| 1 |
<?php |
| 2 |
/** |
| 3 |
* Friends Link Preview |
| 4 |
* |
| 5 |
* This contains the functions for fetching and displaying link previews. |
| 6 |
* |
| 7 |
* @package Friends |
| 8 |
*/ |
| 9 |
|
| 10 |
namespace Friends; |
| 11 |
|
| 12 |
/** |
| 13 |
* This is the class for the link previews of the Friends Plugin. |
| 14 |
* |
| 15 |
* Incoming posts that just contain a link (for example Mastodon statuses that |
| 16 |
* point to a news article) are enriched with the Open Graph metadata of that |
| 17 |
* link so that a preview card can be displayed alongside the post. |
| 18 |
* |
| 19 |
* @since 4.3 |
| 20 |
* |
| 21 |
* @package Friends |
| 22 |
* @author Alex Kirk |
| 23 |
*/ |
| 24 |
class Link_Preview { |
| 25 |
const META = 'link_preview'; |
| 26 |
const META_CHECKED = 'link_preview_checked'; |
| 27 |
const CRON_HOOK = 'friends_fetch_link_preview'; |
| 28 |
|
| 29 |
/** |
| 30 |
* How long to wait before looking at a post again that had no preview. |
| 31 |
*/ |
| 32 |
const RETRY_AFTER = WEEK_IN_SECONDS; |
| 33 |
|
| 34 |
/** |
| 35 |
* How much of the remote document to download while looking for meta tags. |
| 36 |
*/ |
| 37 |
const MAX_RESPONSE_SIZE = 300000; |
| 38 |
|
| 39 |
/** |
| 40 |
* Contains a reference to the Friends class. |
| 41 |
* |
| 42 |
* @var Friends |
| 43 |
*/ |
| 44 |
private $friends = null; |
| 45 |
|
| 46 |
/** |
| 47 |
* Constructor |
| 48 |
* |
| 49 |
* @param Friends $friends A reference to the Friends object. |
| 50 |
*/ |
| 51 |
public function __construct( Friends $friends ) { |
| 52 |
$this->friends = $friends; |
| 53 |
$this->register_hooks(); |
| 54 |
} |
| 55 |
|
| 56 |
/** |
| 57 |
* Register the WordPress hooks |
| 58 |
*/ |
| 59 |
private function register_hooks() { |
| 60 |
add_action( 'friends_retrieved_new_posts', array( $this, 'queue_new_posts' ), 10, 2 ); |
| 61 |
add_action( self::CRON_HOOK, array( $this, 'fetch_link_preview' ) ); |
| 62 |
} |
| 63 |
|
| 64 |
/** |
| 65 |
* Whether link previews should be fetched and displayed. |
| 66 |
* |
| 67 |
* @return bool True if enabled. |
| 68 |
*/ |
| 69 |
public static function is_enabled() { |
| 70 |
return ! get_option( 'friends_disable_link_previews' ); |
| 71 |
} |
| 72 |
|
| 73 |
/** |
| 74 |
* Queue newly retrieved posts for a link preview lookup. |
| 75 |
* |
| 76 |
* @param User_Feed $user_feed The user feed the posts came from. |
| 77 |
* @param array $new_post_ids The ids of the newly created posts. |
| 78 |
*/ |
| 79 |
public function queue_new_posts( $user_feed, $new_post_ids ) { |
| 80 |
if ( ! self::is_enabled() || ! is_array( $new_post_ids ) ) { |
| 81 |
return; |
| 82 |
} |
| 83 |
|
| 84 |
foreach ( $new_post_ids as $post_id ) { |
| 85 |
self::queue( $post_id ); |
| 86 |
} |
| 87 |
} |
| 88 |
|
| 89 |
/** |
| 90 |
* Schedule a link preview lookup for a post. |
| 91 |
* |
| 92 |
* @param int $post_id The post id. |
| 93 |
*/ |
| 94 |
public static function queue( $post_id ) { |
| 95 |
$post_id = intval( $post_id ); |
| 96 |
if ( ! $post_id ) { |
