Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
81.25% |
130 / 160 |
|
64.29% |
9 / 14 |
CRAP | |
0.00% |
0 / 1 |
| PostContentUtils | |
81.25% |
130 / 160 |
|
64.29% |
9 / 14 |
64.83 | |
0.00% |
0 / 1 |
| getContentBody | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
3 | |||
| getPostBody | |
77.78% |
7 / 9 |
|
0.00% |
0 / 1 |
3.10 | |||
| getPostSummaryWrapperFormat | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
3 | |||
| getPostSummary | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| getSegments | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
20 | |||
| getContentWithoutExcludedBlocks | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
3 | |||
| getAudioEnabledBlocks | |
83.33% |
10 / 12 |
|
0.00% |
0 / 1 |
5.12 | |||
| getContentParams | |
100.00% |
31 / 31 |
|
100.00% |
1 / 1 |
7 | |||
| getMetadata | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
2 | |||
| getAllTaxonomiesAndTerms | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
4 | |||
| getAuthorName | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| addMarkerAttribute | |
80.00% |
4 / 5 |
|
0.00% |
0 / 1 |
3.07 | |||
| addMarkerAttributeWithHTMLTagProcessor | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
3 | |||
| addMarkerAttributeWithDOMDocument | |
95.00% |
19 / 20 |
|
0.00% |
0 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | namespace Beyondwords\Wordpress\Component\Post; |
| 6 | |
| 7 | /** |
| 8 | * BeyondWords Post Content Utilities. |
| 9 | * |
| 10 | * @package Beyondwords |
| 11 | * @subpackage Beyondwords/includes |
| 12 | * @author Stuart McAlpine <stu@beyondwords.io> |
| 13 | * @since 3.5.0 |
| 14 | */ |
| 15 | class PostContentUtils |
| 16 | { |
| 17 | public const DATE_FORMAT = 'Y-m-d\TH:i:s\Z'; |
| 18 | |
| 19 | /** |
| 20 | * Get the content "body" param for the audio, ready to be sent to the |
| 21 | * BeyondWords API. |
| 22 | * |
| 23 | * From API version 1.1 the "summary" param is going to be used differently, |
| 24 | * so for WordPress we now prepend the WordPress excerpt to the "body" param. |
| 25 | * |
| 26 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 27 | * |
| 28 | * @since 4.6.0 |
| 29 | * |
| 30 | * @return string The content body param. |
| 31 | */ |
| 32 | public static function getContentBody(int|\WP_Post $post): string|null |
| 33 | { |
| 34 | $post = get_post($post); |
| 35 | |
| 36 | if (!($post instanceof \WP_Post)) { |
| 37 | throw new \Exception(esc_html__('Post Not Found', 'speechkit')); |
| 38 | } |
| 39 | |
| 40 | $summary = PostContentUtils::getPostSummary($post); |
| 41 | $body = PostContentUtils::getPostBody($post); |
| 42 | |
| 43 | if ($summary) { |
| 44 | $format = PostContentUtils::getPostSummaryWrapperFormat($post); |
| 45 | |
| 46 | $body = sprintf($format, $summary) . $body; |
| 47 | } |
| 48 | |
| 49 | return $body; |
| 50 | } |
| 51 | |
| 52 | /** |
| 53 | * Get the post body for the audio content. |
| 54 | * |
| 55 | * @since 3.0.0 |
| 56 | * @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils |
| 57 | * @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false } |
| 58 | * @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody() |
| 59 | * @since 4.6.0 Renamed from PostContentUtils::getBody() to PostContentUtils::getPostBody() |
| 60 | * @since 4.7.0 Remove wpautop filter for block editor API requests. |
| 61 | * @since 5.0.0 Remove SpeechKit-Start shortcode. |
| 62 | * @since 5.0.0 Remove beyondwords_content filter. |
| 63 | * |
| 64 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 65 | * |
| 66 | * @return string The body (the processed $post->post_content). |
| 67 | */ |
