGH_AVAILABILITY = 1 << 2,
};
bool AreExperimentsEnabled(Experiments experiments);
diff --git a/src/flags/general.cpp b/src/flags/general.cpp
index cd2c95c60..37fa17b36 100644
--- a/src/flags/general.cpp
+++ b/src/flags/general.cpp
@@ -131,6 +131,10 @@ DEFINE_uint64(storage_recovery_thread_count,
DEFINE_bool(storage_enable_schema_metadata, false,
"Controls whether metadata should be collected about the resident labels and edge types.");
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+DEFINE_bool(storage_delta_on_identical_property_update, true,
+ "Controls whether updating a property with the same value should create a delta object.");
+
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_bool(telemetry_enabled, false,
"Set to true to enable telemetry. We collect information about the "
diff --git a/src/flags/general.hpp b/src/flags/general.hpp
index a1e8729ab..52f51471d 100644
--- a/src/flags/general.hpp
+++ b/src/flags/general.hpp
@@ -84,6 +84,8 @@ DECLARE_bool(storage_parallel_schema_recovery);
DECLARE_uint64(storage_recovery_thread_count);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_bool(storage_enable_schema_metadata);
+// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
+DECLARE_bool(storage_delta_on_identical_property_update);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_bool(telemetry_enabled);
diff --git a/src/flags/run_time_configurable.cpp b/src/flags/run_time_configurable.cpp
index a42ebd3d0..6c0fc54ac 100644
--- a/src/flags/run_time_configurable.cpp
+++ b/src/flags/run_time_configurable.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -73,11 +73,11 @@ constexpr auto kLogToStderrGFlagsKey = "also_log_to_stderr";
constexpr auto kCartesianProductEnabledSettingKey = "cartesian-product-enabled";
constexpr auto kCartesianProductEnabledGFlagsKey = "cartesian-product-enabled";
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
-std::atomic<double> execution_timeout_sec_; // Local cache-like thing
-
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
-std::atomic<bool> cartesian_product_enabled_{true}; // Local cache-like thing
+// NOLINTBEGIN(cppcoreguidelines-avoid-non-const-global-variables)
+// Local cache-like thing
+std::atomic<double> execution_timeout_sec_;
+std::atomic<bool> cartesian_product_enabled_{true};
+// NOLINTEND(cppcoreguidelines-avoid-non-const-global-variables)
auto ToLLEnum(std::string_view val) {
const auto ll_enum = memgraph::flags::LogLevelToEnum(val);
diff --git a/src/flags/run_time_configurable.hpp b/src/flags/run_time_configurable.hpp
index 944a0539f..b215d6540 100644
--- a/src/flags/run_time_configurable.hpp
+++ b/src/flags/run_time_configurable.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
diff --git a/src/glue/SessionHL.cpp b/src/glue/SessionHL.cpp
index 6c901516c..6a48f15ca 100644
--- a/src/glue/SessionHL.cpp
+++ b/src/glue/SessionHL.cpp
@@ -59,12 +59,14 @@ class TypedValueResultStreamBase {
public:
explicit TypedValueResultStreamBase(memgraph::storage::Storage *storage);
- std::vector<memgraph::communication::bolt::Value> DecodeValues(
- const std::vector<memgraph::query::TypedValue> &values) const;
+ void DecodeValues(const std::vector<memgraph::query::TypedValue> &values);
+
+ auto AccessValues() const -> std::vector<memgraph::communication::bolt::Value> const & { return decoded_values_; }
protected:
// NOTE: Needed only for ToBoltValue conversions
memgraph::storage::Storage *storage_;
+ std::vector<memgraph::communication::bolt::Value> decoded_values_;
};
/// Wrapper around TEncoder which converts TypedValue to Value
@@ -75,16 +77,18 @@ class TypedValueResultStream : public TypedValueResultStreamBase {
TypedValueResultStream(TEncoder *encoder, memgraph::storage::Storage *storage)
: TypedValueResultStreamBase{storage}, encoder_(encoder) {}
- void Result(const std::vector<memgraph::query::TypedValue> &values) { encoder_->MessageRecord(DecodeValues(values)); }
+ void Result(const std::vector<memgraph::query::TypedValue> &values) {
+ DecodeValues(values);
+ encoder_->MessageRecord(AccessValues());
+ }
private:
TEncoder *encoder_;
};
-std::vector<memgraph::communication::bolt::Value> TypedValueResultStreamBase::DecodeValues(
- const std::vector<memgraph::query::TypedValue> &values) const {
- std::vector<memgraph::communication::bolt::Value> decoded_values;
- decoded_values.reserve(values.size());
+void TypedValueResultStreamBase::DecodeValues(const std::vector<memgraph::query::TypedValue> &values) {
+ decoded_values_.reserve(values.size());
+ decoded_values_.clear();
for (const auto &v : values) {
auto maybe_value = memgraph::glue::ToBoltValue(v, storage_, memgraph::storage::View::NEW);
if (maybe_value.HasError()) {
@@ -99,9 +103,8 @@ std::vector<memgraph::communication::bolt::Value> TypedValueResultStreamBase::De
throw memgraph::communication::bolt::ClientError("Unexpected storage error when streaming results.");
}
}
- decoded_values.emplace_back(std::move(*maybe_value));
+ decoded_values_.emplace_back(std::move(*maybe_value));
}
- return decoded_values;
}
TypedValueResultStreamBase::TypedValueResultStreamBase(memgraph::storage::Storage *storage) : storage_(storage) {}
diff --git a/src/memgraph.cpp b/src/memgraph.cpp
index d896bcc4c..9bf50131d 100644
--- a/src/memgraph.cpp
+++ b/src/memgraph.cpp
@@ -332,7 +332,8 @@ int main(int argc, char **argv) {
.durability_directory = FLAGS_data_directory + "/rocksdb_durability",
.wal_directory = FLAGS_data_directory + "/rocksdb_wal"},
.salient.items = {.properties_on_edges = FLAGS_storage_properties_on_edges,
- .enable_schema_metadata = FLAGS_storage_enable_schema_metadata},
+ .enable_schema_metadata = FLAGS_storage_enable_schema_metadata,
+ .delta_on_identical_property_update = FLAGS_storage_delta_on_identical_property_update},
.salient.storage_mode = memgraph::flags::ParseStorageMode()};
spdlog::info("config recover on startup {}, flags {} {}", db_config.durability.recover_on_startup,
FLAGS_storage_recover_on_startup, FLAGS_data_recovery_on_startup);
diff --git a/src/query/db_accessor.hpp b/src/query/db_accessor.hpp
index 915ea9936..ee4988e4a 100644
--- a/src/query/db_accessor.hpp
+++ b/src/query/db_accessor.hpp
@@ -634,6 +634,24 @@ class DbAccessor final {
bool EdgeTypeIndexExists(storage::EdgeTypeId edge_type) const { return accessor_->EdgeTypeIndexExists(edge_type); }
+ bool TextIndexExists(const std::string &index_name) const { return accessor_->TextIndexExists(index_name); }
+
+ void TextIndexAddVertex(const VertexAccessor &vertex) { accessor_->TextIndexAddVertex(vertex.impl_); }
+
+ void TextIndexUpdateVertex(const VertexAccessor &vertex, const std::vector<storage::LabelId> &removed_labels = {}) {
+ accessor_->TextIndexUpdateVertex(vertex.impl_, removed_labels);
+ }
+
+ std::vector<storage::Gid> TextIndexSearch(const std::string &index_name, const std::string &search_query,
+ text_search_mode search_mode) const {
+ return accessor_->TextIndexSearch(index_name, search_query, search_mode);
+ }
+
+ std::string TextIndexAggregate(const std::string &index_name, const std::string &search_query,
+ const std::string &aggregation_query) const {
+ return accessor_->TextIndexAggregate(index_name, search_query, aggregation_query);
+ }
+
std::optional<storage::LabelIndexStats> GetIndexStats(const storage::LabelId &label) const {
return accessor_->GetIndexStats(label);
}
@@ -717,6 +735,12 @@ class DbAccessor final {
return accessor_->DropIndex(edge_type);
}
+ void CreateTextIndex(const std::string &index_name, storage::LabelId label) {
+ accessor_->CreateTextIndex(index_name, label, this);
+ }
+
+ void DropTextIndex(const std::string &index_name) { accessor_->DropTextIndex(index_name); }
+
utils::BasicResult<storage::StorageExistenceConstraintDefinitionError, void> CreateExistenceConstraint(
storage::LabelId label, storage::PropertyId property) {
return accessor_->CreateExistenceConstraint(label, property);
diff --git a/src/query/dump.cpp b/src/query/dump.cpp
index f1dd08c8d..abc147ee8 100644
--- a/src/query/dump.cpp
+++ b/src/query/dump.cpp
@@ -252,6 +252,10 @@ void DumpLabelPropertyIndex(std::ostream *os, query::DbAccessor *dba, storage::L
<< ");";
}
+void DumpTextIndex(std::ostream *os, query::DbAccessor *dba, const std::string &index_name, storage::LabelId label) {
+ *os << "CREATE TEXT INDEX " << EscapeName(index_name) << " ON :" << EscapeName(dba->LabelToName(label)) << ";";
+}
+
void DumpExistenceConstraint(std::ostream *os, query::DbAccessor *dba, storage::LabelId label,
storage::PropertyId property) {
*os << "CREATE CONSTRAINT ON (u:" << EscapeName(dba->LabelToName(label)) << ") ASSERT EXISTS (u."
@@ -286,6 +290,8 @@ PullPlanDump::PullPlanDump(DbAccessor *dba, dbms::DatabaseAccess db_acc)
CreateLabelIndicesPullChunk(),
// Dump all label property indices
CreateLabelPropertyIndicesPullChunk(),
+ // Dump all text indices
+ CreateTextIndicesPullChunk(),
// Dump all existence constraints
CreateExistenceConstraintsPullChunk(),
// Dump all unique constraints
@@ -412,6 +418,34 @@ PullPlanDump::PullChunk PullPlanDump::CreateLabelPropertyIndicesPullChunk() {
};
}
+PullPlanDump::PullChunk PullPlanDump::CreateTextIndicesPullChunk() {
+ // Dump all text indices
+ return [this, global_index = 0U](AnyStream *stream, std::optional<int> n) mutable -> std::optional<size_t> {
+ // Delay the construction of indices vectors
+ if (!indices_info_) {
+ indices_info_.emplace(dba_->ListAllIndices());
+ }
+ const auto &text = indices_info_->text_indices;
+
+ size_t local_counter = 0;
+ while (global_index < text.size() && (!n || local_counter < *n)) {
+ std::ostringstream os;
+ const auto &text_index = text[global_index];
+ DumpTextIndex(&os, dba_, text_index.first, text_index.second);
+ stream->Result({TypedValue(os.str())});
+
+ ++global_index;
+ ++local_counter;
+ }
+
+ if (global_index == text.size()) {
+ return local_counter;
+ }
+
+ return std::nullopt;
+ };
+}
+
PullPlanDump::PullChunk PullPlanDump::CreateExistenceConstraintsPullChunk() {
return [this, global_index = 0U](AnyStream *stream, std::optional<int> n) mutable -> std::optional<size_t> {
// Delay the construction of constraint vectors
diff --git a/src/query/dump.hpp b/src/query/dump.hpp
index 05bd42967..0cf4a82a6 100644
--- a/src/query/dump.hpp
+++ b/src/query/dump.hpp
@@ -55,6 +55,7 @@ struct PullPlanDump {
PullChunk CreateLabelIndicesPullChunk();
PullChunk CreateLabelPropertyIndicesPullChunk();
+ PullChunk CreateTextIndicesPullChunk();
PullChunk CreateExistenceConstraintsPullChunk();
PullChunk CreateUniqueConstraintsPullChunk();
PullChunk CreateInternalIndexPullChunk();
diff --git a/src/query/exceptions.hpp b/src/query/exceptions.hpp
index 147dc8710..a4c25fbae 100644
--- a/src/query/exceptions.hpp
+++ b/src/query/exceptions.hpp
@@ -433,4 +433,17 @@ class MultiDatabaseQueryInMulticommandTxException : public QueryException {
SPECIALIZE_GET_EXCEPTION_NAME(MultiDatabaseQueryInMulticommandTxException)
};
+class TextSearchException : public QueryException {
+ using QueryException::QueryException;
+ SPECIALIZE_GET_EXCEPTION_NAME(TextSearchException)
+};
+
+class TextSearchDisabledException : public TextSearchException {
+ public:
+ TextSearchDisabledException()
+ : TextSearchException(
+ "To use text indices and text search, start Memgraph with the experimental text search feature enabled.") {}
+ SPECIALIZE_GET_EXCEPTION_NAME(TextSearchDisabledException)
+};
+
} // namespace memgraph::query
diff --git a/src/query/frontend/ast/ast.cpp b/src/query/frontend/ast/ast.cpp
index 7da5c09a0..f0d09d453 100644
--- a/src/query/frontend/ast/ast.cpp
+++ b/src/query/frontend/ast/ast.cpp
@@ -189,6 +189,9 @@ constexpr utils::TypeInfo query::IndexQuery::kType{utils::TypeId::AST_INDEX_QUER
constexpr utils::TypeInfo query::EdgeIndexQuery::kType{utils::TypeId::AST_EDGE_INDEX_QUERY, "EdgeIndexQuery",
&query::Query::kType};
+constexpr utils::TypeInfo query::TextIndexQuery::kType{utils::TypeId::AST_TEXT_INDEX_QUERY, "TextIndexQuery",
+ &query::Query::kType};
+
constexpr utils::TypeInfo query::Create::kType{utils::TypeId::AST_CREATE, "Create", &query::Clause::kType};
constexpr utils::TypeInfo query::CallProcedure::kType{utils::TypeId::AST_CALL_PROCEDURE, "CallProcedure",
diff --git a/src/query/frontend/ast/ast.hpp b/src/query/frontend/ast/ast.hpp
index 29f7be3cf..e3d7bc0b2 100644
--- a/src/query/frontend/ast/ast.hpp
+++ b/src/query/frontend/ast/ast.hpp
@@ -2273,6 +2273,37 @@ class EdgeIndexQuery : public memgraph::query::Query {
friend class AstStorage;
};
+class TextIndexQuery : public memgraph::query::Query {
+ public:
+ static const utils::TypeInfo kType;
+ const utils::TypeInfo &GetTypeInfo() const override { return kType; }
+
+ enum class Action { CREATE, DROP };
+
+ TextIndexQuery() = default;
+
+ DEFVISITABLE(QueryVisitor<void>);
+
+ memgraph::query::TextIndexQuery::Action action_;
+ memgraph::query::LabelIx label_;
+ std::string index_name_;
+
+ TextIndexQuery *Clone(AstStorage *storage) const override {
+ TextIndexQuery *object = storage->Create<TextIndexQuery>();
+ object->action_ = action_;
+ object->label_ = storage->GetLabelIx(label_.name);
+ object->index_name_ = index_name_;
+ return object;
+ }
+
+ protected:
+ TextIndexQuery(Action action, LabelIx label, std::string index_name)
+ : action_(action), label_(std::move(label)), index_name_(index_name) {}
+
+ private:
+ friend class AstStorage;
+};
+
class Create : public memgraph::query::Clause {
public:
static const utils::TypeInfo kType;
diff --git a/src/query/frontend/ast/ast_visitor.hpp b/src/query/frontend/ast/ast_visitor.hpp
index bf11878da..cc6aed138 100644
--- a/src/query/frontend/ast/ast_visitor.hpp
+++ b/src/query/frontend/ast/ast_visitor.hpp
@@ -83,6 +83,7 @@ class ExplainQuery;
class ProfileQuery;
class IndexQuery;
class EdgeIndexQuery;
+class TextIndexQuery;
class DatabaseInfoQuery;
class SystemInfoQuery;
class ConstraintQuery;
@@ -144,11 +145,11 @@ class ExpressionVisitor
template <class TResult>
class QueryVisitor
- : public utils::Visitor<TResult, CypherQuery, ExplainQuery, ProfileQuery, IndexQuery, EdgeIndexQuery, AuthQuery,
- DatabaseInfoQuery, SystemInfoQuery, ConstraintQuery, DumpQuery, ReplicationQuery,
- LockPathQuery, FreeMemoryQuery, TriggerQuery, IsolationLevelQuery, CreateSnapshotQuery,
- StreamQuery, SettingQuery, VersionQuery, ShowConfigQuery, TransactionQueueQuery,
- StorageModeQuery, AnalyzeGraphQuery, MultiDatabaseQuery, ShowDatabasesQuery,
- EdgeImportModeQuery, CoordinatorQuery> {};
+ : public utils::Visitor<TResult, CypherQuery, ExplainQuery, ProfileQuery, IndexQuery, EdgeIndexQuery,
+ TextIndexQuery, AuthQuery, DatabaseInfoQuery, SystemInfoQuery, ConstraintQuery, DumpQuery,
+ ReplicationQuery, LockPathQuery, FreeMemoryQuery, TriggerQuery, IsolationLevelQuery,
+ CreateSnapshotQuery, StreamQuery, SettingQuery, VersionQuery, ShowConfigQuery,
+ TransactionQueueQuery, StorageModeQuery, AnalyzeGraphQuery, MultiDatabaseQuery,
+ ShowDatabasesQuery, EdgeImportModeQuery, CoordinatorQuery> {};
} // namespace memgraph::query
diff --git a/src/query/frontend/ast/cypher_main_visitor.cpp b/src/query/frontend/ast/cypher_main_visitor.cpp
index 6da48c97c..35ccb3670 100644
--- a/src/query/frontend/ast/cypher_main_visitor.cpp
+++ b/src/query/frontend/ast/cypher_main_visitor.cpp
@@ -243,6 +243,13 @@ antlrcpp::Any CypherMainVisitor::visitIndexQuery(MemgraphCypher::IndexQueryConte
return index_query;
}
+antlrcpp::Any CypherMainVisitor::visitTextIndexQuery(MemgraphCypher::TextIndexQueryContext *ctx) {
+ MG_ASSERT(ctx->children.size() == 1, "TextIndexQuery should have exactly one child!");
+ auto *text_index_query = std::any_cast<TextIndexQuery *>(ctx->children[0]->accept(this));
+ query_ = text_index_query;
+ return text_index_query;
+}
+
antlrcpp::Any CypherMainVisitor::visitCreateIndex(MemgraphCypher::CreateIndexContext *ctx) {
auto *index_query = storage_->Create<IndexQuery>();
index_query->action_ = IndexQuery::Action::CREATE;
@@ -286,6 +293,21 @@ antlrcpp::Any CypherMainVisitor::visitDropEdgeIndex(MemgraphCypher::DropEdgeInde
return index_query;
}
+antlrcpp::Any CypherMainVisitor::visitCreateTextIndex(MemgraphCypher::CreateTextIndexContext *ctx) {
+ auto *index_query = storage_->Create<TextIndexQuery>();
+ index_query->index_name_ = std::any_cast<std::string>(ctx->indexName()->accept(this));
+ index_query->action_ = TextIndexQuery::Action::CREATE;
+ index_query->label_ = AddLabel(std::any_cast<std::string>(ctx->labelName()->accept(this)));
+ return index_query;
+}
+
+antlrcpp::Any CypherMainVisitor::visitDropTextIndex(MemgraphCypher::DropTextIndexContext *ctx) {
+ auto *index_query = storage_->Create<TextIndexQuery>();
+ index_query->index_name_ = std::any_cast<std::string>(ctx->indexName()->accept(this));
+ index_query->action_ = TextIndexQuery::Action::DROP;
+ return index_query;
+}
+
antlrcpp::Any CypherMainVisitor::visitAuthQuery(MemgraphCypher::AuthQueryContext *ctx) {
MG_ASSERT(ctx->children.size() == 1, "AuthQuery should have exactly one child!");
auto *auth_query = std::any_cast<AuthQuery *>(ctx->children[0]->accept(this));
diff --git a/src/query/frontend/ast/cypher_main_visitor.hpp b/src/query/frontend/ast/cypher_main_visitor.hpp
index 8c65345c8..53738af61 100644
--- a/src/query/frontend/ast/cypher_main_visitor.hpp
+++ b/src/query/frontend/ast/cypher_main_visitor.hpp
@@ -153,6 +153,11 @@ class CypherMainVisitor : public antlropencypher::MemgraphCypherBaseVisitor {
*/
antlrcpp::Any visitEdgeIndexQuery(MemgraphCypher::EdgeIndexQueryContext *ctx) override;
+ /**
+ * @return TextIndexQuery*
+ */
+ antlrcpp::Any visitTextIndexQuery(MemgraphCypher::TextIndexQueryContext *ctx) override;
+
/**
* @return ExplainQuery*
*/
@@ -500,7 +505,7 @@ class CypherMainVisitor : public antlropencypher::MemgraphCypherBaseVisitor {
antlrcpp::Any visitCreateIndex(MemgraphCypher::CreateIndexContext *ctx) override;
/**
- * @return DropIndex*
+ * @return IndexQuery*
*/
antlrcpp::Any visitDropIndex(MemgraphCypher::DropIndexContext *ctx) override;
@@ -514,6 +519,16 @@ class CypherMainVisitor : public antlropencypher::MemgraphCypherBaseVisitor {
*/
antlrcpp::Any visitDropEdgeIndex(MemgraphCypher::DropEdgeIndexContext *ctx) override;
+ /**
+ * @return TextIndexQuery*
+ */
+ antlrcpp::Any visitCreateTextIndex(MemgraphCypher::CreateTextIndexContext *ctx) override;
+
+ /**
+ * @return TextIndexQuery*
+ */
+ antlrcpp::Any visitDropTextIndex(MemgraphCypher::DropTextIndexContext *ctx) override;
+
/**
* @return AuthQuery*
*/
diff --git a/src/query/frontend/opencypher/grammar/Cypher.g4 b/src/query/frontend/opencypher/grammar/Cypher.g4
index 7fa218598..911615314 100644
--- a/src/query/frontend/opencypher/grammar/Cypher.g4
+++ b/src/query/frontend/opencypher/grammar/Cypher.g4
@@ -25,6 +25,7 @@ statement : query ;
query : cypherQuery
| indexQuery
+ | textIndexQuery
| explainQuery
| profileQuery
| databaseInfoQuery
@@ -65,6 +66,8 @@ cypherQuery : singleQuery ( cypherUnion )* ( queryMemoryLimit )? ;
indexQuery : createIndex | dropIndex;
+textIndexQuery : createTextIndex | dropTextIndex;
+
singleQuery : clause ( clause )* ;
cypherUnion : ( UNION ALL singleQuery )
@@ -342,6 +345,12 @@ createIndex : CREATE INDEX ON ':' labelName ( '(' propertyKeyName ')' )? ;
dropIndex : DROP INDEX ON ':' labelName ( '(' propertyKeyName ')' )? ;
+indexName : symbolicName ;
+
+createTextIndex : CREATE TEXT INDEX indexName ON ':' labelName ;
+
+dropTextIndex : DROP TEXT INDEX indexName ;
+
doubleLiteral : FloatingLiteral ;
cypherKeyword : ALL
diff --git a/src/query/frontend/opencypher/grammar/CypherLexer.g4 b/src/query/frontend/opencypher/grammar/CypherLexer.g4
index 3e3c640d6..fb8a30b0f 100644
--- a/src/query/frontend/opencypher/grammar/CypherLexer.g4
+++ b/src/query/frontend/opencypher/grammar/CypherLexer.g4
@@ -131,6 +131,7 @@ SHOW : S H O W ;
SINGLE : S I N G L E ;
STARTS : S T A R T S ;
STORAGE : S T O R A G E ;
+TEXT : T E X T ;
THEN : T H E N ;
TRUE : T R U E ;
UNION : U N I O N ;
diff --git a/src/query/frontend/opencypher/grammar/MemgraphCypher.g4 b/src/query/frontend/opencypher/grammar/MemgraphCypher.g4
index 378310c22..ad15d6213 100644
--- a/src/query/frontend/opencypher/grammar/MemgraphCypher.g4
+++ b/src/query/frontend/opencypher/grammar/MemgraphCypher.g4
@@ -134,6 +134,7 @@ symbolicName : UnescapedSymbolicName
query : cypherQuery
| indexQuery
| edgeIndexQuery
+ | textIndexQuery
| explainQuery
| profileQuery
| databaseInfoQuery
diff --git a/src/query/frontend/semantic/required_privileges.cpp b/src/query/frontend/semantic/required_privileges.cpp
index 15726e3e2..d87fcb10e 100644
--- a/src/query/frontend/semantic/required_privileges.cpp
+++ b/src/query/frontend/semantic/required_privileges.cpp
@@ -29,6 +29,8 @@ class PrivilegeExtractor : public QueryVisitor<void>, public HierarchicalTreeVis
void Visit(EdgeIndexQuery & /*unused*/) override { AddPrivilege(AuthQuery::Privilege::INDEX); }
+ void Visit(TextIndexQuery & /*unused*/) override { AddPrivilege(AuthQuery::Privilege::INDEX); }
+
void Visit(AnalyzeGraphQuery & /*unused*/) override { AddPrivilege(AuthQuery::Privilege::INDEX); }
void Visit(AuthQuery & /*unused*/) override { AddPrivilege(AuthQuery::Privilege::AUTH); }
diff --git a/src/query/frontend/semantic/symbol.hpp b/src/query/frontend/semantic/symbol.hpp
index 0cfb86608..1a5aa2756 100644
--- a/src/query/frontend/semantic/symbol.hpp
+++ b/src/query/frontend/semantic/symbol.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
diff --git a/src/query/frontend/stripped.cpp b/src/query/frontend/stripped.cpp
index 9740cd463..5ea26b041 100644
--- a/src/query/frontend/stripped.cpp
+++ b/src/query/frontend/stripped.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
diff --git a/src/query/interpret/eval.cpp b/src/query/interpret/eval.cpp
index 8bd308420..7c5d838a5 100644
--- a/src/query/interpret/eval.cpp
+++ b/src/query/interpret/eval.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -13,12 +13,12 @@
namespace memgraph::query {
-int64_t EvaluateInt(ExpressionEvaluator *evaluator, Expression *expr, const std::string &what) {
+int64_t EvaluateInt(ExpressionEvaluator *evaluator, Expression *expr, std::string_view what) {
TypedValue value = expr->Accept(*evaluator);
try {
return value.ValueInt();
} catch (TypedValueException &e) {
- throw QueryRuntimeException(what + " must be an int");
+ throw QueryRuntimeException(std::string(what) + " must be an int");
}
}
diff --git a/src/query/interpret/eval.hpp b/src/query/interpret/eval.hpp
index 19fb34d3a..92c410b75 100644
--- a/src/query/interpret/eval.hpp
+++ b/src/query/interpret/eval.hpp
@@ -1216,7 +1216,7 @@ class ExpressionEvaluator : public ExpressionVisitor<TypedValue> {
/// @param what - Name of what's getting evaluated. Used for user feedback (via
/// exception) when the evaluated value is not an int.
/// @throw QueryRuntimeException if expression doesn't evaluate to an int.
-int64_t EvaluateInt(ExpressionEvaluator *evaluator, Expression *expr, const std::string &what);
+int64_t EvaluateInt(ExpressionEvaluator *evaluator, Expression *expr, std::string_view what);
std::optional<size_t> EvaluateMemoryLimit(ExpressionVisitor<TypedValue> &eval, Expression *memory_limit,
size_t memory_scale);
diff --git a/src/query/interpreter.cpp b/src/query/interpreter.cpp
index a5c81cc72..332054485 100644
--- a/src/query/interpreter.cpp
+++ b/src/query/interpreter.cpp
@@ -39,6 +39,7 @@
#include "dbms/dbms_handler.hpp"
#include "dbms/global.hpp"
#include "dbms/inmemory/storage_helper.hpp"
+#include "flags/experimental.hpp"
#include "flags/replication.hpp"
#include "flags/run_time_configurable.hpp"
#include "glue/communication.hpp"
@@ -246,27 +247,6 @@ std::optional<std::string> GetOptionalStringValue(query::Expression *expression,
return {};
};
-bool IsAllShortestPathsQuery(const std::vector<memgraph::query::Clause *> &clauses) {
- for (const auto &clause : clauses) {
- if (clause->GetTypeInfo() != Match::kType) {
- continue;
- }
- auto *match_clause = utils::Downcast<Match>(clause);
- for (const auto &pattern : match_clause->patterns_) {
- for (const auto &atom : pattern->atoms_) {
- if (atom->GetTypeInfo() != EdgeAtom::kType) {
- continue;
- }
- auto *edge_atom = utils::Downcast<EdgeAtom>(atom);
- if (edge_atom->type_ == EdgeAtom::Type::ALL_SHORTEST_PATHS) {
- return true;
- }
- }
- }
- }
- return false;
-}
-
inline auto convertFromCoordinatorToReplicationMode(const CoordinatorQuery::SyncMode &sync_mode)
-> replication_coordination_glue::ReplicationMode {
switch (sync_mode) {
@@ -1733,8 +1713,7 @@ struct PullPlan {
std::shared_ptr<QueryUserOrRole> user_or_role, std::atomic<TransactionStatus> *transaction_status,
std::shared_ptr<utils::AsyncTimer> tx_timer,
TriggerContextCollector *trigger_context_collector = nullptr,
- std::optional<size_t> memory_limit = {}, bool use_monotonic_memory = true,
- FrameChangeCollector *frame_change_collector_ = nullptr);
+ std::optional<size_t> memory_limit = {}, FrameChangeCollector *frame_change_collector_ = nullptr);
std::optional<plan::ProfilingStatsWithTotalTime> Pull(AnyStream *stream, std::optional<int> n,
const std::vector<Symbol> &output_symbols,
@@ -1759,26 +1738,17 @@ struct PullPlan {
// we have to keep track of any unsent results from previous `PullPlan::Pull`
// manually by using this flag.
bool has_unsent_results_ = false;
-
- // In the case of LOAD CSV, we want to use only PoolResource without MonotonicMemoryResource
- // to reuse allocated memory. As LOAD CSV is processing row by row
- // it is possible to reduce memory usage significantly if MemoryResource deals with memory allocation
- // can reuse memory that was allocated on processing the first row on all subsequent rows.
- // This flag signals to `PullPlan::Pull` which MemoryResource to use
- bool use_monotonic_memory_;
};
PullPlan::PullPlan(const std::shared_ptr<PlanWrapper> plan, const Parameters ¶meters, const bool is_profile_query,
DbAccessor *dba, InterpreterContext *interpreter_context, utils::MemoryResource *execution_memory,
std::shared_ptr<QueryUserOrRole> user_or_role, std::atomic<TransactionStatus> *transaction_status,
std::shared_ptr<utils::AsyncTimer> tx_timer, TriggerContextCollector *trigger_context_collector,
- const std::optional<size_t> memory_limit, bool use_monotonic_memory,
- FrameChangeCollector *frame_change_collector)
+ const std::optional<size_t> memory_limit, FrameChangeCollector *frame_change_collector)
: plan_(plan),
cursor_(plan->plan().MakeCursor(execution_memory)),
frame_(plan->symbol_table().max_position(), execution_memory),
- memory_limit_(memory_limit),
- use_monotonic_memory_(use_monotonic_memory) {
+ memory_limit_(memory_limit) {
ctx_.db_accessor = dba;
ctx_.symbol_table = plan->symbol_table();
ctx_.evaluation_context.timestamp = QueryTimestamp();
@@ -1804,6 +1774,7 @@ PullPlan::PullPlan(const std::shared_ptr<PlanWrapper> plan, const Parameters &pa
ctx_.is_profile_query = is_profile_query;
ctx_.trigger_context_collector = trigger_context_collector;
ctx_.frame_change_collector = frame_change_collector;
+ ctx_.evaluation_context.memory = execution_memory;
}
std::optional<plan::ProfilingStatsWithTotalTime> PullPlan::Pull(AnyStream *stream, std::optional<int> n,
@@ -1827,43 +1798,14 @@ std::optional<plan::ProfilingStatsWithTotalTime> PullPlan::Pull(AnyStream *strea
}
}};
- // Set up temporary memory for a single Pull. Initial memory comes from the
- // stack. 256 KiB should fit on the stack and should be more than enough for a
- // single `Pull`.
- static constexpr size_t stack_size = 256UL * 1024UL;
- char stack_data[stack_size];
-
- utils::ResourceWithOutOfMemoryException resource_with_exception;
- utils::MonotonicBufferResource monotonic_memory{&stack_data[0], stack_size, &resource_with_exception};
- std::optional<utils::PoolResource> pool_memory;
- static constexpr auto kMaxBlockPerChunks = 128;
-
- if (!use_monotonic_memory_) {
- pool_memory.emplace(kMaxBlockPerChunks, kExecutionPoolMaxBlockSize, &resource_with_exception,
- &resource_with_exception);
- } else {
- // We can throw on every query because a simple queries for deleting will use only
- // the stack allocated buffer.
- // Also, we want to throw only when the query engine requests more memory and not the storage
- // so we add the exception to the allocator.
- // TODO (mferencevic): Tune the parameters accordingly.
- pool_memory.emplace(kMaxBlockPerChunks, 1024, &monotonic_memory, &resource_with_exception);
- }
-
- ctx_.evaluation_context.memory = &*pool_memory;
-
// Returns true if a result was pulled.
const auto pull_result = [&]() -> bool { return cursor_->Pull(frame_, ctx_); };
- const auto stream_values = [&]() {
- // TODO: The streamed values should also probably use the above memory.
