dotfiles/scripts/.local/bin/vidl

#!/usr/bin/env bash
# download a file if it does not exist in the archive alread
# otherwise just re-link it from the archive

show_help() {
    printf """
vidl: Video downloader

Simple wrapper for youtube-dl (or yt-dlp or similar). 

Usage:  vidl [OPTION] <link>

Point it to a link you want downloaded.

Options:

    -h      Display this help.

    -d      Directory to check for existence of file and archive to if needed.
            If this is passed will not download existing archived files again.

    -f      Directory to download to.

    -c      Clear existing download queue.

    -p      Print out number of remaining items in queue.

    -t      Point to youtube-dl command to use. Can be command or absolute link.
            By default will use \`yt-dlp\` which is a more up to date fork of the 
            youtube-dl program.
"""
}

urls=("$@")

while getopts "t:f:d:hcp" opt; do
    case "$opt" in
    # v)  verbose=1
    # ;;
    f)
        DL_FOLDER="$OPTARG"
        ;;
    t)
        YT_DL_CMD="$OPTARG"
        ;;
    d)
        ARCHIVE_FOLDER="$OPTARG"
        ;;
    c)
        ONLY_DO=clear
        ;;
    p)
        ONLY_DO=remaining
        ;;
    h | \? | *)
        show_help
        exit 0
        ;;
    esac
done
shift $((OPTIND - 1))

get_ytdl_fname() {
    yt_dl_fname=$("$YT_DL_CMD" --get-filename -o "$YT_DL_TITLE" "$*")
    if [ -z "$yt_dl_fname" ]; then
        echo Could not get video filename, error with youtube-dl.
        exit 1
    fi
}

_call_archive() { # 1=new_file, 2=archive_location
    command -v archive >/dev/null 2>&1 || {
        echo "archive command not found."
        exit 1
    }
    archive -d "$2" -a "$1"
}

vid_ext="\(mp4\|avi\|webm\|mkv\|mpe?g\|3gp\|m4a\)"
_findfile() { # 1=directory, 2=file
    find "$1" -type f -name "$2.*" | sed -ne "/$2\.$vid_ext$/Ip"
}

_alreadyexists() { # 1=video_regex
    if [ ! -d "$ARCHIVE_FOLDER" ]; then return 0; fi
    local found
    found=$(_findfile "$ARCHIVE_FOLDER" "$1")
    if [ -n "$found" ]; then
        return 0
    else
        return 1
    fi
}

_should_archive() {
    if [ -n "$ARCHIVE_FOLDER" ]; then
        return 0
    else
        return 1
    fi
}

_download_cmd() {
    "$YT_DL_CMD" \
        -o "$TEMP_FOLDER/$YT_DL_TITLE.%(ext)s" \
        "${YT_DL_OPTS[@]}" \
        "$*"
}

download() { # 1=url
    # # download the video to download folder
    if ! _alreadyexists "$yt_dl_fname" || ! _should_archive; then
        _download_cmd "$*"
        file=$(_findfile "$TEMP_FOLDER" "$yt_dl_fname")
        printf "Moving file %s to dir: %s\n" "$file" "$DL_FOLDER"
        mv "$file" "$DL_FOLDER"

        if _should_archive; then
            # yt-dl never knows the exact filename in advance
            file=$(_findfile "$DL_FOLDER" "$yt_dl_fname")
            if [ -z "$file" ]; then exit 1; fi
            _call_archive "$file" "$ARCHIVE_FOLDER"
        fi
    # only link old file if one exists
    elif _should_archive; then
        archive_file=$(_findfile "$ARCHIVE_FOLDER" "$yt_dl_fname")
        echo "$archive_file" | while read -r file; do
            printf "File %s exists, not downloading duplicate.\n" "$file"
            _call_archive "$file" "$DL_FOLDER"
        done
    fi
}

setup() {
    TEMP_FOLDER="${TEMP_FOLDER:-${HOME}/downloads}"
    DL_FOLDER="${DL_FOLDER:-${XDG_VIDEOS_DIR:-$HOME/videos}/inbox}"
    ARCHIVE_FOLDER="${ARCHIVE_FOLDER:-${XDG_VIDEOS_DIR:-$HOME/videos}/archive}"
    YT_DL_CMD="${YT_DL_CMD:-yt-dlp}"
    yt_default_opts=(-f "best[height\<=1080]" --retries 15 --embed-chapters --embed-subs --sub-lang "en,de,es,fr")
    declare -a YT_DL_OPTS=${YT_DL_OPTS:-( "${yt_default_opts[@]}" )}
    YT_DL_TITLE="${YT_DL_TITLE:-%(channel)s_%(title)s_%(id)s}" # this title needs to be without extension

    data_dir="${XDG_DATA_HOME:-$HOME/.local/share}/vidl"
    cache_dir="${XDG_CACHE_HOME:-$HOME/.cache}/vidl"
    [ ! -d "$data_dir" ] && mkdir -p "$data_dir"
    [ ! -d "$cache_dir" ] && mkdir -p "$cache_dir"

    queue_file="${data_dir}/vidl_queue"
    lock_file="${cache_dir}/vidl_lock"
}

is_in_queue() { # 1=url
    [ -f "$queue_file" ] || return 1
    grep -q "$1" "$queue_file"
}

add_to_queue() { # 1=url
    if is_in_queue "$1"; then return; fi
    echo "$1" >>"$queue_file"
    echo "added $url to queue."
}

remove_from_queue() { # 1=url
    printf "Removing url %s from queue.\n" "$1"
    sed -i.bak -e "\|$1|d" "$queue_file"
}

clear_queue() {
    rm "$queue_file"
}

remove_lock() {
    if ! rmdir "$lock_file"; then
        echo "Failed to remove lock '$lock_file'. Please remove manually before next run."
        exit 1
    fi
}

is_only_instance() {
    if mkdir "$lock_file" 2>/dev/null; then
        trap "remove_lock" EXIT
        return 0
    else
        return 1
    fi
}

print_queue_remaining() {
    if [ ! -f "$queue_file" ]; then
        echo 0
        return
    fi
    wc -l "$queue_file" | cut -f1 -d' '
}

main() {
    setup
    if [ "$ONLY_DO" = "clear" ]; then
        clear_queue
        exit
    elif [ "$ONLY_DO" = "remaining" ]; then
        print_queue_remaining
        exit
    fi

    for url in "$@"; do
        add_to_queue "$url"
    done
    if is_only_instance; then
        echo "Download starting..."
        while read -r line; do
            get_ytdl_fname "$line"
            download "$line"
            remove_from_queue "$line"
        done <"$queue_file"
    else
        echo "Download already running, only adding to queue."
    fi
    echo "Download done..."
}

main "${urls[@]}"