Server-scripts/audiobook_metadata/metadata_remove_genre.sh

#!/usr/bin/env bash

LIBRARY="$1"
MODE="${2:-}"
REPORT="${3:-/home/lucas/audiobook_genre_cleanup.txt}"

if [[ -z "$LIBRARY" ]]; then
    echo "Usage: $0 <library_path> [--delete] [report_file]"
    exit 1
fi

mapfile -t FILES < <(
    find "$LIBRARY" -type f \( \
        -iname "*.mp3" -o \
        -iname "*.m4b" -o \
        -iname "*.m4a" \
    \)
)

TOTAL=${#FILES[@]}
PROCESSED=0
MATCHES=0
MODIFIED=0

{
    echo "Audiobook Genre Cleanup Report"
    echo "Generated: $(date)"
    echo "Library: $LIBRARY"
    echo "Total files: $TOTAL"
    echo
} > "$REPORT"

for file in "${FILES[@]}"; do
    ((PROCESSED++))

    if (( PROCESSED % 100 == 0 )); then
        echo "Progress: $PROCESSED/$TOTAL files processed"
    fi
    genre=$(kid3-cli -c "get genre" "$file" 2>/dev/null)

    [[ -z "$genre" ]] && continue
    cleaned=$(printf '%s\n' "$genre" | sed -E '
        s/[Aa]udio[[:space:]]*[Bb]ooks?//g
        s/\b(Hörb(uch|ücher)|Hoerb(uch|uecher)|Hörspiel(e)?|Hoerspiel(e)?)\b//Ig
 
        s/[Gg]esprochene[[:space:]]*[Ii]nhalte//g
        s/[Ss]poken[[:space:]]*[Ww]ord//g
        s/[Ss]peech//g
        # Normalize genre names
        s/\b(Sci[ -]?Fi|Science Fiction|SF)\b/SciFi/Ig
        s/\bnonfiction\b/NonFiction/Ig
    
        # Normalize separators (&, :, / -> ;)
        s/\b(Sci[ -]?Fi|Science Fiction|SF)\b/Science Fiction/Ig
        s/[[:space:]]*[,&:\/][[:space:]]*/;/g
    
        # Normalize commas
        s/[[:space:]]*;[[:space:]]*/; /g
    
        # Trim leading/trailing punctuation and whitespace
        s/^[[:space:],;:-]+//
        s/[[:space:],;-]+$//
    ')
#           s/.*//;
    
    if [[ "$genre" != "$cleaned" ]]; then
        ((MATCHES++))

        status="MATCH"

        if [[ "$MODE" == "--delete" ]]; then
            kid3-cli \
                -c "tag 2" \
                -c "set genre \"$cleaned\"" \
                -c save \
                "$file"
            ((MODIFIED++))
            status="UPDATED"
        fi

        printf '[%d/%d] [%s] %s | OLD="%s" | NEW="%s"\n' \
            "$PROCESSED" "$TOTAL" "$status" "$file" "$genre" "$cleaned"

        printf '[%d/%d] [%s] %s | OLD="%s" | NEW="%s"\n' \
            "$PROCESSED" "$TOTAL" "$status" "$file" "$genre" "$cleaned" >> "$REPORT"
    fi

done

{
    echo
    echo "========== SUMMARY =========="
    echo "Files scanned : $TOTAL"
    echo "Matches found : $MATCHES"
    echo "Files updated : $MODIFIED"
} >> "$REPORT"

echo
echo "========== SUMMARY =========="
echo "Files scanned : $TOTAL"
echo "Matches found : $MATCHES"
echo "Files updated : $MODIFIED"
echo "Report written to: $REPORT"