| 97 |
return; |
| 98 |
} |
| 99 |
|
| 100 |
if ( wp_next_scheduled( self::CRON_HOOK, array( $post_id ) ) ) { |
| 101 |
return; |
| 102 |
} |
| 103 |
|
| 104 |
wp_schedule_single_event( time(), self::CRON_HOOK, array( $post_id ) ); |
| 105 |
} |
| 106 |
|
| 107 |
/** |
| 108 |
* Cron callback: look up the link preview for a post. |
| 109 |
* |
| 110 |
* @param int $post_id The post id. |
| 111 |
*/ |
| 112 |
public function fetch_link_preview( $post_id ) { |
| 113 |
self::update_link_preview( $post_id ); |
| 114 |
} |
| 115 |
|
| 116 |
/** |
| 117 |
* Get the link preview for a post, scheduling a lookup if we don't have one yet. |
| 118 |
* |
| 119 |
* @param \WP_Post|int|null $post The post (defaults to the current post). |
| 120 |
* |
| 121 |
* @return array|false The link preview data or false if there is none. |
| 122 |
*/ |
| 123 |
public static function get_for_post( $post = null ) { |
| 124 |
if ( ! self::is_enabled() ) { |
| 125 |
return false; |
| 126 |
} |
| 127 |
|
| 128 |
$post = get_post( $post ); |
| 129 |
if ( ! $post ) { |
| 130 |
return false; |
| 131 |
} |
| 132 |
|
| 133 |
$preview = get_post_meta( $post->ID, self::META, true ); |
| 134 |
if ( is_array( $preview ) && ! empty( $preview['url'] ) ) { |
| 135 |
/** |
| 136 |
* Filters the link preview data before it is displayed. |
| 137 |
* |
| 138 |
* @param array $preview The link preview data. |
| 139 |
* @param \WP_Post $post The post the preview belongs to. |
| 140 |
*/ |
| 141 |
return apply_filters( 'friends_link_preview', $preview, $post ); |
| 142 |
} |
| 143 |
|
| 144 |
$checked = intval( get_post_meta( $post->ID, self::META_CHECKED, true ) ); |
| 145 |
if ( ( ! $checked || $checked < time() - self::RETRY_AFTER ) && self::is_supported_post( $post ) ) { |
| 146 |
self::queue( $post->ID ); |
| 147 |
} |
| 148 |
|
| 149 |
return false; |
| 150 |
} |
| 151 |
|
| 152 |
/** |
| 153 |
* Whether a post is a candidate for a link preview. |
| 154 |
* |
| 155 |
* @param \WP_Post $post The post. |
| 156 |
* |
| 157 |
* @return bool True if a preview could be shown for this post. |
| 158 |
*/ |
| 159 |
public static function is_supported_post( $post ) { |
| 160 |
if ( ! in_array( $post->post_type, apply_filters( 'friends_frontend_post_types', array() ), true ) ) { |
| 161 |
return false; |
| 162 |
} |
| 163 |
|
| 164 |
$post_format = get_post_format( $post ); |
| 165 |
if ( ! $post_format ) { |
| 166 |
$post_format = 'standard'; |
| 167 |
} |
| 168 |
|
| 169 |
/** |
| 170 |
* Filters the post formats for which link previews are shown. |
| 171 |
* |
| 172 |
* @param array $post_formats The post formats. |
| 173 |
* @param \WP_Post $post The post. |
| 174 |
*/ |
| 175 |
$post_formats = apply_filters( 'friends_link_preview_post_formats', array( 'status', 'link', 'aside' ), $post ); |
| 176 |
|
| 177 |
return in_array( $post_format, $post_formats, true ); |
| 178 |
} |
| 179 |
|
| 180 |
/** |
| 181 |
* Look up the link preview for a post and store it in post meta. |
| 182 |
* |
| 183 |
