NixOS/workstation/scripts/sub-sync.sh

#! /usr/bin/env nix-shell
#! nix-shell -i bash -p bash fd ripgrep file alass ffmpeg gum

MEDIA_ROOT=("/mnt/pool/multimedia/media/Series" "/mnt/pool/multimedia/media/Movies")
REPLACE_DIR="/mnt/pool/multimedia/media"
SUBTITLE_MIRROR="/mnt/pool/multimedia/backups/subtitles"
RAM_SUB="/dev/shm/sub.srt"

# BACKUPS SUBTITLES
backup_subtitles() {
    while IFS= read -r subtitle; do
        echo "backing up $subtitle"
        dest_dir="$(dirname "$subtitle")"
        dest_dir="${dest_dir/$REPLACE_DIR/$SUBTITLE_MIRROR}"
        mkdir -p "$dest_dir"
        cp "$subtitle" "${subtitle/$REPLACE_DIR/$SUBTITLE_MIRROR}"
    done < <(fd . -tf -e srt --absolute-path "${MEDIA_ROOT[@]}")
}

clean_up() {
    while IFS= read -r directory; do
        echo "cleaning up $directory"
        subtitles=()
        mapfile -d $'\0' subtitles < <(fd . "$directory" -e srt -tf -d 1 -0)

        if [ "${#subtitles[@]}" -lt 2 ]; then
            continue
        fi

        unset base_subtitle
        unset subtitles_group
        for subtitle in "${subtitles[@]}"; do
            group=()
            mapfile -d $'\0' group < <(fd --fixed-strings \
                "$(basename "$subtitle" .srt)" "$directory" \
                -d 1 -tf -0 -e srt)
            for key in "${!group[@]}"; do
                if ! echo "${group[$key]}" | rg -P '\.\d{1,2}(\.\w+(-\w+)?)?\.srt' -q; then
                    unset "group[$key]"
                    continue
                fi
                if [ -z "${group[$key]}" ]; then
                    continue
                fi
                echo "removing $(basename "$subtitle")"
                rm "$subtitle"
            done
        done
    done < <(fd . -td --absolute-path "${MEDIA_ROOT[@]}")
}

rename_languages() {
    while IFS= read -r file; do
        base=$(basename "$file" .eng.srt)
        dir=$(dirname "$file")
        echo "renaming sub $base"
        mv "$file" "$dir/$base.$2.srt"
    done < <(fd . -tf --absolute-path "${MEDIA_ROOT[@]}" -e "$1.srt")
}

sync_subtitles() {
    while IFS= read -r directory; do
        echo "scanning for sync $directory"
        while IFS= read -r subtitle; do
            echo "processing $subtitle"
            video=()
            extension=$(echo "$subtitle" | rg -oP "(\.\w+(-\w+)?)?\.srt")
            basename="$(basename "$subtitle" "$extension")"
            mapfile -d $'\0' video < <(fd "$basename" \
                "$directory" --fixed-strings \
                -e mkv -e mp4 -e avi -e webm -tf -d 1 -0)

            # skips directory if it contains more than 1 video file
            # should never get triggered
            if [ "${#video[@]}" -gt 1 ]; then
                basename "$(dirname "$directory")"
                echo "$(basename "$directory") has many video files: ${#video[@]}"
                continue
            fi

            # update subtitle in ram
            if [ -e "$RAM_SUB" ]; then
                rm "$RAM_SUB"
            fi
            cp "$subtitle" "$RAM_SUB"

            if [ ! -e $RAM_SUB ] && [ ! -e "${video[0]}" ]; then
                continue
            fi
            echo "processing...$subtitle"
            alass-cli "${video[0]}" "$RAM_SUB" "$subtitle"

        done < <(fd . "$directory" -tf -e srt -d 1 --newer "$1")
    done < <(fd . -td --absolute-path "${MEDIA_ROOT[@]}")
}

find_dupes() {
    while IFS= read -r directory; do
        videos=()
        mapfile -d $'\0' videos < <(fd . \
            "$directory" -tf -d 1 -0 \
            -e mkv -e mp4 -e avi -e webm)

        if [ "${#videos[@]}" == 0 ]; then
            if [[ "$directory" != *"Season"* ]]; then
                continue
            fi
            echo "NO FILES ERROR: $directory"
        fi

        if [ "${#videos[@]}" == 1 ]; then
            continue
        fi

        if [ "${#videos[@]}" -gt 1 ]; then
            if [[ "$directory" == *"media/Movies"* ]]; then
                echo "Movie directory has more than a movie"
                continue
            fi
            for episode in "${videos[@]}"; do
                episode_number="$(echo "$episode" |
                    rg -oP "S\d+E\d+(-E\d+)? ")"
                episode_files="$(
                    fd "$episode_number" "$directory" --fixed-strings \
                        -tf -d 1 \
                        -e mkv -e mp4 -e avi -e webm | wc -l
                )"
                if [ "$episode_files" == 1 ]; then
                    continue
                fi
                echo ____________________________
                echo "The episode $episode_number is repeated on"
                echo "$directory"
                fd "$episode_number" "$directory" --fixed-strings \
                    -tf -d 1 \
                    -e mkv -e mp4 -e avi -e webm
            done
        fi

    done < <(fd . -td --absolute-path "${MEDIA_ROOT[@]}")
}

fd . /mnt/pool/multimedia/media/Series/ --owner jawz -x chown sonarr:piracy {}
fd . /mnt/pool/multimedia/media/Movies/ --owner jawz -x chown radarr:piracy {}
fd . "${MEDIA_ROOT[@]}" -td -x chmod 775 {}
fd . "${MEDIA_ROOT[@]}" -tf -x chmod 664 {}

rename_languages eng en
rename_languages spa es
rename_languages mx es_MX

if [ -n "$1" ]; then
    operation=$1
else
    operation=$(gum choose backup clean sync all)
fi
if [  -n "$2" ]; then
    start_time=$2
else
    start_time="$(date '+%Y-%m-%d') 00:00:00"
fi

case $operation in
    backup)
        backup_subtitles
        ;;
    clean)
        clean_up
        ;;
    sync)
        sync_subtitles "$start_time"
        ;;
    dupe)
        find_dupes
        ;;
    all)
        echo "backing up"
        backup_subtitles
        echo "cleaning up"
        clean_up
        echo "syncing"
        sync_subtitles "$start_time"
        ;;
esac