#!/bin/bash
#
## sort photorec results into something usefull

## Set paths
recup_dir="${1%/}"
[ -n "$recup_dir" ] || recup_dir="."
recup_dir="$(realpath "$recup_dir")"
out_dir="$recup_dir/Recovered"
bad_dir="$recup_dir/Corrupt"

## Test path before starting (using current dir if not specified)
for d in $recup_dir/recup*; do
    ### Source: http://stackoverflow.com/a/6364244
    ## Check if the glob gets expanded to existing files.
    ## If not, f here will be exactly the pattern above
    ## and the exists test will evaluate to false.
    [ -e "$d" ] && echo "Found recup folder(s)" || {
        echo "ERROR: No recup folders found"
        echo "Usage: $0 recup_dir"
        exit 1
    }

    ## This is all we needed to know, so we can break after the first iteration
    break
done

# Hard link files into folders by type
for d in $recup_dir/recup*; do
    if [ -d "$d" ]; then
        echo "Linking $d"
        pushd $d >/dev/null
        find -type f | while read k; do
            file="$(basename "$k")"
            src="$(realpath "$k")"
            ext="$(echo "${file##*.}" | tr '[:upper:]' '[:lower:]')"
            ext_dir="$out_dir/$ext"
            if [ "${file##*.}" = "$file" ]; then
                ext_dir="$out_dir/_MISC_"
            elif [ "$ext" = "jpg" ] && [ "${file:0:1}" = "t" ]; then
                ext_dir="$out_dir/jpg-thumbnail"
            fi
            #echo "  $file -> $ext_dir"
            [ -d "$ext_dir" ] || mkdir -p "$ext_dir"
            ln "$src" "$ext_dir"
        done
        popd >/dev/null
    else
        echo "ERROR: '$d' not a directory"
    fi
done

## Check the files output by photorec for corruption
pushd "$out_dir" >/dev/null

# Check archives with 7-Zip
#for d in 7z bz2 gz lzh lzo rar tar xz zip; do
#    if [ -d "$d" ]; then
#        echo "Checking $d files"
#        pushd "$d" >/dev/null
#        for f in *; do
#            if ! 7z t "$f" >/dev/null 2>&1; then
#                #echo "  BAD: $f"
#                [ -d "$bad_dir/$d" ] || mkdir -p "$bad_dir/$d"
#                mv -n "$f" "$bad_dir/$d/$f"
#            fi
#        done
#        popd >/dev/null
#    fi
#done

# Check Audio/Video files with ffprobe
for d in avi flac flv m4a m4p m4v mkv mid mov mp2 mp3 mp4 mpg mpg2 ogg ts vob wav; do
    if [ -d "$d" ]; then
        echo "Checking $d files"
        pushd "$d" >/dev/null
        for f in *; do
            if ! ffprobe "$f" >/dev/null 2>&1; then
                #echo "  BAD: $f"
                [ -d "$bad_dir/$d" ] || mkdir -p "$bad_dir/$d"
                mv -n "$f" "$bad_dir/$d/$f"
            fi
        done
        popd >/dev/null
    fi
done

# Check .doc files with antiword
if [ -d "doc" ]; then
    echo "Checking doc files"
    pushd "doc" >/dev/null
    for f in *doc; do
        if ! antiword "$f" >/dev/null 2>&1; then
            #echo "  BAD: $f"
            [ -d "$bad_dir/doc" ] || mkdir -p "$bad_dir/doc"
            mv -n "$f" "$bad_dir/doc/$f"
        fi
    done
    popd >/dev/null
fi

# Check .docx files with 7z and grep
if [ -d "docx" ]; then
    echo "Checking docx files"
    pushd "docx" >/dev/null
    for f in *docx; do
        if ! 7z l "$f" | grep -q -s "word/document.xml"; then
            #echo "  BAD: $f"
            [ -d "$bad_dir/docx" ] || mkdir -p "$bad_dir/docx"
            mv -n "$f" "$bad_dir/docx/$f"
        fi
    done
    popd >/dev/null
fi

# Sort pictures by date (only for common camera formats)
for d in jpg mrw orf raf raw rw2 tif x3f; do
    if [ -d "$d" ]; then
        echo "Sorting $d files by date"
        pushd "$d" >/dev/null
        for f in *; do
            date_dir="$(date -d "$(stat -c %y "$f")" +"%F")"
            [ -d "$date_dir" ] || mkdir "$date_dir"
            mv -n "$f" "$date_dir/"
        done
        popd >/dev/null
    fi
done

# Sort mov files by encoded date
if [ -d "mov" ]; then
    echo "Sorting mov files by date"
    pushd "mov" >/dev/null
    for f in *mov; do
        enc_date="$(mediainfo "$f" | grep -i "Encoded date" | head -1 | sed -r 's/.*: //')"
        date_dir="$(date -d "$enc_date" +"%F")"
        echo "$date_dir" | grep -E -q -s '^[0-9]{4}-[0-9]{2}-[0-9]{2}$' || date_dir="Unknown Date"
        [ -d "$date_dir" ] || mkdir "$date_dir"
        mv -n "$f" "$date_dir/"
    done
    popd >/dev/null
fi

## sort audio files by tags

## sort matroska files by metadata

## return to original dir
popd >/dev/null