- std::vector<TypedValue> values;
- values.reserve(output_symbols.size());
-
- for (const auto &symbol : output_symbols) {
- values.emplace_back(frame_[symbol]);
+ auto values = std::vector<TypedValue>(output_symbols.size());
+ const auto stream_values = [&] {
+ for (auto const i : ranges::views::iota(0UL, output_symbols.size())) {
+ values[i] = frame_[output_symbols[i]];
}
-
stream->Result(values);
};
@@ -1973,7 +1915,6 @@ PreparedQuery Interpreter::PrepareTransactionQuery(std::string_view query_upper,
std::function<void()> handler;
if (query_upper == "BEGIN") {
- ResetInterpreter();
// TODO: Evaluate doing move(extras). Currently the extras is very small, but this will be important if it ever
// becomes large.
handler = [this, extras = extras] {
@@ -2051,30 +1992,6 @@ inline static void TryCaching(const AstStorage &ast_storage, FrameChangeCollecto
}
}
-bool IsLoadCsvQuery(const std::vector<memgraph::query::Clause *> &clauses) {
- return std::any_of(clauses.begin(), clauses.end(),
- [](memgraph::query::Clause const *clause) { return clause->GetTypeInfo() == LoadCsv::kType; });
-}
-
-bool IsCallBatchedProcedureQuery(const std::vector<memgraph::query::Clause *> &clauses) {
- EvaluationContext evaluation_context;
-
- return std::ranges::any_of(clauses, [&evaluation_context](memgraph::query::Clause *clause) -> bool {
- if (!(clause->GetTypeInfo() == CallProcedure::kType)) return false;
- auto *call_procedure_clause = utils::Downcast<CallProcedure>(clause);
-
- const auto &maybe_found = memgraph::query::procedure::FindProcedure(
- procedure::gModuleRegistry, call_procedure_clause->procedure_name_, evaluation_context.memory);
- if (!maybe_found) {
- throw QueryRuntimeException("There is no procedure named '{}'.", call_procedure_clause->procedure_name_);
- }
- const auto &[module, proc] = *maybe_found;
- if (!proc->info.is_batched) return false;
- spdlog::trace("Using PoolResource for batched query procedure");
- return true;
- });
-}
-
PreparedQuery PrepareCypherQuery(ParsedQuery parsed_query, std::map<std::string, TypedValue> *summary,
InterpreterContext *interpreter_context, CurrentDB ¤t_db,
utils::MemoryResource *execution_memory, std::vector<Notification> *notifications,
@@ -2094,7 +2011,6 @@ PreparedQuery PrepareCypherQuery(ParsedQuery parsed_query, std::map<std::string,
spdlog::info("Running query with memory limit of {}", utils::GetReadableSize(*memory_limit));
}
auto clauses = cypher_query->single_query_->clauses_;
- bool contains_csv = false;
if (std::any_of(clauses.begin(), clauses.end(),
[](const auto *clause) { return clause->GetTypeInfo() == LoadCsv::kType; })) {
notifications->emplace_back(
@@ -2102,13 +2018,8 @@ PreparedQuery PrepareCypherQuery(ParsedQuery parsed_query, std::map<std::string,
"It's important to note that the parser parses the values as strings. It's up to the user to "
"convert the parsed row values to the appropriate type. This can be done using the built-in "
"conversion functions such as ToInteger, ToFloat, ToBoolean etc.");
- contains_csv = true;
}
- // If this is LOAD CSV query, use PoolResource without MonotonicMemoryResource as we want to reuse allocated memory
- auto use_monotonic_memory =
- !contains_csv && !IsCallBatchedProcedureQuery(clauses) && !IsAllShortestPathsQuery(clauses);
-
MG_ASSERT(current_db.execution_db_accessor_, "Cypher query expects a current DB transaction");
auto *dba =
&*current_db
@@ -2147,7 +2058,7 @@ PreparedQuery PrepareCypherQuery(ParsedQuery parsed_query, std::map<std::string,
current_db.trigger_context_collector_ ? &*current_db.trigger_context_collector_ : nullptr;
auto pull_plan = std::make_shared<PullPlan>(
plan, parsed_query.parameters, false, dba, interpreter_context, execution_memory, std::move(user_or_role),
- transaction_status, std::move(tx_timer), trigger_context_collector, memory_limit, use_monotonic_memory,
+ transaction_status, std::move(tx_timer), trigger_context_collector, memory_limit,
frame_change_collector->IsTrackingValues() ? frame_change_collector : nullptr);
return PreparedQuery{std::move(header), std::move(parsed_query.required_privileges),
[pull_plan = std::move(pull_plan), output_symbols = std::move(output_symbols), summary](
@@ -2261,18 +2172,6 @@ PreparedQuery PrepareProfileQuery(ParsedQuery parsed_query, bool in_explicit_tra
auto *cypher_query = utils::Downcast<CypherQuery>(parsed_inner_query.query);
- bool contains_csv = false;
- auto clauses = cypher_query->single_query_->clauses_;
- if (std::any_of(clauses.begin(), clauses.end(),
- [](const auto *clause) { return clause->GetTypeInfo() == LoadCsv::kType; })) {
- contains_csv = true;
- }
-
- // If this is LOAD CSV, BatchedProcedure or AllShortest query, use PoolResource without MonotonicMemoryResource as we
- // want to reuse allocated memory
- auto use_monotonic_memory =
- !contains_csv && !IsCallBatchedProcedureQuery(clauses) && !IsAllShortestPathsQuery(clauses);
-
MG_ASSERT(cypher_query, "Cypher grammar should not allow other queries in PROFILE");
EvaluationContext evaluation_context;
evaluation_context.timestamp = QueryTimestamp();
@@ -2306,14 +2205,14 @@ PreparedQuery PrepareProfileQuery(ParsedQuery parsed_query, bool in_explicit_tra
// We want to execute the query we are profiling lazily, so we delay
// the construction of the corresponding context.
stats_and_total_time = std::optional<plan::ProfilingStatsWithTotalTime>{},
- pull_plan = std::shared_ptr<PullPlanVector>(nullptr), transaction_status, use_monotonic_memory,
- frame_change_collector, tx_timer = std::move(tx_timer)](
- AnyStream *stream, std::optional<int> n) mutable -> std::optional<QueryHandlerResult> {
+ pull_plan = std::shared_ptr<PullPlanVector>(nullptr), transaction_status, frame_change_collector,
+ tx_timer = std::move(tx_timer)](AnyStream *stream,
+ std::optional<int> n) mutable -> std::optional<QueryHandlerResult> {
// No output symbols are given so that nothing is streamed.
if (!stats_and_total_time) {
stats_and_total_time =
PullPlan(plan, parameters, true, dba, interpreter_context, execution_memory, std::move(user_or_role),
- transaction_status, std::move(tx_timer), nullptr, memory_limit, use_monotonic_memory,
+ transaction_status, std::move(tx_timer), nullptr, memory_limit,
frame_change_collector->IsTrackingValues() ? frame_change_collector : nullptr)
.Pull(stream, {}, {}, summary);
pull_plan = std::make_shared<PullPlanVector>(ProfilingStatsToTable(*stats_and_total_time));
@@ -2811,6 +2710,75 @@ PreparedQuery PrepareEdgeIndexQuery(ParsedQuery parsed_query, bool in_explicit_t
RWType::W};
}
+PreparedQuery PrepareTextIndexQuery(ParsedQuery parsed_query, bool in_explicit_transaction,
+ std::vector<Notification> *notifications, CurrentDB ¤t_db) {
+ if (in_explicit_transaction) {
+ throw IndexInMulticommandTxException();
+ }
+
+ auto *text_index_query = utils::Downcast<TextIndexQuery>(parsed_query.query);
+ std::function<void(Notification &)> handler;
+
+ // TODO: we will need transaction for replication
+ MG_ASSERT(current_db.db_acc_, "Text index query expects a current DB");
+ auto &db_acc = *current_db.db_acc_;
+
+ MG_ASSERT(current_db.db_transactional_accessor_, "Text index query expects a current DB transaction");
+ auto *dba = &*current_db.execution_db_accessor_;
+
+ // Creating an index influences computed plan costs.
+ auto invalidate_plan_cache = [plan_cache = db_acc->plan_cache()] {
+ plan_cache->WithLock([&](auto &cache) { cache.reset(); });
+ };
+
+ auto *storage = db_acc->storage();
+ auto label = storage->NameToLabel(text_index_query->label_.name);
+ auto &index_name = text_index_query->index_name_;
+
+ Notification index_notification(SeverityLevel::INFO);
+ switch (text_index_query->action_) {
+ case TextIndexQuery::Action::CREATE: {
+ index_notification.code = NotificationCode::CREATE_INDEX;
+ index_notification.title = fmt::format("Created text index on label {}.", text_index_query->label_.name);
+ // TODO: not just storage + invalidate_plan_cache. Need a DB transaction (for replication)
+ handler = [dba, label, index_name,
+ invalidate_plan_cache = std::move(invalidate_plan_cache)](Notification &index_notification) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw TextSearchDisabledException();
+ }
+ dba->CreateTextIndex(index_name, label);
+ utils::OnScopeExit invalidator(invalidate_plan_cache);
+ };
+ break;
+ }
+ case TextIndexQuery::Action::DROP: {
+ index_notification.code = NotificationCode::DROP_INDEX;
+ index_notification.title = fmt::format("Dropped text index on label {}.", text_index_query->label_.name);
+ // TODO: not just storage + invalidate_plan_cache. Need a DB transaction (for replication)
+ handler = [dba, index_name,
+ invalidate_plan_cache = std::move(invalidate_plan_cache)](Notification &index_notification) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw TextSearchDisabledException();
+ }
+ dba->DropTextIndex(index_name);
+ utils::OnScopeExit invalidator(invalidate_plan_cache);
+ };
+ break;
+ }
+ }
+
+ return PreparedQuery{
+ {},
+ std::move(parsed_query.required_privileges),
+ [handler = std::move(handler), notifications, index_notification = std::move(index_notification)](
+ AnyStream * /*stream*/, std::optional<int> /*unused*/) mutable {
+ handler(index_notification);
+ notifications->push_back(index_notification);
+ return QueryHandlerResult::COMMIT; // TODO: Will need to become COMMIT when we fix replication
+ },
+ RWType::W};
+}
+
PreparedQuery PrepareAuthQuery(ParsedQuery parsed_query, bool in_explicit_transaction,
InterpreterContext *interpreter_context, Interpreter &interpreter) {
if (in_explicit_transaction) {
@@ -3601,7 +3569,7 @@ PreparedQuery PrepareDatabaseInfoQuery(ParsedQuery parsed_query, bool in_explici
}
MG_ASSERT(current_db.db_acc_, "Database info query expects a current DB");
- MG_ASSERT(current_db.db_transactional_accessor_, "Database ifo query expects a current DB transaction");
+ MG_ASSERT(current_db.db_transactional_accessor_, "Database info query expects a current DB transaction");
auto *dba = &*current_db.execution_db_accessor_;
auto *info_query = utils::Downcast<DatabaseInfoQuery>(parsed_query.query);
@@ -3616,10 +3584,11 @@ PreparedQuery PrepareDatabaseInfoQuery(ParsedQuery parsed_query, bool in_explici
const std::string_view label_index_mark{"label"};
const std::string_view label_property_index_mark{"label+property"};
const std::string_view edge_type_index_mark{"edge-type"};
+ const std::string_view text_index_mark{"text"};
auto info = dba->ListAllIndices();
auto storage_acc = database->Access();
std::vector<std::vector<TypedValue>> results;
- results.reserve(info.label.size() + info.label_property.size());
+ results.reserve(info.label.size() + info.label_property.size() + info.text_indices.size());
for (const auto &item : info.label) {
results.push_back({TypedValue(label_index_mark), TypedValue(storage->LabelToName(item)), TypedValue(),
TypedValue(static_cast<int>(storage_acc->ApproximateVertexCount(item)))});
@@ -3634,6 +3603,10 @@ PreparedQuery PrepareDatabaseInfoQuery(ParsedQuery parsed_query, bool in_explici
results.push_back({TypedValue(edge_type_index_mark), TypedValue(storage->EdgeTypeToName(item)), TypedValue(),
TypedValue(static_cast<int>(storage_acc->ApproximateEdgeCount(item)))});
}
+ for (const auto &[index_name, label] : info.text_indices) {
+ results.push_back({TypedValue(fmt::format("{} (name: {})", text_index_mark, index_name)),
+ TypedValue(storage->LabelToName(label)), TypedValue(), TypedValue()});
+ }
std::sort(results.begin(), results.end(), [&label_index_mark](const auto &record_1, const auto &record_2) {
const auto type_1 = record_1[0].ValueString();
const auto type_2 = record_2[0].ValueString();
@@ -4276,6 +4249,7 @@ PreparedQuery PrepareShowDatabasesQuery(ParsedQuery parsed_query, InterpreterCon
std::optional<uint64_t> Interpreter::GetTransactionId() const { return current_transaction_; }
void Interpreter::BeginTransaction(QueryExtras const &extras) {
+ ResetInterpreter();
const auto prepared_query = PrepareTransactionQuery("BEGIN", extras);
prepared_query.query_handler(nullptr, {});
}
@@ -4310,12 +4284,12 @@ Interpreter::PrepareResult Interpreter::Prepare(const std::string &query_string,
const auto upper_case_query = utils::ToUpperCase(query_string);
const auto trimmed_query = utils::Trim(upper_case_query);
if (trimmed_query == "BEGIN" || trimmed_query == "COMMIT" || trimmed_query == "ROLLBACK") {
- auto resource = utils::MonotonicBufferResource(kExecutionMemoryBlockSize);
- auto prepared_query = PrepareTransactionQuery(trimmed_query, extras);
- auto &query_execution =
- query_executions_.emplace_back(QueryExecution::Create(std::move(resource), std::move(prepared_query)));
- std::optional<int> qid =
- in_explicit_transaction_ ? static_cast<int>(query_executions_.size() - 1) : std::optional<int>{};
+ if (trimmed_query == "BEGIN") {
+ ResetInterpreter();
+ }
+ auto &query_execution = query_executions_.emplace_back(QueryExecution::Create());
+ query_execution->prepared_query = PrepareTransactionQuery(trimmed_query, extras);
+ auto qid = in_explicit_transaction_ ? static_cast<int>(query_executions_.size() - 1) : std::optional<int>{};
return {query_execution->prepared_query->header, query_execution->prepared_query->privileges, qid, {}};
}
@@ -4345,35 +4319,8 @@ Interpreter::PrepareResult Interpreter::Prepare(const std::string &query_string,
ParseQuery(query_string, params, &interpreter_context_->ast_cache, interpreter_context_->config.query);
auto parsing_time = parsing_timer.Elapsed().count();
- CypherQuery const *const cypher_query = [&]() -> CypherQuery * {
- if (auto *cypher_query = utils::Downcast<CypherQuery>(parsed_query.query)) {
- return cypher_query;
- }
- if (auto *profile_query = utils::Downcast<ProfileQuery>(parsed_query.query)) {
- return profile_query->cypher_query_;
- }
- return nullptr;
- }(); // IILE
-
- auto const [usePool, hasAllShortestPaths] = [&]() -> std::pair<bool, bool> {
- if (!cypher_query) {
- return {false, false};
- }
- auto const &clauses = cypher_query->single_query_->clauses_;
- bool hasAllShortestPaths = IsAllShortestPathsQuery(clauses);
- // Using PoolResource without MonotonicMemoryResouce for LOAD CSV reduces memory usage.
- bool usePool = hasAllShortestPaths || IsCallBatchedProcedureQuery(clauses) || IsLoadCsvQuery(clauses);
- return {usePool, hasAllShortestPaths};
- }(); // IILE
-
// Setup QueryExecution
- // its MemoryResource is mostly used for allocations done on Frame and storing `row`s
- if (usePool) {
- query_executions_.emplace_back(QueryExecution::Create(utils::PoolResource(128, kExecutionPoolMaxBlockSize)));
- } else {
- query_executions_.emplace_back(QueryExecution::Create(utils::MonotonicBufferResource(kExecutionMemoryBlockSize)));
- }
-
+ query_executions_.emplace_back(QueryExecution::Create());
auto &query_execution = query_executions_.back();
query_execution_ptr = &query_execution;
@@ -4421,13 +4368,15 @@ Interpreter::PrepareResult Interpreter::Prepare(const std::string &query_string,
utils::Downcast<ProfileQuery>(parsed_query.query) || utils::Downcast<DumpQuery>(parsed_query.query) ||
utils::Downcast<TriggerQuery>(parsed_query.query) || utils::Downcast<AnalyzeGraphQuery>(parsed_query.query) ||
utils::Downcast<IndexQuery>(parsed_query.query) || utils::Downcast<EdgeIndexQuery>(parsed_query.query) ||
- utils::Downcast<DatabaseInfoQuery>(parsed_query.query) || utils::Downcast<ConstraintQuery>(parsed_query.query);
+ utils::Downcast<TextIndexQuery>(parsed_query.query) || utils::Downcast<DatabaseInfoQuery>(parsed_query.query) ||
+ utils::Downcast<ConstraintQuery>(parsed_query.query);
if (!in_explicit_transaction_ && requires_db_transaction) {
// TODO: ATM only a single database, will change when we have multiple database transactions
bool could_commit = utils::Downcast<CypherQuery>(parsed_query.query) != nullptr;
bool unique = utils::Downcast<IndexQuery>(parsed_query.query) != nullptr ||
utils::Downcast<EdgeIndexQuery>(parsed_query.query) != nullptr ||
+ utils::Downcast<TextIndexQuery>(parsed_query.query) != nullptr ||
utils::Downcast<ConstraintQuery>(parsed_query.query) != nullptr ||
upper_case_query.find(kSchemaAssert) != std::string::npos;
SetupDatabaseTransaction(could_commit, unique);
@@ -4442,9 +4391,7 @@ Interpreter::PrepareResult Interpreter::Prepare(const std::string &query_string,
utils::Timer planning_timer;
PreparedQuery prepared_query;
- utils::MemoryResource *memory_resource =
- std::visit([](auto &execution_memory) -> utils::MemoryResource * { return &execution_memory; },
- query_execution->execution_memory);
+ utils::MemoryResource *memory_resource = query_execution->execution_memory.resource();
frame_change_collector_.reset();
frame_change_collector_.emplace();
if (utils::Downcast<CypherQuery>(parsed_query.query)) {
@@ -4455,10 +4402,10 @@ Interpreter::PrepareResult Interpreter::Prepare(const std::string &query_string,
prepared_query = PrepareExplainQuery(std::move(parsed_query), &query_execution->summary,
&query_execution->notifications, interpreter_context_, current_db_);
} else if (utils::Downcast<ProfileQuery>(parsed_query.query)) {
- prepared_query = PrepareProfileQuery(std::move(parsed_query), in_explicit_transaction_, &query_execution->summary,
- &query_execution->notifications, interpreter_context_, current_db_,
- &query_execution->execution_memory_with_exception, user_or_role_,
- &transaction_status_, current_timeout_timer_, &*frame_change_collector_);
+ prepared_query =
+ PrepareProfileQuery(std::move(parsed_query), in_explicit_transaction_, &query_execution->summary,
+ &query_execution->notifications, interpreter_context_, current_db_, memory_resource,
+ user_or_role_, &transaction_status_, current_timeout_timer_, &*frame_change_collector_);
} else if (utils::Downcast<DumpQuery>(parsed_query.query)) {
prepared_query = PrepareDumpQuery(std::move(parsed_query), current_db_);
} else if (utils::Downcast<IndexQuery>(parsed_query.query)) {
@@ -4467,6 +4414,9 @@ Interpreter::PrepareResult Interpreter::Prepare(const std::string &query_string,
} else if (utils::Downcast<EdgeIndexQuery>(parsed_query.query)) {
prepared_query = PrepareEdgeIndexQuery(std::move(parsed_query), in_explicit_transaction_,
&query_execution->notifications, current_db_);
+ } else if (utils::Downcast<TextIndexQuery>(parsed_query.query)) {
+ prepared_query = PrepareTextIndexQuery(std::move(parsed_query), in_explicit_transaction_,
+ &query_execution->notifications, current_db_);
} else if (utils::Downcast<AnalyzeGraphQuery>(parsed_query.query)) {
prepared_query = PrepareAnalyzeGraphQuery(std::move(parsed_query), in_explicit_transaction_, current_db_);
} else if (utils::Downcast<AuthQuery>(parsed_query.query)) {
@@ -4660,7 +4610,7 @@ void RunTriggersAfterCommit(dbms::DatabaseAccess db_acc, InterpreterContext *int
std::atomic<TransactionStatus> *transaction_status) {
// Run the triggers
for (const auto &trigger : db_acc->trigger_store()->AfterCommitTriggers().access()) {
- utils::MonotonicBufferResource execution_memory{kExecutionMemoryBlockSize};
+ QueryAllocator execution_memory{};
// create a new transaction for each trigger
auto tx_acc = db_acc->Access();
@@ -4671,7 +4621,7 @@ void RunTriggersAfterCommit(dbms::DatabaseAccess db_acc, InterpreterContext *int
auto trigger_context = original_trigger_context;
trigger_context.AdaptForAccessor(&db_accessor);
try {
- trigger.Execute(&db_accessor, &execution_memory, flags::run_time::GetExecutionTimeout(),
+ trigger.Execute(&db_accessor, execution_memory.resource(), flags::run_time::GetExecutionTimeout(),
&interpreter_context->is_shutting_down, transaction_status, trigger_context);
} catch (const utils::BasicException &exception) {
spdlog::warn("Trigger '{}' failed with exception:\n{}", trigger.Name(), exception.what());
@@ -4825,11 +4775,12 @@ void Interpreter::Commit() {
if (trigger_context) {
// Run the triggers
for (const auto &trigger : db->trigger_store()->BeforeCommitTriggers().access()) {
- utils::MonotonicBufferResource execution_memory{kExecutionMemoryBlockSize};
+ QueryAllocator execution_memory{};
AdvanceCommand();
try {
- trigger.Execute(&*current_db_.execution_db_accessor_, &execution_memory, flags::run_time::GetExecutionTimeout(),
- &interpreter_context_->is_shutting_down, &transaction_status_, *trigger_context);
+ trigger.Execute(&*current_db_.execution_db_accessor_, execution_memory.resource(),
+ flags::run_time::GetExecutionTimeout(), &interpreter_context_->is_shutting_down,
+ &transaction_status_, *trigger_context);
} catch (const utils::BasicException &e) {
throw utils::BasicException(
fmt::format("Trigger '{}' caused the transaction to fail.\nException: {}", trigger.Name(), e.what()));
diff --git a/src/query/interpreter.hpp b/src/query/interpreter.hpp
index f18bd6721..5366b4472 100644
--- a/src/query/interpreter.hpp
+++ b/src/query/interpreter.hpp
@@ -65,6 +65,54 @@ extern const Event SuccessfulQuery;
namespace memgraph::query {
+struct QueryAllocator {
+ QueryAllocator() = default;
+ QueryAllocator(QueryAllocator const &) = delete;
+ QueryAllocator &operator=(QueryAllocator const &) = delete;
+
+ // No move addresses to pool & monotonic fields must be stable
+ QueryAllocator(QueryAllocator &&) = delete;
+ QueryAllocator &operator=(QueryAllocator &&) = delete;
+
+ auto resource() -> utils::MemoryResource * {
+#ifndef MG_MEMORY_PROFILE
+ return &pool;
+#else
+ return upstream_resource();
+#endif
+ }
+ auto resource_without_pool() -> utils::MemoryResource * {
+#ifndef MG_MEMORY_PROFILE
+ return &monotonic;
+#else
+ return upstream_resource();
+#endif
+ }
+ auto resource_without_pool_or_mono() -> utils::MemoryResource * { return upstream_resource(); }
+
+ private:
+ // At least one page to ensure not sharing page with other subsystems
+ static constexpr auto kMonotonicInitialSize = 4UL * 1024UL;
+ // TODO: need to profile to check for good defaults, also maybe PoolResource
+ // needs to be smarter. We expect more reuse of smaller objects than larger
+ // objects. 64*1024B is maybe wasteful, whereas 256*32B maybe sensible.
+ // Depends on number of small objects expected.
+ static constexpr auto kPoolBlockPerChunk = 64UL;
+ static constexpr auto kPoolMaxBlockSize = 1024UL;
+
+ static auto upstream_resource() -> utils::MemoryResource * {
+ // singleton ResourceWithOutOfMemoryException
+ // explicitly backed by NewDeleteResource
+ static auto upstream = utils::ResourceWithOutOfMemoryException{utils::NewDeleteResource()};
+ return &upstream;
+ }
+
+#ifndef MG_MEMORY_PROFILE
+ memgraph::utils::MonotonicBufferResource monotonic{kMonotonicInitialSize, upstream_resource()};
+ memgraph::utils::PoolResource pool{kPoolBlockPerChunk, &monotonic, upstream_resource()};
+#endif
+};
+
struct InterpreterContext;
inline constexpr size_t kExecutionMemoryBlockSize = 1UL * 1024UL * 1024UL;
@@ -304,45 +352,25 @@ class Interpreter final {
}
struct QueryExecution {
- std::variant<utils::MonotonicBufferResource, utils::PoolResource> execution_memory;
- utils::ResourceWithOutOfMemoryException execution_memory_with_exception;
- std::optional<PreparedQuery> prepared_query;
+ QueryAllocator execution_memory; // NOTE: before all other fields which uses this memory
+ std::optional<PreparedQuery> prepared_query;
std::map<std::string, TypedValue> summary;
std::vector<Notification> notifications;
- static auto Create(std::variant<utils::MonotonicBufferResource, utils::PoolResource> memory_resource,
- std::optional<PreparedQuery> prepared_query = std::nullopt) -> std::unique_ptr<QueryExecution> {
- return std::make_unique<QueryExecution>(std::move(memory_resource), std::move(prepared_query));
- }
+ static auto Create() -> std::unique_ptr<QueryExecution> { return std::make_unique<QueryExecution>(); }
- explicit QueryExecution(std::variant<utils::MonotonicBufferResource, utils::PoolResource> memory_resource,
- std::optional<PreparedQuery> prepared_query)
- : execution_memory(std::move(memory_resource)), prepared_query{std::move(prepared_query)} {
- std::visit(
- [&](auto &memory_resource) {
- execution_memory_with_exception = utils::ResourceWithOutOfMemoryException(&memory_resource);
- },
- execution_memory);
- };
+ explicit QueryExecution() = default;
QueryExecution(const QueryExecution &) = delete;
- QueryExecution(QueryExecution &&) = default;
+ QueryExecution(QueryExecution &&) = delete;
QueryExecution &operator=(const QueryExecution &) = delete;
- QueryExecution &operator=(QueryExecution &&) = default;
+ QueryExecution &operator=(QueryExecution &&) = delete;
- ~QueryExecution() {
- // We should always release the execution memory AFTER we
- // destroy the prepared query which is using that instance
- // of execution memory.
- prepared_query.reset();
- std::visit([](auto &memory_resource) { memory_resource.Release(); }, execution_memory);
- }
+ ~QueryExecution() = default;
void CleanRuntimeData() {
- if (prepared_query.has_value()) {
- prepared_query.reset();
- }
+ prepared_query.reset();
notifications.clear();
}
};
@@ -413,9 +441,7 @@ std::map<std::string, TypedValue> Interpreter::Pull(TStream *result_stream, std:
try {
// Wrap the (statically polymorphic) stream type into a common type which
// the handler knows.
- AnyStream stream{result_stream,
- std::visit([](auto &execution_memory) -> utils::MemoryResource * { return &execution_memory; },
- query_execution->execution_memory)};
+ AnyStream stream{result_stream, query_execution->execution_memory.resource()};
const auto maybe_res = query_execution->prepared_query->query_handler(&stream, n);
// Stream is using execution memory of the query_execution which
// can be deleted after its execution so the stream should be cleared
diff --git a/src/query/plan/operator.cpp b/src/query/plan/operator.cpp
index d39ade99f..e6dca1b12 100644
--- a/src/query/plan/operator.cpp
+++ b/src/query/plan/operator.cpp
@@ -32,6 +32,7 @@
#include "spdlog/spdlog.h"
#include "csv/parsing.hpp"
+#include "flags/experimental.hpp"
#include "license/license.hpp"
#include "query/auth_checker.hpp"
#include "query/context.hpp"
@@ -69,6 +70,7 @@
#include "utils/pmr/vector.hpp"
#include "utils/readable_size.hpp"
#include "utils/string.hpp"
+#include "utils/tag.hpp"
#include "utils/temporal.hpp"
#include "utils/typeinfo.hpp"
@@ -266,6 +268,10 @@ VertexAccessor &CreateLocalVertex(const NodeCreationInfo &node_info, Frame *fram
}
MultiPropsInitChecked(&new_node, properties);
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context.db_accessor->TextIndexAddVertex(new_node);
+ }
+
(*frame)[node_info.symbol] = new_node;
return (*frame)[node_info.symbol].ValueVertex();
}
@@ -329,7 +335,7 @@ CreateExpand::CreateExpand(NodeCreationInfo node_info, EdgeCreationInfo edge_inf
ACCEPT_WITH_INPUT(CreateExpand)
UniqueCursorPtr CreateExpand::MakeCursor(utils::MemoryResource *mem) const {
- memgraph::metrics::IncrementCounter(memgraph::metrics::CreateNodeOperator);
+ memgraph::metrics::IncrementCounter(memgraph::metrics::CreateExpandOperator);
return MakeUniqueCursorPtr<CreateExpandCursor>(mem, *this, mem);
}
@@ -865,17 +871,15 @@ bool Expand::ExpandCursor::Pull(Frame &frame, ExecutionContext &context) {
SCOPED_PROFILE_OP_BY_REF(self_);
// A helper function for expanding a node from an edge.
- auto pull_node = [this, &frame](const EdgeAccessor &new_edge, EdgeAtom::Direction direction) {
+ auto pull_node = [this, &frame]<EdgeAtom::Direction direction>(const EdgeAccessor &new_edge,
+ utils::tag_value<direction>) {
if (self_.common_.existing_node) return;
- switch (direction) {
- case EdgeAtom::Direction::IN:
- frame[self_.common_.node_symbol] = new_edge.From();
- break;
- case EdgeAtom::Direction::OUT:
- frame[self_.common_.node_symbol] = new_edge.To();
- break;
- case EdgeAtom::Direction::BOTH:
- LOG_FATAL("Must indicate exact expansion direction here");
+ if constexpr (direction == EdgeAtom::Direction::IN) {
+ frame[self_.common_.node_symbol] = new_edge.From();
+ } else if constexpr (direction == EdgeAtom::Direction::OUT) {
+ frame[self_.common_.node_symbol] = new_edge.To();
+ } else {
+ LOG_FATAL("Must indicate exact expansion direction here");
}
};
@@ -894,7 +898,7 @@ bool Expand::ExpandCursor::Pull(Frame &frame, ExecutionContext &context) {
#endif
frame[self_.common_.edge_symbol] = edge;
- pull_node(edge, EdgeAtom::Direction::IN);
+ pull_node(edge, utils::tag_v<EdgeAtom::Direction::IN>);
return true;
}
@@ -914,7 +918,7 @@ bool Expand::ExpandCursor::Pull(Frame &frame, ExecutionContext &context) {
}
#endif
frame[self_.common_.edge_symbol] = edge;
- pull_node(edge, EdgeAtom::Direction::OUT);
+ pull_node(edge, utils::tag_v<EdgeAtom::Direction::OUT>);
return true;
}
@@ -1008,12 +1012,12 @@ bool Expand::ExpandCursor::InitEdges(Frame &frame, ExecutionContext &context) {
auto existing_node = *expansion_info_.existing_node;
auto edges_result = UnwrapEdgesResult(vertex.InEdges(self_.view_, self_.common_.edge_types, existing_node));
- in_edges_.emplace(edges_result.edges);
+ in_edges_.emplace(std::move(edges_result.edges));
num_expanded_first = edges_result.expanded_count;
}
} else {
auto edges_result = UnwrapEdgesResult(vertex.InEdges(self_.view_, self_.common_.edge_types));
- in_edges_.emplace(edges_result.edges);
+ in_edges_.emplace(std::move(edges_result.edges));
num_expanded_first = edges_result.expanded_count;
}
if (in_edges_) {
@@ -1027,12 +1031,12 @@ bool Expand::ExpandCursor::InitEdges(Frame &frame, ExecutionContext &context) {
if (expansion_info_.existing_node) {
auto existing_node = *expansion_info_.existing_node;
auto edges_result = UnwrapEdgesResult(vertex.OutEdges(self_.view_, self_.common_.edge_types, existing_node));
- out_edges_.emplace(edges_result.edges);
+ out_edges_.emplace(std::move(edges_result.edges));
num_expanded_second = edges_result.expanded_count;
}
} else {
auto edges_result = UnwrapEdgesResult(vertex.OutEdges(self_.view_, self_.common_.edge_types));
- out_edges_.emplace(edges_result.edges);
+ out_edges_.emplace(std::move(edges_result.edges));
num_expanded_second = edges_result.expanded_count;
}
if (out_edges_) {
@@ -1118,14 +1122,14 @@ auto ExpandFromVertex(const VertexAccessor &vertex, EdgeAtom::Direction directio
if (direction != EdgeAtom::Direction::OUT) {
auto edges = UnwrapEdgesResult(vertex.InEdges(view, edge_types)).edges;
- if (edges.begin() != edges.end()) {
+ if (!edges.empty()) {
chain_elements.emplace_back(wrapper(EdgeAtom::Direction::IN, std::move(edges)));
}
}
if (direction != EdgeAtom::Direction::IN) {
auto edges = UnwrapEdgesResult(vertex.OutEdges(view, edge_types)).edges;
- if (edges.begin() != edges.end()) {
+ if (!edges.empty()) {
chain_elements.emplace_back(wrapper(EdgeAtom::Direction::OUT, std::move(edges)));
}
}
@@ -1245,8 +1249,13 @@ class ExpandVariableCursor : public Cursor {
}
// reset the frame value to an empty edge list
- auto *pull_memory = context.evaluation_context.memory;
- frame[self_.common_.edge_symbol] = TypedValue::TVector(pull_memory);
+ if (frame[self_.common_.edge_symbol].IsList()) {
+ // Preserve the list capacity if possible
+ frame[self_.common_.edge_symbol].ValueList().clear();
+ } else {
+ auto *pull_memory = context.evaluation_context.memory;
+ frame[self_.common_.edge_symbol] = TypedValue::TVector(pull_memory);
+ }
return true;
}
@@ -2988,6 +2997,9 @@ bool SetProperty::SetPropertyCursor::Pull(Frame &frame, ExecutionContext &contex
context.trigger_context_collector->RegisterSetObjectProperty(lhs.ValueVertex(), self_.property_,
TypedValue{std::move(old_value)}, TypedValue{rhs});
}
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context.db_accessor->TextIndexUpdateVertex(lhs.ValueVertex());
+ }
break;
}
case TypedValue::Type::Edge: {
@@ -3144,6 +3156,9 @@ void SetPropertiesOnRecord(TRecordAccessor *record, const TypedValue &rhs, SetPr
case TypedValue::Type::Vertex: {
PropertiesMap new_properties = get_props(rhs.ValueVertex());
update_props(new_properties);
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context->db_accessor->TextIndexUpdateVertex(rhs.ValueVertex());
+ }
break;
}
case TypedValue::Type::Map: {
@@ -3201,6 +3216,9 @@ bool SetProperties::SetPropertiesCursor::Pull(Frame &frame, ExecutionContext &co
}
#endif
SetPropertiesOnRecord(&lhs.ValueVertex(), rhs, self_.op_, &context, cached_name_id_);
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context.db_accessor->TextIndexUpdateVertex(lhs.ValueVertex());
+ }
break;
case TypedValue::Type::Edge:
#ifdef MG_ENTERPRISE
@@ -3292,6 +3310,10 @@ bool SetLabels::SetLabelsCursor::Pull(Frame &frame, ExecutionContext &context) {
}
}
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context.db_accessor->TextIndexUpdateVertex(vertex);
+ }
+
return true;
}
@@ -3363,6 +3385,9 @@ bool RemoveProperty::RemovePropertyCursor::Pull(Frame &frame, ExecutionContext &
}
#endif
remove_prop(&lhs.ValueVertex());
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context.db_accessor->TextIndexUpdateVertex(lhs.ValueVertex());
+ }
break;
case TypedValue::Type::Edge:
#ifdef MG_ENTERPRISE
@@ -3455,6 +3480,10 @@ bool RemoveLabels::RemoveLabelsCursor::Pull(Frame &frame, ExecutionContext &cont
}
}
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ context.db_accessor->TextIndexUpdateVertex(vertex, EvaluateLabels(self_.labels_, evaluator, context.db_accessor));
+ }
+
return true;
}
@@ -4475,9 +4504,8 @@ class UnwindCursor : public Cursor {
TypedValue input_value = self_.input_expression_->Accept(evaluator);
if (input_value.type() != TypedValue::Type::List)
throw QueryRuntimeException("Argument of UNWIND must be a list, but '{}' was provided.", input_value.type());
- // Copy the evaluted input_value_list to our vector.
