#!/bin/bash

NAME_REGEX=".*"
LINE_L=1
LINE_R=-1
DIR="."
DLIM=""
HAS_WORD="false"
WORDLIST=""

ER () {
    echo $1
    exit 1
}

while [[ $# > 0 ]]; do
    arg=$1
    shift
    case $arg in
        -name)
            NAME_REGEX=$1
            shift
        ;;
        -l)
            SEG=$1
            if [[ $SEG =~ .*:.* ]]; then
                ## two parameters
                LINE_L=${SEG%:*}
                LINE_R=${SEG##*:}
            else
                LINE_L=$SEG
                LINE_R=$SEG
            fi
            if (($LINE_L == 0)); then
                ER "LINE cannot be 0 sorry QQ"
            fi
            shift
        ;;
        -sep)
            DLIM=$1
            shift
        ;;
        -c)
            WORDLIST=$1
            HAS_WORD="true"
            shift
        ;;
        *)
            DIR=$arg
        ;;
    esac
done

WORDLIST=$(tr "," " " <<< $WORDLIST)

for i in $(find $DIR -type f); do
    if ! [[ $(basename $i) =~ $NAME_REGEX ]]; then
        continue
    fi
    echo "Processing file $(basename $i):"

    ALL=$(wc -l < $i)
    L=$LINE_L
    R=$LINE_R

    if (( $L < 0 )); then
        L=$((ALL+L+1))
        if (( $L < 0 )); then
            L=1
        fi
    fi

    if (( $R < 0 )); then
        R=$((ALL+R+1))
    fi

    if (( $L > $R )); then
        continue
    fi

    if [[ $HAS_WORD == "true" ]]; then

        if [[ $DLIM == "" ]]; then
            CONT=$(sed -n "${L},${R}p" < $i)
            for w in $WORDLIST; do
                echo " Word $w occurs: $(grep -o "$w" <<< "$CONT" | wc -l) times"
            done
        else
            for w in $WORDLIST; do
                T=0
                CONT=$(sed -n "${L},${R}p" < $i)
                while IFS="$DLIM" read -ra ADDR; do
                    for b in "${ADDR[@]}"; do
                        if [[ $b == $w ]]; then
                            ((++T))
                        fi
                    done
                done <<< "$CONT"
                echo " Word $w occurs: $T times"
            done
        fi
    else
        sed -n "${L},${R}p" < $i
    fi
done