| 68 | public static function getPostBody(int|\WP_Post $post): string|null |
| 69 | { |
| 70 | $post = get_post($post); |
| 71 | |
| 72 | if (!($post instanceof \WP_Post)) { |
| 73 | throw new \Exception(esc_html__('Post Not Found', 'speechkit')); |
| 74 | } |
| 75 | |
| 76 | $content = PostContentUtils::getContentWithoutExcludedBlocks($post); |
| 77 | |
| 78 | if (has_blocks($post)) { |
| 79 | // wpautop breaks our HTML markup when block editor paragraphs are empty |
| 80 | remove_filter('the_content', 'wpautop'); |
| 81 | |
| 82 | // But we still want to remove empty lines |
| 83 | $content = preg_replace('/^\h*\v+/m', '', $content); |
| 84 | } |
| 85 | |
| 86 | // Apply the_content filters to handle shortcodes etc |
| 87 | $content = apply_filters('the_content', $content); |
| 88 | |
| 89 | // Trim to remove trailing newlines – common for WordPress content |
| 90 | return trim($content); |
| 91 | } |
| 92 | |
| 93 | /** |
| 94 | * Get the post summary wrapper format. |
| 95 | * |
| 96 | * This is a <div> with optional attributes depending on the BeyondWords |
| 97 | * data of the post. |
| 98 | * |
| 99 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 100 | * |
| 101 | * @since 4.6.0 |
| 102 | * |
| 103 | * @return string The summary wrapper <div>. |
| 104 | */ |
| 105 | public static function getPostSummaryWrapperFormat(int|\WP_Post $post): string |
| 106 | { |
| 107 | $post = get_post($post); |
| 108 | |
| 109 | if (!($post instanceof \WP_Post)) { |
| 110 | throw new \Exception(esc_html__('Post Not Found', 'speechkit')); |
| 111 | } |
| 112 | |
| 113 | $summaryVoiceId = intval(get_post_meta($post->ID, 'beyondwords_summary_voice_id', true)); |
| 114 | |
| 115 | if ($summaryVoiceId > 0) { |
| 116 | return '<div data-beyondwords-summary="true" data-beyondwords-voice-id="' . $summaryVoiceId . '">%s</div>'; |
| 117 | } |
| 118 | |
| 119 | return '<div data-beyondwords-summary="true">%s</div>'; |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * Get the post summary for the audio content. |
| 124 | * |
| 125 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 126 | * |
| 127 | * @since 4.0.0 |
| 128 | * @since 4.6.0 Renamed from PostContentUtils::getSummary() to PostContentUtils::getPostSummary() |
| 129 | * |
| 130 | * @return string The summary. |
| 131 | */ |
| 132 | public static function getPostSummary(int|\WP_Post $post): string|null |
| 133 | { |
| 134 | $post = get_post($post); |
| 135 | |
| 136 | if (!($post instanceof \WP_Post)) { |
| 137 | throw new \Exception(esc_html__('Post Not Found', 'speechkit')); |
| 138 | } |
| 139 | |
| 140 | $summary = null; |
| 141 | |
| 142 | // Optionally send the excerpt to the REST API, if the plugin setting has been checked |
| 143 | $prependExcerpt = get_option('beyondwords_prepend_excerpt'); |
| 144 | |
| 145 | if ($prependExcerpt && has_excerpt($post)) { |
| 146 | // Escape characters |
| 147 | $summary = htmlentities($post->post_excerpt, ENT_QUOTES | ENT_XHTML); |
| 148 | // Apply WordPress filters |
| 149 | $summary = apply_filters('get_the_excerpt', $summary); |
| 150 | // Convert line breaks into paragraphs |
| 151 | $summary = trim(wpautop($summary)); |
| 152 | } |
| 153 | |
| 154 | return $summary; |
| 155 | } |
| 156 | |
| 157 | /** |
| 158 | * Get the segments for the audio content, ready to be sent to the BeyondWords API. |
| 159 | * |
| 160 | * @codeCoverageIgnore |
| 161 | * THIS METHOD IS CURRENTLY NOT IN USE. Segments cannot currently include HTML |
| 162 | * formatting tags such as <strong> and <em> so we do not pass segments, we pass |
| 163 | * a HTML string as the body param instead. |
| 164 | * |
| 165 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 166 | * |
