#!/bin/bash
#script zum durchsuchen von verschiedenen Dateien nach einem Begriff /rb 2018
#erforderlich: ghostscript, libreoffice, tesseract,tesseract-ocr-all, imagemagick, dwg2dxf,unrar
#usage: searchin searchpath mysearch d
if [ -z "$1" ] || [ -z "$2" ];then 
echo please use: 
echo $0 searchpath \"my search\"
echo or for debug:
echo $0 searchpath \"my search\" d
exit 0
fi

search=$2
ret=0
cores=4
found=0
d=$3

echo search for: \"$search\"

IFS=$'\n'
find $1 -type f | while read i; do

fname=`basename -s .$ext "$i"`
ext=`basename ${i##*.}`
uext=`echo  $ext | tr [:upper:] [:lower:]`
fname=`basename -s .$ext $i`
##############################
function searchinPict (){
#####Bild-Dateien durchsuchen#####

 if [[ "$(file "$1")" =~ .*image,* && ($found == 0) ]]; then

test $d && echo search in: "$1"
    if [ -n "$5" ];then
        #tdir=$(mktemp -d)
	tmppath=$6
    else
	#tdir=$5
	tmppath=$1
    fi

    
#    horz=`identify  -format "%h" $1`
#    if [ $horz -lt 200 ];then
     tmpfile=$(mktemp)
     convert -strip -colorspace Gray "$1" -resize 200%  $tmpfile.tif
     tesseract $tmpfile.tif stdout -l deu+eng 2>/dev/null| grep -qi "$2" 
     ret=$? 
     rm $tmpfile $tmpfile.tif
#    else
#     tesseract "$1" stdout -l deu+eng 2>/dev/null | grep -qi "$2" 
#     ret=$?
#    fi
    if [ $ret -eq 0 ]; then 
	echo -e "$2" found in: "$tmppath"
    fi
    #echo tmppath $tmppath
    #rm -rf $tmpdir
    unset tmppath
fi
}
#####PDF-Dateien durchsuchen#####
function searchinPDF () {
#if [ $3 == 'pdf' ];then
if [[ "$(file "$1")" =~ .*PDF.* && ($found == 0) ]]; then
    if [ -n "$5" ];then
        tmpdir=$(mktemp -d)
	tmppath=$6
    else
	tmpdir=$5
	tmppath=$1
    fi
 test $d && echo search in: "$1"
     tmpfile=$(mktemp)
 gs -q -dSAFER -sDEVICE=png16m -dINTERPOLATE -dNumRenderingThreads=$cores \
    -dFirstPage=1 -dLastPage=1 -r300 -o $tmpfile -c 3000000 \
    setvmthreshold -f "$1" -dQUIET -dBATCH -dNOPAUSE \
    -sProcessColorModel=DeviceGray \
    -sColorConversionStrategy=Gray \
    -dOverrideICC 
 tesseract $tmpfile stdout -l deu+eng | grep -qi "$2"
    ret=$?
    if [ $ret -eq 0 ]; then 
	echo -e "$2" found in: "$tmppath"
    fi

    rm -rf $tmpfile  &>/dev/null
    unset tmpfile tmppath tmpdir
fi
}
#####Office-Dateien durchsuchen#####
function searchinOfficefiles () {

if [[ "$(file "$1")" =~ .*Microsoft* && ($found == 0) ]]; then
    if [ -n "$5" ];then
        tdir=$5
	tmppath=$6
    else
	tdir=$(mktemp -d)
	tmppath=$1
    fi
    test $d && echo search in: "$1"
    lowriter --nodefault --nofirststartwizard --nologo --norestore \
    --accept="socket,host=127.0.0.1,port=2002;urp;" --invisible --headless \
    --convert-to html "$1" --outdir $tdir &>/dev/null

    if grep -qir "$2" "$tdir/$4.html";then
        echo -e "$2" found in: "$tmppath"
    fi
    rm -rf "$tdir/$4.html" $tdir &>/dev/null
    unset tdir tmppath
fi
}
#####DWG-Dateien durchsuchen#####
function searchinDWG () {
if [[ "$(file "$1")" =~ .*DWG* && ($found == 0) ]]; then
    if [ -n "$5" ];then
	tdir=$5
	tmppath=$6
    else
	tdir=$(mktemp -d)
	tmppath=$1
    fi
    test $d && echo search in: "$1"
    dwg2dxf $1 $tdir/$4.dxf >>/dev/null
    test $d && echo search in: "$1"
	if grep -qi "$2" $tdir/$4.dxf;then
	    echo -e "$2" found in: "$tmppath"
	fi
    rm -rf  $tdir &>/dev/null
    unset tdir tmppath
fi
}
#####Dateinamen durchsuchen#####
function searchinNAME () {

