#!/bin/bash

SEARCHPATH='/var/data'
KEYWORDS="-e langenscheidt"
TMPPATH='/var/data/tmp'

ARCHIVES=('zip' 'tar' 'gz' 'bz2' 'xz')

MATCHES_NOW=$(find . | grep $KEYWORDS)
MATCHES="$MATCHES\n$MATCHES_NOW"
echo $MATCHES_NOW >> partial-results.txt

ARCHIVE_NO=0
for EXT in $ARCHIVES; do
        echo "Checking for archives (*.$EXT)"
        XZ=$(find "$SEARCHPATH" -iname "*.$EXT")

        for ARCHIVE in $ZIP; do
                echo "Checking $ARCHIVE"

                mkdir -pv "$TMPPATH"
                cd "$TMPPATH"

                case $EXT in
                        zip)
                                unzip $ARCHIVE
                        ;;

                        tar)
                                tar -xf $ARCHIVE
                        ;;

                        gz|bz2)
                                tar -xf $ARCHIVE
                        ;;

                        xz)
                                cat $ARCHIVE | unxz -1 - | tar -x
                        ;;
                esac

                MATCHES_NOW=$(find . | grep $KEYWORDS)
                MATCHES="$MATCHES\n$MATCHES_NOW"
                echo $MATCHES_NOW >> partial-results.txt
                if [ ! -z "$MATCHES_NOW" ]; then
                        echo $MATCHES_NOW > "archive-$ARCHIVE_NO.txt"
                fi
                echo "$ARCHIVE_NO:$ARCHIVE" > "archives.txt"

                rm -rf "$TMPPATH"
                ARCHIVE_NO=$(($ARCHIVE_NO+1))
        done
done

echo $MATCHES > results.txt