- // eval memory != query memory
- input_value_ = input_value.ValueList();
+ // Move the evaluted input_value_list to our vector.
+ input_value_ = std::move(input_value.ValueList());
input_value_it_ = input_value_.begin();
}
@@ -5337,6 +5365,7 @@ class LoadCsvCursor : public Cursor {
"1");
}
did_pull_ = true;
+ reader_->Reset();
}
auto row = reader_->GetNextRow(context.evaluation_context.memory);
diff --git a/src/query/plan/operator.hpp b/src/query/plan/operator.hpp
index fa736cd86..a716fa645 100644
--- a/src/query/plan/operator.hpp
+++ b/src/query/plan/operator.hpp
@@ -76,18 +76,13 @@ using UniqueCursorPtr = std::unique_ptr<Cursor, std::function<void(Cursor *)>>;
template <class TCursor, class... TArgs>
std::unique_ptr<Cursor, std::function<void(Cursor *)>> MakeUniqueCursorPtr(utils::Allocator<TCursor> allocator,
TArgs &&...args) {
- auto *ptr = allocator.allocate(1);
- try {
- auto *cursor = new (ptr) TCursor(std::forward<TArgs>(args)...);
- return std::unique_ptr<Cursor, std::function<void(Cursor *)>>(cursor, [allocator](Cursor *base_ptr) mutable {
- auto *p = static_cast<TCursor *>(base_ptr);
- p->~TCursor();
- allocator.deallocate(p, 1);
- });
- } catch (...) {
- allocator.deallocate(ptr, 1);
- throw;
- }
+ auto *cursor = allocator.template new_object<TCursor>(std::forward<TArgs>(args)...);
+ auto dtr = [allocator](Cursor *base_ptr) mutable {
+ auto *p = static_cast<TCursor *>(base_ptr);
+ allocator.delete_object(p);
+ };
+ // TODO: not std::function
+ return std::unique_ptr<Cursor, std::function<void(Cursor *)>>(cursor, std::move(dtr));
}
class Once;
diff --git a/src/query/plan/vertex_count_cache.hpp b/src/query/plan/vertex_count_cache.hpp
index 802f4e09f..69e002c0a 100644
--- a/src/query/plan/vertex_count_cache.hpp
+++ b/src/query/plan/vertex_count_cache.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
diff --git a/src/query/procedure/mg_procedure_impl.cpp b/src/query/procedure/mg_procedure_impl.cpp
index d6ce3c7b7..a2bc23aa3 100644
--- a/src/query/procedure/mg_procedure_impl.cpp
+++ b/src/query/procedure/mg_procedure_impl.cpp
@@ -23,6 +23,8 @@
#include <utility>
#include <variant>
+#include "flags/experimental.hpp"
+#include "flags/run_time_configurable.hpp"
#include "license/license.hpp"
#include "mg_procedure.h"
#include "module.hpp"
@@ -32,6 +34,7 @@
#include "query/procedure/fmt.hpp"
#include "query/procedure/mg_procedure_helpers.hpp"
#include "query/stream/common.hpp"
+#include "storage/v2/indices/text_index.hpp"
#include "storage/v2/property_value.hpp"
#include "storage/v2/storage_mode.hpp"
#include "storage/v2/view.hpp"
@@ -1843,6 +1846,11 @@ mgp_error mgp_vertex_set_property(struct mgp_vertex *v, const char *property_nam
const auto result = std::visit(
[prop_key, property_value](auto &impl) { return impl.SetProperty(prop_key, ToPropertyValue(*property_value)); },
v->impl);
+ if (memgraph::flags::AreExperimentsEnabled(memgraph::flags::Experiments::TEXT_SEARCH) && !result.HasError()) {
+ auto v_impl = v->getImpl();
+ v->graph->getImpl()->TextIndexUpdateVertex(v_impl);
+ }
+
if (result.HasError()) {
switch (result.GetError()) {
case memgraph::storage::Error::DELETED_OBJECT:
@@ -1899,6 +1907,11 @@ mgp_error mgp_vertex_set_properties(struct mgp_vertex *v, struct mgp_map *proper
}
const auto result = v->getImpl().UpdateProperties(props);
+ if (memgraph::flags::AreExperimentsEnabled(memgraph::flags::Experiments::TEXT_SEARCH) && !result.HasError()) {
+ auto v_impl = v->getImpl();
+ v->graph->getImpl()->TextIndexUpdateVertex(v_impl);
+ }
+
if (result.HasError()) {
switch (result.GetError()) {
case memgraph::storage::Error::DELETED_OBJECT:
@@ -1956,6 +1969,10 @@ mgp_error mgp_vertex_add_label(struct mgp_vertex *v, mgp_label label) {
}
const auto result = std::visit([label_id](auto &impl) { return impl.AddLabel(label_id); }, v->impl);
+ if (memgraph::flags::AreExperimentsEnabled(memgraph::flags::Experiments::TEXT_SEARCH) && !result.HasError()) {
+ auto v_impl = v->getImpl();
+ v->graph->getImpl()->TextIndexUpdateVertex(v_impl);
+ }
if (result.HasError()) {
switch (result.GetError()) {
@@ -1998,6 +2015,10 @@ mgp_error mgp_vertex_remove_label(struct mgp_vertex *v, mgp_label label) {
throw ImmutableObjectException{"Cannot remove a label from an immutable vertex!"};
}
const auto result = std::visit([label_id](auto &impl) { return impl.RemoveLabel(label_id); }, v->impl);
+ if (memgraph::flags::AreExperimentsEnabled(memgraph::flags::Experiments::TEXT_SEARCH) && !result.HasError()) {
+ auto v_impl = v->getImpl();
+ v->graph->getImpl()->TextIndexUpdateVertex(v_impl, {label_id});
+ }
if (result.HasError()) {
switch (result.GetError()) {
@@ -2590,7 +2611,7 @@ mgp_error mgp_edge_iter_properties(mgp_edge *e, mgp_memory *memory, mgp_properti
mgp_error mgp_graph_get_vertex_by_id(mgp_graph *graph, mgp_vertex_id id, mgp_memory *memory, mgp_vertex **result) {
return WrapExceptions(
[graph, id, memory]() -> mgp_vertex * {
- std::optional<memgraph::query::VertexAccessor> maybe_vertex = std::visit(
+ auto maybe_vertex = std::visit(
[graph, id](auto *impl) {
return impl->FindVertex(memgraph::storage::Gid::FromInt(id.as_int), graph->view);
},
@@ -2967,6 +2988,10 @@ mgp_error mgp_graph_create_vertex(struct mgp_graph *graph, mgp_memory *memory, m
}
auto *vertex = std::visit(
[=](auto *impl) { return NewRawMgpObject<mgp_vertex>(memory, impl->InsertVertex(), graph); }, graph->impl);
+ if (memgraph::flags::AreExperimentsEnabled(memgraph::flags::Experiments::TEXT_SEARCH)) {
+ auto v_impl = vertex->getImpl();
+ vertex->graph->getImpl()->TextIndexAddVertex(v_impl);
+ }
auto &ctx = graph->ctx;
ctx->execution_stats[memgraph::query::ExecutionStats::Key::CREATED_NODES] += 1;
@@ -3324,6 +3349,140 @@ mgp_error mgp_graph_delete_edge(struct mgp_graph *graph, mgp_edge *edge) {
});
}
+mgp_error mgp_graph_has_text_index(mgp_graph *graph, const char *index_name, int *result) {
+ return WrapExceptions([graph, index_name, result]() {
+ std::visit(memgraph::utils::Overloaded{
+ [&](memgraph::query::DbAccessor *impl) { *result = impl->TextIndexExists(index_name); },
+ [&](memgraph::query::SubgraphDbAccessor *impl) {
+ *result = impl->GetAccessor()->TextIndexExists(index_name);
+ }},
+ graph->impl);
+ });
+}
+
+mgp_vertex *GetVertexByGid(mgp_graph *graph, memgraph::storage::Gid id, mgp_memory *memory) {
+ auto get_vertex_by_gid = memgraph::utils::Overloaded{
+ [graph, id, memory](memgraph::query::DbAccessor *impl) -> mgp_vertex * {
+ auto maybe_vertex = impl->FindVertex(id, graph->view);
+ if (!maybe_vertex) return nullptr;
+ return NewRawMgpObject<mgp_vertex>(memory, *maybe_vertex, graph);
+ },
+ [graph, id, memory](memgraph::query::SubgraphDbAccessor *impl) -> mgp_vertex * {
+ auto maybe_vertex = impl->FindVertex(id, graph->view);
+ if (!maybe_vertex) return nullptr;
+ return NewRawMgpObject<mgp_vertex>(
+ memory, memgraph::query::SubgraphVertexAccessor(*maybe_vertex, impl->getGraph()), graph);
+ }};
+ return std::visit(get_vertex_by_gid, graph->impl);
+}
+
+void WrapTextSearch(mgp_graph *graph, mgp_memory *memory, mgp_map **result,
+ const std::vector<memgraph::storage::Gid> &vertex_ids = {},
+ const std::optional<std::string> &error_msg = std::nullopt) {
+ if (const auto err = mgp_map_make_empty(memory, result); err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a mgp_map");
+ }
+
+ mgp_value *error_value;
+ if (error_msg.has_value()) {
+ if (const auto err = mgp_value_make_string(error_msg.value().data(), memory, &error_value);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a string mgp_value");
+ }
+ }
+
+ mgp_list *search_results{};
+ if (const auto err = mgp_list_make_empty(vertex_ids.size(), memory, &search_results);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a mgp_list");
+ }
+
+ for (const auto &vertex_id : vertex_ids) {
+ mgp_value *vertex;
+ if (const auto err = mgp_value_make_vertex(GetVertexByGid(graph, vertex_id, memory), &vertex);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a vertex mgp_value");
+ }
+ if (const auto err = mgp_list_append(search_results, vertex); err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error(
+ "Retrieving text search results failed during insertion of the mgp_value into the result list");
+ }
+ }
+
+ mgp_value *search_results_value;
+ if (const auto err = mgp_value_make_list(search_results, &search_results_value);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a list mgp_value");
+ }
+
+ if (error_msg.has_value()) {
+ if (const auto err = mgp_map_insert(*result, "error_msg", error_value); err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text index search error failed during insertion into mgp_map");
+ }
+ return;
+ }
+
+ if (const auto err = mgp_map_insert(*result, "search_results", search_results_value);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text index search results failed during insertion into mgp_map");
+ }
+}
+
+void WrapTextIndexAggregation(mgp_memory *memory, mgp_map **result, const std::string &aggregation_result,
+ const std::optional<std::string> &error_msg = std::nullopt) {
+ if (const auto err = mgp_map_make_empty(memory, result); err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a mgp_map");
+ }
+
+ mgp_value *aggregation_result_or_error_value;
+ if (const auto err = mgp_value_make_string(error_msg.value_or(aggregation_result).data(), memory,
+ &aggregation_result_or_error_value);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text search results failed during creation of a string mgp_value");
+ }
+
+ if (error_msg.has_value()) {
+ if (const auto err = mgp_map_insert(*result, "error_msg", aggregation_result_or_error_value);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text index aggregation error failed during insertion into mgp_map");
+ }
+ return;
+ }
+
+ if (const auto err = mgp_map_insert(*result, "aggregation_results", aggregation_result_or_error_value);
+ err != mgp_error::MGP_ERROR_NO_ERROR) {
+ throw std::logic_error("Retrieving text index aggregation results failed during insertion into mgp_map");
+ }
+}
+
+mgp_error mgp_graph_search_text_index(mgp_graph *graph, const char *index_name, const char *search_query,
+ text_search_mode search_mode, mgp_memory *memory, mgp_map **result) {
+ return WrapExceptions([graph, memory, index_name, search_query, search_mode, result]() {
+ std::vector<memgraph::storage::Gid> found_vertices_ids;
+ std::optional<std::string> error_msg = std::nullopt;
+ try {
+ found_vertices_ids = graph->getImpl()->TextIndexSearch(index_name, search_query, search_mode);
+ } catch (memgraph::query::QueryException &e) {
+ error_msg = e.what();
+ }
+ WrapTextSearch(graph, memory, result, found_vertices_ids, error_msg);
+ });
+}
+
+mgp_error mgp_graph_aggregate_over_text_index(mgp_graph *graph, const char *index_name, const char *search_query,
+ const char *aggregation_query, mgp_memory *memory, mgp_map **result) {
+ return WrapExceptions([graph, memory, index_name, search_query, aggregation_query, result]() {
+ std::string search_results;
+ std::optional<std::string> error_msg = std::nullopt;
+ try {
+ search_results = graph->getImpl()->TextIndexAggregate(index_name, search_query, aggregation_query);
+ } catch (memgraph::query::QueryException &e) {
+ error_msg = e.what();
+ }
+ WrapTextIndexAggregation(memory, result, search_results, error_msg);
+ });
+}
+
#ifdef MG_ENTERPRISE
namespace {
void NextPermitted(mgp_vertices_iterator &it) {
diff --git a/src/query/procedure/mg_procedure_impl.hpp b/src/query/procedure/mg_procedure_impl.hpp
index 17cac4eca..a91b4386c 100644
--- a/src/query/procedure/mg_procedure_impl.hpp
+++ b/src/query/procedure/mg_procedure_impl.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -562,6 +562,13 @@ struct mgp_graph {
memgraph::query::ExecutionContext *ctx;
memgraph::storage::StorageMode storage_mode;
+ memgraph::query::DbAccessor *getImpl() const {
+ return std::visit(
+ memgraph::utils::Overloaded{[](memgraph::query::DbAccessor *impl) { return impl; },
+ [](memgraph::query::SubgraphDbAccessor *impl) { return impl->GetAccessor(); }},
+ this->impl);
+ }
+
static mgp_graph WritableGraph(memgraph::query::DbAccessor &acc, memgraph::storage::View view,
memgraph::query::ExecutionContext &ctx) {
return mgp_graph{&acc, view, &ctx, acc.GetStorageMode()};
diff --git a/src/query/trigger.cpp b/src/query/trigger.cpp
index 437389128..151a33dad 100644
--- a/src/query/trigger.cpp
+++ b/src/query/trigger.cpp
@@ -191,9 +191,9 @@ std::shared_ptr<Trigger::TriggerPlan> Trigger::GetPlan(DbAccessor *db_accessor)
return trigger_plan_;
}
-void Trigger::Execute(DbAccessor *dba, utils::MonotonicBufferResource *execution_memory,
- const double max_execution_time_sec, std::atomic<bool> *is_shutting_down,
- std::atomic<TransactionStatus> *transaction_status, const TriggerContext &context) const {
+void Trigger::Execute(DbAccessor *dba, utils::MemoryResource *execution_memory, const double max_execution_time_sec,
+ std::atomic<bool> *is_shutting_down, std::atomic<TransactionStatus> *transaction_status,
+ const TriggerContext &context) const {
if (!context.ShouldEventTrigger(event_type_)) {
return;
}
@@ -214,22 +214,7 @@ void Trigger::Execute(DbAccessor *dba, utils::MonotonicBufferResource *execution
ctx.is_shutting_down = is_shutting_down;
ctx.transaction_status = transaction_status;
ctx.is_profile_query = false;
-
- // Set up temporary memory for a single Pull. Initial memory comes from the
- // stack. 256 KiB should fit on the stack and should be more than enough for a
- // single `Pull`.
- static constexpr size_t stack_size = 256UL * 1024UL;
- char stack_data[stack_size];
-
- // We can throw on every query because a simple queries for deleting will use only
- // the stack allocated buffer.
- // Also, we want to throw only when the query engine requests more memory and not the storage
- // so we add the exception to the allocator.
- utils::ResourceWithOutOfMemoryException resource_with_exception;
- utils::MonotonicBufferResource monotonic_memory(&stack_data[0], stack_size, &resource_with_exception);
- // TODO (mferencevic): Tune the parameters accordingly.
- utils::PoolResource pool_memory(128, 1024, &monotonic_memory);
- ctx.evaluation_context.memory = &pool_memory;
+ ctx.evaluation_context.memory = execution_memory;
auto cursor = plan.plan().MakeCursor(execution_memory);
Frame frame{plan.symbol_table().max_position(), execution_memory};
diff --git a/src/query/trigger.hpp b/src/query/trigger.hpp
index 91c74579e..24bbf50ee 100644
--- a/src/query/trigger.hpp
+++ b/src/query/trigger.hpp
@@ -39,7 +39,7 @@ struct Trigger {
utils::SkipList<QueryCacheEntry> *query_cache, DbAccessor *db_accessor,
const InterpreterConfig::Query &query_config, std::shared_ptr<QueryUserOrRole> owner);
- void Execute(DbAccessor *dba, utils::MonotonicBufferResource *execution_memory, double max_execution_time_sec,
+ void Execute(DbAccessor *dba, utils::MemoryResource *execution_memory, double max_execution_time_sec,
std::atomic<bool> *is_shutting_down, std::atomic<TransactionStatus> *transaction_status,
const TriggerContext &context) const;
diff --git a/src/storage/v2/CMakeLists.txt b/src/storage/v2/CMakeLists.txt
index ec5108d63..49601eb54 100644
--- a/src/storage/v2/CMakeLists.txt
+++ b/src/storage/v2/CMakeLists.txt
@@ -20,6 +20,7 @@ add_library(mg-storage-v2 STATIC
vertex_info_cache.cpp
storage.cpp
indices/indices.cpp
+ indices/text_index.cpp
all_vertices_iterable.cpp
edges_iterable.cpp
vertices_iterable.cpp
@@ -45,4 +46,5 @@ add_library(mg-storage-v2 STATIC
inmemory/replication/recovery.cpp
)
-target_link_libraries(mg-storage-v2 mg::replication Threads::Threads mg-utils gflags absl::flat_hash_map mg-rpc mg-slk mg-events mg-memory)
+target_include_directories(mg-storage-v2 PUBLIC ${CMAKE_SOURCE_DIR}/include)
+target_link_libraries(mg-storage-v2 mg::replication Threads::Threads mg-utils mg-flags gflags absl::flat_hash_map mg-rpc mg-slk mg-events mg-memory mgcxx_text_search tantivy_text_search)
diff --git a/src/storage/v2/config.hpp b/src/storage/v2/config.hpp
index b2a55a40a..419f29b85 100644
--- a/src/storage/v2/config.hpp
+++ b/src/storage/v2/config.hpp
@@ -37,6 +37,7 @@ struct SalientConfig {
struct Items {
bool properties_on_edges{true};
bool enable_schema_metadata{false};
+ bool delta_on_identical_property_update{true};
friend bool operator==(const Items &lrh, const Items &rhs) = default;
} items;
diff --git a/src/storage/v2/disk/durable_metadata.cpp b/src/storage/v2/disk/durable_metadata.cpp
index fe2c558ae..13d515af2 100644
--- a/src/storage/v2/disk/durable_metadata.cpp
+++ b/src/storage/v2/disk/durable_metadata.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -26,6 +26,7 @@ constexpr const char *kVertexCountDescr = "vertex_count";
constexpr const char *kEdgeDountDescr = "edge_count";
constexpr const char *kLabelIndexStr = "label_index";
constexpr const char *kLabelPropertyIndexStr = "label_property_index";
+constexpr const char *kTextIndexStr = "text_index";
constexpr const char *kExistenceConstraintsStr = "existence_constraints";
constexpr const char *kUniqueConstraintsStr = "unique_constraints";
} // namespace
@@ -144,6 +145,31 @@ bool DurableMetadata::PersistLabelPropertyIndexAndExistenceConstraintDeletion(La
return true;
}
+bool DurableMetadata::PersistTextIndexCreation(const std::string &index_name, LabelId label) {
+ const std::string index_name_label_pair = index_name + "," + label.ToString();
+ if (auto text_index_store = durability_kvstore_.Get(kTextIndexStr); text_index_store.has_value()) {
+ std::string &value = text_index_store.value();
+ value += "|";
+ value += index_name_label_pair;
+ return durability_kvstore_.Put(kTextIndexStr, value);
+ }
+ return durability_kvstore_.Put(kTextIndexStr, index_name_label_pair);
+}
+
+bool DurableMetadata::PersistTextIndexDeletion(const std::string &index_name, LabelId label) {
+ const std::string index_name_label_pair = index_name + "," + label.ToString();
+ if (auto text_index_store = durability_kvstore_.Get(kTextIndexStr); text_index_store.has_value()) {
+ const std::string &value = text_index_store.value();
+ std::vector<std::string> text_indices = utils::Split(value, "|");
+ std::erase(text_indices, index_name_label_pair);
+ if (text_indices.empty()) {
+ return durability_kvstore_.Delete(kTextIndexStr);
+ }
+ return durability_kvstore_.Put(kTextIndexStr, utils::Join(text_indices, "|"));
+ }
+ return true;
+}
+
bool DurableMetadata::PersistUniqueConstraintCreation(LabelId label, const std::set<PropertyId> &properties) {
const std::string entry = utils::GetKeyForUniqueConstraintsDurability(label, properties);
diff --git a/src/storage/v2/disk/durable_metadata.hpp b/src/storage/v2/disk/durable_metadata.hpp
index 168cce469..4aaa8a707 100644
--- a/src/storage/v2/disk/durable_metadata.hpp
+++ b/src/storage/v2/disk/durable_metadata.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -53,6 +53,10 @@ class DurableMetadata {
bool PersistLabelPropertyIndexAndExistenceConstraintDeletion(LabelId label, PropertyId property,
const std::string &key);
+ bool PersistTextIndexCreation(const std::string &index_name, LabelId label);
+
+ bool PersistTextIndexDeletion(const std::string &index_name, LabelId label);
+
bool PersistUniqueConstraintCreation(LabelId label, const std::set<PropertyId> &properties);
bool PersistUniqueConstraintDeletion(LabelId label, const std::set<PropertyId> &properties);
diff --git a/src/storage/v2/disk/storage.cpp b/src/storage/v2/disk/storage.cpp
index 21fa5ecc7..4dbd248f7 100644
--- a/src/storage/v2/disk/storage.cpp
+++ b/src/storage/v2/disk/storage.cpp
@@ -29,6 +29,8 @@
#include <rocksdb/utilities/transaction.h>
#include <rocksdb/utilities/transaction_db.h>
+#include "flags/experimental.hpp"
+#include "flags/run_time_configurable.hpp"
#include "kvstore/kvstore.hpp"
#include "spdlog/spdlog.h"
#include "storage/v2/constraints/unique_constraints.hpp"
@@ -856,6 +858,7 @@ StorageInfo DiskStorage::GetInfo(memgraph::replication_coordination_glue::Replic
const auto &lbl = access->ListAllIndices();
info.label_indices = lbl.label.size();
info.label_property_indices = lbl.label_property.size();
+ info.text_indices = lbl.text_indices.size();
const auto &con = access->ListAllConstraints();
info.existence_constraints = con.existence.size();
info.unique_constraints = con.unique.size();
@@ -1670,6 +1673,18 @@ utils::BasicResult<StorageManipulationError, void> DiskStorage::DiskAccessor::Co
case MetadataDelta::Action::LABEL_PROPERTY_INDEX_STATS_CLEAR: {
throw utils::NotYetImplemented("ClearIndexStats(stats) is not implemented for DiskStorage.");
} break;
+ case MetadataDelta::Action::TEXT_INDEX_CREATE: {
+ const auto &info = md_delta.text_index;
+ if (!disk_storage->durable_metadata_.PersistTextIndexCreation(info.index_name, info.label)) {
+ return StorageManipulationError{PersistenceError{}};
+ }
+ } break;
+ case MetadataDelta::Action::TEXT_INDEX_DROP: {
+ const auto &info = md_delta.text_index;
+ if (!disk_storage->durable_metadata_.PersistTextIndexDeletion(info.index_name, info.label)) {
+ return StorageManipulationError{PersistenceError{}};
+ }
+ } break;
case MetadataDelta::Action::EXISTENCE_CONSTRAINT_CREATE: {
const auto &info = md_delta.label_property;
if (!disk_storage->durable_metadata_.PersistLabelPropertyIndexAndExistenceConstraintCreation(
@@ -1768,6 +1783,9 @@ utils::BasicResult<StorageManipulationError, void> DiskStorage::DiskAccessor::Co
return StorageManipulationError{SerializationError{}};
}
spdlog::trace("rocksdb: Commit successful");
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ disk_storage->indices_.text_index_.Commit();
+ }
is_transaction_active_ = false;
@@ -1886,6 +1904,9 @@ void DiskStorage::DiskAccessor::Abort() {
// query_plan_accumulate_aggregate.cpp
transaction_.disk_transaction_->Rollback();
transaction_.disk_transaction_->ClearSnapshot();
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ storage_->indices_.text_index_.Rollback();
+ }
delete transaction_.disk_transaction_;
transaction_.disk_transaction_ = nullptr;
is_transaction_active_ = false;
@@ -2092,7 +2113,11 @@ IndicesInfo DiskStorage::DiskAccessor::ListAllIndices() const {
auto *disk_label_index = static_cast<DiskLabelIndex *>(on_disk->indices_.label_index_.get());
auto *disk_label_property_index =
static_cast<DiskLabelPropertyIndex *>(on_disk->indices_.label_property_index_.get());
- return {disk_label_index->ListIndices(), disk_label_property_index->ListIndices()};
+ auto &text_index = storage_->indices_.text_index_;
+ return {disk_label_index->ListIndices(),
+ disk_label_property_index->ListIndices(),
+ {/* edge type indices */},
+ text_index.ListIndices()};
}
ConstraintsInfo DiskStorage::DiskAccessor::ListAllConstraints() const {
auto *disk_storage = static_cast<DiskStorage *>(storage_);
diff --git a/src/storage/v2/durability/durability.cpp b/src/storage/v2/durability/durability.cpp
index fbbedbee5..db8bcd93b 100644
--- a/src/storage/v2/durability/durability.cpp
+++ b/src/storage/v2/durability/durability.cpp
@@ -151,7 +151,8 @@ void RecoverConstraints(const RecoveredIndicesAndConstraints::ConstraintsMetadat
void RecoverIndicesAndStats(const RecoveredIndicesAndConstraints::IndicesMetadata &indices_metadata, Indices *indices,
utils::SkipList<Vertex> *vertices, NameIdMapper *name_id_mapper,
- const std::optional<ParallelizedSchemaCreationInfo> ¶llel_exec_info) {
+ const std::optional<ParallelizedSchemaCreationInfo> ¶llel_exec_info,
+ const std::optional<std::filesystem::path> &storage_dir) {
spdlog::info("Recreating indices from metadata.");
// Recover label indices.
@@ -211,6 +212,26 @@ void RecoverIndicesAndStats(const RecoveredIndicesAndConstraints::IndicesMetadat
}
spdlog::info("Edge-type indices are recreated.");
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ // Recover text indices.
+ spdlog::info("Recreating {} text indices from metadata.", indices_metadata.text_indices.size());
+ auto &mem_text_index = indices->text_index_;
+ for (const auto &[index_name, label] : indices_metadata.text_indices) {
+ try {
+ if (!storage_dir.has_value()) {
+ throw RecoveryFailure("There must exist a storage directory in order to recover text indices!");
+ }
+
+ mem_text_index.RecoverIndex(storage_dir.value(), index_name, label, vertices->access(), name_id_mapper);
+ } catch (...) {
+ throw RecoveryFailure("The text index must be created here!");
+ }
+ spdlog::info("Text index {} on :{} is recreated from metadata", index_name,
+ name_id_mapper->IdToName(label.AsUint()));
+ }
+ spdlog::info("Text indices are recreated.");
+ }
+
spdlog::info("Indices are recreated.");
}
@@ -331,8 +352,13 @@ std::optional<RecoveryInfo> Recovery::RecoverData(std::string *uuid, Replication
repl_storage_state.epoch_.SetEpoch(std::move(recovered_snapshot->snapshot_info.epoch_id));
if (!utils::DirExists(wal_directory_)) {
+ std::optional<std::filesystem::path> storage_dir = std::nullopt;
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ storage_dir = config.durability.storage_directory;
+ }
+
RecoverIndicesAndStats(indices_constraints.indices, indices, vertices, name_id_mapper,
- GetParallelExecInfoIndices(recovery_info, config));
+ GetParallelExecInfoIndices(recovery_info, config), storage_dir);
RecoverConstraints(indices_constraints.constraints, constraints, vertices, name_id_mapper,
GetParallelExecInfo(recovery_info, config));
return recovered_snapshot->recovery_info;
@@ -467,8 +493,13 @@ std::optional<RecoveryInfo> Recovery::RecoverData(std::string *uuid, Replication
spdlog::info("All necessary WAL files are loaded successfully.");
}
+ std::optional<std::filesystem::path> storage_dir = std::nullopt;
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ storage_dir = config.durability.storage_directory;
+ }
+
RecoverIndicesAndStats(indices_constraints.indices, indices, vertices, name_id_mapper,
- GetParallelExecInfoIndices(recovery_info, config));
+ GetParallelExecInfoIndices(recovery_info, config), storage_dir);
RecoverConstraints(indices_constraints.constraints, constraints, vertices, name_id_mapper,
GetParallelExecInfo(recovery_info, config));
diff --git a/src/storage/v2/durability/durability.hpp b/src/storage/v2/durability/durability.hpp
index 97e2c7efc..5170b3b04 100644
--- a/src/storage/v2/durability/durability.hpp
+++ b/src/storage/v2/durability/durability.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -102,7 +102,8 @@ std::optional<std::vector<WalDurabilityInfo>> GetWalFiles(const std::filesystem:
/// @throw RecoveryFailure
void RecoverIndicesAndStats(const RecoveredIndicesAndConstraints::IndicesMetadata &indices_metadata, Indices *indices,
utils::SkipList<Vertex> *vertices, NameIdMapper *name_id_mapper,
- const std::optional<ParallelizedSchemaCreationInfo> ¶llel_exec_info = std::nullopt);
+ const std::optional<ParallelizedSchemaCreationInfo> ¶llel_exec_info = std::nullopt,
+ const std::optional<std::filesystem::path> &storage_dir = std::nullopt);
// Helper function used to recover all discovered constraints. The
// constraints must be recovered after the data recovery is done
diff --git a/src/storage/v2/durability/marker.hpp b/src/storage/v2/durability/marker.hpp
index ac0cc074d..18d693e51 100644
--- a/src/storage/v2/durability/marker.hpp
+++ b/src/storage/v2/durability/marker.hpp
@@ -64,6 +64,8 @@ enum class Marker : uint8_t {
DELTA_LABEL_PROPERTY_INDEX_STATS_CLEAR = 0x64,
DELTA_EDGE_TYPE_INDEX_CREATE = 0x65,
DELTA_EDGE_TYPE_INDEX_DROP = 0x66,
+ DELTA_TEXT_INDEX_CREATE = 0x67,
+ DELTA_TEXT_INDEX_DROP = 0x68,
VALUE_FALSE = 0x00,
VALUE_TRUE = 0xff,
@@ -110,6 +112,8 @@ static const Marker kMarkersAll[] = {
Marker::DELTA_LABEL_PROPERTY_INDEX_DROP,
Marker::DELTA_EDGE_TYPE_INDEX_CREATE,
Marker::DELTA_EDGE_TYPE_INDEX_DROP,
+ Marker::DELTA_TEXT_INDEX_CREATE,
+ Marker::DELTA_TEXT_INDEX_DROP,
Marker::DELTA_EXISTENCE_CONSTRAINT_CREATE,
Marker::DELTA_EXISTENCE_CONSTRAINT_DROP,
Marker::DELTA_UNIQUE_CONSTRAINT_CREATE,
diff --git a/src/storage/v2/durability/metadata.hpp b/src/storage/v2/durability/metadata.hpp
index c8ee27b2f..f36fc068d 100644
--- a/src/storage/v2/durability/metadata.hpp
+++ b/src/storage/v2/durability/metadata.hpp
@@ -44,6 +44,7 @@ struct RecoveredIndicesAndConstraints {
std::vector<std::pair<LabelId, LabelIndexStats>> label_stats;
std::vector<std::pair<LabelId, std::pair<PropertyId, LabelPropertyIndexStats>>> label_property_stats;
std::vector<EdgeTypeId> edge;
+ std::vector<std::pair<std::string, LabelId>> text_indices;
} indices;
struct ConstraintsMetadata {
diff --git a/src/storage/v2/durability/serialization.cpp b/src/storage/v2/durability/serialization.cpp
index 28ba64943..becfa7f34 100644
--- a/src/storage/v2/durability/serialization.cpp
+++ b/src/storage/v2/durability/serialization.cpp
@@ -353,6 +353,8 @@ std::optional<PropertyValue> Decoder::ReadPropertyValue() {
case Marker::DELTA_LABEL_PROPERTY_INDEX_DROP:
case Marker::DELTA_EDGE_TYPE_INDEX_CREATE:
case Marker::DELTA_EDGE_TYPE_INDEX_DROP:
+ case Marker::DELTA_TEXT_INDEX_CREATE:
+ case Marker::DELTA_TEXT_INDEX_DROP:
case Marker::DELTA_EXISTENCE_CONSTRAINT_CREATE:
case Marker::DELTA_EXISTENCE_CONSTRAINT_DROP:
case Marker::DELTA_UNIQUE_CONSTRAINT_CREATE:
@@ -459,6 +461,8 @@ bool Decoder::SkipPropertyValue() {
case Marker::DELTA_LABEL_PROPERTY_INDEX_DROP:
case Marker::DELTA_EDGE_TYPE_INDEX_CREATE:
case Marker::DELTA_EDGE_TYPE_INDEX_DROP:
+ case Marker::DELTA_TEXT_INDEX_CREATE:
+ case Marker::DELTA_TEXT_INDEX_DROP:
case Marker::DELTA_EXISTENCE_CONSTRAINT_CREATE:
case Marker::DELTA_EXISTENCE_CONSTRAINT_DROP:
case Marker::DELTA_UNIQUE_CONSTRAINT_CREATE:
diff --git a/src/storage/v2/durability/snapshot.cpp b/src/storage/v2/durability/snapshot.cpp
index 5fea3dfa5..0e3bb96e3 100644
--- a/src/storage/v2/durability/snapshot.cpp
+++ b/src/storage/v2/durability/snapshot.cpp
@@ -13,6 +13,8 @@
#include <thread>
+#include "flags/experimental.hpp"
+#include "flags/run_time_configurable.hpp"
#include "spdlog/spdlog.h"
#include "storage/v2/durability/exceptions.hpp"
#include "storage/v2/durability/paths.hpp"
@@ -2004,6 +2006,24 @@ RecoveredSnapshot LoadSnapshot(const std::filesystem::path &path, utils::SkipLis
spdlog::info("Metadata of edge-type indices are recovered.");
}
+ // Recover text indices.
