thinkrank / includes / seo / class-social-images.php

class-social-images.php in ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO 2.7.0, at includes/seo/class-social-images.php

336 lines 10.3 KB

Raw Download Zip

1	<?php
2	/**
3	* The set of images a page offers to Open Graph consumers.
4	*
5	* @package ThinkRank
6	* @since 2.7.0
7	*/
8
9	declare(strict_types=1);
10
11	namespace ThinkRank\SEO;
12
13	if (!defined('ABSPATH')) {
14	exit;
15	}
16
17	/**
18	* Collects the secondary images that follow the primary `og:image`.
19	*
20	* Open Graph allows several `og:image` tags on one page and consumers treat
21	* the first as primary: Facebook lets the sharer pick between them, and others
22	* fall back down the list when the first fails validation (too small, wrong
23	* aspect, 404 behind a CDN). ThinkRank published exactly one, so a post whose
24	* social image failed validation shared with no image at all rather than with
25	* its second-best one.
26	*
27	* This produces everything after the primary, in the order the issue asked
28	* for: the featured image (when a per-post override displaced it), then images
29	* found in the content. The primary itself is resolved elsewhere and passed in,
30	* so there is one answer to "what is the main image" and this cannot disagree
31	* with it.
32	*
33	* Twitter and schema are deliberately not involved. A Twitter card renders one
34	* image and schema's `image` identifies the page's primary; handing either a
35	* list would change what they mean, not enrich it.
36	*
37	* @since 2.7.0
38	*/
39	class Social_Images {
40
41	/**
42	* Total `og:image` tags a page may carry, primary included.
43	*
44	* Four is enough to give a sharer a real choice without turning the head
45	* into a gallery; consumers read the list in order and stop at the first
46	* that validates, so length past this point buys nothing.
47	*
48	* @since 2.7.0
49	* @var int
50	*/
51	public const DEFAULT_LIMIT = 4;
52
53	/**
54	* Every image after the primary, ready to emit.
55	*
56	* @since 2.7.0
57	*
58	* @param int $post_id Post being rendered.
59	* @param string $primary_url The `og:image` already resolved for the page.
60	* @return array<int, array<string, mixed>> Each with url, and width/height/
61	* type/alt where they are known.
62	*/
63	public static function additional(int $post_id, string $primary_url): array {
64	if ($post_id <= 0) {
65	return [];
66	}
67
68	/**
69	* Filter how many og:image tags a page may carry in total.
70	*
71	* @since 2.7.0
72	*
73	* @param int $limit Maximum tags, primary included.
74	* @param int $post_id Post being rendered.
75	*/
76	$limit = (int) apply_filters('thinkrank_og_image_limit', self::DEFAULT_LIMIT, $post_id);
77
78	// One slot is the primary's, so nothing below can be offered unless the
79	// site asked for at least two.
80	if ($limit < 2) {
81	return [];
82	}
83
84	// The primary occupies the first slot whether or not it was resolvable
85	// here, so it seeds the seen-list and can never be offered twice.
86	$seen = [];
87
88	if ('' !== $primary_url) {
89	$seen[self::fingerprint($primary_url)] = true;
90	}
91
92	$images = [];
93
94	foreach (self::candidates($post_id) as $url) {
95	if (count($images) >= $limit - 1) {
96	break;
97	}
98
99	$url = self::usable($url);
100
101	if ('' === $url) {
102	continue;
103	}
104
105	$print = self::fingerprint($url);
106
107	if (isset($seen[$print])) {
108	continue;
109	}
110
111	$seen[$print] = true;
112	$images[] = self::describe($url, $post_id);
113	}
114
115	/**
116	* Filter the secondary Open Graph images for a page.
117	*
118	* @since 2.7.0
119	*
120	* @param array $images Secondary images, primary excluded.
121	* @param int $post_id Post being rendered.
122	*/
123	return (array) apply_filters('thinkrank_og_extra_images', $images, $post_id);
124	}
125
126	/**
127	* Candidate URLs, in the order they should be offered.
128	*
129	* @since 2.7.0
130	* @param int $post_id Post being rendered.
131	* @return array<int, string>
132	*/
133	private static function candidates(int $post_id): array {
134	$urls = [];
135
136	// The featured image. Usually this IS the primary and the fingerprint
137	// check drops it; it matters when a per-post social override displaced
138	// it, which is exactly the case where a second image is worth offering.
139	$thumbnail_id = (int) get_post_thumbnail_id($post_id);
140
141	if ($thumbnail_id) {
142	$src = wp_get_attachment_image_src($thumbnail_id, 'large');
143
144	if (is_array($src) && !empty($src[0])) {
145	$urls[] = (string) $src[0];
146	}
147	}
148
149	foreach (self::from_content($post_id) as $url) {
150	$urls[] = $url;
151	}
152
153	return $urls;
154	}
155
156	/**
157	* Image URLs in the post body, in document order.
158	*
159	* Reads the stored content rather than running it through `the_content`.
160	* This is called while the document head is being written, and rendering
161	* the body there would run every shortcode and block on the page for the
