tursom/memgraph
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Extend text index operations

Browse Source
This commit is contained in:
Ante Pušić 2024-01-10 01:23:28 +01:00
parent 00275f5736
commit ba581f58ef
4 changed files with 70 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/query/db_accessor.hpp
View File

@ -636,7 +636,7 @@ class DbAccessor final {
utils::BasicResult<storage::StorageIndexDefinitionError, void> CreateTextIndex(std::string index_name,
storage::LabelId label) {
return accessor_->CreateTextIndex(index_name, label);
return accessor_->CreateTextIndex(index_name, label, this);
}
utils::BasicResult<storage::StorageIndexDefinitionError, void> DropTextIndex(std::string index_name) {

59
src/storage/v2/indices/text_index.hpp
View File

@ -9,6 +9,7 @@
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#include "query/db_accessor.hpp"
#include "storage/v2/mgcxx_mock.hpp"
namespace memgraph::storage {
@ -28,6 +29,8 @@ class TextIndex {
void UpdateOnRemoveLabel(LabelId removed_label, Vertex *vertex_after_update, const Transaction &tx) {}
void UpdateOnSetProperty(PropertyId property, const PropertyValue &value, Vertex *vertex, const Transaction &tx) {}
std::vector<mgcxx_mock::text_search::IndexContext *> GetApplicableTextIndices(Vertex *vertex, Storage *storage) {
std::vector<mgcxx_mock::text_search::IndexContext *> applicable_text_indices;
for (const auto &label : vertex->labels) {
@ -38,52 +41,54 @@ class TextIndex {
return applicable_text_indices;
}
/// @throw std::bad_alloc
bool CreateIndex(std::string index_name, LabelId label,
const std::optional<durability::ParallelizedSchemaCreationInfo> &parallel_exec_info) {
auto index_config = mgcxx_mock::text_search::IndexConfig{.mappings = "TODO antepusic"};
bool CreateIndex(std::string index_name, LabelId label, memgraph::query::DbAccessor *db) {
auto index_config = mgcxx_mock::text_search::IndexConfig{
.mappings = "TODO devise the mapping by reading the indexable nodes' properties"};
auto new_index = mgcxx_mock::text_search::Mock::create_index(index_name, index_config);
index_[index_name] = new_index;
label_to_index_[label] = index_name;
return true;
// TODO add documents to index
// TODO add documents (indexable nodes) to index
}
bool DropIndex(std::string index_name) {
mgcxx_mock::text_search::Mock::drop_index(index_name);
index_.erase(index_name);
std::erase_if(label_to_index_, [index_name](const auto &item) { return item.second == index_name; });
return true;
}
bool IndexExists(std::string index_name) { return index_.contains(index_name); }
std::vector<Gid> Search(std::string index_name, std::string search_query) {
auto input = mgcxx_mock::text_search::SearchInput{.search_query = search_query, .return_fields = {"metadata.gid"}};
// Basic check for search fields in the query (Tantivy syntax delimits them with `:` to the right)
if (search_query.find(":") == std::string::npos) {
input.search_fields = {"data"};
auto input = mgcxx_mock::text_search::SearchInput {
.search_query = search_query, .return_fields = {{"metadata.gid"}};
// Basic check for search fields in the query (Tantivy syntax delimits them with a `:` to the right)
if (search_query.find(":") == std::string::npos) {
input.search_fields = {"data"};
}
std::vector<Gid> found_nodes;
for (const auto &doc : mgcxx_mock::text_search::Mock::search(index_.at(index_name), input).docs) {
found_nodes.push_back(storage::Gid::FromString(doc.data));
}
return found_nodes;
}
std::vector<Gid> found_nodes;
for (const auto &doc : mgcxx_mock::text_search::Mock::search(index_.at(index_name), input).docs) {
found_nodes.push_back(storage::Gid::FromString(doc.data));
std::vector<std::string> ListIndices() {
std::vector<std::string> ret;
ret.reserve(index_.size());
for (const auto &item : index_) {
ret.push_back(item.first);
}
return ret;
}
return found_nodes;
}
std::vector<std::string> ListIndices() {
std::vector<std::string> ret;
ret.reserve(index_.size());
for (const auto &item : index_) {
ret.push_back(item.first);
}
return ret;
}
uint64_t ApproximateVertexCount(std::string index_name) { return 10; }
uint64_t ApproximateVertexCount(std::string index_name) { return 10; }
std::map<std::string, mgcxx_mock::text_search::IndexContext> index_;
std::map<LabelId, std::string> label_to_index_;
};
std::map<std::string, mgcxx_mock::text_search::IndexContext> index_;
std::map<LabelId, std::string> label_to_index_;
};
} // namespace memgraph::storage

7
src/storage/v2/storage.hpp
View File

@ -268,10 +268,9 @@ class Storage {
virtual utils::BasicResult<StorageIndexDefinitionError, void> DropIndex(LabelId label, PropertyId property) = 0;
virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateTextIndex(std::string index_name,
LabelId label) {
// TODO: pass vertices to CreateIndex
storage_->indices_.text_index_->CreateIndex(index_name, label, std::nullopt);
virtual utils::BasicResult<StorageIndexDefinitionError, void> CreateTextIndex(std::string index_name, LabelId label,
DbAccessor *db) {
storage_->indices_.text_index_->CreateIndex(index_name, label, db);
return {};
}