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ auto size = snapshot.ReadUint();
+ if (!size) throw RecoveryFailure("Couldn't recover the number of text indices!");
+ spdlog::info("Recovering metadata of {} text indices.", *size);
+ for (uint64_t i = 0; i < *size; ++i) {
+ auto index_name = snapshot.ReadString();
+ if (!index_name.has_value()) throw RecoveryFailure("Couldn't read text index name!");
+ auto label = snapshot.ReadUint();
+ if (!label) throw RecoveryFailure("Couldn't read text index label!");
+ AddRecoveredIndexConstraint(&indices_constraints.indices.text_indices,
+ {index_name.value(), get_label_from_id(*label)}, "The text index already exists!");
+ SPDLOG_TRACE("Recovered metadata of text index {} for :{}", index_name.value(),
+ name_id_mapper->IdToName(snapshot_id_map.at(*label)));
+ }
+ spdlog::info("Metadata of text indices are recovered.");
+ }
+
spdlog::info("Metadata of indices are recovered.");
}
@@ -2493,6 +2513,16 @@ void CreateSnapshot(Storage *storage, Transaction *transaction, const std::files
write_mapping(item);
}
}
+
+ // Write text indices.
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ auto text_indices = storage->indices_.text_index_.ListIndices();
+ snapshot.WriteUint(text_indices.size());
+ for (const auto &[index_name, label] : text_indices) {
+ snapshot.WriteString(index_name);
+ write_mapping(label);
+ }
+ }
}
// Write constraints.
diff --git a/src/storage/v2/durability/storage_global_operation.hpp b/src/storage/v2/durability/storage_global_operation.hpp
index 7dd635e9d..d9c77b6c6 100644
--- a/src/storage/v2/durability/storage_global_operation.hpp
+++ b/src/storage/v2/durability/storage_global_operation.hpp
@@ -25,6 +25,8 @@ enum class StorageMetadataOperation {
LABEL_PROPERTY_INDEX_STATS_CLEAR,
EDGE_TYPE_INDEX_CREATE,
EDGE_TYPE_INDEX_DROP,
+ TEXT_INDEX_CREATE,
+ TEXT_INDEX_DROP,
EXISTENCE_CONSTRAINT_CREATE,
EXISTENCE_CONSTRAINT_DROP,
UNIQUE_CONSTRAINT_CREATE,
diff --git a/src/storage/v2/durability/wal.cpp b/src/storage/v2/durability/wal.cpp
index 5c40ab1c5..c684d818c 100644
--- a/src/storage/v2/durability/wal.cpp
+++ b/src/storage/v2/durability/wal.cpp
@@ -99,6 +99,10 @@ Marker OperationToMarker(StorageMetadataOperation operation) {
return Marker::DELTA_EDGE_TYPE_INDEX_CREATE;
case StorageMetadataOperation::EDGE_TYPE_INDEX_DROP:
return Marker::DELTA_EDGE_TYPE_INDEX_DROP;
+ case StorageMetadataOperation::TEXT_INDEX_CREATE:
+ return Marker::DELTA_TEXT_INDEX_CREATE;
+ case StorageMetadataOperation::TEXT_INDEX_DROP:
+ return Marker::DELTA_TEXT_INDEX_DROP;
case StorageMetadataOperation::EXISTENCE_CONSTRAINT_CREATE:
return Marker::DELTA_EXISTENCE_CONSTRAINT_CREATE;
case StorageMetadataOperation::EXISTENCE_CONSTRAINT_DROP:
@@ -172,6 +176,10 @@ WalDeltaData::Type MarkerToWalDeltaDataType(Marker marker) {
return WalDeltaData::Type::LABEL_PROPERTY_INDEX_CREATE;
case Marker::DELTA_LABEL_PROPERTY_INDEX_DROP:
return WalDeltaData::Type::LABEL_PROPERTY_INDEX_DROP;
+ case Marker::DELTA_TEXT_INDEX_CREATE:
+ return WalDeltaData::Type::TEXT_INDEX_CREATE;
+ case Marker::DELTA_TEXT_INDEX_DROP:
+ return WalDeltaData::Type::TEXT_INDEX_DROP;
case Marker::DELTA_LABEL_PROPERTY_INDEX_STATS_SET:
return WalDeltaData::Type::LABEL_PROPERTY_INDEX_STATS_SET;
case Marker::DELTA_LABEL_PROPERTY_INDEX_STATS_CLEAR:
@@ -382,6 +390,21 @@ WalDeltaData ReadSkipWalDeltaData(BaseDecoder *decoder) {
if (!decoder->SkipString()) throw RecoveryFailure("Invalid WAL data!");
}
}
+ break;
+ }
+ case WalDeltaData::Type::TEXT_INDEX_CREATE:
+ case WalDeltaData::Type::TEXT_INDEX_DROP: {
+ if constexpr (read_data) {
+ auto index_name = decoder->ReadString();
+ if (!index_name) throw RecoveryFailure("Invalid WAL data!");
+ delta.operation_text.index_name = std::move(*index_name);
+ auto label = decoder->ReadString();
+ if (!label) throw RecoveryFailure("Invalid WAL data!");
+ delta.operation_text.label = std::move(*label);
+ } else {
+ if (!decoder->SkipString() || !decoder->SkipString()) throw RecoveryFailure("Invalid WAL data!");
+ }
+ break;
}
}
@@ -529,6 +552,12 @@ bool operator==(const WalDeltaData &a, const WalDeltaData &b) {
case WalDeltaData::Type::LABEL_PROPERTY_INDEX_CREATE:
case WalDeltaData::Type::LABEL_PROPERTY_INDEX_DROP:
+ case WalDeltaData::Type::TEXT_INDEX_CREATE:
+ return a.operation_text.index_name == b.operation_text.index_name &&
+ a.operation_text.label == b.operation_text.label;
+ case WalDeltaData::Type::TEXT_INDEX_DROP:
+ return a.operation_text.index_name == b.operation_text.index_name &&
+ a.operation_text.label == b.operation_text.label;
case WalDeltaData::Type::EXISTENCE_CONSTRAINT_CREATE:
case WalDeltaData::Type::EXISTENCE_CONSTRAINT_DROP:
return a.operation_label_property.label == b.operation_label_property.label &&
@@ -675,7 +704,8 @@ void EncodeTransactionEnd(BaseEncoder *encoder, uint64_t timestamp) {
}
void EncodeOperation(BaseEncoder *encoder, NameIdMapper *name_id_mapper, StorageMetadataOperation operation,
- LabelId label, const std::set<PropertyId> &properties, const LabelIndexStats &stats,
+ const std::optional<std::string> text_index_name, LabelId label,
+ const std::set<PropertyId> &properties, const LabelIndexStats &stats,
const LabelPropertyIndexStats &property_stats, uint64_t timestamp) {
encoder->WriteMarker(Marker::SECTION_DELTA);
encoder->WriteUint(timestamp);
@@ -731,6 +761,14 @@ void EncodeOperation(BaseEncoder *encoder, NameIdMapper *name_id_mapper, Storage
case StorageMetadataOperation::EDGE_TYPE_INDEX_DROP: {
MG_ASSERT(false, "Invalid function call!");
}
+ case StorageMetadataOperation::TEXT_INDEX_CREATE:
+ case StorageMetadataOperation::TEXT_INDEX_DROP: {
+ MG_ASSERT(text_index_name.has_value(), "Text indices must be named!");
+ encoder->WriteMarker(OperationToMarker(operation));
+ encoder->WriteString(text_index_name.value());
+ encoder->WriteString(name_id_mapper->IdToName(label.AsUint()));
+ break;
+ }
}
}
@@ -752,6 +790,8 @@ void EncodeOperation(BaseEncoder *encoder, NameIdMapper *name_id_mapper, Storage
case StorageMetadataOperation::LABEL_INDEX_STATS_SET:
case StorageMetadataOperation::LABEL_PROPERTY_INDEX_CREATE:
case StorageMetadataOperation::LABEL_PROPERTY_INDEX_DROP:
+ case StorageMetadataOperation::TEXT_INDEX_CREATE:
+ case StorageMetadataOperation::TEXT_INDEX_DROP:
case StorageMetadataOperation::EXISTENCE_CONSTRAINT_CREATE:
case StorageMetadataOperation::EXISTENCE_CONSTRAINT_DROP:
case StorageMetadataOperation::LABEL_PROPERTY_INDEX_STATS_SET:
@@ -1000,6 +1040,20 @@ RecoveryInfo LoadWal(const std::filesystem::path &path, RecoveredIndicesAndConst
"The label index stats doesn't exist!");
break;
}
+ case WalDeltaData::Type::TEXT_INDEX_CREATE: {
+ auto index_name = delta.operation_text.index_name;
+ auto label = LabelId::FromUint(name_id_mapper->NameToId(delta.operation_text.label));
+ AddRecoveredIndexConstraint(&indices_constraints->indices.text_indices, {index_name, label},
+ "The text index already exists!");
+ break;
+ }
+ case WalDeltaData::Type::TEXT_INDEX_DROP: {
+ auto index_name = delta.operation_text.index_name;
+ auto label = LabelId::FromUint(name_id_mapper->NameToId(delta.operation_text.label));
+ RemoveRecoveredIndexConstraint(&indices_constraints->indices.text_indices, {index_name, label},
+ "The text index doesn't exist!");
+ break;
+ }
case WalDeltaData::Type::EXISTENCE_CONSTRAINT_CREATE: {
auto label_id = LabelId::FromUint(name_id_mapper->NameToId(delta.operation_label_property.label));
auto property_id = PropertyId::FromUint(name_id_mapper->NameToId(delta.operation_label_property.property));
@@ -1148,10 +1202,11 @@ void WalFile::AppendTransactionEnd(uint64_t timestamp) {
UpdateStats(timestamp);
}
-void WalFile::AppendOperation(StorageMetadataOperation operation, LabelId label, const std::set<PropertyId> &properties,
- const LabelIndexStats &stats, const LabelPropertyIndexStats &property_stats,
- uint64_t timestamp) {
- EncodeOperation(&wal_, name_id_mapper_, operation, label, properties, stats, property_stats, timestamp);
+void WalFile::AppendOperation(StorageMetadataOperation operation, const std::optional<std::string> text_index_name,
+ LabelId label, const std::set<PropertyId> &properties, const LabelIndexStats &stats,
+ const LabelPropertyIndexStats &property_stats, uint64_t timestamp) {
+ EncodeOperation(&wal_, name_id_mapper_, operation, text_index_name, label, properties, stats, property_stats,
+ timestamp);
UpdateStats(timestamp);
}
diff --git a/src/storage/v2/durability/wal.hpp b/src/storage/v2/durability/wal.hpp
index 516487e0d..4990e6979 100644
--- a/src/storage/v2/durability/wal.hpp
+++ b/src/storage/v2/durability/wal.hpp
@@ -69,6 +69,8 @@ struct WalDeltaData {
LABEL_PROPERTY_INDEX_STATS_CLEAR,
EDGE_INDEX_CREATE,
EDGE_INDEX_DROP,
+ TEXT_INDEX_CREATE,
+ TEXT_INDEX_DROP,
EXISTENCE_CONSTRAINT_CREATE,
EXISTENCE_CONSTRAINT_DROP,
UNIQUE_CONSTRAINT_CREATE,
@@ -127,6 +129,11 @@ struct WalDeltaData {
std::string property;
std::string stats;
} operation_label_property_stats;
+
+ struct {
+ std::string index_name;
+ std::string label;
+ } operation_text;
};
bool operator==(const WalDeltaData &a, const WalDeltaData &b);
@@ -163,6 +170,8 @@ constexpr bool IsWalDeltaDataTypeTransactionEndVersion15(const WalDeltaData::Typ
case WalDeltaData::Type::LABEL_PROPERTY_INDEX_STATS_CLEAR:
case WalDeltaData::Type::EDGE_INDEX_CREATE:
case WalDeltaData::Type::EDGE_INDEX_DROP:
+ case WalDeltaData::Type::TEXT_INDEX_CREATE:
+ case WalDeltaData::Type::TEXT_INDEX_DROP:
case WalDeltaData::Type::EXISTENCE_CONSTRAINT_CREATE:
case WalDeltaData::Type::EXISTENCE_CONSTRAINT_DROP:
case WalDeltaData::Type::UNIQUE_CONSTRAINT_CREATE:
@@ -213,7 +222,8 @@ void EncodeTransactionEnd(BaseEncoder *encoder, uint64_t timestamp);
/// Function used to encode non-transactional operation.
void EncodeOperation(BaseEncoder *encoder, NameIdMapper *name_id_mapper, StorageMetadataOperation operation,
- LabelId label, const std::set<PropertyId> &properties, const LabelIndexStats &stats,
+ const std::optional<std::string> text_index_name, LabelId label,
+ const std::set<PropertyId> &properties, const LabelIndexStats &stats,
const LabelPropertyIndexStats &property_stats, uint64_t timestamp);
void EncodeOperation(BaseEncoder *encoder, NameIdMapper *name_id_mapper, StorageMetadataOperation operation,
@@ -248,8 +258,9 @@ class WalFile {
void AppendTransactionEnd(uint64_t timestamp);
- void AppendOperation(StorageMetadataOperation operation, LabelId label, const std::set<PropertyId> &properties,
- const LabelIndexStats &stats, const LabelPropertyIndexStats &property_stats, uint64_t timestamp);
+ void AppendOperation(StorageMetadataOperation operation, const std::optional<std::string> text_index_name,
+ LabelId label, const std::set<PropertyId> &properties, const LabelIndexStats &stats,
+ const LabelPropertyIndexStats &property_stats, uint64_t timestamp);
void AppendOperation(StorageMetadataOperation operation, EdgeTypeId edge_type, uint64_t timestamp);
diff --git a/src/storage/v2/edge_accessor.cpp b/src/storage/v2/edge_accessor.cpp
index 62a9f4bcd..ba354371e 100644
--- a/src/storage/v2/edge_accessor.cpp
+++ b/src/storage/v2/edge_accessor.cpp
@@ -130,9 +130,13 @@ Result<storage::PropertyValue> EdgeAccessor::SetProperty(PropertyId property, co
if (edge_.ptr->deleted) return Error::DELETED_OBJECT;
using ReturnType = decltype(edge_.ptr->properties.GetProperty(property));
std::optional<ReturnType> current_value;
+ const bool skip_duplicate_write = !storage_->config_.salient.items.delta_on_identical_property_update;
utils::AtomicMemoryBlock atomic_memory_block{
- [¤t_value, &property, &value, transaction = transaction_, edge = edge_]() {
+ [¤t_value, &property, &value, transaction = transaction_, edge = edge_, skip_duplicate_write]() {
current_value.emplace(edge.ptr->properties.GetProperty(property));
+ if (skip_duplicate_write && current_value == value) {
+ return;
+ }
// We could skip setting the value if the previous one is the same to the new
// one. This would save some memory as a delta would not be created as well as
// avoid copying the value. The reason we are not doing that is because the
@@ -184,12 +188,14 @@ Result<std::vector<std::tuple<PropertyId, PropertyValue, PropertyValue>>> EdgeAc
if (edge_.ptr->deleted) return Error::DELETED_OBJECT;
+ const bool skip_duplicate_write = !storage_->config_.salient.items.delta_on_identical_property_update;
using ReturnType = decltype(edge_.ptr->properties.UpdateProperties(properties));
std::optional<ReturnType> id_old_new_change;
utils::AtomicMemoryBlock atomic_memory_block{
- [transaction_ = transaction_, edge_ = edge_, &properties, &id_old_new_change]() {
+ [transaction_ = transaction_, edge_ = edge_, &properties, &id_old_new_change, skip_duplicate_write]() {
id_old_new_change.emplace(edge_.ptr->properties.UpdateProperties(properties));
for (auto &[property, old_value, new_value] : *id_old_new_change) {
+ if (skip_duplicate_write && old_value == new_value) continue;
CreateAndLinkDelta(transaction_, edge_.ptr, Delta::SetPropertyTag(), property, std::move(old_value));
}
}};
diff --git a/src/storage/v2/indices/indices.cpp b/src/storage/v2/indices/indices.cpp
index 6068f888f..1cbde2362 100644
--- a/src/storage/v2/indices/indices.cpp
+++ b/src/storage/v2/indices/indices.cpp
@@ -16,6 +16,7 @@
#include "storage/v2/inmemory/edge_type_index.hpp"
#include "storage/v2/inmemory/label_index.hpp"
#include "storage/v2/inmemory/label_property_index.hpp"
+#include "storage/v2/storage.hpp"
namespace memgraph::storage {
diff --git a/src/storage/v2/indices/indices.hpp b/src/storage/v2/indices/indices.hpp
index 40cff577f..6f1bc44db 100644
--- a/src/storage/v2/indices/indices.hpp
+++ b/src/storage/v2/indices/indices.hpp
@@ -18,6 +18,7 @@
#include "storage/v2/indices/edge_type_index.hpp"
#include "storage/v2/indices/label_index.hpp"
#include "storage/v2/indices/label_property_index.hpp"
+#include "storage/v2/indices/text_index.hpp"
#include "storage/v2/storage_mode.hpp"
namespace memgraph::storage {
@@ -31,12 +32,12 @@ struct Indices {
Indices &operator=(Indices &&) = delete;
~Indices() = default;
- /// This function should be called from garbage collection to clean-up the
+ /// This function should be called from garbage collection to clean up the
/// index.
/// TODO: unused in disk indices
void RemoveObsoleteEntries(uint64_t oldest_active_start_timestamp, std::stop_token token) const;
- /// Surgical removal of entries that was inserted this transaction
+ /// Surgical removal of entries that were inserted in this transaction
/// TODO: unused in disk indices
void AbortEntries(LabelId labelId, std::span<Vertex *const> vertices, uint64_t exact_start_timestamp) const;
void AbortEntries(PropertyId property, std::span<std::pair<PropertyValue, Vertex *> const> vertices,
@@ -71,6 +72,7 @@ struct Indices {
std::unique_ptr<LabelIndex> label_index_;
std::unique_ptr<LabelPropertyIndex> label_property_index_;
std::unique_ptr<EdgeTypeIndex> edge_type_index_;
+ mutable TextIndex text_index_;
};
} // namespace memgraph::storage
diff --git a/src/storage/v2/indices/text_index.cpp b/src/storage/v2/indices/text_index.cpp
new file mode 100644
index 000000000..1c9488097
--- /dev/null
+++ b/src/storage/v2/indices/text_index.cpp
@@ -0,0 +1,430 @@
+// Copyright 2024 Memgraph Ltd.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+// License, and you may not use this file except in compliance with the Business Source License.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+#include "storage/v2/indices/text_index.hpp"
+#include "flags/experimental.hpp"
+#include "flags/run_time_configurable.hpp"
+#include "query/db_accessor.hpp"
+#include "storage/v2/view.hpp"
+#include "text_search.hpp"
+
+namespace memgraph::storage {
+
+std::string GetPropertyName(PropertyId prop_id, memgraph::query::DbAccessor *db) { return db->PropertyToName(prop_id); }
+
+std::string GetPropertyName(PropertyId prop_id, NameIdMapper *name_id_mapper) {
+ return name_id_mapper->IdToName(prop_id.AsUint());
+}
+
+inline std::string TextIndex::MakeIndexPath(const std::filesystem::path &storage_dir, const std::string &index_name) {
+ return (storage_dir / kTextIndicesDirectory / index_name).string();
+}
+
+void TextIndex::CreateEmptyIndex(const std::filesystem::path &storage_dir, const std::string &index_name,
+ LabelId label) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ if (index_.contains(index_name)) {
+ throw query::TextSearchException("Text index \"{}\" already exists.", index_name);
+ }
+
+ try {
+ nlohmann::json mappings = {};
+ mappings["properties"] = {};
+ mappings["properties"]["metadata"] = {{"type", "json"}, {"fast", true}, {"stored", true}, {"text", true}};
+ mappings["properties"]["data"] = {{"type", "json"}, {"fast", true}, {"stored", true}, {"text", true}};
+ mappings["properties"]["all"] = {{"type", "text"}, {"fast", true}, {"stored", true}, {"text", true}};
+
+ index_.emplace(index_name, TextIndexData{.context_ = mgcxx::text_search::create_index(
+ MakeIndexPath(storage_dir, index_name),
+ mgcxx::text_search::IndexConfig{.mappings = mappings.dump()}),
+ .scope_ = label});
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+ label_to_index_.emplace(label, index_name);
+}
+
+template <typename T>
+nlohmann::json TextIndex::SerializeProperties(const std::map<PropertyId, PropertyValue> &properties, T *name_resolver) {
+ nlohmann::json serialized_properties = nlohmann::json::value_t::object;
+ for (const auto &[prop_id, prop_value] : properties) {
+ switch (prop_value.type()) {
+ case PropertyValue::Type::Bool:
+ serialized_properties[GetPropertyName(prop_id, name_resolver)] = prop_value.ValueBool();
+ break;
+ case PropertyValue::Type::Int:
+ serialized_properties[GetPropertyName(prop_id, name_resolver)] = prop_value.ValueInt();
+ break;
+ case PropertyValue::Type::Double:
+ serialized_properties[GetPropertyName(prop_id, name_resolver)] = prop_value.ValueDouble();
+ break;
+ case PropertyValue::Type::String:
+ serialized_properties[GetPropertyName(prop_id, name_resolver)] = prop_value.ValueString();
+ break;
+ case PropertyValue::Type::Null:
+ case PropertyValue::Type::List:
+ case PropertyValue::Type::Map:
+ case PropertyValue::Type::TemporalData:
+ default:
+ continue;
+ }
+ }
+
+ return serialized_properties;
+}
+
+std::string TextIndex::StringifyProperties(const std::map<PropertyId, PropertyValue> &properties) {
+ std::vector<std::string> indexable_properties_as_string;
+ for (const auto &[_, prop_value] : properties) {
+ switch (prop_value.type()) {
+ case PropertyValue::Type::Bool:
+ indexable_properties_as_string.push_back(prop_value.ValueBool() ? "true" : "false");
+ break;
+ case PropertyValue::Type::Int:
+ indexable_properties_as_string.push_back(std::to_string(prop_value.ValueInt()));
+ break;
+ case PropertyValue::Type::Double:
+ indexable_properties_as_string.push_back(std::to_string(prop_value.ValueDouble()));
+ break;
+ case PropertyValue::Type::String:
+ indexable_properties_as_string.push_back(prop_value.ValueString());
+ break;
+ // NOTE: As the following types aren‘t indexed in Tantivy, they don’t appear in the property value string either.
+ case PropertyValue::Type::Null:
+ case PropertyValue::Type::List:
+ case PropertyValue::Type::Map:
+ case PropertyValue::Type::TemporalData:
+ default:
+ continue;
+ }
+ }
+ return utils::Join(indexable_properties_as_string, " ");
+}
+
+std::vector<mgcxx::text_search::Context *> TextIndex::GetApplicableTextIndices(const std::vector<LabelId> &labels) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ std::vector<mgcxx::text_search::Context *> applicable_text_indices;
+ for (const auto &label : labels) {
+ if (label_to_index_.contains(label)) {
+ applicable_text_indices.push_back(&index_.at(label_to_index_.at(label)).context_);
+ }
+ }
+ return applicable_text_indices;
+}
+
+void TextIndex::LoadNodeToTextIndices(const std::int64_t gid, const nlohmann::json &properties,
+ const std::string &property_values_as_str,
+ const std::vector<mgcxx::text_search::Context *> &applicable_text_indices) {
+ if (applicable_text_indices.empty()) {
+ return;
+ }
+
+ // NOTE: Text indexes are presently all-property indices. If we allow text indexes restricted to specific properties,
+ // an indexable document should be created for each applicable index.
+ nlohmann::json document = {};
+ document["data"] = properties;
+ document["all"] = property_values_as_str;
+ document["metadata"] = {};
+ document["metadata"]["gid"] = gid;
+ document["metadata"]["deleted"] = false;
+ document["metadata"]["is_node"] = true;
+
+ for (auto *index_context : applicable_text_indices) {
+ try {
+ mgcxx::text_search::add_document(
+ *index_context,
+ mgcxx::text_search::DocumentInput{
+ .data = document.dump(-1, ' ', false, nlohmann::json::error_handler_t::replace)},
+ kDoSkipCommit);
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+ }
+}
+
+void TextIndex::CommitLoadedNodes(mgcxx::text_search::Context &index_context) {
+ // As CREATE TEXT INDEX (...) queries don’t accumulate deltas, db_transactional_accessor_->Commit() does not reach
+ // the code area where changes to indices are committed. To get around that without needing to commit text indices
+ // after every such query, we commit here.
+ try {
+ mgcxx::text_search::commit(index_context);
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+}
+
+void TextIndex::AddNode(
+ Vertex *vertex_after_update, NameIdMapper *name_id_mapper,
+ const std::optional<std::vector<mgcxx::text_search::Context *>> &maybe_applicable_text_indices) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ auto applicable_text_indices =
+ maybe_applicable_text_indices.value_or(GetApplicableTextIndices(vertex_after_update->labels));
+ if (applicable_text_indices.empty()) {
+ return;
+ }
+
+ auto vertex_properties = vertex_after_update->properties.Properties();
+ LoadNodeToTextIndices(vertex_after_update->gid.AsInt(), SerializeProperties(vertex_properties, name_id_mapper),
+ StringifyProperties(vertex_properties), applicable_text_indices);
+}
+
+void TextIndex::UpdateNode(Vertex *vertex_after_update, NameIdMapper *name_id_mapper,
+ const std::vector<LabelId> &removed_labels) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ if (!removed_labels.empty()) {
+ auto indexes_to_remove_node_from = GetApplicableTextIndices(removed_labels);
+ RemoveNode(vertex_after_update, indexes_to_remove_node_from);
+ }
+
+ auto applicable_text_indices = GetApplicableTextIndices(vertex_after_update->labels);
+ if (applicable_text_indices.empty()) return;
+ RemoveNode(vertex_after_update, applicable_text_indices);
+ AddNode(vertex_after_update, name_id_mapper, applicable_text_indices);
+}
+
+void TextIndex::RemoveNode(
+ Vertex *vertex_after_update,
+ const std::optional<std::vector<mgcxx::text_search::Context *>> &maybe_applicable_text_indices) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ auto search_node_to_be_deleted =
+ mgcxx::text_search::SearchInput{.search_query = fmt::format("metadata.gid:{}", vertex_after_update->gid.AsInt())};
+
+ for (auto *index_context :
+ maybe_applicable_text_indices.value_or(GetApplicableTextIndices(vertex_after_update->labels))) {
+ try {
+ mgcxx::text_search::delete_document(*index_context, search_node_to_be_deleted, kDoSkipCommit);
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+ }
+}
+
+void TextIndex::CreateIndex(const std::filesystem::path &storage_dir, const std::string &index_name, LabelId label,
+ memgraph::query::DbAccessor *db) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ CreateEmptyIndex(storage_dir, index_name, label);
+
+ for (const auto &v : db->Vertices(View::NEW)) {
+ if (!v.HasLabel(View::NEW, label).GetValue()) {
+ continue;
+ }
+
+ auto vertex_properties = v.Properties(View::NEW).GetValue();
+ LoadNodeToTextIndices(v.Gid().AsInt(), SerializeProperties(vertex_properties, db),
+ StringifyProperties(vertex_properties), {&index_.at(index_name).context_});
+ }
+
+ CommitLoadedNodes(index_.at(index_name).context_);
+}
+
+void TextIndex::RecoverIndex(const std::filesystem::path &storage_dir, const std::string &index_name, LabelId label,
+ memgraph::utils::SkipList<Vertex>::Accessor vertices, NameIdMapper *name_id_mapper) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ // Clear Tantivy-internal files if they exist from previous sessions
+ std::filesystem::remove_all(storage_dir / kTextIndicesDirectory / index_name);
+
+ CreateEmptyIndex(storage_dir, index_name, label);
+
+ for (const auto &v : vertices) {
+ if (std::find(v.labels.begin(), v.labels.end(), label) == v.labels.end()) {
+ continue;
+ }
+
+ auto vertex_properties = v.properties.Properties();
+ LoadNodeToTextIndices(v.gid.AsInt(), SerializeProperties(vertex_properties, name_id_mapper),
+ StringifyProperties(vertex_properties), {&index_.at(index_name).context_});
+ }
+
+ CommitLoadedNodes(index_.at(index_name).context_);
+}
+
+LabelId TextIndex::DropIndex(const std::filesystem::path &storage_dir, const std::string &index_name) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ if (!index_.contains(index_name)) {
+ throw query::TextSearchException("Text index \"{}\" doesn’t exist.", index_name);
+ }
+
+ try {
+ mgcxx::text_search::drop_index(MakeIndexPath(storage_dir, index_name));
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+ auto deleted_index_label = index_.at(index_name).scope_;
+
+ index_.erase(index_name);
+ std::erase_if(label_to_index_, [index_name](const auto &item) { return item.second == index_name; });
+
+ return deleted_index_label;
+}
+
+bool TextIndex::IndexExists(const std::string &index_name) const { return index_.contains(index_name); }
+
+mgcxx::text_search::SearchOutput TextIndex::SearchGivenProperties(const std::string &index_name,
+ const std::string &search_query) {
+ try {
+ return mgcxx::text_search::search(
+ index_.at(index_name).context_,
+ mgcxx::text_search::SearchInput{.search_query = search_query, .return_fields = {"metadata"}});
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+
+ return mgcxx::text_search::SearchOutput{};
+}
+
+mgcxx::text_search::SearchOutput TextIndex::RegexSearch(const std::string &index_name,
+ const std::string &search_query) {
+ try {
+ return mgcxx::text_search::regex_search(
+ index_.at(index_name).context_,
+ mgcxx::text_search::SearchInput{
+ .search_fields = {"all"}, .search_query = search_query, .return_fields = {"metadata"}});
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+
+ return mgcxx::text_search::SearchOutput{};
+}
+
+mgcxx::text_search::SearchOutput TextIndex::SearchAllProperties(const std::string &index_name,
+ const std::string &search_query) {
+ try {
+ return mgcxx::text_search::search(
+ index_.at(index_name).context_,
+ mgcxx::text_search::SearchInput{
+ .search_fields = {"all"}, .search_query = search_query, .return_fields = {"metadata"}});
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+
+ return mgcxx::text_search::SearchOutput{};
+}
+
+std::vector<Gid> TextIndex::Search(const std::string &index_name, const std::string &search_query,
+ text_search_mode search_mode) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ if (!index_.contains(index_name)) {
+ throw query::TextSearchException("Text index \"{}\" doesn’t exist.", index_name);
+ }
+
+ mgcxx::text_search::SearchOutput search_results;
+ switch (search_mode) {
+ case text_search_mode::SPECIFIED_PROPERTIES:
+ search_results = SearchGivenProperties(index_name, search_query);
+ break;
+ case text_search_mode::REGEX:
+ search_results = RegexSearch(index_name, search_query);
+ break;
+ case text_search_mode::ALL_PROPERTIES:
+ search_results = SearchAllProperties(index_name, search_query);
+ break;
+ default:
+ throw query::TextSearchException(
+ "Unsupported search mode: please use one of text_search.search, text_search.search_all, or "
+ "text_search.regex_search.");
+ }
+
+ std::vector<Gid> found_nodes;
+ for (const auto &doc : search_results.docs) {
+ // The CXX .data() method (https://cxx.rs/binding/string.html) may overestimate string length, causing JSON parsing
+ // errors downstream. We prevent this by resizing the converted string with the correctly-working .length() method.
+ std::string doc_string = doc.data.data();
+ doc_string.resize(doc.data.length());
+ auto doc_json = nlohmann::json::parse(doc_string);
+ found_nodes.push_back(storage::Gid::FromString(doc_json["metadata"]["gid"].dump()));
+ }
+ return found_nodes;
+}
+
+std::string TextIndex::Aggregate(const std::string &index_name, const std::string &search_query,
+ const std::string &aggregation_query) {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ if (!index_.contains(index_name)) {
+ throw query::TextSearchException("Text index \"{}\" doesn’t exist.", index_name);
+ }
+
+ mgcxx::text_search::DocumentOutput aggregation_result;
+ try {
+ aggregation_result = mgcxx::text_search::aggregate(
+ index_.at(index_name).context_,
+ mgcxx::text_search::SearchInput{
+ .search_fields = {"all"}, .search_query = search_query, .aggregation_query = aggregation_query});
+
+ } catch (const std::exception &e) {
+ throw query::TextSearchException("Tantivy error: {}", e.what());
+ }
+ // The CXX .data() method (https://cxx.rs/binding/string.html) may overestimate string length, causing JSON parsing
+ // errors downstream. We prevent this by resizing the converted string with the correctly-working .length() method.