| 167 | * @since 4.0.0 |
| 168 | */ |
| 169 | public static function getSegments(int|\WP_Post $post): array |
| 170 | { |
| 171 | if (! has_blocks($post)) { |
| 172 | return []; |
| 173 | } |
| 174 | |
| 175 | $titleSegment = (object) [ |
| 176 | 'section' => 'title', |
| 177 | 'text' => get_the_title($post), |
| 178 | ]; |
| 179 | |
| 180 | $summarySegment = (object) [ |
| 181 | 'section' => 'summary', |
| 182 | 'text' => PostContentUtils::getPostSummary($post), |
| 183 | ]; |
| 184 | |
| 185 | $blocks = PostContentUtils::getAudioEnabledBlocks($post); |
| 186 | |
| 187 | $bodySegments = array_map(function ($block) { |
| 188 | $marker = null; |
| 189 | |
| 190 | if (isset($block['attrs']) && isset($block['attrs']['beyondwordsMarker'])) { |
| 191 | $marker = $block['attrs']['beyondwordsMarker']; |
| 192 | } |
| 193 | |
| 194 | return (object) [ |
| 195 | 'section' => 'body', |
| 196 | 'marker' => $marker, |
| 197 | 'text' => trim(render_block($block)), |
| 198 | ]; |
| 199 | }, $blocks); |
| 200 | |
| 201 | // Merge title, summary and body segments |
| 202 | $segments = array_values(array_merge([$titleSegment], [$summarySegment], $bodySegments)); |
| 203 | |
| 204 | // Remove any segments with empty text |
| 205 | $segments = array_values(array_filter($segments, fn($segment) => ! empty($segment::text))); |
| 206 | |
| 207 | return $segments; |
| 208 | } |
| 209 | |
| 210 | /** |
| 211 | * Get the post content without blocks which have been filtered. |
| 212 | * |
| 213 | * We have added buttons into the Gutenberg editor to optionally exclude selected |
| 214 | * blocks from the source text for audio. |
| 215 | * |
| 216 | * This method filters all blocks, removing any which have been excluded. |
| 217 | * |
| 218 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 219 | * |
| 220 | * @since 3.8.0 |
| 221 | * @since 4.0.0 Replace for loop with array_reduce |
| 222 | * |
| 223 | * @return string The post body without excluded blocks. |
| 224 | */ |
| 225 | public static function getContentWithoutExcludedBlocks(int|\WP_Post $post): string |
| 226 | { |
| 227 | if (! has_blocks($post)) { |
| 228 | return trim($post->post_content); |
| 229 | } |
| 230 | |
| 231 | $blocks = parse_blocks($post->post_content); |
| 232 | $output = ''; |
| 233 | |
| 234 | $blocks = PostContentUtils::getAudioEnabledBlocks($post); |
| 235 | |
| 236 | foreach ($blocks as $block) { |
| 237 | $marker = $block['attrs']['beyondwordsMarker'] ?? ''; |
| 238 | |
| 239 | $output .= PostContentUtils::addMarkerAttribute( |
| 240 | render_block($block), |
| 241 | $marker |
| 242 | ); |
| 243 | } |
| 244 | |
| 245 | return $output; |
| 246 | } |
| 247 | |
| 248 | /** |
| 249 | * Get audio-enabled blocks. |
| 250 | * |
| 251 | * @param int|\WP_Post $post The WordPress post ID, or post object. |
| 252 | * |
| 253 | * @since 4.0.0 |
| 254 | * @since 5.0.0 Remove beyondwords_post_audio_enabled_blocks filter. |
| 255 | * |
| 256 | * @return array The blocks. |
| 257 | */ |
| 258 | public static function getAudioEnabledBlocks(int|\WP_Post $post): array |
| 259 | { |
| 260 | $post = get_post($post); |
| 261 | |
| 262 | if (! ($post instanceof \WP_Post)) { |
| 263 | return []; |
| 264 | } |
| 265 | |
| 266 | if (! has_blocks($post)) { |
| 267 | return []; |
| 268 | } |
| 269 | |
| 270 | $allBlocks = parse_blocks($post->post_content); |
| 271 | |
| 272 | return array_filter($allBlocks, function ($block) { |
| 273 | $enabled = true; |
| 274 | |
| 275 | if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) { |
| 276 | $enabled = (bool) $block['attrs']['beyondwordsAudio']; |
| 277 | } |
| 278 | |
| 279 | return $enabled; |
| 280 | }); |
| 281 | } |
| 282 | |
| 283 | /** |
