AE_rerun_GMN-D2.sh

#!/bin/bash

DIR=$VSIM_HOME/test/GMN-dataset-2

echo "Clean cached files"
rm $DIR/with_fp-refined/vSim.json
rm -rf $DIR/with_fp-refined/neg_*.csv
rm -rf $DIR/with_fp-refined/pos_*.csv

if [[ $1 == 'full' ]]; then
        echo "Full rerun selected"
        cd $DIR
        rm -rf ba_be_cache.pkl
        rm -rf logs fp-refined
        rm -rf cache
        rm -rf dump
        find . -name "*.df.pkl" | xargs rm -rf
        find . -name "*.csv.pkl" | xargs rm -rf

        mkdir -p logs

        cd $VSIM_HOME

        echo "Running value extraction (This can take several hours)"
        python $VSIM_HOME/scripts/run_se_cmds.py $DIR/se_cmds.txt

        if [ ! -d "$DIR/dump" ]; then
            echo "No dumped files. Exit"
            exit 1
        fi

        echo "Running fingerprint generation (This can take several hours)"
        python $DIR/fp-gen.py
        if [ ! -d "$DIR/fp-refined" ]; then
            echo "No refined fingerprints. Exit"
            exit 1
        fi
fi

echo "Running comparison"
echo "This process needs to create a ba_be_cache.pkl, which can take about half an hour and 11GB disk space."
python $DIR/test_pairwise_match.py

echo "There should be a newly generated pos_*.csv and neg_*.csv"

cd $DIR/with_fp-refined
ls pos_* | xargs python ../vSim2GMN_csv.py
ls neg_* | xargs python ../vSim2GMN_csv.py
cd $VSIM_HOME

python $DIR/their_evaluate.py
echo "There should be a newly generated vSim.json"

echo "Show result of Table V"
python test/GMN-dataset-2/show_table.py test/GMN-dataset-2/with_fp-refined/vSim.json