+ std::string result_string = aggregation_result.data.data();
+ result_string.resize(aggregation_result.data.length());
+ return result_string;
+}
+
+void TextIndex::Commit() {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ for (auto &[_, index_data] : index_) {
+ mgcxx::text_search::commit(index_data.context_);
+ }
+}
+
+void TextIndex::Rollback() {
+ if (!flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ throw query::TextSearchDisabledException();
+ }
+
+ for (auto &[_, index_data] : index_) {
+ mgcxx::text_search::rollback(index_data.context_);
+ }
+}
+
+std::vector<std::pair<std::string, LabelId>> TextIndex::ListIndices() const {
+ std::vector<std::pair<std::string, LabelId>> ret;
+ ret.reserve(index_.size());
+ for (const auto &[index_name, index_data] : index_) {
+ ret.emplace_back(index_name, index_data.scope_);
+ }
+ return ret;
+}
+
+} // namespace memgraph::storage
diff --git a/src/storage/v2/indices/text_index.hpp b/src/storage/v2/indices/text_index.hpp
new file mode 100644
index 000000000..af4748c6e
--- /dev/null
+++ b/src/storage/v2/indices/text_index.hpp
@@ -0,0 +1,105 @@
+// Copyright 2024 Memgraph Ltd.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+// License, and you may not use this file except in compliance with the Business Source License.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+#pragma once
+
+#include <json/json.hpp>
+#include "mg_procedure.h"
+#include "storage/v2/id_types.hpp"
+#include "storage/v2/name_id_mapper.hpp"
+#include "storage/v2/vertex.hpp"
+#include "text_search.hpp"
+
+namespace memgraph::query {
+class DbAccessor;
+}
+
+namespace memgraph::storage {
+struct TextIndexData {
+ mgcxx::text_search::Context context_;
+ LabelId scope_;
+};
+
+class TextIndex {
+ private:
+ static constexpr bool kDoSkipCommit = true;
+ static constexpr std::string_view kTextIndicesDirectory = "text_indices";
+
+ inline std::string MakeIndexPath(const std::filesystem::path &storage_dir, const std::string &index_name);
+
+ void CreateEmptyIndex(const std::filesystem::path &storage_dir, const std::string &index_name, LabelId label);
+
+ template <typename T>
+ nlohmann::json SerializeProperties(const std::map<PropertyId, PropertyValue> &properties, T *name_resolver);
+
+ std::string StringifyProperties(const std::map<PropertyId, PropertyValue> &properties);
+
+ std::vector<mgcxx::text_search::Context *> GetApplicableTextIndices(const std::vector<LabelId> &labels);
+
+ void LoadNodeToTextIndices(const std::int64_t gid, const nlohmann::json &properties,
+ const std::string &property_values_as_str,
+ const std::vector<mgcxx::text_search::Context *> &applicable_text_indices);
+
+ void CommitLoadedNodes(mgcxx::text_search::Context &index_context);
+
+ mgcxx::text_search::SearchOutput SearchGivenProperties(const std::string &index_name,
+ const std::string &search_query);
+
+ mgcxx::text_search::SearchOutput RegexSearch(const std::string &index_name, const std::string &search_query);
+
+ mgcxx::text_search::SearchOutput SearchAllProperties(const std::string &index_name, const std::string &search_query);
+
+ public:
+ TextIndex() = default;
+
+ TextIndex(const TextIndex &) = delete;
+ TextIndex(TextIndex &&) = delete;
+ TextIndex &operator=(const TextIndex &) = delete;
+ TextIndex &operator=(TextIndex &&) = delete;
+
+ ~TextIndex() = default;
+
+ std::map<std::string, TextIndexData> index_;
+ std::map<LabelId, std::string> label_to_index_;
+
+ void AddNode(
+ Vertex *vertex, NameIdMapper *name_id_mapper,
+ const std::optional<std::vector<mgcxx::text_search::Context *>> &maybe_applicable_text_indices = std::nullopt);
+
+ void UpdateNode(Vertex *vertex, NameIdMapper *name_id_mapper, const std::vector<LabelId> &removed_labels = {});
+
+ void RemoveNode(
+ Vertex *vertex,
+ const std::optional<std::vector<mgcxx::text_search::Context *>> &maybe_applicable_text_indices = std::nullopt);
+
+ void CreateIndex(const std::filesystem::path &storage_dir, const std::string &index_name, LabelId label,
+ memgraph::query::DbAccessor *db);
+
+ void RecoverIndex(const std::filesystem::path &storage_dir, const std::string &index_name, LabelId label,
+ memgraph::utils::SkipList<Vertex>::Accessor vertices, NameIdMapper *name_id_mapper);
+
+ LabelId DropIndex(const std::filesystem::path &storage_dir, const std::string &index_name);
+
+ bool IndexExists(const std::string &index_name) const;
+
+ std::vector<Gid> Search(const std::string &index_name, const std::string &search_query, text_search_mode search_mode);
+
+ std::string Aggregate(const std::string &index_name, const std::string &search_query,
+ const std::string &aggregation_query);
+
+ void Commit();
+
+ void Rollback();
+
+ std::vector<std::pair<std::string, LabelId>> ListIndices() const;
+};
+
+} // namespace memgraph::storage
diff --git a/src/storage/v2/inmemory/storage.cpp b/src/storage/v2/inmemory/storage.cpp
index dab56750b..dbaa56ce2 100644
--- a/src/storage/v2/inmemory/storage.cpp
+++ b/src/storage/v2/inmemory/storage.cpp
@@ -15,6 +15,8 @@
#include <functional>
#include <optional>
#include "dbms/constants.hpp"
+#include "flags/experimental.hpp"
+#include "flags/run_time_configurable.hpp"
#include "memory/global_memory_control.hpp"
#include "storage/v2/durability/durability.hpp"
#include "storage/v2/durability/snapshot.hpp"
@@ -890,6 +892,10 @@ utils::BasicResult<StorageManipulationError, void> InMemoryStorage::InMemoryAcce
commit_timestamp_.reset(); // We have aborted, hence we have not committed
return StorageManipulationError{*unique_constraint_violation};
}
+
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ mem_storage->indices_.text_index_.Commit();
+ }
}
is_transaction_active_ = false;
@@ -1213,6 +1219,9 @@ void InMemoryStorage::InMemoryAccessor::Abort() {
for (auto const &[property, prop_vertices] : property_cleanup) {
storage_->indices_.AbortEntries(property, prop_vertices, transaction_.start_timestamp);
}
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ storage_->indices_.text_index_.Rollback();
+ }
// VERTICES
{
@@ -1846,6 +1855,7 @@ StorageInfo InMemoryStorage::GetInfo(memgraph::replication_coordination_glue::Re
const auto &lbl = access->ListAllIndices();
info.label_indices = lbl.label.size();
info.label_property_indices = lbl.label_property.size();
+ info.text_indices = lbl.text_indices.size();
const auto &con = access->ListAllConstraints();
info.existence_constraints = con.existence.size();
info.unique_constraints = con.unique.size();
@@ -2107,6 +2117,16 @@ bool InMemoryStorage::AppendToWal(const Transaction &transaction, uint64_t final
AppendToWalDataDefinition(durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_STATS_CLEAR, info.label,
final_commit_timestamp);
} break;
+ case MetadataDelta::Action::TEXT_INDEX_CREATE: {
+ const auto &info = md_delta.text_index;
+ AppendToWalDataDefinition(durability::StorageMetadataOperation::TEXT_INDEX_CREATE, info.index_name, info.label,
+ final_commit_timestamp);
+ } break;
+ case MetadataDelta::Action::TEXT_INDEX_DROP: {
+ const auto &info = md_delta.text_index;
+ AppendToWalDataDefinition(durability::StorageMetadataOperation::TEXT_INDEX_DROP, info.index_name, info.label,
+ final_commit_timestamp);
+ } break;
case MetadataDelta::Action::EXISTENCE_CONSTRAINT_CREATE: {
const auto &info = md_delta.label_property;
AppendToWalDataDefinition(durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_CREATE, info.label,
@@ -2137,11 +2157,13 @@ bool InMemoryStorage::AppendToWal(const Transaction &transaction, uint64_t final
return repl_storage_state_.FinalizeTransaction(final_commit_timestamp, this, std::move(db_acc));
}
-void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOperation operation, LabelId label,
+void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOperation operation,
+ const std::optional<std::string> text_index_name, LabelId label,
const std::set<PropertyId> &properties, LabelIndexStats stats,
LabelPropertyIndexStats property_stats,
uint64_t final_commit_timestamp) {
- wal_file_->AppendOperation(operation, label, properties, stats, property_stats, final_commit_timestamp);
+ wal_file_->AppendOperation(operation, text_index_name, label, properties, stats, property_stats,
+ final_commit_timestamp);
repl_storage_state_.AppendOperation(operation, label, properties, stats, property_stats, final_commit_timestamp);
}
@@ -2155,12 +2177,13 @@ void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOpera
const std::set<PropertyId> &properties,
LabelPropertyIndexStats property_stats,
uint64_t final_commit_timestamp) {
- return AppendToWalDataDefinition(operation, label, properties, {}, property_stats, final_commit_timestamp);
+ return AppendToWalDataDefinition(operation, std::nullopt, label, properties, {}, property_stats,
+ final_commit_timestamp);
}
void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOperation operation, LabelId label,
LabelIndexStats stats, uint64_t final_commit_timestamp) {
- return AppendToWalDataDefinition(operation, label, {}, stats, {}, final_commit_timestamp);
+ return AppendToWalDataDefinition(operation, std::nullopt, label, {}, stats, {}, final_commit_timestamp);
}
void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOperation operation, LabelId label,
@@ -2174,6 +2197,12 @@ void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOpera
return AppendToWalDataDefinition(operation, label, {}, {}, final_commit_timestamp);
}
+void InMemoryStorage::AppendToWalDataDefinition(durability::StorageMetadataOperation operation,
+ const std::optional<std::string> text_index_name, LabelId label,
+ uint64_t final_commit_timestamp) {
+ return AppendToWalDataDefinition(operation, text_index_name, label, {}, {}, {}, final_commit_timestamp);
+}
+
utils::BasicResult<InMemoryStorage::CreateSnapshotError> InMemoryStorage::CreateSnapshot(
memgraph::replication_coordination_glue::ReplicationRole replication_role) {
using memgraph::replication_coordination_glue::ReplicationRole;
@@ -2301,7 +2330,9 @@ IndicesInfo InMemoryStorage::InMemoryAccessor::ListAllIndices() const {
auto *mem_label_property_index =
static_cast<InMemoryLabelPropertyIndex *>(in_memory->indices_.label_property_index_.get());
auto *mem_edge_type_index = static_cast<InMemoryEdgeTypeIndex *>(in_memory->indices_.edge_type_index_.get());
- return {mem_label_index->ListIndices(), mem_label_property_index->ListIndices(), mem_edge_type_index->ListIndices()};
+ auto &text_index = storage_->indices_.text_index_;
+ return {mem_label_index->ListIndices(), mem_label_property_index->ListIndices(), mem_edge_type_index->ListIndices(),
+ text_index.ListIndices()};
}
ConstraintsInfo InMemoryStorage::InMemoryAccessor::ListAllConstraints() const {
const auto *mem_storage = static_cast<InMemoryStorage *>(storage_);
diff --git a/src/storage/v2/inmemory/storage.hpp b/src/storage/v2/inmemory/storage.hpp
index 6d10e0fbd..6532471f3 100644
--- a/src/storage/v2/inmemory/storage.hpp
+++ b/src/storage/v2/inmemory/storage.hpp
@@ -398,7 +398,7 @@ class InMemoryStorage final : public Storage {
StorageInfo GetBaseInfo() override;
StorageInfo GetInfo(memgraph::replication_coordination_glue::ReplicationRole replication_role) override;
- /// Return true in all cases excepted if any sync replicas have not sent confirmation.
+ /// Return true in all cases except if any sync replicas have not sent confirmation.
[[nodiscard]] bool AppendToWal(const Transaction &transaction, uint64_t final_commit_timestamp,
DatabaseAccessProtector db_acc);
void AppendToWalDataDefinition(durability::StorageMetadataOperation operation, LabelId label,
@@ -412,9 +412,13 @@ class InMemoryStorage final : public Storage {
void AppendToWalDataDefinition(durability::StorageMetadataOperation operation, LabelId label,
const std::set<PropertyId> &properties, LabelPropertyIndexStats property_stats,
uint64_t final_commit_timestamp);
- void AppendToWalDataDefinition(durability::StorageMetadataOperation operation, LabelId label,
+ void AppendToWalDataDefinition(durability::StorageMetadataOperation operation,
+ const std::optional<std::string> text_index_name, LabelId label,
const std::set<PropertyId> &properties, LabelIndexStats stats,
LabelPropertyIndexStats property_stats, uint64_t final_commit_timestamp);
+ void AppendToWalDataDefinition(durability::StorageMetadataOperation operation,
+ const std::optional<std::string> text_index_name, LabelId label,
+ uint64_t final_commit_timestamp);
uint64_t CommitTimestamp(std::optional<uint64_t> desired_commit_timestamp = {});
diff --git a/src/storage/v2/metadata_delta.hpp b/src/storage/v2/metadata_delta.hpp
index b34966a62..e4616161d 100644
--- a/src/storage/v2/metadata_delta.hpp
+++ b/src/storage/v2/metadata_delta.hpp
@@ -37,6 +37,8 @@ struct MetadataDelta {
LABEL_PROPERTY_INDEX_STATS_CLEAR,
EDGE_INDEX_CREATE,
EDGE_INDEX_DROP,
+ TEXT_INDEX_CREATE,
+ TEXT_INDEX_DROP,
EXISTENCE_CONSTRAINT_CREATE,
EXISTENCE_CONSTRAINT_DROP,
UNIQUE_CONSTRAINT_CREATE,
@@ -63,6 +65,10 @@ struct MetadataDelta {
} edge_index_create;
static constexpr struct EdgeIndexDrop {
} edge_index_drop;
+ static constexpr struct TextIndexCreate {
+ } text_index_create;
+ static constexpr struct TextIndexDrop {
+ } text_index_drop;
static constexpr struct ExistenceConstraintCreate {
} existence_constraint_create;
static constexpr struct ExistenceConstraintDrop {
@@ -98,6 +104,12 @@ struct MetadataDelta {
MetadataDelta(EdgeIndexDrop /*tag*/, EdgeTypeId edge_type) : action(Action::EDGE_INDEX_DROP), edge_type(edge_type) {}
+ MetadataDelta(TextIndexCreate /*tag*/, std::string index_name, LabelId label)
+ : action(Action::TEXT_INDEX_CREATE), text_index{index_name, label} {}
+
+ MetadataDelta(TextIndexDrop /*tag*/, std::string index_name, LabelId label)
+ : action(Action::TEXT_INDEX_DROP), text_index{index_name, label} {}
+
MetadataDelta(ExistenceConstraintCreate /*tag*/, LabelId label, PropertyId property)
: action(Action::EXISTENCE_CONSTRAINT_CREATE), label_property{label, property} {}
@@ -127,6 +139,8 @@ struct MetadataDelta {
case Action::LABEL_PROPERTY_INDEX_STATS_CLEAR:
case Action::EDGE_INDEX_CREATE:
case Action::EDGE_INDEX_DROP:
+ case Action::TEXT_INDEX_CREATE:
+ case Action::TEXT_INDEX_DROP:
case Action::EXISTENCE_CONSTRAINT_CREATE:
case Action::EXISTENCE_CONSTRAINT_DROP:
break;
@@ -164,6 +178,11 @@ struct MetadataDelta {
PropertyId property;
LabelPropertyIndexStats stats;
} label_property_stats;
+
+ struct {
+ std::string index_name;
+ LabelId label;
+ } text_index;
};
};
diff --git a/src/storage/v2/property_store.cpp b/src/storage/v2/property_store.cpp
index adf3440a2..0cfee0f98 100644
--- a/src/storage/v2/property_store.cpp
+++ b/src/storage/v2/property_store.cpp
@@ -118,7 +118,7 @@ enum class Type : uint8_t {
STRING = 0x50,
LIST = 0x60,
MAP = 0x70,
- TEMPORAL_DATA = 0x80
+ TEMPORAL_DATA = 0x80,
};
const uint8_t kMaskType = 0xf0;
diff --git a/src/storage/v2/property_value.hpp b/src/storage/v2/property_value.hpp
index e48be008a..161ad151a 100644
--- a/src/storage/v2/property_value.hpp
+++ b/src/storage/v2/property_value.hpp
@@ -92,7 +92,28 @@ class PropertyValue {
// TODO: Implement copy assignment operators for primitive types.
// TODO: Implement copy and move assignment operators for non-primitive types.
- ~PropertyValue() { DestroyValue(); }
+ ~PropertyValue() {
+ switch (type_) {
+ // destructor for primitive types does nothing
+ case Type::Null:
+ case Type::Bool:
+ case Type::Int:
+ case Type::Double:
+ case Type::TemporalData:
+ return;
+
+ // destructor for non primitive types since we used placement new
+ case Type::String:
+ std::destroy_at(&string_v.val_);
+ return;
+ case Type::List:
+ std::destroy_at(&list_v.val_);
+ return;
+ case Type::Map:
+ std::destroy_at(&map_v.val_);
+ return;
+ }
+ }
Type type() const { return type_; }
@@ -189,8 +210,6 @@ class PropertyValue {
}
private:
- void DestroyValue() noexcept;
-
// NOTE: this may look strange but it is for better data layout
// https://eel.is/c++draft/class.union#general-note-1
union {
@@ -357,13 +376,13 @@ inline PropertyValue::PropertyValue(const PropertyValue &other) : type_(other.ty
this->double_v.val_ = other.double_v.val_;
return;
case Type::String:
- new (&string_v.val_) std::string(other.string_v.val_);
+ std::construct_at(&string_v.val_, other.string_v.val_);
return;
case Type::List:
- new (&list_v.val_) std::vector<PropertyValue>(other.list_v.val_);
+ std::construct_at(&list_v.val_, other.list_v.val_);
return;
case Type::Map:
- new (&map_v.val_) std::map<std::string, PropertyValue>(other.map_v.val_);
+ std::construct_at(&map_v.val_, other.map_v.val_);
return;
case Type::TemporalData:
this->temporal_data_v.val_ = other.temporal_data_v.val_;
@@ -371,7 +390,7 @@ inline PropertyValue::PropertyValue(const PropertyValue &other) : type_(other.ty
}
}
-inline PropertyValue::PropertyValue(PropertyValue &&other) noexcept : type_(std::exchange(other.type_, Type::Null)) {
+inline PropertyValue::PropertyValue(PropertyValue &&other) noexcept : type_(other.type_) {
switch (type_) {
case Type::Null:
break;
@@ -386,15 +405,12 @@ inline PropertyValue::PropertyValue(PropertyValue &&other) noexcept : type_(std:
break;
case Type::String:
std::construct_at(&string_v.val_, std::move(other.string_v.val_));
- std::destroy_at(&other.string_v.val_);
break;
case Type::List:
std::construct_at(&list_v.val_, std::move(other.list_v.val_));
- std::destroy_at(&other.list_v.val_);
break;
case Type::Map:
std::construct_at(&map_v.val_, std::move(other.map_v.val_));
- std::destroy_at(&other.map_v.val_);
break;
case Type::TemporalData:
temporal_data_v.val_ = other.temporal_data_v.val_;
@@ -403,38 +419,88 @@ inline PropertyValue::PropertyValue(PropertyValue &&other) noexcept : type_(std:
}
inline PropertyValue &PropertyValue::operator=(const PropertyValue &other) {
- if (this == &other) return *this;
+ if (type_ == other.type_) {
+ if (this == &other) return *this;
+ switch (other.type_) {
+ case Type::Null:
+ break;
+ case Type::Bool:
+ bool_v.val_ = other.bool_v.val_;
+ break;
+ case Type::Int:
+ int_v.val_ = other.int_v.val_;
+ break;
+ case Type::Double:
+ double_v.val_ = other.double_v.val_;
+ break;
+ case Type::String:
+ string_v.val_ = other.string_v.val_;
+ break;
+ case Type::List:
+ list_v.val_ = other.list_v.val_;
+ break;
+ case Type::Map:
+ map_v.val_ = other.map_v.val_;
+ break;
+ case Type::TemporalData:
+ temporal_data_v.val_ = other.temporal_data_v.val_;
+ break;
+ }
+ return *this;
+ } else {
+ // destroy
+ switch (type_) {
+ case Type::Null:
+ break;
+ case Type::Bool:
+ break;
+ case Type::Int:
+ break;
+ case Type::Double:
+ break;
+ case Type::String:
+ std::destroy_at(&string_v.val_);
+ break;
+ case Type::List:
+ std::destroy_at(&list_v.val_);
+ break;
+ case Type::Map:
+ std::destroy_at(&map_v.val_);
+ break;
+ case Type::TemporalData:
+ break;
+ }
+ // construct
+ auto *new_this = std::launder(this);
+ switch (other.type_) {
+ case Type::Null:
+ break;
+ case Type::Bool:
+ new_this->bool_v.val_ = other.bool_v.val_;
+ break;
+ case Type::Int:
+ new_this->int_v.val_ = other.int_v.val_;
+ break;
+ case Type::Double:
+ new_this->double_v.val_ = other.double_v.val_;
+ break;
+ case Type::String:
+ std::construct_at(&new_this->string_v.val_, other.string_v.val_);
+ break;
+ case Type::List:
+ std::construct_at(&new_this->list_v.val_, other.list_v.val_);
+ break;
+ case Type::Map:
+ std::construct_at(&new_this->map_v.val_, other.map_v.val_);
+ break;
+ case Type::TemporalData:
+ new_this->temporal_data_v.val_ = other.temporal_data_v.val_;
+ break;
+ }
- DestroyValue();
- type_ = other.type_;
-
- switch (other.type_) {
- case Type::Null:
- break;
- case Type::Bool:
- this->bool_v.val_ = other.bool_v.val_;
- break;
- case Type::Int:
- this->int_v.val_ = other.int_v.val_;
- break;
- case Type::Double:
- this->double_v.val_ = other.double_v.val_;
- break;
- case Type::String:
- new (&string_v.val_) std::string(other.string_v.val_);
- break;
- case Type::List:
- new (&list_v.val_) std::vector<PropertyValue>(other.list_v.val_);
- break;
- case Type::Map:
- new (&map_v.val_) std::map<std::string, PropertyValue>(other.map_v.val_);
- break;
- case Type::TemporalData:
- this->temporal_data_v.val_ = other.temporal_data_v.val_;
- break;
+ new_this->type_ = other.type_;
+ return *new_this;
}
-
- return *this;
}
inline PropertyValue &PropertyValue::operator=(PropertyValue &&other) noexcept {
@@ -456,48 +522,71 @@ inline PropertyValue &PropertyValue::operator=(PropertyValue &&other) noexcept {
break;
case Type::String:
string_v.val_ = std::move(other.string_v.val_);
- std::destroy_at(&other.string_v.val_);
break;
case Type::List:
list_v.val_ = std::move(other.list_v.val_);
- std::destroy_at(&other.list_v.val_);
break;
case Type::Map:
map_v.val_ = std::move(other.map_v.val_);
- std::destroy_at(&other.map_v.val_);
break;
case Type::TemporalData:
temporal_data_v.val_ = other.temporal_data_v.val_;
break;
}
- other.type_ = Type::Null;
return *this;
} else {
- std::destroy_at(this);
- return *std::construct_at(std::launder(this), std::move(other));
- }
-}
+ // destroy
+ switch (type_) {
+ case Type::Null:
+ break;
+ case Type::Bool:
+ break;
+ case Type::Int:
+ break;
+ case Type::Double:
+ break;
+ case Type::String:
+ std::destroy_at(&string_v.val_);
+ break;
+ case Type::List:
+ std::destroy_at(&list_v.val_);
+ break;
+ case Type::Map:
+ std::destroy_at(&map_v.val_);
+ break;
+ case Type::TemporalData:
+ break;
+ }
+ // construct (no need to destroy moved from type)
+ auto *new_this = std::launder(this);
+ switch (other.type_) {
+ case Type::Null:
+ break;
+ case Type::Bool:
+ new_this->bool_v.val_ = other.bool_v.val_;
+ break;
+ case Type::Int:
+ new_this->int_v.val_ = other.int_v.val_;
+ break;
+ case Type::Double:
+ new_this->double_v.val_ = other.double_v.val_;
+ break;
+ case Type::String:
+ std::construct_at(&new_this->string_v.val_, std::move(other.string_v.val_));
+ break;
+ case Type::List:
+ std::construct_at(&new_this->list_v.val_, std::move(other.list_v.val_));
+ break;
+ case Type::Map:
+ std::construct_at(&new_this->map_v.val_, std::move(other.map_v.val_));
+ break;
+ case Type::TemporalData:
+ new_this->temporal_data_v.val_ = other.temporal_data_v.val_;
+ break;
+ }
-inline void PropertyValue::DestroyValue() noexcept {
- switch (std::exchange(type_, Type::Null)) {
- // destructor for primitive types does nothing
- case Type::Null:
- case Type::Bool:
- case Type::Int:
- case Type::Double:
- case Type::TemporalData:
- return;
-
- // destructor for non primitive types since we used placement new
- case Type::String:
- std::destroy_at(&string_v.val_);
- return;
- case Type::List:
- std::destroy_at(&list_v.val_);
- return;
- case Type::Map:
- std::destroy_at(&map_v.val_);
- return;
+ new_this->type_ = other.type_;
+ return *new_this;
}
}
diff --git a/src/storage/v2/replication/replication_client.cpp b/src/storage/v2/replication/replication_client.cpp
index 3c1081206..ee1394fdb 100644
--- a/src/storage/v2/replication/replication_client.cpp
+++ b/src/storage/v2/replication/replication_client.cpp
@@ -13,6 +13,7 @@
#include "replication/replication_client.hpp"
#include "storage/v2/inmemory/storage.hpp"
+#include "storage/v2/replication/enums.hpp"
#include "storage/v2/storage.hpp"
#include "utils/exceptions.hpp"
#include "utils/on_scope_exit.hpp"
@@ -247,11 +248,13 @@ bool ReplicationStorageClient::FinalizeTransactionReplication(Storage *storage,
MG_ASSERT(replica_stream_, "Missing stream for transaction deltas");
try {
auto response = replica_stream_->Finalize();
- return replica_state_.WithLock([storage, &response, db_acc = std::move(db_acc), this](auto &state) mutable {
+ // NOLINTNEXTLINE
+ return replica_state_.WithLock([storage, response, db_acc = std::move(db_acc), this](auto &state) mutable {
replica_stream_.reset();
if (!response.success || state == replication::ReplicaState::RECOVERY) {
state = replication::ReplicaState::RECOVERY;
- client_.thread_pool_.AddTask([storage, &response, db_acc = std::move(db_acc), this] {
+ // NOLINTNEXTLINE
+ client_.thread_pool_.AddTask([storage, response, db_acc = std::move(db_acc), this] {
this->RecoverReplica(response.current_commit_timestamp, storage);
});
return false;
@@ -403,8 +406,9 @@ void ReplicaStream::AppendOperation(durability::StorageMetadataOperation operati
const std::set<PropertyId> &properties, const LabelIndexStats &stats,
const LabelPropertyIndexStats &property_stats, uint64_t timestamp) {
replication::Encoder encoder(stream_.GetBuilder());
- EncodeOperation(&encoder, storage_->name_id_mapper_.get(), operation, label, properties, stats, property_stats,
- timestamp);
+ // NOTE: Text search doesn’t have replication in scope yet (Phases 1 and 2) -> text index name not sent here
+ EncodeOperation(&encoder, storage_->name_id_mapper_.get(), operation, std::nullopt, label, properties, stats,
+ property_stats, timestamp);
}
void ReplicaStream::AppendOperation(durability::StorageMetadataOperation operation, EdgeTypeId edge_type,
diff --git a/src/storage/v2/storage.cpp b/src/storage/v2/storage.cpp
index 536a504a0..db4bec8be 100644
--- a/src/storage/v2/storage.cpp
+++ b/src/storage/v2/storage.cpp
@@ -13,6 +13,8 @@
#include "absl/container/flat_hash_set.h"
#include "spdlog/spdlog.h"
+#include "flags/experimental.hpp"
+#include "flags/run_time_configurable.hpp"
#include "storage/v2/disk/name_id_mapper.hpp"
#include "storage/v2/storage.hpp"
#include "storage/v2/transaction.hpp"
@@ -273,6 +275,12 @@ Storage::Accessor::DetachDelete(std::vector<VertexAccessor *> nodes, std::vector
return maybe_deleted_vertices.GetError();
}
+ if (flags::AreExperimentsEnabled(flags::Experiments::TEXT_SEARCH)) {
+ for (auto *node : nodes_to_delete) {
+ storage_->indices_.text_index_.RemoveNode(node);
+ }
+ }
+
auto deleted_vertices = maybe_deleted_vertices.GetValue();
return std::make_optional<ReturnType>(std::move(deleted_vertices), std::move(deleted_edges));
@@ -543,4 +551,19 @@ void Storage::Accessor::MarkEdgeAsDeleted(Edge *edge) {
}
}
+void Storage::Accessor::CreateTextIndex(const std::string &index_name, LabelId label, query::DbAccessor *db) {
+ MG_ASSERT(unique_guard_.owns_lock(), "Creating a text index requires unique access to storage!");
+ storage_->indices_.text_index_.CreateIndex(storage_->config_.durability.storage_directory, index_name, label, db);
+ transaction_.md_deltas.emplace_back(MetadataDelta::text_index_create, index_name, label);
+ memgraph::metrics::IncrementCounter(memgraph::metrics::ActiveTextIndices);
+}
+
+void Storage::Accessor::DropTextIndex(const std::string &index_name) {
+ MG_ASSERT(unique_guard_.owns_lock(), "Dropping a text index requires unique access to storage!");
+ auto deleted_index_label =
+ storage_->indices_.text_index_.DropIndex(storage_->config_.durability.storage_directory, index_name);
+ transaction_.md_deltas.emplace_back(MetadataDelta::text_index_drop, index_name, deleted_index_label);
+ memgraph::metrics::DecrementCounter(memgraph::metrics::ActiveTextIndices);
+}
+
} // namespace memgraph::storage
diff --git a/src/storage/v2/storage.hpp b/src/storage/v2/storage.hpp
index 58936bd56..a4436b1b7 100644
--- a/src/storage/v2/storage.hpp
+++ b/src/storage/v2/storage.hpp
@@ -20,6 +20,7 @@
#include "io/network/endpoint.hpp"
#include "kvstore/kvstore.hpp"
+#include "mg_procedure.h"
#include "query/exceptions.hpp"
#include "replication/config.hpp"
#include "replication/replication_server.hpp"
@@ -53,6 +54,7 @@ extern const Event SnapshotCreationLatency_us;
extern const Event ActiveLabelIndices;
extern const Event ActiveLabelPropertyIndices;
+extern const Event ActiveTextIndices;
} // namespace memgraph::metrics
namespace memgraph::storage {
@@ -63,6 +65,7 @@ struct IndicesInfo {
std::vector<LabelId> label;
std::vector<std::pair<LabelId, PropertyId>> label_property;
std::vector<EdgeTypeId> edge_type;
+ std::vector<std::pair<std::string, LabelId>> text_indices;
};
struct ConstraintsInfo {
@@ -78,6 +81,7 @@ struct StorageInfo {
uint64_t disk_usage;
uint64_t label_indices;
uint64_t label_property_indices;
+ uint64_t text_indices;
uint64_t existence_constraints;
uint64_t unique_constraints;
StorageMode storage_mode;
@@ -95,6 +99,7 @@ static inline nlohmann::json ToJson(const StorageInfo &info) {
res["disk"] = info.disk_usage;
res["label_indices"] = info.label_indices;
res["label_prop_indices"] = info.label_property_indices;
+ res["text_indices"] = info.text_indices;
res["existence_constraints"] = info.existence_constraints;
res["unique_constraints"] = info.unique_constraints;
res["storage_mode"] = storage::StorageModeToString(info.storage_mode);
@@ -232,6 +237,28 @@ class Storage {
virtual bool EdgeTypeIndexExists(EdgeTypeId edge_type) const = 0;
+ bool TextIndexExists(const std::string &index_name) const {
+ return storage_->indices_.text_index_.IndexExists(index_name);
+ }
+
+ void TextIndexAddVertex(const VertexAccessor &vertex) {
+ storage_->indices_.text_index_.AddNode(vertex.vertex_, storage_->name_id_mapper_.get());
+ }
+
+ void TextIndexUpdateVertex(const VertexAccessor &vertex, const std::vector<LabelId> &removed_labels = {}) {
+ storage_->indices_.text_index_.UpdateNode(vertex.vertex_, storage_->name_id_mapper_.get(), removed_labels);
+ }
+
+ std::vector<Gid> TextIndexSearch(const std::string &index_name, const std::string &search_query,
+ text_search_mode search_mode) const {
+ return storage_->indices_.text_index_.Search(index_name, search_query, search_mode);
+ }
+
+ std::string TextIndexAggregate(const std::string &index_name, const std::string &search_query,
+ const std::string &aggregation_query) const {
+ return storage_->indices_.text_index_.Aggregate(index_name, search_query, aggregation_query);
+ }
+
virtual IndicesInfo ListAllIndices() const = 0;
virtual ConstraintsInfo ListAllConstraints() const = 0;
@@ -284,6 +311,10 @@ class Storage {
virtual utils::BasicResult<StorageIndexDefinitionError, void> DropIndex(EdgeTypeId edge_type) = 0;
+ void CreateTextIndex(const std::string &index_name, LabelId label, query::DbAccessor *db);
+
+ void DropTextIndex(const std::string &index_name);
+
virtual utils::BasicResult<StorageExistenceConstraintDefinitionError, void> CreateExistenceConstraint(
LabelId label, PropertyId property) = 0;
diff --git a/src/storage/v2/vertex_accessor.cpp b/src/storage/v2/vertex_accessor.cpp
index 7d78070a8..83dcc003b 100644
--- a/src/storage/v2/vertex_accessor.cpp
+++ b/src/storage/v2/vertex_accessor.cpp
@@ -261,20 +261,31 @@ Result<PropertyValue> VertexAccessor::SetProperty(PropertyId property, const Pro
if (vertex_->deleted) return Error::DELETED_OBJECT;
- auto current_value = vertex_->properties.GetProperty(property);
- // We could skip setting the value if the previous one is the same to the new
- // one. This would save some memory as a delta would not be created as well as
- // avoid copying the value. The reason we are not doing that is because the
- // current code always follows the logical pattern of "create a delta" and
- // "modify in-place". Additionally, the created delta will make other
- // transactions get a SERIALIZATION_ERROR.
-
+ PropertyValue current_value;
+ const bool skip_duplicate_write = !storage_->config_.salient.items.delta_on_identical_property_update;
utils::AtomicMemoryBlock atomic_memory_block{
- [transaction = transaction_, vertex = vertex_, &value, &property, ¤t_value]() {
+ [transaction = transaction_, vertex = vertex_, &value, &property, ¤t_value, skip_duplicate_write]() {
+ current_value = vertex->properties.GetProperty(property);
+ // We could skip setting the value if the previous one is the same to the new
+ // one. This would save some memory as a delta would not be created as well as
+ // avoid copying the value. The reason we are not doing that is because the
+ // current code always follows the logical pattern of "create a delta" and
+ // "modify in-place". Additionally, the created delta will make other
+ // transactions get a SERIALIZATION_ERROR.
