← All changes | app/Services/RemoteUrlParser.php +99 -3 2.5.0 → 2.10.0 View file →

		@@ -26,8 +26,78 @@
26	26
27	27	return self::$instance->getInfoFromRemoteUrl($url);
28	28	}
29	29
	30	+ public static function extractIframeThumbnail(&$html)
	31	+ {
	32	+ if (!is_string($html)) {
	33	+ return '';
	34	+ }
	35	+ $html = self::sanitizeOembedHtml($html);
	36	+ if (!preg_match('/<iframe\s[^>]*\bsrc\s*=\s*([\'"])(.*?)\1/i', $html, $matches)) {
	37	+ return '';
	38	+ }
	39	+
	40	+ $src = sanitize_url(html_entity_decode($matches[2], ENT_QUOTES | ENT_HTML5, 'UTF-8'));
	41	+ if (!$src) {
	42	+ return '';
	43	+ }
	44	+
	45	+ if (preg_match('#^https?://(?:[\w-]+\.)?youtube\.com/embed/([^?/]+)#i', $src, $ytMatch)) {
	46	+ return self::bestYoutubeThumbnail($ytMatch[1]);
	47	+ }
	48	+
	49	+ $providers = [
	50	+ '#^https?://player\.vimeo\.com/video/([^?/]+).*$#i' => 'https://vumbnail.com/$1.jpg',
	51	+ '#^https?://fast\.wistia\.net/embed/iframe/([^?/]+).*$#i' => 'https://fast.wistia.net/embed/medias/$1/swatch',
	52	+ '#^https?://(?:www\.)?dailymotion\.com/(?:embed/video|player\.html\?video=)/?([^?/&]+).*$#i' => 'https://www.dailymotion.com/thumbnail/video/$1',
	53	+ ];
	54	+
	55	+ foreach ($providers as $pattern => $template) {
	56	+ $thumb = preg_replace($pattern, $template, $src, 1, $count);
	57	+ if ($count) {
	58	+ return $thumb;
	59	+ }
	60	+ }
	61	+
	62	+ $parsed = self::parse($src);
	63	+ return (!is_wp_error($parsed) && !empty($parsed['image'])) ? $parsed['image'] : '';
	64	+ }
	65	+
	66	+ /**
	67	+ * Pick the best YouTube thumbnail for a feed preview.
	68	+ *
	69	+ * Runs on the feed-save path, so it stays cheap: a single HEAD probe for
	70	+ * the HD WebP frame (~30-56% smaller than JPG, sharp 16:9), falling back
	71	+ * to hqdefault.jpg — the one universally present rung (maxres/sd and even
	72	+ * hqdefault.webp 404 for non-HD or legacy uploads). The result is stored
	73	+ * on the feed and cached, so the probe is paid once per video.
	74	+ */
	75	+ protected static function bestYoutubeThumbnail($videoId)
	76	+ {
	77	+ $videoId = sanitize_text_field($videoId);
	78	+ if (!$videoId) {
	79	+ return '';
	80	+ }
	81	+
	82	+ $fallback = 'https://img.youtube.com/vi/' . $videoId . '/hqdefault.jpg';
	83	+
	84	+ $cacheKey = 'fcom_yt_thumb_' . md5($videoId);
	85	+ $cached = get_transient($cacheKey);
	86	+ if ($cached !== false) {
	87	+ return $cached;
	88	+ }
	89	+
	90	+ $maxRes = 'https://i.ytimg.com/vi_webp/' . $videoId . '/maxresdefault.webp';
	91	+ $response = wp_remote_head($maxRes, ['timeout' => 1.5, 'redirection' => 0]);
	92	+ if (!is_wp_error($response) && wp_remote_retrieve_response_code($response) === 200) {
	93	+ set_transient($cacheKey, $maxRes, WEEK_IN_SECONDS);
	94	+ return $maxRes;
	95	+ }
	96	+
	97	+ return $fallback;
	98	+ }
	99	+
30	100	public function getOembed($url)
31	101	{
32	102	$data = (new \WP_oEmbed())->get_data($url, [
33	103	'discover' => false
		@@ -38,22 +108,38 @@
38	108	}
39	109
40	110	$data = (array)$data;
41	111
	112	+ $provider = strtolower(Arr::get($data, 'provider_name'));
	113	+
	114	+ $image = Arr::get($data, 'thumbnail_url');
	115	+ if ($provider === 'youtube') {
	116	+ $image = self::bestYoutubeThumbnail(self::getYoutubeVideoId($url)) ?: $image;
	117	+ }
	118	+
42	119	return array_filter([
43	120	'title' => Arr::get($data, 'title'),
44	121	'author_name' => Arr::get($data, 'author_name'),
45	122	'type' => 'oembed',
46		- 'provider' => strtolower(Arr::get($data, 'provider_name')),
	123	+ 'provider' => $provider,
47	124	'content_type' => Arr::get($data, 'type'),
48	125	'url' => $url,
49	126	'html' => self::sanitizeOembedHtml(Arr::get($data, 'html')),
50		- 'image' => Arr::get($data, 'thumbnail_url'),
	127	+ 'image' => $image,
51	128	]);
52	129	}
53	130
54		- private static function sanitizeOembedHtml($html)
	131	+ protected static function getYoutubeVideoId($url)
55	132	{
	133	+ if (preg_match('#(?:youtu\.be/|youtube\.com/(?:embed/|v/|live/|shorts/|watch\?v=))([a-zA-Z0-9_-]+)#i', (string)$url, $match)) {
	134	+ return $match[1];
	135	+ }
	136	+
	137	+ return '';
	138	+ }
	139	+
	140	+ public static function sanitizeOembedHtml($html)
	141	+ {
56	142	if (empty($html)) {
57	143	return $html;
58	144	}
59	145
		@@ -72,8 +158,12 @@
72	158	'sandbox' => true,
73	159	];
74	160	}
75	161
	162	+ // Strip <script>/<style> incl. contents; wp_kses keeps their inner text otherwise.
	163	+ $html = preg_replace('#<(script|style)\b[^>]*>.*?</\1>#is', '', $html);
	164	+ $html = preg_replace('#<(script|style)\b[^>]*/?>#i', '', $html);
	165	+
76	166	return wp_kses($html, $allowed, ['https']);
77	167	}
78	168
79	169	public function getInfoFromRemoteUrl($url)
		@@ -88,8 +178,14 @@
88	178	$cachedReponse = wp_cache_get($cacheKey, 'fluent-community');
89	179
90	180	if ($cachedReponse) {
91	181	return $cachedReponse;
	182	+ }
	183	+
	184	+ $preempted = apply_filters('fluent_community/preview_metadata_pre_fetch', null, $url);
	185	+ if ($preempted && is_array($preempted)) {
	186	+ wp_cache_set($cacheKey, $preempted, 'fluent-community', apply_filters('rest_url_details_cache_expiration', HOUR_IN_SECONDS)); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound
	187	+ return $preempted;
92	188	}
93	189
94	190	$remote_url_response = $this->getRemoteBody($url);
95	191	if (is_wp_error($remote_url_response) || empty($remote_url_response)) {