a="$(tr [A-Z] [a-z] <<< "$1")"
b="$(tr [A-Z] [a-z] <<< "$2")"
test $d && echo search in: "$1"

if [[ "$a" =~ "$b" ]]; then
    echo -e "$2" found in: "$1"
    found=1
    unset tmppath a b
fi
}
#####Ascii-Dateien durchsuchen#####
function searchinASCII () {

if [[ "$(file "$1")" =~ ': ASCII text'$ && ($found == 0) ]]; then
    if [ -n "$5" ];then
	tmppath=$6
    else
	tmppath=$1
    fi
    test $d && echo search in: "$1"
	if grep -qi "$2" $1;then
	    echo -e "$2" found in: "$tmppath"
	fi
    unset tmppath
fi
}
#####GZIP-Dateien durchsuchen#####
function searchinGZIP () {

if [[ "$(file "$1")" =~ .*gzip* && ($found == 0) ]]; then

    if [[ "$(gzip -l $1)" ]];then &>/dev/null

    test $d && echo search in: "$1"
    tdir=$(mktemp -d)
    gunzip -c "$1" >$tdir/$4
    z=`find  $tdir -type f`
    ext=`basename ${z##*.}`
    fname=`basename -s .$ext "$z"`
    searchinNAME $z $2 $uext $fname $tdir $1
    searchinOfficefiles $z $2 $ext $fname $tdir $1
    searchinPict $z $2 $ext $fname $tdir $1 
    searchinPDF $z $2 $ext $fname $tdir $1
    searchinASCII $z $2 $ext $fname $tdir $1
    searchinDWG $z $2 $uext $fname $tdir $1

    rm -rf $tdir 
    unset tdir z buf
    fi
fi
    
}
#####Zip-Dateien durchsuchen?####
function searchinZIP () { 
if [[ "$(file "$1")" =~ .*Zip.* && ($found == 0) ]]; then
test $d && echo search in: "$1"
    if [[ "$(unzip -t $1)" ]];then
    	tdir=$(mktemp -d)
        unzip ---jqod $tdir $1
        find $tdir -type f | while read z;do
        ext=`basename ${z##*.}`
        fname=`basename -s .$ext "$z"`
        searchinNAME $z $2 $uext $fname $tdir $1
        searchinPict $z $2 $ext $fname $tdir $1
        searchinPDF $z $2 $ext $fname $tdir $1
        searchinOfficefiles $z $2 $ext $fname $tdir $1
        searchinASCII $z $2 $uext $fname $tdir $1
        searchinDWG $z $2 $uext $fname $tdir $1

    done
  fi
  rm -rf $tdir
  unset tdir z
fi
}
#####RAR-Dateien durchsuchen?####
function searchinRAR () { 
if [[ "$(file "$1")" =~ .*RAR* && ($found == 0) ]]; then
test $d && echo search in: "$1"
    if [[ "$(unrar -t $1)" ]];then
    	tdir=$(mktemp -d)
        unrar -inul e $1 $tdir
        find $tdir -type f | while read z;do
        ext=`basename ${z##*.}`
        fname=`basename -s .$ext "$z"`
        searchinNAME $z $2 $uext $fname $tdir $1
        searchinPict $z $2 $ext $fname $tdir $1
        searchinPDF $z $2 $ext $fname $tdir $1
        searchinOfficefiles $z $2 $ext $fname $tdir $1
        searchinASCII $z $2 $uext $fname $tdir $1
        searchinDWG $z $2 $uext $fname $tdir $1

    done
  fi
  rm -rf $tdir
  unset tdir z
fi
}
##################################

searchinNAME $i $search #$uext $fname
searchinPict $i $search $uext $fname
searchinPDF $i $search $uext $fname
searchinOfficefiles $i $search $uext $fname 
searchinASCII $i $search $uext $fname
searchinZIP $i $search $uext $fname 
searchinGZIP $i $search $uext $fname
searchinRAR $i $search $uext $fname
searchinDWG $i $search $uext $fname

found=0
#################################
done
#rm -rf /tmp/tmp.*
#ls /tmp/tmp.*
unset fname ext uext search ret d i found