#!/bin/bash

INPUT_FILE=""
OUTPUT_FILE=""
KEEP_ALL=1
FILE_REGEX=()
CHECK_FILES=0

CMD_OPTS=$(getopt \
    -o hi:o:Imc \
    -l help,input-file:,output-file:,inspirals,merged-triggers,check-files \
    -n pycbc_copy_output_map \
    -- "$@")

eval set -- "$CMD_OPTS"

while true
do
    case "$1" in
	-h|--help)
	    echo "usage: pycbc_copy_output_map [-h] -i input_map -o output_map [optional arguments]"
	    echo
	    echo "required arguments:"
	    echo "  -i, --input-file        the .map file that will be copied from"
	    echo "  -o, --output-file       the name of the output file to be created"
	    echo
	    echo "optional arguments:"
	    echo "  -h, --help              show this help message and exit"
	    echo "  -I, --inspirals         if given all entries will no longer be copied,"
	    echo "                            only inspirals and any other specified file types."
	    echo "  -m, --merged-triggers   if given all entries will no longer be copied,"
	    echo "                            only merged-triggers and any other specified file types."
	    echo "  -c, --check-files       check if the files exist with non-zero size before copying."
	    echo
	    exit 0 ;;
	-i|--input-file)
	    INPUT_FILE=$(readlink -f "$2")
	    shift 2 ;;
	-o|--output-file)
	    OUTPUT_FILE=$(readlink -f "$2")
	    shift 2 ;;
	-I|--inspirals)
	    KEEP_ALL=0
	    FILE_REGEX+=("1-INSPIRAL")
	    shift 1 ;;
	-m|--merged-triggers)
	    KEEP_ALL=0
	    FILE_REGEX+=("1-HDF_TRIGGER_MERGE")
	    shift 1 ;;
	-c|--check-files)
	    CHECK_FILES=1
	    shift 1 ;;
	--)
	    break ;;
    esac
done

if [ "x$INPUT_FILE" == "x" ]
then
    echo "-i, --input-file must be specified, see -h, --help for options"
    exit 1
fi

if [ "x$OUTPUT_FILE" == "x" ]
then
    echo "-o, --output-file must be specified, see -h, --help for options"
    exit 1
fi

if [ ! -s "$INPUT_FILE" ]
then
    echo "$INPUT_FILE does not exist or is empty."
    exit 1
elif [[ "$INPUT_FILE" != *.map ]]
then
    echo "$INPUT_FILE is not a .map file."
    exit 1
fi

if [ -f "$OUTPUT_FILE" ]
then
    echo "$OUTPUT_FILE already exists."
    exit 1
fi

touch "$OUTPUT_FILE"
ERROR=""
MESSAGE=""

while read -r LINE
do
    KEEP_FILE=0
    ARRAY=($LINE)
    if (( ${#ARRAY[@]} != 3 ))
    then
	ERROR="${ERROR}A line contained ${#ARRAY[@]} space seperated entries.\n" 
	ERROR="${ERROR}A map file should contain 3 per line.\n"
	break
    elif [[ "${ARRAY[1]}" =~ \.dax$ ]] || [[ "${ARRAY[1]}" =~ \.dax\.map$ ]] || [[ "${ARRAY[1]}" =~ \.tc\.txt$ ]]
    then
	MESSAGE="${MESSAGE}Skipping subdax file: ${ARRAY[1]}\n"
	continue
    elif (( KEEP_ALL == 1 ))
    then
	KEEP_FILE=1
    fi

    for REGEX in ${FILE_REGEX[@]}
    do
	if [[ "${ARRAY[0]}" =~ $REGEX ]]
	then
	    KEEP_FILE=1
	    break
	fi
    done

    if (( KEEP_FILE != 1 ))
    then
	continue
    elif (( CHECK_FILES == 0 ))
    then
	echo "$LINE"
	continue
    elif [ -s "${ARRAY[1]}" ]
    then
	echo "$LINE"
    fi

done < "$INPUT_FILE" >> $OUTPUT_FILE

if [ "x${MESSAGE}" != "x" ]
then
    printf "$MESSAGE"
fi

if [ "x${ERROR}" != "x" ]
then
    printf "$ERROR"
    exit 1
fi