* @param int $post_id The post id. |
| 184 |
* @param bool $force Whether to look it up even if we already did. |
| 185 |
* |
| 186 |
* @return array|false The link preview data or false if there is none. |
| 187 |
*/ |
| 188 |
public static function update_link_preview( $post_id, $force = false ) { |
| 189 |
if ( ! self::is_enabled() ) { |
| 190 |
return false; |
| 191 |
} |
| 192 |
|
| 193 |
$post = get_post( $post_id ); |
| 194 |
if ( ! $post || ! self::is_supported_post( $post ) ) { |
| 195 |
return false; |
| 196 |
} |
| 197 |
|
| 198 |
if ( ! $force ) { |
| 199 |
$checked = intval( get_post_meta( $post->ID, self::META_CHECKED, true ) ); |
| 200 |
if ( $checked && $checked > time() - self::RETRY_AFTER ) { |
| 201 |
$preview = get_post_meta( $post->ID, self::META, true ); |
| 202 |
return is_array( $preview ) && ! empty( $preview['url'] ) ? $preview : false; |
| 203 |
} |
| 204 |
} |
| 205 |
|
| 206 |
update_post_meta( $post->ID, self::META_CHECKED, time() ); |
| 207 |
|
| 208 |
$url = self::extract_url( $post ); |
| 209 |
if ( ! $url ) { |
| 210 |
update_post_meta( $post->ID, self::META, array() ); |
| 211 |
return false; |
| 212 |
} |
| 213 |
|
| 214 |
$preview = self::fetch( $url ); |
| 215 |
if ( is_wp_error( $preview ) ) { |
| 216 |
do_action( 'friends_link_preview_failed', $url, $preview, $post ); |
| 217 |
update_post_meta( $post->ID, self::META, array() ); |
| 218 |
return false; |
| 219 |
} |
| 220 |
|
| 221 |
update_post_meta( $post->ID, self::META, $preview ); |
| 222 |
|
| 223 |
return $preview; |
| 224 |
} |
| 225 |
|
| 226 |
/** |
| 227 |
* Find the URL in a post that a preview should be shown for. |
| 228 |
* |
| 229 |
* Mirrors what Mastodon does: the first link in the post that is not a |
| 230 |
* mention or a hashtag, and only if the post doesn't already show media. |
| 231 |
* |
| 232 |
* @param \WP_Post $post The post. |
| 233 |
* |
| 234 |
* @return string|false The url or false if there is none. |
| 235 |
*/ |
| 236 |
public static function extract_url( $post ) { |
| 237 |
$content = $post->post_content; |
| 238 |
|
| 239 |
if ( preg_match( '/<(?:img|video|audio|iframe|embed|object)[\s\/>]/i', $content ) ) { |
| 240 |
// The post already displays media, no need for a preview card. |
| 241 |
return false; |
| 242 |
} |
| 243 |
|
| 244 |
if ( ! preg_match_all( '/<a\s((?>[^>"\']+|"[^"]*"|\'[^\']*\')*)>/i', $content, $matches ) ) { |
| 245 |
return false; |
| 246 |
} |
| 247 |
|
| 248 |
$own_host = wp_parse_url( $post->guid, PHP_URL_HOST ); |
| 249 |
|
| 250 |
foreach ( $matches[1] as $attributes ) { |
| 251 |
if ( ! preg_match( '/\bhref\s*=\s*(["\'])(.*?)\1/is', $attributes, $m ) ) { |
| 252 |
continue; |
| 253 |
} |
| 254 |
$url = html_entity_decode( trim( $m[2] ), ENT_QUOTES, get_bloginfo( 'charset' ) ); |
| 255 |
|
| 256 |
if ( preg_match( '/\bclass\s*=\s*(["\'])(.*?)\1/is', $attributes, $m ) ) { |
| 257 |
$classes = preg_split( '/\s+/', strtolower( $m[2] ) ); |
| 258 |
if ( array_intersect( $classes, array( 'mention', 'hashtag', 'u-url' ) ) ) { |