61
src/storage/v2/vertex_accessor.cpp
View File

@ -15,6 +15,8 @@
#include <tuple>
#include <utility>
#include <fmt/format.h>
#include "flags/run_time_configurable.hpp"
#include "query/exceptions.hpp"
#include "storage/v2/disk/storage.hpp"
@ -276,12 +278,13 @@ Result<PropertyValue> VertexAccessor::SetProperty(PropertyId property, const Pro
CreateAndLinkDelta(transaction, vertex, Delta::SetPropertyTag(), property, current_value);
if (flags::run_time::GetTextSearchEnabled()) {
for (const auto *index_context : storage->indices_.text_index_->GetApplicableTextIndices(vertex, storage)) {
auto search_input = mgcxx_mock::text_search::SearchInput{};
auto search_input = mgcxx_mock::text_search::SearchInput{
.search_query = fmt::format("metadata.gid:{}", vertex->gid.AsInt()), .return_fields = {"data"}};
auto search_result = mgcxx_mock::text_search::Mock::search(*index_context, search_input);
mgcxx_mock::text_search::Mock::delete_document(*index_context, search_input, true);
// parse result to JSON, set property in JSON and convert to string
auto new_properties = search_result.docs[0].data;
auto new_properties = search_result.docs[0].data; // TODO (pending real Tantivy results): parse result to
// JSON, set property and convert back to string
auto new_properties_document = mgcxx_mock::text_search::DocumentInput{.data = new_properties};
mgcxx_mock::text_search::Mock::add(*index_context, new_properties_document, true);
}
@ -321,7 +324,9 @@ Result<bool> VertexAccessor::InitProperties(const std::map<storage::PropertyId,
}
if (flags::run_time::GetTextSearchEnabled()) {
for (const auto *index_context : storage->indices_.text_index_->GetApplicableTextIndices(vertex, storage)) {
auto new_properties_document = mgcxx_mock::text_search::DocumentInput{}; // empty properties
auto new_properties_document =
mgcxx_mock::text_search::DocumentInput{}; // TODO (pending real Tantivy operation): create a JSON, set
// properties and convert to string
mgcxx_mock::text_search::Mock::add(*index_context, new_properties_document, true);
}
}
@ -362,12 +367,13 @@ Result<std::vector<std::tuple<PropertyId, PropertyValue, PropertyValue>>> Vertex
id_old_new_change.emplace(vertex->properties.UpdateProperties(properties));
if (flags::run_time::GetTextSearchEnabled()) {
for (const auto *index_context : storage->indices_.text_index_->GetApplicableTextIndices(vertex, storage)) {
auto search_input = mgcxx_mock::text_search::SearchInput{};
auto search_input = mgcxx_mock::text_search::SearchInput{
.search_query = fmt::format("metadata.gid:{}", vertex->gid.AsInt()), .return_fields = {"data"}};
auto search_result = mgcxx_mock::text_search::Mock::search(*index_context, search_input);
mgcxx_mock::text_search::Mock::delete_document(*index_context, search_input, true);
// parse result to JSON, set property in JSON and convert to string
auto new_properties = search_result.docs[0].data;
auto new_properties = search_result.docs[0].data; // TODO (pending real Tantivy results): parse result to
// JSON, set property and convert back to string
auto new_properties_document = mgcxx_mock::text_search::DocumentInput{.data = new_properties};
mgcxx_mock::text_search::Mock::add(*index_context, new_properties_document, true);
}
@ -404,27 +410,28 @@ Result<std::map<PropertyId, PropertyValue>> VertexAccessor::ClearProperties() {
using ReturnType = decltype(vertex_->properties.Properties());
std::optional<ReturnType> properties;
utils::AtomicMemoryBlock atomic_memory_block{
[storage = storage_, transaction = transaction_, vertex = vertex_, &properties]() {
properties.emplace(vertex->properties.Properties());
if (!properties.has_value()) {
return;
}
for (const auto &[property, value] : *properties) {
CreateAndLinkDelta(transaction, vertex, Delta::SetPropertyTag(), property, value);
storage->indices_.UpdateOnSetProperty(property, PropertyValue(), vertex, *transaction);
transaction->constraint_verification_info.RemovedProperty(vertex);
transaction->manyDeltasCache.Invalidate(vertex, property);
}
utils::AtomicMemoryBlock atomic_memory_block{[storage = storage_, transaction = transaction_, vertex = vertex_,
&properties]() {
properties.emplace(vertex->properties.Properties());
if (!properties.has_value()) {
return;
}
for (const auto &[property, value] : *properties) {
CreateAndLinkDelta(transaction, vertex, Delta::SetPropertyTag(), property, value);
storage->indices_.UpdateOnSetProperty(property, PropertyValue(), vertex, *transaction);
transaction->constraint_verification_info.RemovedProperty(vertex);
transaction->manyDeltasCache.Invalidate(vertex, property);
}
vertex->properties.ClearProperties();
if (flags::run_time::GetTextSearchEnabled()) {
for (const auto *index_context : storage->indices_.text_index_->GetApplicableTextIndices(vertex, storage)) {
auto search_input = mgcxx_mock::text_search::SearchInput{};
mgcxx_mock::text_search::Mock::delete_document(*index_context, search_input, true);
}
}
}};
vertex->properties.ClearProperties();
if (flags::run_time::GetTextSearchEnabled()) {
for (const auto *index_context : storage->indices_.text_index_->GetApplicableTextIndices(vertex, storage)) {
auto search_input =
mgcxx_mock::text_search::SearchInput{.search_query = fmt::format("metadata.gid:{}", vertex->gid.AsInt())};
mgcxx_mock::text_search::Mock::delete_document(*index_context, search_input, true);
}
}
}};
std::invoke(atomic_memory_block);
return properties.has_value() ? std::move(properties.value()) : ReturnType{};