| 284 | * Get the body param we pass to the API. |
| 285 | * |
| 286 | * @since 3.0.0 Introduced as getBodyJson. |
| 287 | * @since 3.3.0 Added metadata to aid custom playlist generation. |
| 288 | * @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils. |
| 289 | * @since 3.10.4 Rename `published_at` API param to `publish_date`. |
| 290 | * @since 4.0.0 Use new API params. |
| 291 | * @since 4.0.3 Ensure `image_url` is always a string. |
| 292 | * @since 4.3.0 Rename from getBodyJson to getContentParams. |
| 293 | * @since 4.6.0 Remove summary param & prepend body with summary. |
| 294 | * @since 5.0.0 Remove beyondwords_body_params filter. |
| 295 | * @since 6.0.0 Cast return value to string. |
| 296 | * |
| 297 | * @static |
| 298 | * @param int $postId WordPress Post ID. |
| 299 | * |
| 300 | * @return string JSON endoded params. |
| 301 | **/ |
| 302 | public static function getContentParams(int $postId): array|string |
| 303 | { |
| 304 | $body = [ |
| 305 | 'type' => 'auto_segment', |
| 306 | 'title' => get_the_title($postId), |
| 307 | 'body' => PostContentUtils::getContentBody($postId), |
| 308 | 'source_url' => get_the_permalink($postId), |
| 309 | 'source_id' => strval($postId), |
| 310 | 'author' => PostContentUtils::getAuthorName($postId), |
| 311 | 'image_url' => strval(wp_get_original_image_url(get_post_thumbnail_id($postId))), |
| 312 | 'metadata' => PostContentUtils::getMetadata($postId), |
| 313 | 'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId), |
| 314 | ]; |
| 315 | |
| 316 | $status = get_post_status($postId); |
| 317 | |
| 318 | /* |
| 319 | * If the post status is draft/pending then we explicity send |
| 320 | * { published: false } to the BeyondWords API, to prevent the |
| 321 | * generated audio from being published in playlists. |
| 322 | * |
| 323 | * We also omit { publish_date } because get_post_time() returns `false` |
| 324 | * for posts which are "Pending Review". |
| 325 | */ |
| 326 | if (in_array($status, ['draft', 'pending'])) { |
| 327 | $body['published'] = false; |
| 328 | unset($body['publish_date']); |
| 329 | } elseif (get_option('beyondwords_project_auto_publish_enabled')) { |
| 330 | $body['published'] = true; |
| 331 | } |
| 332 | |
| 333 | $languageCode = get_post_meta($postId, 'beyondwords_language_code', true); |
| 334 | |
| 335 | if ($languageCode) { |
| 336 | $body['language'] = $languageCode; |
| 337 | } |
| 338 | |
| 339 | $bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true)); |
| 340 | |
| 341 | if ($bodyVoiceId > 0) { |
| 342 | $body['body_voice_id'] = $bodyVoiceId; |
| 343 | } |
| 344 | |
| 345 | $titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true)); |
| 346 | |
| 347 | if ($titleVoiceId > 0) { |
| 348 | $body['title_voice_id'] = $titleVoiceId; |
| 349 | } |
| 350 | |
| 351 | $summaryVoiceId = intval(get_post_meta($postId, 'beyondwords_summary_voice_id', true)); |
| 352 | |
| 353 | if ($summaryVoiceId > 0) { |
| 354 | $body['summary_voice_id'] = $summaryVoiceId; |
| 355 | } |
| 356 | |
| 357 | /** |
| 358 | * Filters the params we send to the BeyondWords API 'content' endpoint. |
| 359 | * |
| 360 | * @since 4.0.0 Introduced as beyondwords_body_params |
| 361 | * @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params |
| 362 | * |
| 363 | * @param array $body The params we send to the BeyondWords API. |
| 364 | * @param array $postId WordPress post ID. |
| 365 | */ |
| 366 | $body = apply_filters('beyondwords_content_params', $body, $postId); |
| 367 | |
| 368 | return (string) wp_json_encode($body); |
| 369 | } |
| 370 | |
| 371 | /** |
| 372 | * Get the post metadata to send with BeyondWords API requests. |
| 373 | * |