| 259 |
continue; |
| 260 |
} |
| 261 |
} |
| 262 |
|
| 263 |
if ( preg_match( '/\brel\s*=\s*(["\'])(.*?)\1/is', $attributes, $m ) ) { |
| 264 |
$rels = preg_split( '/\s+/', strtolower( $m[2] ) ); |
| 265 |
if ( array_intersect( $rels, array( 'tag', 'author' ) ) ) { |
| 266 |
continue; |
| 267 |
} |
| 268 |
} |
| 269 |
|
| 270 |
$parsed_url = wp_parse_url( $url ); |
| 271 |
if ( ! isset( $parsed_url['host'] ) || ! isset( $parsed_url['scheme'] ) ) { |
| 272 |
continue; |
| 273 |
} |
| 274 |
if ( ! in_array( strtolower( $parsed_url['scheme'] ), array( 'http', 'https' ), true ) ) { |
| 275 |
continue; |
| 276 |
} |
| 277 |
if ( $own_host && strtolower( $parsed_url['host'] ) === strtolower( $own_host ) ) { |
| 278 |
// Don't preview the post itself or its neighbors on the same site. |
| 279 |
continue; |
| 280 |
} |
| 281 |
if ( preg_match( '/\.(?:jpe?g|png|gif|webp|avif|svg|mp4|webm|mp3|ogg|pdf)$/i', $parsed_url['path'] ?? '' ) ) { |
| 282 |
continue; |
| 283 |
} |
| 284 |
|
| 285 |
/** |
| 286 |
* Filters the URL a link preview will be fetched for. |
| 287 |
* |
| 288 |
* Return false to not show a link preview for this post. |
| 289 |
* |
| 290 |
* @param string $url The url. |
| 291 |
* @param \WP_Post $post The post. |
| 292 |
*/ |
| 293 |
return apply_filters( 'friends_link_preview_url', $url, $post ); |
| 294 |
} |
| 295 |
|
| 296 |
return false; |
| 297 |
} |
| 298 |
|
| 299 |
/** |
| 300 |
* Download a URL and extract its Open Graph metadata. |
| 301 |
* |
| 302 |
* @param string $url The url. |
| 303 |
* |
| 304 |
* @return array|\WP_Error The link preview data or an error. |
| 305 |
*/ |
| 306 |
public static function fetch( $url ) { |
| 307 |
$response = wp_safe_remote_get( |
| 308 |
$url, |
| 309 |
array( |
| 310 |
'timeout' => 10, |
| 311 |
'redirection' => 3, |
| 312 |
'limit_response_size' => self::MAX_RESPONSE_SIZE, |
| 313 |
'headers' => array( |
| 314 |
'Accept' => 'text/html,application/xhtml+xml', |
| 315 |
), |
| 316 |
) |
| 317 |
); |
| 318 |
|
| 319 |
if ( is_wp_error( $response ) ) { |
| 320 |
return $response; |
| 321 |
} |
| 322 |
|
| 323 |
$response_code = wp_remote_retrieve_response_code( $response ); |
| 324 |
if ( $response_code < 200 || $response_code >= 300 ) { |
| 325 |
return new \WP_Error( 'friends_link_preview_http_error', wp_remote_retrieve_response_message( $response ), $response_code ); |
| 326 |
} |
| 327 |
|
| 328 |
$content_type = wp_remote_retrieve_header( $response, 'content-type' ); |
| 329 |
if ( $content_type && false === stripos( $content_type, 'html' ) ) { |
| 330 |
return new \WP_Error( 'friends_link_preview_not_html', $content_type ); |
| 331 |
} |
| 332 |
|
| 333 |
$body = wp_remote_retrieve_body( $response ); |
| 334 |
if ( ! $body ) { |
| 335 |
return new \WP_Error( 'friends_link_preview_empty_body' ); |
| 336 |
} |
| 337 |
|
| 338 |
return self::parse( $body, self::get_final_url( $response, $url ), $url ); |
| 339 |
} |
| 340 |
|
| 341 |
/** |
| 342 |
* Get the URL a request ended up at after following redirects. |
