dlytsubs.sh

#!/usr/bin/env bash

# Replace the target with your output of `which yt-dlp`
alias yt-dlp="/opt/homebrew/bin/yt-dlp"

usage() {
    cat <<EOF
Usage: $(basename "${BASH_SOURCE[0]}") [-t] [-l LANGUAGE_TAG] URL

A simple script to download YouTube subtitles with yt-dlp, and optionally
convert to plain text.

Available options:

-h, --help      Print this help and exit.
-l, --lang      Filter by language tag as defined in RFC 5646, only one tag is
                permitted.
-t, --text      Output a plain text file along with the subtitles file.
EOF
    exit
}

get_params() {
    text=0
    while :; do
        case "${1-}" in
        -h | --help) usage;;
        -t | --text) text=1;;
        -l | --lang) extra_lang="${2-}"; shift;;
        -?*) echo "Unknown option: $1"; exit 1;;
        *) break ;;
        esac
        shift
    done
    args=("$@")
    if [ ${#args[@]} -ne 1 ]; then echo "Must provide one and only one URL"; exit 1; fi
    return 0
}

list_subs() {
    url="$1"
    list_subs=$(yt-dlp --list-subs "$url" | grep "vtt")
    if [ -z "$extra_lang" ];
    then
        echo "No designated language; falls back to English and Chinese"
        autocaps=$(echo "$list_subs" | grep " from " | grep "^zh-Hans" | awk '{print "[autocaps]", $1, $2}'
echo "$list_subs" | grep " from " | grep "^en" | awk '{print "[autocaps]", $1, $2}'
)
        subtitles=$(echo "$list_subs" | grep -v " from " | grep "^zh-Hans" | awk '{print "[subtitles]", $1, $2}'
echo "$list_subs" | grep -v " from " | grep "^en" | awk '{print "[subtitles]", $1, $2 }'
)
    else
        echo "Filtering by:" "$extra_lang"
        autocaps=$(echo "$list_subs" | grep " from " | grep "^$extra_lang" | awk '{print "[autocaps]", $1, $2}'
)
        subtitles=$(echo "$list_subs" | grep -v " from " | grep "^$extra_lang" | awk '{print "[subtitles]", $1, $2}'
)
    fi
    opts=()
    while IFS= read -r l; do opts+=("$l"); done <<EOF
$autocaps
$subtitles
EOF
    select lang in "${opts[@]}"; do
        code=$(echo "$lang" | cut -f2 -d' ')
        if echo "$lang" | grep -q "auto";
        then cmd="--write-auto-subs";
        else cmd="--write-subs";
        fi
        break
    done
}

dl_subs() {
    yt-dlp "$1" --skip-download --sub-langs "$2" --output subs "$3"
}

vtt_to_txt() {
    tail +4 "$1" |\
    grep -v " --> " |\
    grep . |\
    sed -E 's!<([0-9]{2}:[0-9]{2}:[0-9]{2}\.[0-9]{3}|c|\/c)>!!g' |\
    uniq
}

get_params "$@"
list_subs "${args[0]}"
dl_subs "$cmd" "$code" "$url"
if [ $text -eq 1 ]; then vtt_to_txt "subs.$code.vtt" > "text.$code.txt"; fi