+ if (skip_duplicate_write && current_value == value) {
+ return true;
+ }
+
CreateAndLinkDelta(transaction, vertex, Delta::SetPropertyTag(), property, current_value);
vertex->properties.SetProperty(property, value);
+
+ return false;
}};
- std::invoke(atomic_memory_block);
+ const bool early_exit = std::invoke(atomic_memory_block);
+
+ if (early_exit) {
+ return std::move(current_value);
+ }
if (transaction_->constraint_verification_info) {
if (!value.IsNull()) {
@@ -339,27 +350,29 @@ Result<std::vector<std::tuple<PropertyId, PropertyValue, PropertyValue>>> Vertex
if (vertex_->deleted) return Error::DELETED_OBJECT;
+ const bool skip_duplicate_update = storage_->config_.salient.items.delta_on_identical_property_update;
using ReturnType = decltype(vertex_->properties.UpdateProperties(properties));
std::optional<ReturnType> id_old_new_change;
- utils::AtomicMemoryBlock atomic_memory_block{
- [storage = storage_, transaction = transaction_, vertex = vertex_, &properties, &id_old_new_change]() {
- id_old_new_change.emplace(vertex->properties.UpdateProperties(properties));
- if (!id_old_new_change.has_value()) {
- return;
+ utils::AtomicMemoryBlock atomic_memory_block{[storage = storage_, transaction = transaction_, vertex = vertex_,
+ &properties, &id_old_new_change, skip_duplicate_update]() {
+ id_old_new_change.emplace(vertex->properties.UpdateProperties(properties));
+ if (!id_old_new_change.has_value()) {
+ return;
+ }
+ for (auto &[id, old_value, new_value] : *id_old_new_change) {
+ storage->indices_.UpdateOnSetProperty(id, new_value, vertex, *transaction);
+ if (skip_duplicate_update && old_value == new_value) continue;
+ CreateAndLinkDelta(transaction, vertex, Delta::SetPropertyTag(), id, std::move(old_value));
+ transaction->manyDeltasCache.Invalidate(vertex, id);
+ if (transaction->constraint_verification_info) {
+ if (!new_value.IsNull()) {
+ transaction->constraint_verification_info->AddedProperty(vertex);
+ } else {
+ transaction->constraint_verification_info->RemovedProperty(vertex);
}
- for (auto &[id, old_value, new_value] : *id_old_new_change) {
- storage->indices_.UpdateOnSetProperty(id, new_value, vertex, *transaction);
- CreateAndLinkDelta(transaction, vertex, Delta::SetPropertyTag(), id, std::move(old_value));
- transaction->manyDeltasCache.Invalidate(vertex, id);
- if (transaction->constraint_verification_info) {
- if (!new_value.IsNull()) {
- transaction->constraint_verification_info->AddedProperty(vertex);
- } else {
- transaction->constraint_verification_info->RemovedProperty(vertex);
- }
- }
- }
- }};
+ }
+ }
+ }};
std::invoke(atomic_memory_block);
return id_old_new_change.has_value() ? std::move(id_old_new_change.value()) : ReturnType{};
diff --git a/src/utils/atomic_memory_block.hpp b/src/utils/atomic_memory_block.hpp
index 31a3cf3a9..5ae2aab35 100644
--- a/src/utils/atomic_memory_block.hpp
+++ b/src/utils/atomic_memory_block.hpp
@@ -29,12 +29,10 @@ class [[nodiscard]] AtomicMemoryBlock {
AtomicMemoryBlock &operator=(AtomicMemoryBlock &&) = delete;
~AtomicMemoryBlock() = default;
- void operator()() {
- {
- utils::MemoryTracker::OutOfMemoryExceptionBlocker oom_blocker;
- function_();
- }
- total_memory_tracker.DoCheck();
+ auto operator()() -> std::invoke_result_t<Callable> {
+ auto check_on_exit = OnScopeExit{[&] { total_memory_tracker.DoCheck(); }};
+ utils::MemoryTracker::OutOfMemoryExceptionBlocker oom_blocker;
+ return function_();
}
private:
diff --git a/src/utils/event_counter.cpp b/src/utils/event_counter.cpp
index da396ff56..f457a6762 100644
--- a/src/utils/event_counter.cpp
+++ b/src/utils/event_counter.cpp
@@ -61,6 +61,7 @@
\
M(ActiveLabelIndices, Index, "Number of active label indices in the system.") \
M(ActiveLabelPropertyIndices, Index, "Number of active label property indices in the system.") \
+ M(ActiveTextIndices, Index, "Number of active text indices in the system.") \
\
M(StreamsCreated, Stream, "Number of Streams created.") \
M(MessagesConsumed, Stream, "Number of consumed streamed messages.") \
diff --git a/src/utils/memory.cpp b/src/utils/memory.cpp
index d09f70fc3..6b1f26c11 100644
--- a/src/utils/memory.cpp
+++ b/src/utils/memory.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -150,128 +150,133 @@ void *MonotonicBufferResource::DoAllocate(size_t bytes, size_t alignment) {
namespace impl {
-Pool::Pool(size_t block_size, unsigned char blocks_per_chunk, MemoryResource *memory)
- : blocks_per_chunk_(blocks_per_chunk), block_size_(block_size), chunks_(memory) {}
-
-Pool::~Pool() { MG_ASSERT(chunks_.empty(), "You need to call Release before destruction!"); }
-
-void *Pool::Allocate() {
- auto allocate_block_from_chunk = [this](Chunk *chunk) {
- unsigned char *available_block = chunk->data + (chunk->first_available_block_ix * block_size_);
- // Update free-list pointer (index in our case) by reading "next" from the
- // available_block.
- chunk->first_available_block_ix = *available_block;
- --chunk->blocks_available;
- return available_block;
- };
- if (last_alloc_chunk_ && last_alloc_chunk_->blocks_available > 0U)
- return allocate_block_from_chunk(last_alloc_chunk_);
- // Find a Chunk with available memory.
- for (auto &chunk : chunks_) {
- if (chunk.blocks_available > 0U) {
- last_alloc_chunk_ = &chunk;
- return allocate_block_from_chunk(last_alloc_chunk_);
- }
- }
- // We haven't found a Chunk with available memory, so allocate a new one.
- if (block_size_ > std::numeric_limits<size_t>::max() / blocks_per_chunk_) throw BadAlloc("Allocation size overflow");
- size_t data_size = blocks_per_chunk_ * block_size_;
+Pool::Pool(size_t block_size, unsigned char blocks_per_chunk, MemoryResource *chunk_memory)
+ : blocks_per_chunk_(blocks_per_chunk), block_size_(block_size), chunks_(chunk_memory) {
// Use the next pow2 of block_size_ as alignment, so that we cover alignment
// requests between 1 and block_size_. Users of this class should make sure
// that requested alignment of particular blocks is never greater than the
// block itself.
- size_t alignment = Ceil2(block_size_);
- if (alignment < block_size_) throw BadAlloc("Allocation alignment overflow");
- auto *data = reinterpret_cast<unsigned char *>(GetUpstreamResource()->Allocate(data_size, alignment));
- // Form a free-list of blocks in data.
- for (unsigned char i = 0U; i < blocks_per_chunk_; ++i) {
- *(data + (i * block_size_)) = i + 1U;
- }
- Chunk chunk{data, 0, blocks_per_chunk_};
- // Insert the big block in the sorted position.
- auto it = std::lower_bound(chunks_.begin(), chunks_.end(), chunk,
- [](const auto &a, const auto &b) { return a.data < b.data; });
- try {
- it = chunks_.insert(it, chunk);
- } catch (...) {
- GetUpstreamResource()->Deallocate(data, data_size, alignment);
- throw;
- }
+ if (block_size_ > std::numeric_limits<size_t>::max() / blocks_per_chunk_) throw BadAlloc("Allocation size overflow");
+}
- last_alloc_chunk_ = &*it;
- last_dealloc_chunk_ = &*it;
- return allocate_block_from_chunk(last_alloc_chunk_);
+Pool::~Pool() {
+ if (!chunks_.empty()) {
+ auto *resource = GetUpstreamResource();
+ auto const dataSize = blocks_per_chunk_ * block_size_;
+ auto const alignment = Ceil2(block_size_);
+ for (auto &chunk : chunks_) {
+ resource->Deallocate(chunk.raw_data, dataSize, alignment);
+ }
+ chunks_.clear();
+ }
+ free_list_ = nullptr;
+}
+
+void *Pool::Allocate() {
+ if (!free_list_) [[unlikely]] {
+ // need new chunk
+ auto const data_size = blocks_per_chunk_ * block_size_;
+ auto const alignment = Ceil2(block_size_);
+ auto *resource = GetUpstreamResource();
+ auto *data = reinterpret_cast<std::byte *>(resource->Allocate(data_size, alignment));
+ try {
+ auto &new_chunk = chunks_.emplace_front(data);
+ free_list_ = new_chunk.build_freelist(block_size_, blocks_per_chunk_);
+ } catch (...) {
+ resource->Deallocate(data, data_size, alignment);
+ throw;
+ }
+ }
+ return std::exchange(free_list_, *reinterpret_cast<std::byte **>(free_list_));
}
void Pool::Deallocate(void *p) {
- MG_ASSERT(last_dealloc_chunk_, "No chunk to deallocate");
- MG_ASSERT(!chunks_.empty(),
- "Expected a call to Deallocate after at least a "
- "single Allocate has been done.");
- auto is_in_chunk = [this, p](const Chunk &chunk) {
- auto ptr = reinterpret_cast<uintptr_t>(p);
- size_t data_size = blocks_per_chunk_ * block_size_;
- return reinterpret_cast<uintptr_t>(chunk.data) <= ptr && ptr < reinterpret_cast<uintptr_t>(chunk.data + data_size);
- };
- auto deallocate_block_from_chunk = [this, p](Chunk *chunk) {
- // NOTE: This check is not enough to cover all double-free issues.
- MG_ASSERT(chunk->blocks_available < blocks_per_chunk_,
- "Deallocating more blocks than a chunk can contain, possibly a "
- "double-free situation or we have a bug in the allocator.");
- // Link the block into the free-list
- auto *block = reinterpret_cast<unsigned char *>(p);
- *block = chunk->first_available_block_ix;
- chunk->first_available_block_ix = (block - chunk->data) / block_size_;
- chunk->blocks_available++;
- };
- if (is_in_chunk(*last_dealloc_chunk_)) {
- deallocate_block_from_chunk(last_dealloc_chunk_);
- return;
- }
-
- // Find the chunk which served this allocation
- Chunk chunk{reinterpret_cast<unsigned char *>(p) - blocks_per_chunk_ * block_size_, 0, 0};
- auto it = std::lower_bound(chunks_.begin(), chunks_.end(), chunk,
- [](const auto &a, const auto &b) { return a.data <= b.data; });
- MG_ASSERT(it != chunks_.end(), "Failed deallocation in utils::Pool");
- MG_ASSERT(is_in_chunk(*it), "Failed deallocation in utils::Pool");
-
- // Update last_alloc_chunk_ as well because it now has a free block.
- // Additionally this corresponds with C++ pattern of allocations and
- // deallocations being done in reverse order.
- last_alloc_chunk_ = &*it;
- last_dealloc_chunk_ = &*it;
- deallocate_block_from_chunk(last_dealloc_chunk_);
- // TODO: We could release the Chunk to upstream memory
-}
-
-void Pool::Release() {
- for (auto &chunk : chunks_) {
- size_t data_size = blocks_per_chunk_ * block_size_;
- size_t alignment = Ceil2(block_size_);
- GetUpstreamResource()->Deallocate(chunk.data, data_size, alignment);
- }
- chunks_.clear();
- last_alloc_chunk_ = nullptr;
- last_dealloc_chunk_ = nullptr;
+ *reinterpret_cast<std::byte **>(p) = std::exchange(free_list_, reinterpret_cast<std::byte *>(p));
}
} // namespace impl
-PoolResource::PoolResource(size_t max_blocks_per_chunk, size_t max_block_size, MemoryResource *memory_pools,
- MemoryResource *memory_unpooled)
- : pools_(memory_pools),
- unpooled_(memory_unpooled),
- max_blocks_per_chunk_(std::min(max_blocks_per_chunk, static_cast<size_t>(impl::Pool::MaxBlocksInChunk()))),
- max_block_size_(max_block_size) {
- MG_ASSERT(max_blocks_per_chunk_ > 0U, "Invalid number of blocks per chunk");
- MG_ASSERT(max_block_size_ > 0U, "Invalid size of block");
+struct NullMemoryResourceImpl final : public MemoryResource {
+ NullMemoryResourceImpl() = default;
+ NullMemoryResourceImpl(NullMemoryResourceImpl const &) = default;
+ NullMemoryResourceImpl &operator=(NullMemoryResourceImpl const &) = default;
+ NullMemoryResourceImpl(NullMemoryResourceImpl &&) = default;
+ NullMemoryResourceImpl &operator=(NullMemoryResourceImpl &&) = default;
+ ~NullMemoryResourceImpl() override = default;
+
+ private:
+ void *DoAllocate(size_t /*bytes*/, size_t /*alignment*/) override {
+ throw BadAlloc{"NullMemoryResource doesn't allocate"};
+ }
+ void DoDeallocate(void * /*p*/, size_t /*bytes*/, size_t /*alignment*/) override {
+ throw BadAlloc{"NullMemoryResource doesn't deallocate"};
+ }
+ bool DoIsEqual(MemoryResource const &other) const noexcept override {
+ return dynamic_cast<NullMemoryResourceImpl const *>(&other) != nullptr;
+ }
+};
+
+MemoryResource *NullMemoryResource() noexcept {
+ static auto res = NullMemoryResourceImpl{};
+ return &res;
}
+namespace impl {
+
+/// 1 bit sensitivity test
+static_assert(bin_index<1>(9U) == 0);
+static_assert(bin_index<1>(10U) == 0);
+static_assert(bin_index<1>(11U) == 0);
+static_assert(bin_index<1>(12U) == 0);
+static_assert(bin_index<1>(13U) == 0);
+static_assert(bin_index<1>(14U) == 0);
+static_assert(bin_index<1>(15U) == 0);
+static_assert(bin_index<1>(16U) == 0);
+
+static_assert(bin_index<1>(17U) == 1);
+static_assert(bin_index<1>(18U) == 1);
+static_assert(bin_index<1>(19U) == 1);
+static_assert(bin_index<1>(20U) == 1);
+static_assert(bin_index<1>(21U) == 1);
+static_assert(bin_index<1>(22U) == 1);
+static_assert(bin_index<1>(23U) == 1);
+static_assert(bin_index<1>(24U) == 1);
+static_assert(bin_index<1>(25U) == 1);
+static_assert(bin_index<1>(26U) == 1);
+static_assert(bin_index<1>(27U) == 1);
+static_assert(bin_index<1>(28U) == 1);
+static_assert(bin_index<1>(29U) == 1);
+static_assert(bin_index<1>(30U) == 1);
+static_assert(bin_index<1>(31U) == 1);
+static_assert(bin_index<1>(32U) == 1);
+
+/// 2 bit sensitivity test
+
+static_assert(bin_index<2>(9U) == 0);
+static_assert(bin_index<2>(10U) == 0);
+static_assert(bin_index<2>(11U) == 0);
+static_assert(bin_index<2>(12U) == 0);
+
+static_assert(bin_index<2>(13U) == 1);
+static_assert(bin_index<2>(14U) == 1);
+static_assert(bin_index<2>(15U) == 1);
+static_assert(bin_index<2>(16U) == 1);
+
+static_assert(bin_index<2>(17U) == 2);
+static_assert(bin_index<2>(18U) == 2);
+static_assert(bin_index<2>(19U) == 2);
+static_assert(bin_index<2>(20U) == 2);
+static_assert(bin_index<2>(21U) == 2);
+static_assert(bin_index<2>(22U) == 2);
+static_assert(bin_index<2>(23U) == 2);
+static_assert(bin_index<2>(24U) == 2);
+
+} // namespace impl
+
void *PoolResource::DoAllocate(size_t bytes, size_t alignment) {
// Take the max of `bytes` and `alignment` so that we simplify handling
// alignment requests.
- size_t block_size = std::max(bytes, alignment);
+ size_t block_size = std::max({bytes, alignment, 1UL});
// Check that we have received a regular allocation request with non-padded
// structs/classes in play. These will always have
// `sizeof(T) % alignof(T) == 0`. Special requests which don't have that
@@ -279,80 +284,36 @@ void *PoolResource::DoAllocate(size_t bytes, size_t alignment) {
// have to write a general-purpose allocator which has to behave as complex
// as malloc/free.
if (block_size % alignment != 0) throw BadAlloc("Requested bytes must be a multiple of alignment");
- if (block_size > max_block_size_) {
- // Allocate a big block.
- BigBlock big_block{bytes, alignment, GetUpstreamResourceBlocks()->Allocate(bytes, alignment)};
- // Insert the big block in the sorted position.
- auto it = std::lower_bound(unpooled_.begin(), unpooled_.end(), big_block,
- [](const auto &a, const auto &b) { return a.data < b.data; });
- try {
- unpooled_.insert(it, big_block);
- } catch (...) {
- GetUpstreamResourceBlocks()->Deallocate(big_block.data, bytes, alignment);
- throw;
- }
- return big_block.data;
- }
- // Allocate a regular block, first check if last_alloc_pool_ is suitable.
- if (last_alloc_pool_ && last_alloc_pool_->GetBlockSize() == block_size) {
- return last_alloc_pool_->Allocate();
- }
- // Find the pool with greater or equal block_size.
- impl::Pool pool(block_size, max_blocks_per_chunk_, GetUpstreamResource());
- auto it = std::lower_bound(pools_.begin(), pools_.end(), pool,
- [](const auto &a, const auto &b) { return a.GetBlockSize() < b.GetBlockSize(); });
- if (it != pools_.end() && it->GetBlockSize() == block_size) {
- last_alloc_pool_ = &*it;
- last_dealloc_pool_ = &*it;
- return it->Allocate();
- }
- // We don't have a pool for this block_size, so insert it in the sorted
- // position.
- it = pools_.emplace(it, std::move(pool));
- last_alloc_pool_ = &*it;
- last_dealloc_pool_ = &*it;
- return it->Allocate();
-}
+ if (block_size <= 64) {
+ return mini_pools_[(block_size - 1UL) / 8UL].Allocate();
+ }
+ if (block_size <= 128) {
+ return pools_3bit_.allocate(block_size);
+ }
+ if (block_size <= 512) {
+ return pools_4bit_.allocate(block_size);
+ }
+ if (block_size <= 1024) {
+ return pools_5bit_.allocate(block_size);
+ }
+ return unpooled_memory_->Allocate(bytes, alignment);
+}
void PoolResource::DoDeallocate(void *p, size_t bytes, size_t alignment) {
- size_t block_size = std::max(bytes, alignment);
- MG_ASSERT(block_size % alignment == 0,
- "PoolResource shouldn't serve allocation requests where bytes aren't "
- "a multiple of alignment");
- if (block_size > max_block_size_) {
- // Deallocate a big block.
- BigBlock big_block{bytes, alignment, p};
- auto it = std::lower_bound(unpooled_.begin(), unpooled_.end(), big_block,
- [](const auto &a, const auto &b) { return a.data < b.data; });
- MG_ASSERT(it != unpooled_.end(), "Failed deallocation");
- MG_ASSERT(it->data == p && it->bytes == bytes && it->alignment == alignment, "Failed deallocation");
- unpooled_.erase(it);
- GetUpstreamResourceBlocks()->Deallocate(p, bytes, alignment);
- return;
+ size_t block_size = std::max({bytes, alignment, 1UL});
+ DMG_ASSERT(block_size % alignment == 0);
+
+ if (block_size <= 64) {
+ mini_pools_[(block_size - 1UL) / 8UL].Deallocate(p);
+ } else if (block_size <= 128) {
+ pools_3bit_.deallocate(p, block_size);
+ } else if (block_size <= 512) {
+ pools_4bit_.deallocate(p, block_size);
+ } else if (block_size <= 1024) {
+ pools_5bit_.deallocate(p, block_size);
+ } else {
+ unpooled_memory_->Deallocate(p, bytes, alignment);
}
- // Deallocate a regular block, first check if last_dealloc_pool_ is suitable.
- if (last_dealloc_pool_ && last_dealloc_pool_->GetBlockSize() == block_size) return last_dealloc_pool_->Deallocate(p);
- // Find the pool with equal block_size.
- impl::Pool pool(block_size, max_blocks_per_chunk_, GetUpstreamResource());
- auto it = std::lower_bound(pools_.begin(), pools_.end(), pool,
- [](const auto &a, const auto &b) { return a.GetBlockSize() < b.GetBlockSize(); });
- MG_ASSERT(it != pools_.end(), "Failed deallocation");
- MG_ASSERT(it->GetBlockSize() == block_size, "Failed deallocation");
- last_alloc_pool_ = &*it;
- last_dealloc_pool_ = &*it;
- return it->Deallocate(p);
}
-
-void PoolResource::Release() {
- for (auto &pool : pools_) pool.Release();
- pools_.clear();
- for (auto &big_block : unpooled_)
- GetUpstreamResourceBlocks()->Deallocate(big_block.data, big_block.bytes, big_block.alignment);
- unpooled_.clear();
- last_alloc_pool_ = nullptr;
- last_dealloc_pool_ = nullptr;
-}
-
-// PoolResource END
-
+bool PoolResource::DoIsEqual(MemoryResource const &other) const noexcept { return this == &other; }
} // namespace memgraph::utils
diff --git a/src/utils/memory.hpp b/src/utils/memory.hpp
index 225a3b6a1..8ff6c3523 100644
--- a/src/utils/memory.hpp
+++ b/src/utils/memory.hpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -15,7 +15,11 @@
#pragma once
+#include <climits>
#include <cstddef>
+#include <cstdint>
+#include <forward_list>
+#include <list>
#include <memory>
#include <mutex>
#include <new>
@@ -248,6 +252,8 @@ bool operator!=(const Allocator<T> &a, const Allocator<U> &b) {
return !(a == b);
}
+auto NullMemoryResource() noexcept -> MemoryResource *;
+
/// Wraps std::pmr::memory_resource for use with out MemoryResource
class StdMemoryResource final : public MemoryResource {
public:
@@ -380,37 +386,45 @@ class MonotonicBufferResource final : public MemoryResource {
namespace impl {
+template <class T>
+using AList = std::forward_list<T, Allocator<T>>;
+
template <class T>
using AVector = std::vector<T, Allocator<T>>;
/// Holds a number of Chunks each serving blocks of particular size. When a
-/// Chunk runs out of available blocks, a new Chunk is allocated. The naming is
-/// taken from `libstdc++` implementation, but the implementation details are
-/// more similar to `FixedAllocator` described in "Small Object Allocation" from
-/// "Modern C++ Design".
+/// Chunk runs out of available blocks, a new Chunk is allocated.
class Pool final {
/// Holds a pointer into a chunk of memory which consists of equal sized
- /// blocks. Each Chunk can handle `std::numeric_limits<unsigned char>::max()`
- /// number of blocks. Blocks form a "free-list", where each unused block has
- /// an embedded index to the next unused block.
+ /// blocks. Blocks form a "free-list"
struct Chunk {
- unsigned char *data;
- unsigned char first_available_block_ix;
- unsigned char blocks_available;
+ // TODO: make blocks_per_chunk a per chunk thing (ie. allow chunk growth)
+ std::byte *raw_data;
+ explicit Chunk(std::byte *rawData) : raw_data(rawData) {}
+ std::byte *build_freelist(std::size_t block_size, std::size_t blocks_in_chunk) {
+ auto current = raw_data;
+ std::byte *prev = nullptr;
+ auto end = current + (blocks_in_chunk * block_size);
+ while (current != end) {
+ std::byte **list_entry = reinterpret_cast<std::byte **>(current);
+ *list_entry = std::exchange(prev, current);
+ current += block_size;
+ }
+ DMG_ASSERT(prev != nullptr);
+ return prev;
+ }
};
- unsigned char blocks_per_chunk_;
- size_t block_size_;
- AVector<Chunk> chunks_;
- Chunk *last_alloc_chunk_{nullptr};
- Chunk *last_dealloc_chunk_{nullptr};
+ std::byte *free_list_{nullptr};
+ uint8_t blocks_per_chunk_{};
+ std::size_t block_size_{};
+
+ AList<Chunk> chunks_; // TODO: do ourself so we can do fast Release (detect monotonic, do nothing)
public:
- static constexpr auto MaxBlocksInChunk() {
- return std::numeric_limits<decltype(Chunk::first_available_block_ix)>::max();
- }
+ static constexpr auto MaxBlocksInChunk = std::numeric_limits<decltype(blocks_per_chunk_)>::max();
- Pool(size_t block_size, unsigned char blocks_per_chunk, MemoryResource *memory);
+ Pool(size_t block_size, unsigned char blocks_per_chunk, MemoryResource *chunk_memory);
Pool(const Pool &) = delete;
Pool &operator=(const Pool &) = delete;
@@ -430,8 +444,147 @@ class Pool final {
void *Allocate();
void Deallocate(void *p);
+};
- void Release();
+// C++ overloads for clz
+constexpr auto clz(unsigned int x) { return __builtin_clz(x); }
+constexpr auto clz(unsigned long x) { return __builtin_clzl(x); }
+constexpr auto clz(unsigned long long x) { return __builtin_clzll(x); }
+
+template <typename T>
+constexpr auto bits_sizeof = sizeof(T) * CHAR_BIT;
+
+/// 0-based bit index of the most significant bit assumed that `n` != 0
+template <typename T>
+constexpr auto msb_index(T n) {
+ return bits_sizeof<T> - clz(n) - T(1);
+}
+
+/* This function will in O(1) time provide a bin index based on:
+ * B - the number of most significant bits to be sensitive to
+ * LB - the value that should be considered below the consideration for bin index of 0 (LB is exclusive)
+ *
+ * lets say we were:
+ * - sensitive to two bits (B == 2)
+ * - lowest bin is for 8 (LB == 8)
+ *
+ * our bin indexes would look like:
+ * 0 - 0000'1100 12
+ * 1 - 0001'0000 16
+ * 2 - 0001'1000 24
+ * 3 - 0010'0000 32
+ * 4 - 0011'0000 48
+ * 5 - 0100'0000 64
+ * 6 - 0110'0000 96
+ * 7 - 1000'0000 128
+ * 8 - 1100'0000 192
+ * ...
+ *
+ * Example:
+ * Given n == 70, we want to return the bin index to the first value which is
+ * larger than n.
+ * bin_index<2,8>(70) => 6, as 64 (index 5) < 70 and 70 <= 96 (index 6)
+ */
+template <std::size_t B = 2, std::size_t LB = 8>
+constexpr std::size_t bin_index(std::size_t n) {
+ static_assert(B >= 1U, "Needs to be sensitive to at least one bit");
+ static_assert(LB != 0U, "Lower bound need to be non-zero");
+ DMG_ASSERT(n > LB);
+
+ // We will alway be sensitive to at least the MSB
+ // exponent tells us how many bits we need to use to select within a level
+ constexpr auto kExponent = B - 1U;
+ // 2^exponent gives the size of each level
+ constexpr auto kSize = 1U << kExponent;
+ // offset help adjust results down to be inline with bin_index(LB) == 0
+ constexpr auto kOffset = msb_index(LB);
+
+ auto const msb_idx = msb_index(n);
+ DMG_ASSERT(msb_idx != 0);
+
+ auto const mask = (1u << msb_idx) - 1u;
+ auto const under = n & mask;
+ auto const selector = under >> (msb_idx - kExponent);
+
+ auto const rest = under & (mask >> kExponent);
+ auto const no_overflow = rest == 0U;
+
+ auto const msb_level = kSize * (msb_idx - kOffset);
+ return msb_level + selector - no_overflow;
+}
+
+// This is the inverse opperation for bin_index
+// bin_size(bin_index(X)-1) < X <= bin_size(bin_index(X))
+template <std::size_t B = 2, std::size_t LB = 8>
+std::size_t bin_size(std::size_t idx) {
+ constexpr auto kExponent = B - 1U;
+ constexpr auto kSize = 1U << kExponent;
+ constexpr auto kOffset = msb_index(LB);
+
+ // no need to optimise `/` or `%` compiler can see `kSize` is a power of 2
+ auto const level = (idx + 1) / kSize;
+ auto const sub_level = (idx + 1) % kSize;
+ return (1U << (level + kOffset)) | (sub_level << (level + kOffset - kExponent));
+}
+
+template <std::size_t Bits, std::size_t LB, std::size_t UB>
+struct MultiPool {
+ static_assert(LB < UB, "lower bound must be less than upper bound");
+ static_assert(IsPow2(LB) && IsPow2(UB), "Design untested for non powers of 2");
+ static_assert((LB << Bits) % sizeof(void *) == 0, "Smallest pool must have space and alignment for freelist");
+
+ // upper bound is inclusive
+ static bool is_size_handled(std::size_t size) { return LB < size && size <= UB; }
+ static bool is_above_upper_bound(std::size_t size) { return UB < size; }
+
+ static constexpr auto n_bins = bin_index<Bits, LB>(UB) + 1U;
+
+ MultiPool(uint8_t blocks_per_chunk, MemoryResource *memory, MemoryResource *internal_memory)
+ : blocks_per_chunk_{blocks_per_chunk}, memory_{memory}, internal_memory_{internal_memory} {}
+
+ ~MultiPool() {
+ if (pools_) {
+ auto pool_alloc = Allocator<Pool>(internal_memory_);
+ for (auto i = 0U; i != n_bins; ++i) {
+ pool_alloc.destroy(&pools_[i]);
+ }
+ pool_alloc.deallocate(pools_, n_bins);
+ }
+ }
+
+ void *allocate(std::size_t bytes) {
+ auto idx = bin_index<Bits, LB>(bytes);
+ if (!pools_) [[unlikely]] {
+ initialise_pools();
+ }
+ return pools_[idx].Allocate();
+ }
+
+ void deallocate(void *ptr, std::size_t bytes) {
+ auto idx = bin_index<Bits, LB>(bytes);
+ pools_[idx].Deallocate(ptr);
+ }
+
+ private:
+ void initialise_pools() {
+ auto pool_alloc = Allocator<Pool>(internal_memory_);
+ auto pools = pool_alloc.allocate(n_bins);
+ try {
+ for (auto i = 0U; i != n_bins; ++i) {
+ auto block_size = bin_size<Bits, LB>(i);
+ pool_alloc.construct(&pools[i], block_size, blocks_per_chunk_, memory_);
+ }
+ pools_ = pools;
+ } catch (...) {
+ pool_alloc.deallocate(pools, n_bins);
+ throw;
+ }
+ }
+
+ Pool *pools_{};
+ uint8_t blocks_per_chunk_{};
+ MemoryResource *memory_{};
+ MemoryResource *internal_memory_{};
};
} // namespace impl
@@ -442,8 +595,6 @@ class Pool final {
///
/// This class has the following properties with regards to memory management.
///
-/// * All allocated memory will be freed upon destruction, even if Deallocate
-/// has not been called for some of the allocated blocks.
/// * It consists of a collection of impl::Pool instances, each serving
/// requests for different block sizes. Each impl::Pool manages a collection
/// of impl::Pool::Chunk instances which are divided into blocks of uniform
@@ -452,91 +603,46 @@ class Pool final {
/// arbitrary alignment requests. Each requested block size must be a
/// multiple of alignment or smaller than the alignment value.
/// * An allocation request within the limits of the maximum block size will
-/// find a Pool serving the requested size. If there's no Pool serving such
-/// a request, a new one is instantiated.
+/// find a Pool serving the requested size. Some requests will share a larger
+/// pool size.
/// * When a Pool exhausts its Chunk, a new one is allocated with the size for
/// the maximum number of blocks.
/// * Allocation requests which exceed the maximum block size will be
/// forwarded to upstream MemoryResource.
-/// * Maximum block size and maximum number of blocks per chunk can be tuned
-/// by passing the arguments to the constructor.
+/// * Maximum number of blocks per chunk can be tuned by passing the
+/// arguments to the constructor.
+
class PoolResource final : public MemoryResource {
public:
- /// Construct with given max_blocks_per_chunk, max_block_size and upstream
- /// memory.
- ///
- /// The implementation will use std::min(max_blocks_per_chunk,
- /// impl::Pool::MaxBlocksInChunk()) as the real maximum number of blocks per
- /// chunk. Allocation requests exceeding max_block_size are simply forwarded
- /// to upstream memory.
- PoolResource(size_t max_blocks_per_chunk, size_t max_block_size, MemoryResource *memory_pools = NewDeleteResource(),
- MemoryResource *memory_unpooled = NewDeleteResource());
-
- PoolResource(const PoolResource &) = delete;
- PoolResource &operator=(const PoolResource &) = delete;
-
- PoolResource(PoolResource &&) = default;
- PoolResource &operator=(PoolResource &&) = default;
-
- ~PoolResource() override { Release(); }
-
- MemoryResource *GetUpstreamResource() const { return pools_.get_allocator().GetMemoryResource(); }
- MemoryResource *GetUpstreamResourceBlocks() const { return unpooled_.get_allocator().GetMemoryResource(); }
-
- /// Release all allocated memory.
- void Release();
+ PoolResource(uint8_t blocks_per_chunk, MemoryResource *memory = NewDeleteResource(),
+ MemoryResource *internal_memory = NewDeleteResource())
+ : mini_pools_{
+ impl::Pool{8, blocks_per_chunk, memory},
+ impl::Pool{16, blocks_per_chunk, memory},
+ impl::Pool{24, blocks_per_chunk, memory},
+ impl::Pool{32, blocks_per_chunk, memory},
+ impl::Pool{40, blocks_per_chunk, memory},
+ impl::Pool{48, blocks_per_chunk, memory},
+ impl::Pool{56, blocks_per_chunk, memory},
+ impl::Pool{64, blocks_per_chunk, memory},
+ },
+ pools_3bit_(blocks_per_chunk, memory, internal_memory),
+ pools_4bit_(blocks_per_chunk, memory, internal_memory),
+ pools_5bit_(blocks_per_chunk, memory, internal_memory),
+ unpooled_memory_{internal_memory} {}
+ ~PoolResource() override = default;
private:
- // Big block larger than max_block_size_, doesn't go into a pool.
- struct BigBlock {
- size_t bytes;
- size_t alignment;
- void *data;
- };
-
- // TODO: Potential memory optimization is replacing `std::vector` with our
- // custom vector implementation which doesn't store a `MemoryResource *`.
- // Currently we have vectors for `pools_` and `unpooled_`, as well as each
- // `impl::Pool` stores a `chunks_` vector.
-
- // Pools are sorted by bound_size_, ascending.
- impl::AVector<impl::Pool> pools_;
- impl::Pool *last_alloc_pool_{nullptr};
- impl::Pool *last_dealloc_pool_{nullptr};
- // Unpooled BigBlocks are sorted by data pointer.
- impl::AVector<BigBlock> unpooled_;
- size_t max_blocks_per_chunk_;
- size_t max_block_size_;
-
void *DoAllocate(size_t bytes, size_t alignment) override;
-
void DoDeallocate(void *p, size_t bytes, size_t alignment) override;
-
- bool DoIsEqual(const MemoryResource &other) const noexcept override { return this == &other; }
-};
-
-/// Like PoolResource but uses SpinLock for thread safe usage.
