memgraph/tests/public_benchmark/ldbc/mg

#!/bin/bash

function print_help () {
    echo "Usage: $0 [OPTION]"
    echo "Optional arguments:"
    echo -e "  -h|--help -> Prints help."
    echo -e "  --scale-factor Positive_Integer -> Defines the dataset size."
    echo -e "  --transform-dataset -> Run just transform dataset (SNB -> Memgraph Snapshot)."
    echo -e "  --copy-dataset -> Just copy dataset into the Memgraph snapshots path."
    echo -e "  --run -> Just run Memgraph."
}

set -e
script_dir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"

memgraph_dir="${script_dir}/../../.."
# TODO: pass as an argument
memgraph_build_dir="${memgraph_dir}/build"
loadable_snapshot_path="${memgraph_build_dir}/snapshots/default/3000_01_01__01_01_01_00000"

# Read the arguments.
scale_factor=1
run_all=true
transform_dataset=false
copy_dataset=false
run_memgraph=false
while [[ $# -gt 0 ]]
do
    case $1 in
        -h|--help)
        print_help
        exit 1
        ;;
        --scale-factor)
        scale_factor=$2
        shift
        ;;
        --transform-dataset)
        run_all=false
        transform_dataset=true
        ;;
        --copy-dataset)
        run_all=false
        copy_dataset=true
        ;;
        --run)
        run_all=false
        run_memgraph=true
        ;;
        *)
        # unknown option
        ;;
    esac
    shift # past argument or value
done

snapshot_path="${script_dir}/memgraph_snapshots/snb_scale_${scale_factor}.snapshot"
csv_folder="${script_dir}/neo4j_csv_dataset_scale_${scale_factor}"

# Transform dataset into MemGraph Snapshot.
if [[ ${run_all} = true ]] || [[ ${transform_dataset} = true ]] ; then
    rm ${snapshot_path}
    ${memgraph_dir}/tools/csv_to_snapshot -o ${snapshot_path} \
        --nodes ${csv_folder}/comment_0_0.csv \
        --nodes ${csv_folder}/forum_0_0.csv \
        --nodes ${csv_folder}/organisation_0_0.csv \
        --nodes ${csv_folder}/person_0_0.csv \
        --nodes ${csv_folder}/place_0_0.csv \
        --nodes ${csv_folder}/post_0_0.csv \
        --nodes ${csv_folder}/tag_0_0.csv \
        --nodes ${csv_folder}/tagclass_0_0.csv \
        --relationships ${csv_folder}/comment_hasCreator_person_0_0.csv \
        --relationships ${csv_folder}/comment_hasTag_tag_0_0.csv \
        --relationships ${csv_folder}/comment_isLocatedIn_place_0_0.csv \
        --relationships ${csv_folder}/comment_replyOf_comment_0_0.csv \
        --relationships ${csv_folder}/comment_replyOf_post_0_0.csv \
        --relationships ${csv_folder}/forum_containerOf_post_0_0.csv \
        --relationships ${csv_folder}/forum_hasMember_person_0_0.csv \
        --relationships ${csv_folder}/forum_hasModerator_person_0_0.csv \
        --relationships ${csv_folder}/forum_hasTag_tag_0_0.csv \
        --relationships ${csv_folder}/organisation_isLocatedIn_place_0_0.csv \
        --relationships ${csv_folder}/person_hasInterest_tag_0_0.csv \
        --relationships ${csv_folder}/person_isLocatedIn_place_0_0.csv \
        --relationships ${csv_folder}/person_knows_person_0_0.csv \
        --relationships ${csv_folder}/person_likes_comment_0_0.csv \
        --relationships ${csv_folder}/person_likes_post_0_0.csv \
        --relationships ${csv_folder}/person_studyAt_organisation_0_0.csv \
        --relationships ${csv_folder}/person_workAt_organisation_0_0.csv \
        --relationships ${csv_folder}/place_isPartOf_place_0_0.csv \
        --relationships ${csv_folder}/post_hasCreator_person_0_0.csv \
        --relationships ${csv_folder}/post_hasTag_tag_0_0.csv \
        --relationships ${csv_folder}/post_isLocatedIn_place_0_0.csv \
        --relationships ${csv_folder}/tag_hasType_tagclass_0_0.csv \
        --relationships ${csv_folder}/tagclass_isSubclassOf_tagclass_0_0.csv \
        --csv-delimiter "|" --array-delimiter ";"
    echo "Dataset transformed."
fi

# Copy the dataset.
if [[ ${run_all} = true ]] || [[ ${copy_dataset} = true ]] ; then
    cp ${snapshot_path} ${loadable_snapshot_path}
    echo "Dataset copied."
fi

# Run MemGraph.
if [[ ${run_all} = true ]] || [[ ${run_memgraph} = true ]] ; then
    ${memgraph_build_dir}/memgraph -flagfile ${memgraph_dir}/config/public_benchmark_ldbc.conf 2>&1 &
    memgraph_pid=$!
    sleep 200 # TODO: replace this with something that is going to work in all cases
              #       not just in SNB scale 1 case

    # Create indexes.
    cd ${script_dir}
    if [ ! -d "ve3" ]; then
        virtualenv -p python3 ve3 || command_fail "Virtualenv setup failed."
        source ve3/bin/activate
        pip install -r ${script_dir}/requirements_3.txt
    fi
    source ve3/bin/activate
    python index_creation.py ${script_dir}/ldbc-snb-impls/snb-interactive-neo4j/scripts/indexCreation.neo4j

    # On Ctrl-C stop Memgraph.
    trap ctrl_c INT
    function ctrl_c() {
        kill -9 ${memgraph_pid}
        exit 0
    }
    while true; do
        sleep 1
    done
fi