162	* sake of a few `src` attributes.
163	*
164	* @since 2.7.0
165	* @param int $post_id Post being rendered.
166	* @return array<int, string>
167	*/
168	private static function from_content(int $post_id): array {
169	$content = (string) get_post_field('post_content', $post_id);
170
171	if ('' === $content \|\| false === stripos($content, '<img')) {
172	return [];
173	}
174
175	if (!preg_match_all('/<img\b[^>]*>/i', $content, $tags)) {
176	return [];
177	}
178
179	$urls = [];
180
181	foreach ($tags[0] as $tag) {
182	if (preg_match('/\bsrc\s=\s["\']([^"\']+)["\']/i', $tag, $match)) {
183	$urls[] = $match[1];
184	}
185	}
186
187	return $urls;
188	}
189
190	/**
191	* A URL Open Graph can actually use, or '' to skip it.
192	*
193	* @since 2.7.0
194	* @param string $url Raw candidate.
195	* @return string
196	*/
197	private static function usable(string $url): string {
198	$url = trim($url);
199
200	if ('' === $url) {
201	return '';
202	}
203
204	// Inline data and blob URLs are not fetchable by a crawler, and a
205	// protocol-relative or root-relative src cannot be resolved without
206	// guessing the host it belongs to.
207	$scheme = wp_parse_url($url, PHP_URL_SCHEME);
208
209	if (!is_string($scheme) \|\| !in_array(strtolower($scheme), ['http', 'https'], true)) {
210	return '';
211	}
212
213	return (string) esc_url_raw($url);
214	}
215
216	/**
217	* An image plus whatever is known about it.
218	*
219	* Dimensions and type are only claimed for an attachment on this site.
220	* Guessing them for a remote image would publish numbers a consumer uses
221	* to lay out a card before it has fetched the file.
222	*
223	* @since 2.7.0
224	*
225	* @param string $url Image URL.
226	* @param int $post_id Post being rendered, for the alt fallback.
227	* @return array<string, mixed>
228	*/
229	private static function describe(string $url, int $post_id): array {
230	$image = ['url' => $url];
231
232	$attachment_id = (int) attachment_url_to_postid($url);
233
234	if ($attachment_id) {
235	$meta = wp_get_attachment_metadata($attachment_id);
236
237	// Vector uploads report 0x0; publishing that as a dimension is
238	// invalid, so the companions are omitted rather than zeroed.
239	$width = is_array($meta) && isset($meta['width']) ? (int) $meta['width'] : 0;
240	$height = is_array($meta) && isset($meta['height']) ? (int) $meta['height'] : 0;
241
242	if ($width > 0 && $height > 0) {
243	$image['width'] = $width;
244	$image['height'] = $height;
245	}
246
247	$type = (string) get_post_mime_type($attachment_id);
248
249	if ('' !== $type) {
250	$image['type'] = $type;
251	}
252
253	$alt = trim((string) get_post_meta($attachment_id, '_wp_attachment_image_alt', true));
254
255	if ('' !== $alt) {
256	$image['alt'] = $alt;
257	}
258	}
259
260	if (!isset($image['alt'])) {
261	$alt = self::alt_from_content($url, $post_id);
262
263	if ('' !== $alt) {
264	$image['alt'] = $alt;
265	}
266	}
267
268	return $image;
269	}
270
271	/**
272	* The `alt` the author wrote on this image in the body.
273	*
274	* The attachment's own alt text wins where there is one, but an image
275	* inserted with a different alt — or one hosted elsewhere, which has no
276	* attachment at all — still has the author's words on it.
277	*
278	* @since 2.7.0
279	*
280	* @param string $url Image URL.
281	* @param int $post_id Post being rendered.
282	* @return string
283	*/
284	private static function alt_from_content(string $url, int $post_id): string {
285	$content = (string) get_post_field('post_content', $post_id);
286
287	if ('' === $content) {
288	return '';
289	}
290
291	if (!preg_match_all('/<img\b[^>]*>/i', $content, $tags)) {
292	return '';
293	}
294
295	foreach ($tags[0] as $tag) {
296	if (!preg_match('/\bsrc\s=\s["\']([^"\']+)["\']/i', $tag, $src)) {
297	continue;
298	}
299
300	if (self::fingerprint(trim($src[1])) !== self::fingerprint($url)) {
301	continue;
302	}
303
304	if (preg_match('/\balt\s=\s["\']([^"\']*)["\']/i', $tag, $alt)) {
305	return trim($alt[1]);
306	}
307	}
308
309	return '';
310	}
311
312	/**
313	* An identity for a URL that survives WordPress's size derivatives.
314	*
315	* `logo.jpg`, `logo-1024x768.jpg` and the `large` derivative of the same
316	* upload are one image to a reader, and offering all three as alternatives
317	* is noise. Scheme and query are dropped for the same reason — the same
318	* file over http and https is not a second option.
319	*
320	* @since 2.7.0
321	* @param string $url Image URL.
322	* @return string
323	*/
324	private static function fingerprint(string $url): string {
325	$parts = wp_parse_url($url);
326
327	$host = isset($parts['host']) ? strtolower((string) $parts['host']) : '';
328	$path = isset($parts['path']) ? (string) $parts['path'] : $url;
329
330	// Strip a trailing -WxH that WordPress appends to a resized copy.
331	$path = (string) preg_replace('/-\d+x\d+(\.[A-Za-z0-9]+)$/', '$1', $path);
332
333	return $host . $path;
334	}
335	}
336