csvmake

#!/bin/bash
# executes taxdata Python scripts through the specified last stage

function usage {
    echo "USAGE: ./csvmake DATATYPE LASTSTAGE"
    echo "       where DATATYPE can be puf or cps"
    echo "         and LASTSTAGE can be 1 or 2 or 3"
    echo "       Note: run ./csvmake in top-level taxdata directory"
    echo "Execution time will be roughly one hour when LASTSTAGE is 2 or 3" 
    exit 1
}

function gitdiffs {
    git diff --quiet
    if [ $? -eq 0 ]; then
        echo "NO 'git diff' OUTPUT ==> NO CHANGES IN FILES"
    else
        echo "SOME 'git diff' OUTPUT ==> CHANGES IN FILES"
        echo "USE 'git status' TO SEE IF CSV OUTPUT FILES CHANGED"
    fi
}

# process command-line arguments
if [ $# -ne 2 ]; then
    echo "ERROR: number of command-line arguments is not two"
    usage
fi
DTYPE=$1
if [ $DTYPE != "puf" ] && [ $DTYPE != "cps" ]; then
    echo "ERROR: DATATYPE is neither puf nor cps"
    usage
fi
LSTAGE=$2
if [ $LSTAGE -ne 1 ] && [ $LSTAGE -ne 2 ] && [ $LSTAGE -ne 3 ]; then
    echo "ERROR: LASTSTAGE is neither 1 nor 2 nor 3"
    usage
fi

# TODO TEMPORARY CODE BELOW
if [ $DTYPE == "cps" ]; then
    echo "ERROR: DATATYPE cannot be cps yet"
    usage
fi
# TODO TEMPORARY CODE ABOVE

# execute $DTYPE_data script before taking steps
DP=$DTYPE"_data"
cd $DP
echo "`date` : $DP START"
python finalprep.py
if [ $? -ne 0 ]; then
    echo "ERROR: executing $DP_data/finalprep.py script"
    gitdiffs
    exit 1
fi
echo "`date` : $DP FINISH"
cd ..

# execute stage1 scripts
cd stage1
echo "`date` : stage1 START"
python stage1.py
if [ $? -ne 0 ]; then
    echo "ERROR: executing stage1.py script"
    gitdiffs
    exit 1
fi
python factors_finalprep.py
if [ $? -ne 0 ]; then
    echo "ERROR: executing stage1/finalprep.py script"
    gitdiffs
    exit 1
fi
echo "`date` : stage1 FINISH"
cd ..

# consider stopping after stage1 scripts have executed
if [ $LSTAGE -eq 1 ]; then
    echo "STOPPING AFTER stage1"
    gitdiffs
    exit 0
fi

# execute $DTYPE_stage2 script
S2=$DTYPE"_stage2"
cd $S2
echo "`date` : $S2 START"
python stage2.py
RC=$?
rm -f *.pyc
if [ $RC -ne 0 ]; then
    echo "ERROR: executing $S2/stage2.py script"
    gitdiffs
    exit 1
fi
echo "`date` : $S2 FINISH"
cd ..

# consider stopping after $DTYPE_stage2 script has executed
if [ $LSTAGE -eq 2 ]; then
    echo "STOPPING AFTER $S2"
    gitdiffs
    exit 0
fi

# execute $DTYPE_stage3 script
S3=$DTYPE"_stage3"
cd $S3
echo "`date` : $S3 START"
python stage3.py
if [ $? -ne 0 ]; then
    echo "ERROR: executing $S3/stage3.py script"
    gitdiffs
    exit 1
fi
echo "`date` : $S3 FINISH"
cd ..

# consider stopping after $DTYPE_stage3 script has executed
if [ $LSTAGE -eq 3 ]; then
    echo "STOPPING AFTER $S3"
    gitdiffs
    exit 0
fi

exit 0