#!/usr/bin/env bash

set -a
export SOURCE_HOME=`dirname "$0"`
export SOURCE_HOME=`cd "$SOURCE_HOME/bigflow"; pwd`
TARGET_DIR=$1
[ -z $TARGET_DIR ] && TARGET_DIR=.

ERROR_LOG=$PWD/error_logs
rm -rf $ERROR_LOG
mkdir -p $ERROR_LOG
FAILED=0

#environment
#export PIPELINE_TYPE=hadoop
#export HDFS_ROOT_PATH=hdfs:///app/inf/dataflow/unittest
#export PARALLEL_RUN=1

function run_test() {
    local dir
    local filename
    dir=`dirname $1`
    filename=`basename $1`
    pushd $dir > /dev/null
    local LOGFILE=$PWD/${filename%.py}.test_log
    rm -rf $LOGFILE
    echo "Running test: $1" | tee -a $LOGFILE

    bash -c '"$SOURCE_HOME"/bin/pyrun $filename | tee ${LOGFILE}.stdout; exit ${PIPESTATUS[0]}' >$LOGFILE  2>&1
    FAILED=$?

    if [ $(stat -c %s ${LOGFILE}.stdout) -ne 0 ]; then
        echo [FAIL] stdout should have been empty.
        echo Unexpected stdout in $1. The output is:
        cat ${LOGFILE}.stdout
        FAILED=-1
    else
        echo [OK] As expected, stdout is empty.
    fi

    rm ${LOGFILE}.stdout
    cat $LOGFILE | grep "Warning"

    # Fail and exit on the first test failure.
    if [[ $FAILED != 0 ]]; then
        cat $LOGFILE | grep -v "^[0-9][0-9]*" # filter all lines starting with a number.
        echo -en "\033[31m"  # Red
        echo "Had test failures; see logs at $ERROR_LOG/${filename%.py}.test_log"
        echo -en "\033[0m"  # No color
        cp $LOGFILE $ERROR_LOG
    fi

    popd > /dev/null

    rm -rf $LOGFILE
    #[ $FAILED -ne 0 ] && exit -1
}

function pp {
    echo "run $1"
}

function parallel_run_all_tests {
    # nohup $SOURCE_HOME/bin/bigflow_server --service_port=60019 > /dev/null 2>$1 < /dev/null &
    # SERVER_PID=$!
    # echo "PID: $SERVER_PID"
    local SLEEP_SEC=1
    local PARALLEL_NUM=20
    for target in `find ${TARGET_DIR} -name '*_test.py' -not -path '*/\.*' | grep -v python_runtime`; do
        run_test $target &
        while [ $(jobs | wc -l) -ge ${PARALLEL_NUM} ]
        do
            sleep ${SLEEP_SEC}s
            if [ -e $ERROR_LOG/* ]; then
                exit -1
            fi
        done
    done
    wait
    if [ -e $ERROR_LOG/* ]; then
        exit -1
    fi
}

function run_all_tests {
    # nohup $SOURCE_HOME/bin/bigflow_server --service_port=60019 > /dev/null 2>$1 < /dev/null &
    # SERVER_PID=$!
    # echo "PID: $SERVER_PID"
    for target in `find ${TARGET_DIR} -name '*_test.py' -not -path '*/\.*' | grep -v python_runtime`; do
        { # try

            run_test $target

        } || { # catch
            echo $target >> test_status.log
        }
    done
}

echo "Running Bigflow Python tests. Output is in $LOG_FILE."

if [ "$PARALLEL_RUN" != "" ]; then
    parallel_run_all_tests "$1"
else
    run_all_tests
fi

# wait all background job finishes if we already have failures.
wait

if [ -d ${ERROR_LOG} -a "$(ls -A ${ERROR_LOG})" ]; then
    # only display the last 500 lines of log in err_log
    tail -n 500 $ERROR_LOG/*
    exit -1
else
    echo -en "\033[32m"  # Green
    echo "Tests passed."
    echo -en "\033[0m"  # No color
fi

echo "Testing with Python version:"
${SOURCE_HOME}/bin/pyrun --version 1>/dev/null