vtt-embed/vtt-embed.sh

#!/bin/sh

# © DJ Chase, 2024. Licensed under the Academic Free License version 3.0.

readonly HTTP_FOUND='302 Found'
readonly HTTP_BAD_REQUEST='400 Bad Request'
readonly HTTP_NOT_FOUND='404 Not Found'
readonly HTTP_METHOD_NOT_ALLOWED='405 Method Not Allowed'
# SC2155 is about munging return codes, which is fine here
# shellcheck disable=SC2155
readonly HTML_PAGE_START="$(
	dirname "${DOCUMENT_ROOT}${SCRIPT_NAME}"
)/.page-start.html"
# shellcheck disable=SC2155
readonly HTML_PAGE_END="$(
	dirname "${DOCUMENT_ROOT}${SCRIPT_NAME}"
)/.page-end.html"

percentDecode() \
{
	# usage: foo | percentDecode

	# shellcheck disable=SC2059
	printf "$(sed 's/+/ /g; s/%/\\x/g' | tr -d '\r')"
}

htmlEncode() \
{
	# usage: foo | htmlEncode

	sed "
		s_&_&amp\;_g;
		s_'_&#39\;_g;
		s_\"_&quot\;_g;
		s_<_&lt\;_g;
		s_>_&gt\;_g
	"
}

error() \
{
	# usage: error <status_code> [<message>]

	printf 'Status: %s\n' "$1"

	if [ "$2" ]
	then
		printf 'Content-type: text/plain; charset=utf-8\n\n'

		[ "$REQUEST_METHOD" = 'HEAD' ] && exit 0

		printf '%s\n\n' "$1" "$2"
	else
		printf '\n'
	fi

	# exit 0 so that the webserver doesn't return "Internal Server Error"
	exit 0
}


if [ "$REQUEST_METHOD" = 'GET' ] || [ "$REQUEST_METHOD" = 'HEAD' ]
then
	if [ "$QUERY_STRING" ]
	then
		format="$(
			printf '%s\n' "$QUERY_STRING" \
			| sed 's/&/\n/g' \
			| grep '^format=' \
			| cut -d = -f 2- \
			| percentDecode
		)"
		: "${format:=best}"
		pageURL="$(
			printf '%s\n' "$QUERY_STRING" \
			| sed 's/&/\n/g' \
			| grep '^url=' \
			| cut -d = -f 2- \
			| percentDecode
		)"

		[ "$pageURL" ] \
		|| error "$HTTP_BAD_REQUEST" \
			'You need to include the `url` parameter'

		ytdlOutput="$(
			youtube-dl --format "$format" --get-url -- "$pageURL" \
			2>&1
		)"

		# shellcheck disable=SC2181
		if [ "$?" -eq 0 ]
		then
			printf 'Status: %s\n' "$HTTP_FOUND"

			# `head -n 1` in case someone sets a format of
			# 'bestaudio+bestvideo' or similar, since we can only
			# redirect to one URL at a time
			printf 'Location: %s\n\n' "$(
				printf '%s\n' "$ytdlOutput" \
				| sed '/^$/d' \
				| head -n 1
			)"
		else
			error "$HTTP_NOT_FOUND" "$ytdlOutput"
		fi
	else
		printf 'Content-Type: text/html; charset=utf-8\n\n'

		[ "$REQUEST_METHOD" = 'HEAD' ] && exit 0

		cat "$HTML_PAGE_START" "$HTML_PAGE_END"
	fi

elif [ "$REQUEST_METHOD" = 'POST' ]
then
	postBody="$(tr -d '\r' | sed 's/&/\n/g')"
	url="$(
		# field is already percent encoded
		printf '%s' "$postBody" \
		| grep '^url=' \
		| cut -d = -f 2- \
		| htmlEncode
	)"
	scriptURL="$(
		printf 'https://%s%s' "$SERVER_NAME" "$SCRIPT_NAME" \
		| htmlEncode
	)"

	printf 'Content-Type: text/html; charset=utf-8\n\n'
	cat "$HTML_PAGE_START" - "$HTML_PAGE_END" <<EOF
<hr/>

<p>Use this code to embed the video:</p>

<pre><code>&lt;video controls&gt;
    &lt;source media="(prefers-reduced-data: reduce)" src="$scriptURL?format=worst&amp;url=$url"&gt;
    &lt;source src="$scriptURL?url=$url"&gt;
    &lt;track src="https://example.com/captions.vtt" kind="captions" srclang="en" label="English"&gt;
    &lt;p&gt;It looks like your browser doesn’t support HTML video. You can still
        &lt;a href="$scriptURL?url=$url"&gt;download
        the video&lt;/a&gt; and watch it locally. You can also
        &lt;a href="https://example.com/captions.vtt"&gt;download the captions
        file&lt;/a&gt; if your video player can import it.
    &lt;/p&gt;
&lt;/video&gt;</code></pre>

<p>Remember to replace <code>https://example.com/captions.vtt</code> with the
link to your captions file (WebVTT). See
<a href="https://dj-chase.com/unlisted/vtt-embed-example-cc.vtt">the demo
caption file</a> if you’d like an example of what that looks like.</p>
EOF
else
	printf 'Status: %s\n' "$HTTP_METHOD_NOT_ALLOWED"
	printf 'Allow: GET, POST, HEAD\n\n'
fi