-class SynchronizedPoolResource final : public MemoryResource {
- public:
- SynchronizedPoolResource(size_t max_blocks_per_chunk, size_t max_block_size,
- MemoryResource *memory = NewDeleteResource())
- : pool_memory_(max_blocks_per_chunk, max_block_size, memory) {}
+ bool DoIsEqual(MemoryResource const &other) const noexcept override;
private:
- PoolResource pool_memory_;
- SpinLock lock_;
-
- void *DoAllocate(size_t bytes, size_t alignment) override {
- std::lock_guard<SpinLock> guard(lock_);
- return pool_memory_.Allocate(bytes, alignment);
- }
-
- void DoDeallocate(void *p, size_t bytes, size_t alignment) override {
- std::lock_guard<SpinLock> guard(lock_);
- pool_memory_.Deallocate(p, bytes, alignment);
- }
-
- bool DoIsEqual(const MemoryResource &other) const noexcept override { return this == &other; }
+ std::array<impl::Pool, 8> mini_pools_;
+ impl::MultiPool<3, 64, 128> pools_3bit_;
+ impl::MultiPool<4, 128, 512> pools_4bit_;
+ impl::MultiPool<5, 512, 1024> pools_5bit_;
+ MemoryResource *unpooled_memory_;
};
class MemoryTrackingResource final : public utils::MemoryResource {
diff --git a/src/utils/on_scope_exit.hpp b/src/utils/on_scope_exit.hpp
index a5398b017..114f1c370 100644
--- a/src/utils/on_scope_exit.hpp
+++ b/src/utils/on_scope_exit.hpp
@@ -35,7 +35,7 @@ namespace memgraph::utils {
* // long block of code, might throw an exception
* }
*/
-template <typename Callable>
+template <std::invocable Callable>
class [[nodiscard]] OnScopeExit {
public:
template <typename U>
@@ -46,7 +46,7 @@ class [[nodiscard]] OnScopeExit {
OnScopeExit &operator=(OnScopeExit const &) = delete;
OnScopeExit &operator=(OnScopeExit &&) = delete;
~OnScopeExit() {
- if (doCall_) function_();
+ if (doCall_) std::invoke(std::move(function_));
}
void Disable() { doCall_ = false; }
@@ -57,5 +57,4 @@ class [[nodiscard]] OnScopeExit {
};
template <typename Callable>
OnScopeExit(Callable &&) -> OnScopeExit<Callable>;
-
} // namespace memgraph::utils
diff --git a/src/utils/tag.hpp b/src/utils/tag.hpp
new file mode 100644
index 000000000..dfd8c8f81
--- /dev/null
+++ b/src/utils/tag.hpp
@@ -0,0 +1,32 @@
+// Copyright 2024 Memgraph Ltd.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+// License, and you may not use this file except in compliance with the Business Source License.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+#pragma once
+
+namespace memgraph::utils {
+
+template <typename T>
+struct tag_type {
+ using type = T;
+};
+
+template <auto V>
+struct tag_value {
+ static constexpr auto value = V;
+};
+
+template <typename T>
+auto tag_t = tag_type<T>{};
+
+template <auto V>
+auto tag_v = tag_value<V>{};
+
+} // namespace memgraph::utils
diff --git a/src/utils/typeinfo.hpp b/src/utils/typeinfo.hpp
index 77910f731..aeb62d2c1 100644
--- a/src/utils/typeinfo.hpp
+++ b/src/utils/typeinfo.hpp
@@ -187,6 +187,7 @@ enum class TypeId : uint64_t {
AST_PROFILE_QUERY,
AST_INDEX_QUERY,
AST_EDGE_INDEX_QUERY,
+ AST_TEXT_INDEX_QUERY,
AST_CREATE,
AST_CALL_PROCEDURE,
AST_MATCH,
diff --git a/tests/benchmark/query/execution.cpp b/tests/benchmark/query/execution.cpp
index d49b14fc3..1d65cdb93 100644
--- a/tests/benchmark/query/execution.cpp
+++ b/tests/benchmark/query/execution.cpp
@@ -55,12 +55,12 @@ class NewDeleteResource final {
};
class PoolResource final {
- memgraph::utils::PoolResource memory_{128, 4 * 1024};
+ memgraph::utils::PoolResource memory_{128};
public:
memgraph::utils::MemoryResource *get() { return &memory_; }
- void Reset() { memory_.Release(); }
+ void Reset() {}
};
static void AddVertices(memgraph::storage::Storage *db, int vertex_count) {
diff --git a/tests/benchmark/skip_list_vs_stl.cpp b/tests/benchmark/skip_list_vs_stl.cpp
index 1a17e56e1..9a856822f 100644
--- a/tests/benchmark/skip_list_vs_stl.cpp
+++ b/tests/benchmark/skip_list_vs_stl.cpp
@@ -1,4 +1,4 @@
-// Copyright 2022 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -101,8 +101,7 @@ class StdSetWithPoolAllocatorInsertFixture : public benchmark::Fixture {
}
protected:
- memgraph::utils::PoolResource memory_{256U /* max_blocks_per_chunk */, 1024U /* max_block_size */,
- memgraph::utils::NewDeleteResource()};
+ memgraph::utils::PoolResource memory_{128U /* max_blocks_per_chunk */, memgraph::utils::NewDeleteResource()};
std::set<uint64_t, std::less<>, memgraph::utils::Allocator<uint64_t>> container{&memory_};
memgraph::utils::SpinLock lock;
};
@@ -208,8 +207,7 @@ class StdSetWithPoolAllocatorFindFixture : public benchmark::Fixture {
}
protected:
- memgraph::utils::PoolResource memory_{256U /* max_blocks_per_chunk */, 1024U /* max_block_size */,
- memgraph::utils::NewDeleteResource()};
+ memgraph::utils::PoolResource memory_{128U /* max_blocks_per_chunk */, memgraph::utils::NewDeleteResource()};
std::set<uint64_t, std::less<>, memgraph::utils::Allocator<uint64_t>> container{&memory_};
memgraph::utils::SpinLock lock;
};
@@ -325,8 +323,7 @@ class StdMapWithPoolAllocatorInsertFixture : public benchmark::Fixture {
}
protected:
- memgraph::utils::PoolResource memory_{256U /* max_blocks_per_chunk */, 1024U /* max_block_size */,
- memgraph::utils::NewDeleteResource()};
+ memgraph::utils::PoolResource memory_{128U /* max_blocks_per_chunk */, memgraph::utils::NewDeleteResource()};
std::map<uint64_t, uint64_t, std::less<>, memgraph::utils::Allocator<std::pair<const uint64_t, uint64_t>>> container{
&memory_};
memgraph::utils::SpinLock lock;
@@ -433,8 +430,7 @@ class StdMapWithPoolAllocatorFindFixture : public benchmark::Fixture {
}
protected:
- memgraph::utils::PoolResource memory_{256U /* max_blocks_per_chunk */, 1024U /* max_block_size */,
- memgraph::utils::NewDeleteResource()};
+ memgraph::utils::PoolResource memory_{128U /* max_blocks_per_chunk */, memgraph::utils::NewDeleteResource()};
std::map<uint64_t, uint64_t, std::less<>, memgraph::utils::Allocator<std::pair<const uint64_t, uint64_t>>> container{
&memory_};
memgraph::utils::SpinLock lock;
diff --git a/tests/code_analysis/clang_tidy.sh b/tests/code_analysis/clang_tidy.sh
new file mode 100755
index 000000000..f8bdfc252
--- /dev/null
+++ b/tests/code_analysis/clang_tidy.sh
@@ -0,0 +1,23 @@
+#!/bin/bash
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+PROJECT_ROOT="$SCRIPT_DIR/../.."
+BASE_BRANCH="origin/master"
+THREADS=${THREADS:-$(nproc)}
+
+if [[ "$#" -gt 0 ]]; then
+ case "$1" in
+ --base-branch)
+ BASE_BRANCH=$2
+ ;;
+ *)
+ echo "Error: Unknown flag '$1'"
+ exit 1
+ ;;
+ esac
+fi
+
+cd $PROJECT_ROOT
+git diff -U0 $BASE_BRANCH... -- src | ./tools/github/clang-tidy/clang-tidy-diff.py -p 1 -j $THREADS -path build -regex ".+\.cpp" | tee ./build/clang_tidy_output.txt
+# Fail if any warning is reported
+! cat ./build/clang_tidy_output.txt | ./tools/github/clang-tidy/grep_error_lines.sh > /dev/null
+cd $SCRIPT_DIR
diff --git a/tests/code_analysis/python_code_analysis.sh b/tests/code_analysis/python_code_analysis.sh
new file mode 100755
index 000000000..500a873d1
--- /dev/null
+++ b/tests/code_analysis/python_code_analysis.sh
@@ -0,0 +1,27 @@
+#!/bin/bash
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+PROJECT_ROOT="$SCRIPT_DIR/../.."
+BASE_BRANCH="origin/master"
+
+if [[ "$#" -gt 0 ]]; then
+ case "$1" in
+ --base-branch)
+ BASE_BRANCH=$2
+ ;;
+ *)
+ echo "Error: Unknown flag '$1'"
+ exit 1
+ ;;
+ esac
+fi
+
+cd $PROJECT_ROOT
+CHANGED_FILES=$(git diff -U0 $BASE_BRANCH... --name-only --diff-filter=d)
+for file in ${CHANGED_FILES}; do
+ echo ${file}
+ if [[ ${file} == *.py ]]; then
+ python3 -m black --check --diff ${file}
+ python3 -m isort --profile black --check-only --diff ${file}
+ fi
+done
+cd $SCRIPT_DIR
diff --git a/tests/e2e/CMakeLists.txt b/tests/e2e/CMakeLists.txt
index 1876074ee..60743676d 100644
--- a/tests/e2e/CMakeLists.txt
+++ b/tests/e2e/CMakeLists.txt
@@ -77,6 +77,7 @@ add_subdirectory(garbage_collection)
add_subdirectory(query_planning)
add_subdirectory(awesome_functions)
add_subdirectory(high_availability)
+add_subdirectory(concurrency)
add_subdirectory(replication_experimental)
diff --git a/tests/e2e/concurrency/CMakeLists.txt b/tests/e2e/concurrency/CMakeLists.txt
new file mode 100644
index 000000000..f981a2537
--- /dev/null
+++ b/tests/e2e/concurrency/CMakeLists.txt
@@ -0,0 +1,6 @@
+function(copy_concurrency_e2e_python_files FILE_NAME)
+ copy_e2e_python_files(concurrency ${FILE_NAME})
+endfunction()
+
+copy_concurrency_e2e_python_files(common.py)
+copy_concurrency_e2e_python_files(concurrency.py)
diff --git a/tests/e2e/concurrency/common.py b/tests/e2e/concurrency/common.py
new file mode 100644
index 000000000..208278929
--- /dev/null
+++ b/tests/e2e/concurrency/common.py
@@ -0,0 +1,60 @@
+# Copyright 2023 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import typing
+
+import mgclient
+import pytest
+
+
+def execute_and_fetch_all(cursor: mgclient.Cursor, query: str, params: dict = {}) -> typing.List[tuple]:
+ cursor.execute(query, params)
+ return cursor.fetchall()
+
+
+def execute_and_fetch_all_with_commit(
+ connection: mgclient.Connection, query: str, params: dict = {}
+) -> typing.List[tuple]:
+ cursor = connection.cursor()
+ cursor.execute(query, params)
+ results = cursor.fetchall()
+ connection.commit()
+ return results
+
+
+@pytest.fixture
+def first_connection(**kwargs) -> mgclient.Connection:
+ connection = mgclient.connect(host="localhost", port=7687, **kwargs)
+ connection.autocommit = True
+ cursor = connection.cursor()
+ execute_and_fetch_all(cursor, "USE DATABASE memgraph")
+ try:
+ execute_and_fetch_all(cursor, "DROP DATABASE clean")
+ except:
+ pass
+ execute_and_fetch_all(cursor, "MATCH (n) DETACH DELETE n")
+ connection.autocommit = False
+ yield connection
+
+
+@pytest.fixture
+def second_connection(**kwargs) -> mgclient.Connection:
+ connection = mgclient.connect(host="localhost", port=7687, **kwargs)
+ connection.autocommit = True
+ cursor = connection.cursor()
+ execute_and_fetch_all(cursor, "USE DATABASE memgraph")
+ try:
+ execute_and_fetch_all(cursor, "DROP DATABASE clean")
+ except:
+ pass
+ execute_and_fetch_all(cursor, "MATCH (n) DETACH DELETE n")
+ connection.autocommit = False
+ yield connection
diff --git a/tests/e2e/concurrency/concurrency.py b/tests/e2e/concurrency/concurrency.py
new file mode 100644
index 000000000..7961c1984
--- /dev/null
+++ b/tests/e2e/concurrency/concurrency.py
@@ -0,0 +1,57 @@
+# Copyright 2023 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import sys
+
+import pytest
+from common import execute_and_fetch_all, first_connection, second_connection
+
+
+def test_concurrency_if_no_delta_on_same_node_property_update(first_connection, second_connection):
+ m1c = first_connection.cursor()
+ m2c = second_connection.cursor()
+
+ execute_and_fetch_all(m1c, "CREATE (:Node {prop: 1})")
+ first_connection.commit()
+
+ test_has_error = False
+ try:
+ m1c.execute("MATCH (n) SET n.prop = 1")
+ m2c.execute("MATCH (n) SET n.prop = 1")
+ first_connection.commit()
+ second_connection.commit()
+ except Exception as e:
+ test_has_error = True
+
+ assert test_has_error is False
+
+
+def test_concurrency_if_no_delta_on_same_edge_property_update(first_connection, second_connection):
+ m1c = first_connection.cursor()
+ m2c = second_connection.cursor()
+
+ execute_and_fetch_all(m1c, "CREATE ()-[:TYPE {prop: 1}]->()")
+ first_connection.commit()
+
+ test_has_error = False
+ try:
+ m1c.execute("MATCH (n)-[r]->(m) SET r.prop = 1")
+ m2c.execute("MATCH (n)-[r]->(m) SET n.prop = 1")
+ first_connection.commit()
+ second_connection.commit()
+ except Exception as e:
+ test_has_error = True
+
+ assert test_has_error is False
+
+
+if __name__ == "__main__":
+ sys.exit(pytest.main([__file__, "-rA"]))
diff --git a/tests/e2e/concurrency/workloads.yaml b/tests/e2e/concurrency/workloads.yaml
new file mode 100644
index 000000000..839090538
--- /dev/null
+++ b/tests/e2e/concurrency/workloads.yaml
@@ -0,0 +1,14 @@
+concurrency_cluster: &concurrency_cluster
+ cluster:
+ main:
+ args: ["--bolt-port", "7687", "--log-level=TRACE", "--storage-delta-on-identical-property-update=false"]
+ log_file: "concurrency.log"
+ setup_queries: []
+ validation_queries: []
+
+
+workloads:
+ - name: "Concurrency"
+ binary: "tests/e2e/pytest_runner.sh"
+ args: ["concurrency/concurrency.py"]
+ <<: *concurrency_cluster
diff --git a/tests/e2e/configuration/default_config.py b/tests/e2e/configuration/default_config.py
index 65a850f0b..11435da65 100644
--- a/tests/e2e/configuration/default_config.py
+++ b/tests/e2e/configuration/default_config.py
@@ -141,6 +141,11 @@ startup_config_dict = {
"1",
"The time duration between two replica checks/pings. If < 1, replicas will NOT be checked at all. NOTE: The MAIN instance allocates a new thread for each REPLICA.",
),
+ "storage_delta_on_identical_property_update": (
+ "true",
+ "true",
+ "Controls whether updating a property with the same value should create a delta object.",
+ ),
"storage_gc_cycle_sec": ("30", "30", "Storage garbage collector interval (in seconds)."),
"storage_python_gc_cycle_sec": ("180", "180", "Storage python full garbage collection interval (in seconds)."),
"storage_items_per_batch": (
@@ -221,6 +226,6 @@ startup_config_dict = {
"experimental_enabled": (
"",
"",
- "Experimental features to be used, comma seperated. Options [system-replication, high-availability]",
+ "Experimental features to be used, comma-separated. Options [system-replication, text-search, high-availability]",
),
}
diff --git a/tests/e2e/memory/workloads.yaml b/tests/e2e/memory/workloads.yaml
index bf29e484c..c043e03d8 100644
--- a/tests/e2e/memory/workloads.yaml
+++ b/tests/e2e/memory/workloads.yaml
@@ -52,26 +52,26 @@ in_memory_query_limit_cluster: &in_memory_query_limit_cluster
setup_queries: []
validation_queries: []
-args_450_MiB_limit: &args_450_MiB_limit
+args_350_MiB_limit: &args_350_MiB_limit
- "--bolt-port"
- *bolt_port
- - "--memory-limit=450"
+ - "--memory-limit=350"
- "--storage-gc-cycle-sec=180"
- "--log-level=INFO"
-in_memory_450_MiB_limit_cluster: &in_memory_450_MiB_limit_cluster
+in_memory_350_MiB_limit_cluster: &in_memory_350_MiB_limit_cluster
cluster:
main:
- args: *args_450_MiB_limit
+ args: *args_350_MiB_limit
log_file: "memory-e2e.log"
setup_queries: []
validation_queries: []
-disk_450_MiB_limit_cluster: &disk_450_MiB_limit_cluster
+disk_350_MiB_limit_cluster: &disk_350_MiB_limit_cluster
cluster:
main:
- args: *args_450_MiB_limit
+ args: *args_350_MiB_limit
log_file: "memory-e2e.log"
setup_queries: []
validation_queries: []
@@ -192,22 +192,22 @@ workloads:
- name: "Memory control for accumulation"
binary: "tests/e2e/memory/memgraph__e2e__memory__limit_accumulation"
args: ["--bolt-port", *bolt_port]
- <<: *in_memory_450_MiB_limit_cluster
+ <<: *in_memory_350_MiB_limit_cluster
- name: "Memory control for accumulation on disk storage"
binary: "tests/e2e/memory/memgraph__e2e__memory__limit_accumulation"
args: ["--bolt-port", *bolt_port]
- <<: *disk_450_MiB_limit_cluster
+ <<: *disk_350_MiB_limit_cluster
- name: "Memory control for edge create"
binary: "tests/e2e/memory/memgraph__e2e__memory__limit_edge_create"
args: ["--bolt-port", *bolt_port]
- <<: *in_memory_450_MiB_limit_cluster
+ <<: *in_memory_350_MiB_limit_cluster
- name: "Memory control for edge create on disk storage"
binary: "tests/e2e/memory/memgraph__e2e__memory__limit_edge_create"
args: ["--bolt-port", *bolt_port]
- <<: *disk_450_MiB_limit_cluster
+ <<: *disk_350_MiB_limit_cluster
- name: "Memory control for memory limit global thread alloc"
binary: "tests/e2e/memory/memgraph__e2e__memory_limit_global_thread_alloc_proc"
diff --git a/tests/e2e/module_file_manager/module_file_manager.cpp b/tests/e2e/module_file_manager/module_file_manager.cpp
index 20641b9d7..73508d81b 100644
--- a/tests/e2e/module_file_manager/module_file_manager.cpp
+++ b/tests/e2e/module_file_manager/module_file_manager.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -126,7 +126,7 @@ std::filesystem::path CreateModuleFile(auto &client, const std::string_view file
}
std::string GetModuleFileQuery(const std::filesystem::path &path) {
- return fmt::format("CALL mg.get_module_file({}) YIELD content", path);
+ return fmt::format("CALL mg.get_module_file('{}') YIELD content", path);
}
std::string GetModuleFile(auto &client, const std::filesystem::path &path) {
@@ -141,7 +141,7 @@ std::string GetModuleFile(auto &client, const std::filesystem::path &path) {
}
std::string UpdateModuleFileQuery(const std::filesystem::path &path, const std::string_view content) {
- return fmt::format("CALL mg.update_module_file({}, '{}')", path, content);
+ return fmt::format("CALL mg.update_module_file('{}', '{}')", path, content);
}
void UpdateModuleFile(auto &client, const std::filesystem::path &path, const std::string_view content) {
@@ -151,7 +151,7 @@ void UpdateModuleFile(auto &client, const std::filesystem::path &path, const std
}
std::string DeleteModuleFileQuery(const std::filesystem::path &path) {
- return fmt::format("CALL mg.delete_module_file({})", path);
+ return fmt::format("CALL mg.delete_module_file('{}')", path);
}
void DeleteModuleFile(auto &client, const std::filesystem::path &path) {
diff --git a/tests/e2e/run.sh b/tests/e2e/run.sh
index 88b70ae32..9ee4babe9 100755
--- a/tests/e2e/run.sh
+++ b/tests/e2e/run.sh
@@ -25,7 +25,7 @@ if [ "$#" -eq 0 ]; then
# NOTE: If you want to run all tests under specific folder/section just
# replace the dot (root directory below) with the folder name, e.g.
# `--workloads-root-directory replication`.
- python3 runner.py --workloads-root-directory "$SCRIPT_DIR/../../build"
+ python3 runner.py --workloads-root-directory "$SCRIPT_DIR/../../build/tests/e2e"
elif [ "$#" -eq 1 ]; then
if [ "$1" == "-h" ] || [ "$1" == "--help" ]; then
print_help
@@ -34,7 +34,7 @@ elif [ "$#" -eq 1 ]; then
# NOTE: --workload-name comes from each individual folder/section
# workloads.yaml file. E.g. `streams/workloads.yaml` has a list of
# `workloads:` and each workload has it's `-name`.
- python3 runner.py --workloads-root-directory "$SCRIPT_DIR/../../build" --workload-name "$1"
+ python3 runner.py --workloads-root-directory "$SCRIPT_DIR/../../build/tests/e2e" --workload-name "$1"
else
print_help
fi
diff --git a/tests/e2e/runner.py b/tests/e2e/runner.py
index ae022d4d8..282629b20 100755
--- a/tests/e2e/runner.py
+++ b/tests/e2e/runner.py
@@ -40,6 +40,9 @@ def load_args():
def load_workloads(root_directory):
workloads = []
for file in Path(root_directory).rglob("*.yaml"):
+ # 8.03.2024. - Skip streams e2e tests
+ if str(file).endswith("/streams/workloads.yaml"):
+ continue
with open(file, "r") as f:
workloads.extend(yaml.load(f, Loader=yaml.FullLoader)["workloads"])
return workloads
diff --git a/tests/e2e/streams/conftest.py b/tests/e2e/streams/conftest.py
index 1bf3544c2..2dadfcb28 100644
--- a/tests/e2e/streams/conftest.py
+++ b/tests/e2e/streams/conftest.py
@@ -9,6 +9,7 @@
# by the Apache License, Version 2.0, included in the file
# licenses/APL.txt.
+# import os
import pulsar
import pytest
from common import NAME, PULSAR_SERVICE_URL, connect, execute_and_fetch_all
@@ -20,6 +21,9 @@ import requests
# To run these test locally a running Kafka sever is necessery. The test tries
# to connect on localhost:9092.
+# KAFKA_HOSTNAME=os.getenv("KAFKA_HOSTNAME", "localhost")
+# PULSAR_HOSTNAME=os.getenv("PULSAR_HOSTNAME", "localhost")
+# PULSAR_PORT="6652" if PULSAR_HOSTNAME == "localhost" else "8080"
@pytest.fixture(autouse=True)
def connection():
diff --git a/tests/e2e/streams/kafka_streams_tests.py b/tests/e2e/streams/kafka_streams_tests.py
index b988a6c26..17decdc0f 100755
--- a/tests/e2e/streams/kafka_streams_tests.py
+++ b/tests/e2e/streams/kafka_streams_tests.py
@@ -11,6 +11,7 @@
# by the Apache License, Version 2.0, included in the file
# licenses/APL.txt.
+# import os
import sys
import time
from multiprocessing import Process
@@ -23,7 +24,7 @@ from mg_utils import mg_sleep_and_assert
TRANSFORMATIONS_TO_CHECK_C = ["c_transformations.empty_transformation"]
TRANSFORMATIONS_TO_CHECK_PY = ["kafka_transform.simple", "kafka_transform.with_parameters"]
KAFKA_PRODUCER_SENDING_MSG_DEFAULT_TIMEOUT = 60
-
+# KAFKA_HOSTNAME=os.getenv("KAFKA_HOSTNAME", "localhost")
@pytest.mark.parametrize("transformation", TRANSFORMATIONS_TO_CHECK_PY)
def test_simple(kafka_producer, kafka_topics, connection, transformation):
@@ -162,7 +163,7 @@ def test_show_streams(kafka_topics, connection):
complex_values_stream = "complex_values"
common.create_stream(
- cursor, default_values_stream, kafka_topics[0], "kafka_transform.simple", bootstrap_servers="'localhost:29092'"
+ cursor, default_values_stream, kafka_topics[0], "kafka_transform.simple", bootstrap_servers="'localhost:29092'"
)
common.create_stream(
cursor,
diff --git a/tests/e2e/streams/pulsar_streams_tests.py b/tests/e2e/streams/pulsar_streams_tests.py
index cf52416cb..49aa773e6 100755
--- a/tests/e2e/streams/pulsar_streams_tests.py
+++ b/tests/e2e/streams/pulsar_streams_tests.py
@@ -11,6 +11,7 @@
# by the Apache License, Version 2.0, included in the file
# licenses/APL.txt.
+# import os
import sys
import time
from multiprocessing import Process, Value
@@ -20,6 +21,7 @@ import mgclient
import pytest
TRANSFORMATIONS_TO_CHECK = ["pulsar_transform.simple", "pulsar_transform.with_parameters"]
+# PULSAR_HOSTNAME=os.getenv("PULSAR_HOSTNAME", "127.0.0.1")
def check_vertex_exists_with_topic_and_payload(cursor, topic, payload_byte):
diff --git a/tests/e2e/text_search/CMakeLists.txt b/tests/e2e/text_search/CMakeLists.txt
new file mode 100644
index 000000000..db2af7a11
--- /dev/null
+++ b/tests/e2e/text_search/CMakeLists.txt
@@ -0,0 +1,6 @@
+function(copy_text_search_e2e_python_files FILE_NAME)
+ copy_e2e_python_files(text_search ${FILE_NAME})
+endfunction()
+
+copy_text_search_e2e_python_files(common.py)
+copy_text_search_e2e_python_files(test_text_search.py)
diff --git a/tests/e2e/text_search/common.py b/tests/e2e/text_search/common.py
new file mode 100644
index 000000000..0f28351d3
--- /dev/null
+++ b/tests/e2e/text_search/common.py
@@ -0,0 +1,87 @@
+# Copyright 2023 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import typing
+
+import mgclient
+import pytest
+from gqlalchemy import Memgraph
+
+
+def execute_and_fetch_all(cursor: mgclient.Cursor, query: str, params: dict = {}) -> typing.List[tuple]:
+ cursor.execute(query, params)
+ return cursor.fetchall()
+
+
+@pytest.fixture
+def connect(**kwargs) -> mgclient.Connection:
+ connection = mgclient.connect(host="localhost", port=7687, **kwargs)
+ connection.autocommit = True
+ cursor = connection.cursor()
+ execute_and_fetch_all(cursor, """USE DATABASE memgraph""")
+ try:
+ execute_and_fetch_all(cursor, """DROP DATABASE clean""")
+ except:
+ pass
+ execute_and_fetch_all(cursor, """MATCH (n) DETACH DELETE n""")
+ yield connection
+
+
+@pytest.fixture
+def memgraph(**kwargs) -> Memgraph:
+ memgraph = Memgraph()
+
+ yield memgraph
+
+ memgraph.drop_database()
+ memgraph.drop_indexes()
+
+
+@pytest.fixture
+def memgraph_with_text_indexed_data(**kwargs) -> Memgraph:
+ memgraph = Memgraph()
+
+ memgraph.execute(
+ """CREATE (:Document {title: "Rules2024", version: 1, fulltext: "random works", date: date("2023-11-14")});"""
+ )
+ memgraph.execute(
+ """CREATE (:Document {title: "Rules2023", version: 9, fulltext: "text Rules2024", date: date("2023-11-14")});"""
+ )
+ memgraph.execute(
+ """CREATE (:Document:Revision {title: "Rules2024", version: 2, fulltext: "random words", date: date("2023-12-15")});"""
+ )
+ memgraph.execute("""CREATE (:Revision {title: "OperationSchema", version: 3, date: date("2023-10-01")});""")
+ memgraph.execute("""CREATE TEXT INDEX complianceDocuments ON :Document;""")
+
+ yield memgraph
+
+ memgraph.execute("""DROP TEXT INDEX complianceDocuments;""")
+ memgraph.drop_database()
+ memgraph.drop_indexes()
+
+
+@pytest.fixture
+def memgraph_with_mixed_data(**kwargs) -> Memgraph:
+ memgraph = Memgraph()
+
+ memgraph.execute(
+ """CREATE (:Document:Revision {title: "Rules2024", version: 1, date: date("2023-11-14"), contents: "Lorem ipsum dolor sit amet"});"""
+ )
+ memgraph.execute(
+ """CREATE (:Revision {title: "Rules2024", version: 2, date: date("2023-12-15"), contents: "consectetur adipiscing elit"});"""
+ )
+ memgraph.execute("""CREATE TEXT INDEX complianceDocuments ON :Document;""")
+
+ yield memgraph
+
+ memgraph.execute("""DROP TEXT INDEX complianceDocuments;""")
+ memgraph.drop_database()
+ memgraph.drop_indexes()
diff --git a/tests/e2e/text_search/test_text_search.py b/tests/e2e/text_search/test_text_search.py
new file mode 100644
index 000000000..8d538d464
--- /dev/null
+++ b/tests/e2e/text_search/test_text_search.py
@@ -0,0 +1,206 @@
+# Copyright 2024 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import json
+import re
+import sys
+
+import gqlalchemy
+import mgclient
+import pytest
+from common import memgraph, memgraph_with_mixed_data, memgraph_with_text_indexed_data
+
+GET_RULES_2024_DOCUMENT = """CALL libtext.search("complianceDocuments", "data.title:Rules2024") YIELD node
+ RETURN node.title AS title, node.version AS version
+ ORDER BY version ASC, title ASC;"""
+
+
+def test_create_index(memgraph):
+ memgraph.execute("""CREATE TEXT INDEX exampleIndex ON :Document;""")
+
+ index_info = memgraph.execute_and_fetch("""SHOW INDEX INFO""")
+
+ assert list(index_info) == [
+ {"index type": "text (name: exampleIndex)", "label": "Document", "property": None, "count": None}
+ ]
+
+
+def test_drop_index(memgraph):
+ memgraph.execute("""DROP TEXT INDEX exampleIndex;""")
+
+ index_info = memgraph.execute_and_fetch("""SHOW INDEX INFO""")
+
+ assert list(index_info) == []
+
+
+def test_create_existing_index(memgraph):
+ memgraph.execute("""CREATE TEXT INDEX duplicatedIndex ON :Document;""")
+ with pytest.raises(
+ gqlalchemy.exceptions.GQLAlchemyDatabaseError, match='Text index "duplicatedIndex" already exists.'
+ ) as _:
+ memgraph.execute("""CREATE TEXT INDEX duplicatedIndex ON :Document;""")
+ memgraph.execute("""DROP TEXT INDEX duplicatedIndex;""") # cleanup
+
+
+def test_drop_nonexistent_index(memgraph):
+ with pytest.raises(
+ gqlalchemy.exceptions.GQLAlchemyDatabaseError, match='Text index "noSuchIndex" doesn’t exist.'
+ ) as _:
+ memgraph.execute("""DROP TEXT INDEX noSuchIndex;""")
+
+
+def test_text_search_given_property(memgraph_with_text_indexed_data):
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(GET_RULES_2024_DOCUMENT))
+
+ assert len(result) == 2 and result == [{"title": "Rules2024", "version": 1}, {"title": "Rules2024", "version": 2}]
+
+
+def test_text_search_all_properties(memgraph_with_text_indexed_data):
+ SEARCH_QUERY = "Rules2024"
+
+ ALL_PROPERTIES_QUERY = f"""CALL libtext.search_all("complianceDocuments", "{SEARCH_QUERY}") YIELD node
+ RETURN node
+ ORDER BY node.version ASC, node.title ASC;"""
+
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(ALL_PROPERTIES_QUERY))
+ result_nodes = [record["node"] for record in result]
+
+ assert len(result) == 3 and (
+ result_nodes[0].title == SEARCH_QUERY
+ and result_nodes[1].title == SEARCH_QUERY
+ and SEARCH_QUERY in result_nodes[2].fulltext
+ )
+
+
+def test_regex_text_search(memgraph_with_text_indexed_data):
+ REGEX_QUERY = """CALL libtext.regex_search("complianceDocuments", "wor.*s") YIELD node
+ RETURN node
+ ORDER BY node.version ASC, node.title ASC;"""
+
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(REGEX_QUERY))
+
+ assert (
+ len(result) == 2
+ and re.search("wor.*s", result[0]["node"].fulltext)
+ and re.search("wor.*s", result[1]["node"].fulltext)
+ # In this test, all values matching the regex string are found in the .node property only ^
+ )
+
+
+def test_text_search_aggregate(memgraph_with_text_indexed_data):
+ input_aggregation = json.dumps({"count": {"value_count": {"field": "metadata.gid"}}}, separators=(",", ":"))
+ expected_aggregation = json.dumps({"count": {"value": 2.0}}, separators=(",", ":"))
+
+ AGGREGATION_QUERY = f"""CALL libtext.aggregate("complianceDocuments", "data.title:Rules2024", '{input_aggregation}')
+ YIELD aggregation
+ RETURN aggregation;"""
+
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(AGGREGATION_QUERY))
+
+ assert len(result) == 1 and result[0]["aggregation"] == expected_aggregation
+
+
+def test_text_search_query_boolean(memgraph_with_text_indexed_data):
+ BOOLEAN_QUERY = """CALL libtext.search("complianceDocuments", "(data.title:Rules2023 OR data.title:Rules2024) AND data.fulltext:words") YIELD node
+ RETURN node.title AS title, node.version AS version
+ ORDER BY version ASC, title ASC;"""
+
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(BOOLEAN_QUERY))
+
+ assert len(result) == 1 and result == [{"title": "Rules2024", "version": 2}]
+
+
+def test_create_indexed_node(memgraph_with_text_indexed_data):
+ memgraph_with_text_indexed_data.execute("""CREATE (:Document {title: "Rules2024", version: 3});""")
+
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(GET_RULES_2024_DOCUMENT))
+
+ assert len(result) == 3 and result == [
+ {"title": "Rules2024", "version": 1},
+ {"title": "Rules2024", "version": 2},
+ {"title": "Rules2024", "version": 3},
+ ]
+
+
+def test_delete_indexed_node(memgraph_with_text_indexed_data):
+ memgraph_with_text_indexed_data.execute("""MATCH (n:Document {title: "Rules2024", version: 2}) DETACH DELETE n;""")
+
+ result = list(memgraph_with_text_indexed_data.execute_and_fetch(GET_RULES_2024_DOCUMENT))
+
+ assert len(result) == 1 and result == [{"title": "Rules2024", "version": 1}]
+
+
+def test_add_indexed_label(memgraph_with_mixed_data):
+ memgraph_with_mixed_data.execute("""MATCH (n:Revision {version:2}) SET n:Document;""")
+
+ result = list(memgraph_with_mixed_data.execute_and_fetch(GET_RULES_2024_DOCUMENT))
+
+ assert len(result) == 2 and result == [{"title": "Rules2024", "version": 1}, {"title": "Rules2024", "version": 2}]
+
+
+def test_remove_indexed_label(memgraph_with_mixed_data):
+ memgraph_with_mixed_data.execute("""MATCH (n:Document {version: 1}) REMOVE n:Document;""")
+
+ result = list(memgraph_with_mixed_data.execute_and_fetch(GET_RULES_2024_DOCUMENT))
+
+ assert len(result) == 0
+
+
+def test_update_text_property_of_indexed_node(memgraph_with_text_indexed_data):
+ memgraph_with_text_indexed_data.execute("""MATCH (n:Document {version:1}) SET n.title = "Rules2030";""")
+
+ result = list(
+ memgraph_with_text_indexed_data.execute_and_fetch(
+ """CALL libtext.search("complianceDocuments", "data.title:Rules2030") YIELD node
+ RETURN node.title AS title, node.version AS version
+ ORDER BY version ASC, title ASC;"""
+ )
+ )
+
+ assert len(result) == 1 and result == [{"title": "Rules2030", "version": 1}]
+
+
+def test_add_unindexable_property_to_indexed_node(memgraph_with_text_indexed_data):
+ try:
+ memgraph_with_text_indexed_data.execute("""MATCH (n:Document {version:1}) SET n.randomList = [2, 3, 4, 5];""")
+ except Exception:
+ assert False
+
+
+def test_remove_indexable_property_from_indexed_node(memgraph_with_text_indexed_data):
+ try:
+ memgraph_with_text_indexed_data.execute(
+ """MATCH (n:Document {version:1}) REMOVE n.title, n.version, n.fulltext, n.date;"""
+ )
+ except Exception:
+ assert False
+
+
+def test_remove_unindexable_property_from_indexed_node(memgraph_with_text_indexed_data):
+ try:
+ memgraph_with_text_indexed_data.execute_and_fetch(
+ """MATCH (n:Document {date: date("2023-12-15")}) REMOVE n.date;"""
+ )
+ except Exception:
+ assert False
+
+
+def test_text_search_nonexistent_index(memgraph_with_text_indexed_data):
+ NONEXISTENT_INDEX_QUERY = """CALL libtext.search("noSuchIndex", "data.fulltext:words") YIELD node
+ RETURN node.title AS title, node.version AS version
+ ORDER BY version ASC, title ASC;"""
+
+ with pytest.raises(mgclient.DatabaseError, match='Text index "noSuchIndex" doesn’t exist.') as _:
+ list(memgraph_with_text_indexed_data.execute_and_fetch(NONEXISTENT_INDEX_QUERY))
+
+
+if __name__ == "__main__":
+ sys.exit(pytest.main([__file__, "-rA"]))
diff --git a/tests/e2e/text_search/test_text_search_disabled.py b/tests/e2e/text_search/test_text_search_disabled.py
new file mode 100644
index 000000000..064f7b409
--- /dev/null
+++ b/tests/e2e/text_search/test_text_search_disabled.py
@@ -0,0 +1,69 @@
+# Copyright 2024 Memgraph Ltd.