| 343 |
* |
| 344 |
* @param array $response The response of wp_safe_remote_get. |
| 345 |
* @param string $url The url that was requested. |
| 346 |
* |
| 347 |
* @return string The final url. |
| 348 |
*/ |
| 349 |
private static function get_final_url( $response, $url ) { |
| 350 |
if ( ! isset( $response['http_response'] ) || ! is_object( $response['http_response'] ) ) { |
| 351 |
return $url; |
| 352 |
} |
| 353 |
if ( ! method_exists( $response['http_response'], 'get_response_object' ) ) { |
| 354 |
return $url; |
| 355 |
} |
| 356 |
$response_object = $response['http_response']->get_response_object(); |
| 357 |
if ( isset( $response_object->url ) && $response_object->url ) { |
| 358 |
return $response_object->url; |
| 359 |
} |
| 360 |
|
| 361 |
return $url; |
| 362 |
} |
| 363 |
|
| 364 |
/** |
| 365 |
* Extract the Open Graph metadata from an HTML document. |
| 366 |
* |
| 367 |
* @param string $html The HTML document. |
| 368 |
* @param string $final_url The url the document was loaded from. |
| 369 |
* @param string $url The url as it appeared in the post. |
| 370 |
* |
| 371 |
* @return array|\WP_Error The link preview data or an error. |
| 372 |
*/ |
| 373 |
public static function parse( $html, $final_url, $url = null ) { |
| 374 |
if ( is_null( $url ) ) { |
| 375 |
$url = $final_url; |
| 376 |
} |
| 377 |
|
| 378 |
$head = $html; |
| 379 |
if ( preg_match( '/<head\b[^>]*>(.*?)<\/head>/is', $html, $m ) ) { |
| 380 |
$head = $m[1]; |
| 381 |
} |
| 382 |
|
| 383 |
$meta = array(); |
| 384 |
if ( preg_match_all( '/<meta\s((?>[^>"\']+|"[^"]*"|\'[^\']*\')*)\/?>/is', $head, $matches ) ) { |
| 385 |
foreach ( $matches[1] as $attributes ) { |
| 386 |
if ( ! preg_match( '/\b(?:property|name)\s*=\s*(["\'])\s*(.*?)\s*\1/is', $attributes, $m ) ) { |
| 387 |
continue; |
| 388 |
} |
| 389 |
$key = strtolower( $m[2] ); |
| 390 |
if ( isset( $meta[ $key ] ) ) { |
| 391 |
continue; |
| 392 |
} |
| 393 |
if ( ! preg_match( '/\bcontent\s*=\s*(["\'])(.*?)\1/is', $attributes, $m ) ) { |
| 394 |
continue; |
| 395 |
} |
| 396 |
$meta[ $key ] = trim( html_entity_decode( $m[2], ENT_QUOTES, 'UTF-8' ) ); |
| 397 |
} |
| 398 |
} |
| 399 |
|
| 400 |
$title = self::first_of( $meta, array( 'og:title', 'twitter:title' ) ); |
| 401 |
if ( ! $title && preg_match( '/<title\b[^>]*>(.*?)<\/title>/is', $head, $m ) ) { |
| 402 |
$title = trim( html_entity_decode( wp_strip_all_tags( $m[1] ), ENT_QUOTES, 'UTF-8' ) ); |
| 403 |
} |
| 404 |
|
| 405 |
$description = self::first_of( $meta, array( 'og:description', 'twitter:description', 'description' ) ); |
| 406 |
$image = self::first_of( $meta, array( 'og:image:secure_url', 'og:image:url', 'og:image', 'twitter:image', 'twitter:image:src' ) ); |
| 407 |
$site_name = self::first_of( $meta, array( 'og:site_name', 'application-name' ) ); |
| 408 |
|
| 409 |
if ( $image ) { |
| 410 |
$image = \WP_Http::make_absolute_url( $image, $final_url ); |
| 411 |
$scheme = wp_parse_url( $image, PHP_URL_SCHEME ); |
| 412 |
