Linux_post_install/bin/pos-media-yt-subtitles

#!/usr/bin/env bash
set -euo pipefail
# POS: media yt-subtitles — Extract subtitles/captions from a URL (yt-dlp)
# POS_FLAGS: --lang --format --auto-only --output --list-subs --no-playlist --dry-run

source "$(dirname "$0")/../lib/common.sh" 2>/dev/null || source "$(dirname "$0")/common.sh"
source "$(dirname "$0")/../lib/yt-lib.sh" 2>/dev/null || source "$(dirname "$0")/yt-lib.sh"

# --dry-run / --list-subs are parsed before the deps guards (below): dry-run
# previews the yt-dlp command without requiring yt-dlp to be installed.
DRY_RUN=0
LIST_SUBS=0
for arg in "$@"; do
    [ "$arg" = "--dry-run" ] && DRY_RUN=1
    [ "$arg" = "--list-subs" ] && LIST_SUBS=1
done

# Deps guards sit before -h|--help. Subtitles needs only yt-dlp (no ffmpeg).
# dry-run skips the deps check entirely (preview needs no binary).
if [ "$DRY_RUN" -eq 0 ]; then
    command -v yt-dlp &>/dev/null || err "yt-dlp not found — install it with: sudo apt install yt-dlp"
fi

LANG_SPEC=""
FMT="srt"
AUTO_ONLY=0
OUT_DIR=""
NO_PLAYLIST=0

usage() {
    cat <<EOF
Usage: pos media yt subtitles [options] <url>

Extract subtitles/captions from a URL via yt-dlp. Fetches manual captions and
auto-generated captions by default.

Options:
  --lang <list>         Subtitle languages (comma-separated), default: best
  --format <fmt>        Output format: srt (default), vtt, or txt
  --auto-only           Only auto-generated captions (no manual subs)
  -o, --output <dir>    Output directory (default: current directory)
  --list-subs           List available subtitles and exit (probe only)
  --no-playlist         Download only the single video
  --dry-run             Print the yt-dlp command without running it
  -h, --help            This help

Examples:
  pos media yt subtitles https://youtube.com/watch?v=dQw4w9WgXcQ
  pos media yt subtitles --lang en,ar https://youtube.com/watch?v=dQw4w9WgXcQ
  pos media yt subtitles --format txt https://youtu.be/dQw4w9WgXcQ
  pos media yt subtitles --list-subs https://youtube.com/watch?v=dQw4w9WgXcQ
EOF
    exit 0
}

# convert_srt_to_txt <srt-file> — strip SRT timestamps, seq numbers, HTML tags
convert_srt_to_txt() {
    local srt_file="$1" txt_file="${1%.srt}.txt"
    sed '/^[0-9][0-9]:[0-9][0-9]/d; /^$/d; s/<[^>]*>//g' "$srt_file" > "$txt_file"
}

URL=""
while [[ $# -gt 0 ]]; do
    case "$1" in
        -h|--help) usage ;;
        --lang)
            [ $# -ge 2 ] || err "--lang needs a value"
            LANG_SPEC="$2"; shift 2 ;;
        --format)
            [ $# -ge 2 ] || err "--format needs a value"
            case "$2" in
                srt|vtt|txt) FMT="$2" ;;
                *) err "unknown format '$2' (use srt, vtt, or txt)" ;;
            esac
            shift 2 ;;
        --auto-only) AUTO_ONLY=1; shift ;;
        -o|--output)
            [ $# -ge 2 ] || err "--output needs a value"
            OUT_DIR="$2"; shift 2 ;;
        --list-subs) LIST_SUBS=1; shift ;;
        --no-playlist) NO_PLAYLIST=1; shift ;;
        --dry-run) DRY_RUN=1; shift ;;
        -*) err "Unknown option: $1 (see --help)" ;;
        *)
            [ -z "$URL" ] && URL="$1" && shift || err "Unexpected argument: $1" ;;
    esac
done
[ -n "$URL" ] || err "missing URL (see --help)"

yt_validate_url "$URL" 2>/dev/null || err "not a valid URL: $URL (must start with http:// or https://)"

# ── Build the yt-dlp command ─────────────────────────────────
if [ "$LIST_SUBS" -eq 1 ]; then
    args=(yt-dlp --list-subs "$URL")
    if [ "$DRY_RUN" -eq 1 ]; then
        yt_echo_cmd "${args[@]:1}"
        exit 0
    fi
    spawn "listing subtitles" "${args[@]}"
    exit 0
fi

# Output template
OUT_TEMPLATE=""
if [ -n "$OUT_DIR" ]; then
    OUT_TEMPLATE="$OUT_DIR/%(title)s.%(sub_lang)s.%(ext)s"
else
    OUT_TEMPLATE="./%(title)s.%(sub_lang)s.%(ext)s"
fi

args=(yt-dlp)
if [ "$AUTO_ONLY" -eq 1 ]; then
    args+=(--write-auto-subs)
else
    args+=(--write-subs --write-auto-subs)
fi

if [ -n "$LANG_SPEC" ]; then
    args+=(--sub-langs "$LANG_SPEC")
else
    args+=(--sub-langs best)
fi

case "$FMT" in
    vtt) args+=(--sub-format vtt) ;;
    txt) args+=(--sub-format srt) ;;
    *)   args+=(--sub-format srt) ;;
esac

[ "$NO_PLAYLIST" -eq 1 ] && args+=(--no-playlist)

args+=(-o "$OUT_TEMPLATE" "$URL")

if [ "$DRY_RUN" -eq 1 ]; then
    yt_echo_cmd "${args[@]:1}"
    exit 0
fi

# ── Execute ───────────────────────────────────────────────────
set +e
out="$(mktemp)"
err_tmp="$(mktemp)"
"${args[@]}" >"$out" 2>"$err_tmp"
rc=$?
set -e

if [ "$rc" -ne 0 ]; then
    # Unavailable-subtitle detection: exit!=0 AND stderr matches no-subs patterns
    if grep -qiE 'no subtitles|no captions|unable to extract|subtitles not available' "$err_tmp"; then
        rm -f "$out" "$err_tmp"
        err "unavailable subtitles for this video (try --list-subs to check)"
    fi
    # Re-surface the real yt-dlp error otherwise
    sed 's/^/    /' "$err_tmp" >&2
    rm -f "$out" "$err_tmp"
    exit "$rc"
fi
rm -f "$out" "$err_tmp"

# ── Post-conversion for txt format ────────────────────────────
if [ "$FMT" = "txt" ]; then
    search_dir="."
    [ -n "$OUT_DIR" ] && search_dir="$OUT_DIR"
    while IFS= read -r -d '' f; do
        case "$f" in
            *.srt) convert_srt_to_txt "$f" ;;
        esac
    done < <(find "$search_dir" -maxdepth 1 -name '*.srt' -print0 2>/dev/null || true)
fi