+#
+# Use of this software is governed by the Business Source License
+# included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
+# License, and you may not use this file except in compliance with the Business Source License.
+#
+# As of the Change Date specified in that file, in accordance with
+# the Business Source License, use of this software will be governed
+# by the Apache License, Version 2.0, included in the file
+# licenses/APL.txt.
+
+import json
+import sys
+
+import gqlalchemy
+import pytest
+from common import memgraph
+
+TEXT_SEARCH_DISABLED_ERROR = (
+ "To use text indices and text search, start Memgraph with the experimental text search feature enabled."
+)
+
+
+def test_create_index(memgraph):
+ with pytest.raises(gqlalchemy.exceptions.GQLAlchemyDatabaseError, match=TEXT_SEARCH_DISABLED_ERROR) as _:
+ memgraph.execute("""CREATE TEXT INDEX exampleIndex ON :Document;""")
+
+
+def test_drop_index(memgraph):
+ with pytest.raises(gqlalchemy.exceptions.GQLAlchemyDatabaseError, match=TEXT_SEARCH_DISABLED_ERROR) as _:
+ memgraph.execute("""DROP TEXT INDEX exampleIndex;""")
+
+
+def test_text_search_given_property(memgraph):
+ with pytest.raises(gqlalchemy.exceptions.GQLAlchemyDatabaseError, match=TEXT_SEARCH_DISABLED_ERROR) as _:
+ memgraph.execute(
+ """CALL libtext.search("complianceDocuments", "data.title:Rules2024") YIELD node
+ RETURN node;"""
+ )
+
+
+def test_text_search_all_properties(memgraph):
+ with pytest.raises(gqlalchemy.exceptions.GQLAlchemyDatabaseError, match=TEXT_SEARCH_DISABLED_ERROR) as _:
+ memgraph.execute(
+ """CALL libtext.search_all("complianceDocuments", "Rules2024") YIELD node
+ RETURN node;"""
+ )
+
+
+def test_regex_text_search(memgraph):
+ with pytest.raises(gqlalchemy.exceptions.GQLAlchemyDatabaseError, match=TEXT_SEARCH_DISABLED_ERROR) as _:
+ memgraph.execute(
+ """CALL libtext.regex_search("complianceDocuments", "wor.*s") YIELD node
+ RETURN node;"""
+ )
+
+
+def test_text_search_aggregate(memgraph):
+ with pytest.raises(gqlalchemy.exceptions.GQLAlchemyDatabaseError, match=TEXT_SEARCH_DISABLED_ERROR) as _:
+ input_aggregation = json.dumps({"count": {"value_count": {"field": "metadata.gid"}}}, separators=(",", ":"))
+
+ memgraph.execute(
+ f"""CALL libtext.aggregate("complianceDocuments", "wor.*s", '{input_aggregation}') YIELD aggregation
+ RETURN aggregation;"""
+ )
+
+
+if __name__ == "__main__":
+ sys.exit(pytest.main([__file__, "-rA"]))
diff --git a/tests/e2e/text_search/workloads.yaml b/tests/e2e/text_search/workloads.yaml
new file mode 100644
index 000000000..5b1640715
--- /dev/null
+++ b/tests/e2e/text_search/workloads.yaml
@@ -0,0 +1,33 @@
+text_search_cluster: &text_search_cluster
+ cluster:
+ main:
+ args:
+ [
+ "--bolt-port",
+ "7687",
+ "--log-level=TRACE",
+ "--experimental-enabled=text-search",
+ ]
+ log_file: "text_search.log"
+ setup_queries: []
+ validation_queries: []
+
+text_search_disabled_cluster: &text_search_disabled_cluster
+ cluster:
+ main:
+ args: ["--bolt-port", "7687", "--log-level=TRACE"]
+ log_file: "text_search.log"
+ setup_queries: []
+ validation_queries: []
+
+workloads:
+ - name: "Test behavior of text search in Memgraph"
+ binary: "tests/e2e/pytest_runner.sh"
+ proc: "tests/e2e/text_search/query_modules/"
+ args: ["text_search/test_text_search.py"]
+ <<: *text_search_cluster
+ - name: "Test behavior of text search in Memgraph when disabled"
+ binary: "tests/e2e/pytest_runner.sh"
+ proc: "tests/e2e/text_search/query_modules/"
+ args: ["text_search/test_text_search_disabled.py"]
+ <<: *text_search_disabled_cluster
diff --git a/tests/jepsen/jepsen_0.3.0.patch b/tests/jepsen/jepsen_0.3.0.patch
deleted file mode 100644
index be47cc8b4..000000000
--- a/tests/jepsen/jepsen_0.3.0.patch
+++ /dev/null
@@ -1,13 +0,0 @@
-diff --git a/docker/control/Dockerfile b/docker/control/Dockerfile
-index 6b2d3c0e..195a7a60 100644
---- a/docker/control/Dockerfile
-+++ b/docker/control/Dockerfile
-@@ -7,7 +7,7 @@ ENV LEIN_ROOT true
- # Jepsen dependencies
- #
- RUN apt-get -y -q update && \
-- apt-get install -qy openjdk-17-jdk-headless \
-+ apt-get install -qy ca-certificates-java openjdk-17-jdk-headless \
- libjna-java \
- vim \
- emacs \
diff --git a/tests/jepsen/project.clj b/tests/jepsen/project.clj
index 506bcff52..445e57b41 100644
--- a/tests/jepsen/project.clj
+++ b/tests/jepsen/project.clj
@@ -5,10 +5,10 @@
:url "https://github.com/memgraph/memgraph/blob/master/release/LICENSE_ENTERPRISE.md"}
:main jepsen.memgraph.core
:dependencies [[org.clojure/clojure "1.10.0"]
- ;; 0.2.4-SNAPSHOT but 0.3.0, for more -> https://clojars.org/jepsen/versions
- [jepsen "0.2.4-SNAPSHOT"]
+ ;; Details under https://clojars.org/jepsen/versions.
+ [jepsen "0.3.5-SNAPSHOT"]
[gorillalabs/neo4j-clj "4.1.0"]]
:profiles {:test {:dependencies [#_[org.neo4j.test/neo4j-harness "4.1.0"]]}}
- ;; required to run 0.3.0
- ; :aot :all
+ ;; The below line is required to run after Jepsen 0.3.0.
+ :aot :all
:repl-options {:init-ns jepsen.memgraph.core})
diff --git a/tests/jepsen/run.sh b/tests/jepsen/run.sh
index a1587c8a1..040491c3f 100755
--- a/tests/jepsen/run.sh
+++ b/tests/jepsen/run.sh
@@ -2,12 +2,10 @@
set -Eeuo pipefail
script_dir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
-MEMGRAPH_BINARY_PATH="../../build/memgraph"
+MEMGRAPH_BUILD_PATH="$script_dir/../../build"
+MEMGRAPH_BINARY_PATH="$MEMGRAPH_BUILD_PATH/memgraph"
# NOTE: Jepsen Git tags are not consistent, there are: 0.2.4, v0.3.0, 0.3.2, ...
-# NOTE: On Ubuntu 22.04 v0.3.2 uses non-existing docker compose --compatibility flag.
-# NOTE: On Ubuntu 22.04 v0.3.0 and v0.3.1 seems to be runnable.
-# TODO(gitbuda): Make sure Memgraph can be testes with Jepsen >= 0.3.0
-JEPSEN_VERSION="${JEPSEN_VERSION:-0.2.4}"
+JEPSEN_VERSION="${JEPSEN_VERSION:-v0.3.5}"
JEPSEN_ACTIVE_NODES_NO=5
CONTROL_LEIN_RUN_ARGS="test-all --node-configs resources/node-config.edn"
CONTROL_LEIN_RUN_STDOUT_LOGS=1
@@ -24,7 +22,7 @@ PRINT_CONTEXT() {
HELP_EXIT() {
echo ""
- echo "HELP: $0 help|cluster-up|cluster-refresh|cluster-cleanup|cluster-dealloc|mgbuild|test|test-all-individually [args]"
+ echo "HELP: $0 help|cluster-up|cluster-refresh|cluster-nodes-cleanup|cluster-dealloc|mgbuild|test|test-all-individually [args]"
echo ""
echo " test args --binary MEMGRAPH_BINARY_PATH"
echo " --ignore-run-stdout-logs Ignore lein run stdout logs."
@@ -43,24 +41,18 @@ INFO() {
/bin/echo -e "\e[104m\e[97m[INFO]\e[49m\e[39m" "$@"
}
+if [[ "$#" -lt 1 || "$1" == "-h" || "$1" == "--help" ]]; then
+ HELP_EXIT
+fi
+
if ! command -v docker > /dev/null 2>&1 || ! command -v docker-compose > /dev/null 2>&1; then
ERROR "docker and docker-compose have to be installed."
exit 1
fi
if [ ! -d "$script_dir/jepsen" ]; then
+ # TODO(deda): install apt get docker-compose-plugin on all build machines.
git clone https://github.com/jepsen-io/jepsen.git -b "$JEPSEN_VERSION" "$script_dir/jepsen"
- if [ "$JEPSEN_VERSION" == "v0.3.0" ]; then
- if [ -f "$script_dir/jepsen_0.3.0.patch" ]; then
- cd "$script_dir/jepsen"
- git apply "$script_dir/jepsen_0.3.0.patch"
- cd "$script_dir"
- fi
- fi
-fi
-
-if [ "$#" -lt 1 ]; then
- HELP_EXIT
fi
PROCESS_ARGS() {
@@ -199,7 +191,7 @@ CLUSTER_UP() {
}
CLUSTER_DEALLOC() {
- ps=$(docker ps --filter name=jepsen* --filter status=running -q)
+ ps=$(docker ps -a --filter name=jepsen* -q)
if [[ ! -z ${ps} ]]; then
echo "Killing ${ps}"
docker rm -f ${ps}
@@ -213,6 +205,7 @@ CLUSTER_DEALLOC() {
else
echo "No Jepsen containers detected!"
fi
+ echo "Cluster dealloc DONE"
}
# Initialize testing context by copying source/binary files. Inside CI,
@@ -239,7 +232,7 @@ case $1 in
CLUSTER_DEALLOC
;;
- cluster-cleanup)
+ cluster-nodes-cleanup)
jepsen_control_exec="docker exec jepsen-control bash -c"
INFO "Deleting /jepsen/memgraph/store/* on jepsen-control"
$jepsen_control_exec "rm -rf /jepsen/memgraph/store/*"
@@ -252,9 +245,13 @@ case $1 in
;;
mgbuild)
+ PROCESS_ARGS "$@"
PRINT_CONTEXT
+ # docker cp -L mgbuild_debian-12:/memgraph/build/memgraph "${MEMGRAPH_BUILD_PATH}/"
+ # NOTE: mgconsole is interesting inside jepsen container to inspect Memgraph state.
+ # docker cp -L mgbuild_debian-12:/usr/local/bin/mgconsole "${MEMGRAPH_BUILD_PATH}/"
echo ""
- echo "TODO(gitbuda): Build memgraph for Debian 10 via memgraph/memgraph-builder"
+ echo "TODO(gitbuda): Build memgraph for Jepsen (on v0.3.5 for Debian 12) via memgraph/memgraph-builder"
exit 1
;;
diff --git a/tests/jepsen/src/jepsen/memgraph/support.clj b/tests/jepsen/src/jepsen/memgraph/support.clj
index b7846f8ce..aec2aadb1 100644
--- a/tests/jepsen/src/jepsen/memgraph/support.clj
+++ b/tests/jepsen/src/jepsen/memgraph/support.clj
@@ -25,6 +25,7 @@
:--storage-recover-on-startup
:--storage-wal-enabled
:--storage-snapshot-interval-sec 300
+ :--replication-restore-state-on-startup
:--storage-properties-on-edges))
(defn stop-node!
diff --git a/tests/mgbench/runners.py b/tests/mgbench/runners.py
index e1f52b696..155ceac06 100644
--- a/tests/mgbench/runners.py
+++ b/tests/mgbench/runners.py
@@ -416,6 +416,7 @@ class Memgraph(BaseRunner):
def __init__(self, benchmark_context: BenchmarkContext):
super().__init__(benchmark_context=benchmark_context)
self._memgraph_binary = benchmark_context.vendor_binary
+ self._bolt_num_workers = benchmark_context.num_workers_for_benchmark
self._performance_tracking = benchmark_context.performance_tracking
self._directory = tempfile.TemporaryDirectory(dir=benchmark_context.temporary_directory)
self._vendor_args = benchmark_context.vendor_args
@@ -440,6 +441,7 @@ class Memgraph(BaseRunner):
kwargs["bolt_port"] = self._bolt_port
kwargs["data_directory"] = data_directory
kwargs["storage_properties_on_edges"] = True
+ kwargs["bolt_num_workers"] = self._bolt_num_workers
for key, value in self._vendor_args.items():
kwargs[key] = value
return _convert_args_to_flags(self._memgraph_binary, **kwargs)
diff --git a/tests/setup.sh b/tests/setup.sh
index 7cab86db6..b91eea7c3 100755
--- a/tests/setup.sh
+++ b/tests/setup.sh
@@ -37,6 +37,7 @@ pip --timeout 1000 install "pulsar-client==3.1.0"
for pkg in "${PIP_DEPS[@]}"; do
pip --timeout 1000 install "$pkg"
done
+pip --timeout 1000 install "networkx==2.4"
# Install mgclient from source becasue of full flexibility.
pushd "$DIR/../libs/pymgclient" > /dev/null
diff --git a/tests/unit/database_get_info.cpp b/tests/unit/database_get_info.cpp
index be6885b37..2547378c6 100644
--- a/tests/unit/database_get_info.cpp
+++ b/tests/unit/database_get_info.cpp
@@ -197,8 +197,8 @@ TYPED_TEST(InfoTest, InfoCheck) {
ASSERT_EQ(info.storage_info.vertex_count, 5);
ASSERT_EQ(info.storage_info.edge_count, 2);
ASSERT_EQ(info.storage_info.average_degree, 0.8);
- ASSERT_GT(info.storage_info.memory_res, 10'000'000); // 250MB < > 10MB
- ASSERT_LT(info.storage_info.memory_res, 250'000'000);
+ ASSERT_GT(info.storage_info.memory_res, 10'000'000); // 300MB < > 10MB
+ ASSERT_LT(info.storage_info.memory_res, 300'000'000);
ASSERT_GT(info.storage_info.disk_usage, 100); // 1MB < > 100B
ASSERT_LT(info.storage_info.disk_usage, 1000'000);
ASSERT_EQ(info.storage_info.label_indices, 1);
diff --git a/tests/unit/property_value_v2.cpp b/tests/unit/property_value_v2.cpp
index aba322ce7..28937598e 100644
--- a/tests/unit/property_value_v2.cpp
+++ b/tests/unit/property_value_v2.cpp
@@ -1,4 +1,4 @@
-// Copyright 2022 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -570,7 +570,6 @@ TEST(PropertyValue, MoveConstructor) {
for (auto &item : data) {
memgraph::storage::PropertyValue copy(item);
memgraph::storage::PropertyValue pv(std::move(item));
- ASSERT_EQ(item.type(), memgraph::storage::PropertyValue::Type::Null);
ASSERT_EQ(pv.type(), copy.type());
switch (copy.type()) {
case memgraph::storage::PropertyValue::Type::Null:
@@ -668,7 +667,6 @@ TEST(PropertyValue, MoveAssignment) {
memgraph::storage::PropertyValue copy(item);
memgraph::storage::PropertyValue pv(123);
pv = std::move(item);
- ASSERT_EQ(item.type(), memgraph::storage::PropertyValue::Type::Null);
ASSERT_EQ(pv.type(), copy.type());
switch (copy.type()) {
case memgraph::storage::PropertyValue::Type::Null:
diff --git a/tests/unit/query_dump.cpp b/tests/unit/query_dump.cpp
index a2ca2864d..2dd1e7ac7 100644
--- a/tests/unit/query_dump.cpp
+++ b/tests/unit/query_dump.cpp
@@ -71,6 +71,11 @@ struct DatabaseState {
std::string property;
};
+ struct TextItem {
+ std::string index_name;
+ std::string label;
+ };
+
struct LabelPropertiesItem {
std::string label;
std::set<std::string, std::less<>> properties;
@@ -80,6 +85,7 @@ struct DatabaseState {
std::set<Edge> edges;
std::set<LabelItem> label_indices;
std::set<LabelPropertyItem> label_property_indices;
+ std::set<TextItem> text_indices;
std::set<LabelPropertyItem> existence_constraints;
std::set<LabelPropertiesItem> unique_constraints;
};
@@ -106,6 +112,10 @@ bool operator<(const DatabaseState::LabelPropertyItem &first, const DatabaseStat
return first.property < second.property;
}
+bool operator<(const DatabaseState::TextItem &first, const DatabaseState::TextItem &second) {
+ return first.index_name < second.index_name && first.label < second.label;
+}
+
bool operator<(const DatabaseState::LabelPropertiesItem &first, const DatabaseState::LabelPropertiesItem &second) {
if (first.label != second.label) return first.label < second.label;
return first.properties < second.properties;
@@ -128,6 +138,10 @@ bool operator==(const DatabaseState::LabelPropertyItem &first, const DatabaseSta
return first.label == second.label && first.property == second.property;
}
+bool operator==(const DatabaseState::TextItem &first, const DatabaseState::TextItem &second) {
+ return first.index_name == second.index_name && first.label == second.label;
+}
+
bool operator==(const DatabaseState::LabelPropertiesItem &first, const DatabaseState::LabelPropertiesItem &second) {
return first.label == second.label && first.properties == second.properties;
}
@@ -185,6 +199,7 @@ DatabaseState GetState(memgraph::storage::Storage *db) {
// Capture all indices
std::set<DatabaseState::LabelItem> label_indices;
std::set<DatabaseState::LabelPropertyItem> label_property_indices;
+ std::set<DatabaseState::TextItem> text_indices;
{
auto info = dba->ListAllIndices();
for (const auto &item : info.label) {
@@ -193,6 +208,9 @@ DatabaseState GetState(memgraph::storage::Storage *db) {
for (const auto &item : info.label_property) {
label_property_indices.insert({dba->LabelToName(item.first), dba->PropertyToName(item.second)});
}
+ for (const auto &item : info.text_indices) {
+ text_indices.insert({item.first, dba->LabelToName(item.second)});
+ }
}
// Capture all constraints
@@ -212,7 +230,8 @@ DatabaseState GetState(memgraph::storage::Storage *db) {
}
}
- return {vertices, edges, label_indices, label_property_indices, existence_constraints, unique_constraints};
+ return {vertices, edges, label_indices, label_property_indices, text_indices, existence_constraints,
+ unique_constraints};
}
auto Execute(memgraph::query::InterpreterContext *context, memgraph::dbms::DatabaseAccess db,
diff --git a/tests/unit/storage_v2_decoder_encoder.cpp b/tests/unit/storage_v2_decoder_encoder.cpp
index 15db49b1c..0264e2287 100644
--- a/tests/unit/storage_v2_decoder_encoder.cpp
+++ b/tests/unit/storage_v2_decoder_encoder.cpp
@@ -358,6 +358,8 @@ TEST_F(DecoderEncoderTest, PropertyValueInvalidMarker) {
case memgraph::storage::durability::Marker::DELTA_LABEL_PROPERTY_INDEX_STATS_CLEAR:
case memgraph::storage::durability::Marker::DELTA_EDGE_TYPE_INDEX_CREATE:
case memgraph::storage::durability::Marker::DELTA_EDGE_TYPE_INDEX_DROP:
+ case memgraph::storage::durability::Marker::DELTA_TEXT_INDEX_CREATE:
+ case memgraph::storage::durability::Marker::DELTA_TEXT_INDEX_DROP:
case memgraph::storage::durability::Marker::DELTA_EXISTENCE_CONSTRAINT_CREATE:
case memgraph::storage::durability::Marker::DELTA_EXISTENCE_CONSTRAINT_DROP:
case memgraph::storage::durability::Marker::DELTA_UNIQUE_CONSTRAINT_CREATE:
diff --git a/tests/unit/storage_v2_get_info.cpp b/tests/unit/storage_v2_get_info.cpp
index 71dbc1a8d..ee5c1bb62 100644
--- a/tests/unit/storage_v2_get_info.cpp
+++ b/tests/unit/storage_v2_get_info.cpp
@@ -146,6 +146,7 @@ TYPED_TEST(InfoTest, InfoCheck) {
ASSERT_LT(info.disk_usage, 1000'000);
ASSERT_EQ(info.label_indices, 1);
ASSERT_EQ(info.label_property_indices, 1);
+ ASSERT_EQ(info.text_indices, 0);
ASSERT_EQ(info.existence_constraints, 0);
ASSERT_EQ(info.unique_constraints, 2);
ASSERT_EQ(info.storage_mode, this->mode);
diff --git a/tests/unit/storage_v2_wal_file.cpp b/tests/unit/storage_v2_wal_file.cpp
index 4094090f5..a94b20590 100644
--- a/tests/unit/storage_v2_wal_file.cpp
+++ b/tests/unit/storage_v2_wal_file.cpp
@@ -53,6 +53,10 @@ memgraph::storage::durability::WalDeltaData::Type StorageMetadataOperationToWalD
return memgraph::storage::durability::WalDeltaData::Type::LABEL_PROPERTY_INDEX_STATS_SET;
case memgraph::storage::durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_STATS_CLEAR:
return memgraph::storage::durability::WalDeltaData::Type::LABEL_PROPERTY_INDEX_STATS_CLEAR;
+ case memgraph::storage::durability::StorageMetadataOperation::TEXT_INDEX_CREATE:
+ return memgraph::storage::durability::WalDeltaData::Type::TEXT_INDEX_CREATE;
+ case memgraph::storage::durability::StorageMetadataOperation::TEXT_INDEX_DROP:
+ return memgraph::storage::durability::WalDeltaData::Type::TEXT_INDEX_DROP;
case memgraph::storage::durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_CREATE:
return memgraph::storage::durability::WalDeltaData::Type::EXISTENCE_CONSTRAINT_CREATE;
case memgraph::storage::durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_DROP:
@@ -252,7 +256,7 @@ class DeltaGenerator final {
ASSERT_TRUE(false) << "Unexpected statistics operation!";
}
}
- wal_file_.AppendOperation(operation, label_id, property_ids, l_stats, lp_stats, timestamp_);
+ wal_file_.AppendOperation(operation, std::nullopt, label_id, property_ids, l_stats, lp_stats, timestamp_);
if (valid_) {
UpdateStats(timestamp_, 1);
memgraph::storage::durability::WalDeltaData data;
@@ -271,6 +275,8 @@ class DeltaGenerator final {
break;
case memgraph::storage::durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_CREATE:
case memgraph::storage::durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_DROP:
+ case memgraph::storage::durability::StorageMetadataOperation::TEXT_INDEX_CREATE:
+ case memgraph::storage::durability::StorageMetadataOperation::TEXT_INDEX_DROP:
case memgraph::storage::durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_CREATE:
case memgraph::storage::durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_DROP:
data.operation_label_property.label = label;
@@ -313,6 +319,8 @@ class DeltaGenerator final {
case memgraph::storage::durability::StorageMetadataOperation::LABEL_INDEX_STATS_SET:
case memgraph::storage::durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_CREATE:
case memgraph::storage::durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_DROP:
+ case memgraph::storage::durability::StorageMetadataOperation::TEXT_INDEX_CREATE:
+ case memgraph::storage::durability::StorageMetadataOperation::TEXT_INDEX_DROP:
case memgraph::storage::durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_CREATE:
case memgraph::storage::durability::StorageMetadataOperation::EXISTENCE_CONSTRAINT_DROP:;
case memgraph::storage::durability::StorageMetadataOperation::LABEL_PROPERTY_INDEX_STATS_SET:
diff --git a/tests/unit/utils_memory.cpp b/tests/unit/utils_memory.cpp
index 5173a5f7b..e46c6c1f9 100644
--- a/tests/unit/utils_memory.cpp
+++ b/tests/unit/utils_memory.cpp
@@ -1,4 +1,4 @@
-// Copyright 2023 Memgraph Ltd.
+// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -195,134 +195,6 @@ TEST(MonotonicBufferResource, AllocationWithInitialBufferOnStack) {
}
}
-// NOLINTNEXTLINE(hicpp-special-member-functions)
-TEST(PoolResource, SingleSmallBlockAllocations) {
- TestMemory test_mem;
- const size_t max_blocks_per_chunk = 3U;
- const size_t max_block_size = 64U;
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, max_block_size, &test_mem);
- // Fill the first chunk.
- CheckAllocation(&mem, 64U, 1U);
- // May allocate more than once due to bookkeeping.
- EXPECT_GE(test_mem.new_count_, 1U);
- // Reset tracking and continue filling the first chunk.
- test_mem.new_count_ = 0U;
- CheckAllocation(&mem, 64U, 64U);
- CheckAllocation(&mem, 64U);
- EXPECT_EQ(test_mem.new_count_, 0U);
- // Reset tracking and fill the second chunk
- test_mem.new_count_ = 0U;
- CheckAllocation(&mem, 64U, 32U);
- auto *ptr1 = CheckAllocation(&mem, 32U, 64U); // this will become 64b block
- auto *ptr2 = CheckAllocation(&mem, 64U, 32U);
- // We expect one allocation for chunk and at most one for bookkeeping.
- EXPECT_TRUE(test_mem.new_count_ >= 1U && test_mem.new_count_ <= 2U);
- test_mem.delete_count_ = 0U;
- mem.Deallocate(ptr1, 32U, 64U);
- mem.Deallocate(ptr2, 64U, 32U);
- EXPECT_EQ(test_mem.delete_count_, 0U);
- mem.Release();
- EXPECT_GE(test_mem.delete_count_, 2U);
- CheckAllocation(&mem, 64U, 1U);
-}
-
-// NOLINTNEXTLINE(hicpp-special-member-functions)
-TEST(PoolResource, MultipleSmallBlockAllocations) {
- TestMemory test_mem;
- const size_t max_blocks_per_chunk = 1U;
- const size_t max_block_size = 64U;
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, max_block_size, &test_mem);
- CheckAllocation(&mem, 64U);
- CheckAllocation(&mem, 18U, 2U);
- CheckAllocation(&mem, 24U, 8U);
- // May allocate more than once per chunk due to bookkeeping.
- EXPECT_GE(test_mem.new_count_, 3U);
- // Reset tracking and fill the second chunk
- test_mem.new_count_ = 0U;
- CheckAllocation(&mem, 64U);
- CheckAllocation(&mem, 18U, 2U);
- CheckAllocation(&mem, 24U, 8U);
- // We expect one allocation for chunk and at most one for bookkeeping.
- EXPECT_TRUE(test_mem.new_count_ >= 3U && test_mem.new_count_ <= 6U);
- mem.Release();
- EXPECT_GE(test_mem.delete_count_, 6U);
- CheckAllocation(&mem, 64U);
-}
-
-// NOLINTNEXTLINE(hicpp-special-member-functions)
-TEST(PoolResource, BigBlockAllocations) {
- TestMemory test_mem;
- TestMemory test_mem_unpooled;
- const size_t max_blocks_per_chunk = 3U;
- const size_t max_block_size = 64U;
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, max_block_size, &test_mem, &test_mem_unpooled);
- CheckAllocation(&mem, max_block_size + 1, 1U);
- // May allocate more than once per block due to bookkeeping.
- EXPECT_GE(test_mem_unpooled.new_count_, 1U);
- CheckAllocation(&mem, max_block_size + 1, 1U);
- EXPECT_GE(test_mem_unpooled.new_count_, 2U);
- auto *ptr = CheckAllocation(&mem, max_block_size * 2, 1U);
- EXPECT_GE(test_mem_unpooled.new_count_, 3U);
- mem.Deallocate(ptr, max_block_size * 2, 1U);
- EXPECT_GE(test_mem_unpooled.delete_count_, 1U);
- mem.Release();
- EXPECT_GE(test_mem_unpooled.delete_count_, 3U);
- CheckAllocation(&mem, max_block_size + 1, 1U);
-}
-
-// NOLINTNEXTLINE(hicpp-special-member-functions)
-TEST(PoolResource, BlockSizeIsNotMultipleOfAlignment) {
- const size_t max_blocks_per_chunk = 3U;
- const size_t max_block_size = 64U;
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, max_block_size);
- EXPECT_THROW(mem.Allocate(64U, 24U), std::bad_alloc);
- EXPECT_THROW(mem.Allocate(63U), std::bad_alloc);
- EXPECT_THROW(mem.Allocate(max_block_size + 1, max_block_size), std::bad_alloc);
-}
-
-// NOLINTNEXTLINE(hicpp-special-member-functions)
-TEST(PoolResource, AllocationWithOverflow) {
- {
- const size_t max_blocks_per_chunk = 2U;
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, std::numeric_limits<size_t>::max());
- EXPECT_THROW(mem.Allocate(std::numeric_limits<size_t>::max(), 1U), std::bad_alloc);
- // Throws because initial chunk block is aligned to
- // memgraph::utils::Ceil2(block_size), which wraps in this case.
- EXPECT_THROW(mem.Allocate((std::numeric_limits<size_t>::max() - 1U) / max_blocks_per_chunk, 1U), std::bad_alloc);
- }
- {
- const size_t max_blocks_per_chunk = memgraph::utils::impl::Pool::MaxBlocksInChunk();
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, std::numeric_limits<size_t>::max());
- EXPECT_THROW(mem.Allocate(std::numeric_limits<size_t>::max(), 1U), std::bad_alloc);
- // Throws because initial chunk block is aligned to
- // memgraph::utils::Ceil2(block_size), which wraps in this case.
- EXPECT_THROW(mem.Allocate((std::numeric_limits<size_t>::max() - 1U) / max_blocks_per_chunk, 1U), std::bad_alloc);
- }
-}
-
-TEST(PoolResource, BlockDeallocation) {
- TestMemory test_mem;
- const size_t max_blocks_per_chunk = 2U;
- const size_t max_block_size = 64U;
- memgraph::utils::PoolResource mem(max_blocks_per_chunk, max_block_size, &test_mem);
- auto *ptr = CheckAllocation(&mem, max_block_size);
- test_mem.new_count_ = 0U;
- // Do another allocation before deallocating `ptr`, so that we are sure that
- // the chunk of 2 blocks is still alive and therefore `ptr` may be reused when
- // it's deallocated. If we deallocate now, the implementation may choose to
- // free the whole chunk, and we do not want that for the purposes of this
- // test.
- CheckAllocation(&mem, max_block_size);
- EXPECT_EQ(test_mem.new_count_, 0U);
- EXPECT_EQ(test_mem.delete_count_, 0U);
- mem.Deallocate(ptr, max_block_size);
- EXPECT_EQ(test_mem.delete_count_, 0U);
- // CheckAllocation(&mem, max_block_size) will fail as PoolResource should
- // reuse free blocks.
- EXPECT_EQ(ptr, mem.Allocate(max_block_size));
- EXPECT_EQ(test_mem.new_count_, 0U);
-}
-
class AllocationTrackingMemory final : public memgraph::utils::MemoryResource {
public:
std::vector<size_t> allocated_sizes_;
diff --git a/tools/github/clang-tidy/clang-tidy-diff.py b/tools/github/clang-tidy/clang-tidy-diff.py
index 1bdf1da25..609b2eedb 100755
--- a/tools/github/clang-tidy/clang-tidy-diff.py
+++ b/tools/github/clang-tidy/clang-tidy-diff.py
@@ -250,7 +250,6 @@ def main():
common_clang_tidy_args.append("-extra-arg=%s" % arg)
for arg in args.extra_arg_before:
common_clang_tidy_args.append("-extra-arg-before=%s" % arg)
-
for name in lines_by_file:
line_filter_json = json.dumps([{"name": name, "lines": lines_by_file[name]}], separators=(",", ":"))
@@ -266,7 +265,6 @@ def main():
command.extend(common_clang_tidy_args)
command.append(name)
command.extend(clang_tidy_args)
-
task_queue.put(command)
# Wait for all threads to be done.