if ( ! in_array( strtolower( (string) $scheme ), array( 'http', 'https' ), true ) ) { |
| 413 |
$image = ''; |
| 414 |
} |
| 415 |
} |
| 416 |
|
| 417 |
if ( ! $title && ! $description && ! $image ) { |
| 418 |
return new \WP_Error( 'friends_link_preview_no_metadata', $url ); |
| 419 |
} |
| 420 |
|
| 421 |
$preview = array( |
| 422 |
'url' => $url, |
| 423 |
'host' => self::pretty_host( $final_url ), |
| 424 |
'title' => self::shorten( sanitize_text_field( $title ), 200 ), |
| 425 |
'description' => self::shorten( sanitize_text_field( $description ), 400 ), |
| 426 |
'image' => $image ? esc_url_raw( $image ) : '', |
| 427 |
'site_name' => self::shorten( sanitize_text_field( $site_name ), 100 ), |
| 428 |
); |
| 429 |
|
| 430 |
foreach ( array( 'width', 'height' ) as $dimension ) { |
| 431 |
$value = isset( $meta[ 'og:image:' . $dimension ] ) ? intval( $meta[ 'og:image:' . $dimension ] ) : 0; |
| 432 |
if ( $value > 0 ) { |
| 433 |
$preview[ 'image_' . $dimension ] = $value; |
| 434 |
} |
| 435 |
} |
| 436 |
|
| 437 |
return $preview; |
| 438 |
} |
| 439 |
|
| 440 |
/** |
| 441 |
* Get the first non-empty value of a list of keys. |
| 442 |
* |
| 443 |
* @param array $meta The metadata. |
| 444 |
* @param array $keys The keys to look at, in order. |
| 445 |
* |
| 446 |
* @return string The value or an empty string. |
| 447 |
*/ |
| 448 |
private static function first_of( $meta, $keys ) { |
| 449 |
foreach ( $keys as $key ) { |
| 450 |
if ( ! empty( $meta[ $key ] ) ) { |
| 451 |
return $meta[ $key ]; |
| 452 |
} |
| 453 |
} |
| 454 |
|
| 455 |
return ''; |
| 456 |
} |
| 457 |
|
| 458 |
/** |
| 459 |
* Shorten a string to a maximum length. |
| 460 |
* |
| 461 |
* @param string $text The text. |
| 462 |
* @param int $length The maximum length. |
| 463 |
* |
| 464 |
* @return string The shortened text. |
| 465 |
*/ |
| 466 |
private static function shorten( $text, $length ) { |
| 467 |
if ( mb_strlen( $text ) <= $length ) { |
| 468 |
return $text; |
| 469 |
} |
| 470 |
|
| 471 |
return trim( mb_substr( $text, 0, $length - 1 ) ) . '…'; |
| 472 |
} |
| 473 |
|
| 474 |
/** |
| 475 |
* Render the link preview card for the current post. |
| 476 |
* |
| 477 |
* @return string The HTML of the card or an empty string. |
| 478 |
*/ |
| 479 |
public static function render() { |
| 480 |
if ( ! self::is_enabled() ) { |
| 481 |
return ''; |
| 482 |
} |
| 483 |
|
| 484 |
ob_start(); |
| 485 |
Friends::template_loader()->get_template_part( 'frontend/parts/link-preview', get_post_format(), array() ); |
| 486 |
return ob_get_clean(); |
| 487 |
} |
| 488 |
|
| 489 |
/** |
| 490 |
* Get the hostname of a URL without a leading www. |
| 491 |
* |
| 492 |
* @param string $url The url. |
| 493 |
* |
| 494 |
* @return string The hostname. |
| 495 |
*/ |
| 496 |
public static function pretty_host( $url ) { |
| 497 |
$host = wp_parse_url( $url, PHP_URL_HOST ); |
| 498 |
if ( ! $host ) { |
| 499 |
return ''; |
| 500 |
} |
| 501 |
|
| 502 |
return preg_replace( '/^www\./i', '', $host ); |
| 503 |
} |
| 504 |
} |
| 505 |
|