| 374 | * The metadata key is defined by the BeyondWords API as "A custom object |
| 375 | * for storing meta information". |
| 376 | * |
| 377 | * The metadata values are used to create filters for playlists in the |
| 378 | * BeyondWords dashboard. |
| 379 | * |
| 380 | * We currently only include taxonomies by default, and the output of this |
| 381 | * method can be filtered using the `beyondwords_post_metadata` filter. |
| 382 | * |
| 383 | * @since 3.3.0 |
| 384 | * @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils. |
| 385 | * @since 5.0.0 Remove beyondwords_post_metadata filter. |
| 386 | * |
| 387 | * @param int $postId Post ID. |
| 388 | * |
| 389 | * @return object The metadata object (empty if no metadata). |
| 390 | */ |
| 391 | public static function getMetadata(int $postId): array|object |
| 392 | { |
| 393 | $metadata = new \stdClass(); |
| 394 | |
| 395 | $taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId); |
| 396 | |
| 397 | if (count((array)$taxonomy)) { |
| 398 | $metadata->taxonomy = $taxonomy; |
| 399 | } |
| 400 | |
| 401 | return $metadata; |
| 402 | } |
| 403 | |
| 404 | /** |
| 405 | * Get all taxonomies, and their selected terms, for a post. |
| 406 | * |
| 407 | * Returns an associative array of taxonomy names and terms. |
| 408 | * |
| 409 | * For example: |
| 410 | * |
| 411 | * array( |
| 412 | * "categories" => array("Category 1"), |
| 413 | * "post_tag" => array("Tag 1", "Tag 2", "Tag 3"), |
| 414 | * ) |
| 415 | * |
| 416 | * @since 3.3.0 |
| 417 | * @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils |
| 418 | * |
| 419 | * @param int $postId Post ID. |
| 420 | * |
| 421 | * @return object The taxonomies object (empty if no taxonomies). |
| 422 | */ |
| 423 | public static function getAllTaxonomiesAndTerms(int $postId): array|object |
| 424 | { |
| 425 | $postType = get_post_type($postId); |
| 426 | |
| 427 | $postTypeTaxonomies = get_object_taxonomies($postType); |
| 428 | |
| 429 | $taxonomies = new \stdClass(); |
| 430 | |
| 431 | foreach ($postTypeTaxonomies as $postTypeTaxonomy) { |
| 432 | $terms = get_the_terms($postId, $postTypeTaxonomy); |
| 433 | |
| 434 | if (! empty($terms) && ! is_wp_error($terms)) { |
| 435 | $taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name'); |
| 436 | } |
| 437 | } |
| 438 | |
| 439 | return $taxonomies; |
| 440 | } |
| 441 | |
| 442 | /** |
| 443 | * Get author name for a post. |
| 444 | * |
| 445 | * @since 3.10.4 |
| 446 | * |
| 447 | * @param int $postId Post ID. |
| 448 | */ |
| 449 | public static function getAuthorName(int $postId): string |
| 450 | { |
| 451 | $authorId = get_post_field('post_author', $postId); |
| 452 | |
| 453 | return get_the_author_meta('display_name', $authorId); |
| 454 | } |
| 455 | |
| 456 | /** |
| 457 | * Add data-beyondwords-marker attribute to the root elements in a HTML |
| 458 | * string (typically the rendered HTML of a single block). |
| 459 | * |
| 460 | * Checks to see whether we can use WP_HTML_Tag_Processor, or whether we |
| 461 | * fall back to using DOMDocument to add the marker. |
| 462 | * |
| 463 | * @since 4.2.2 |
| 464 | * |
| 465 | * @param string $html HTML. |
| 466 | * @param string $marker Marker UUID. |
| 467 | * |
| 468 | * @return string HTML. |
| 469 | */ |
| 470 | public static function addMarkerAttribute(string $html, string $marker): string |
| 471 | { |
| 472 | if (! $marker) { |
| 473 | return $html; |
| 474 | } |
| 475 | |
| 476 | // Prefer WP_HTML_Tag_Processor, introduced in WordPress 6.2 |
| 477 | if (class_exists('WP_HTML_Tag_Processor')) { |
| 478 | return PostContentUtils::addMarkerAttributeWithHTMLTagProcessor($html, $marker); |
| 479 | } else { |
| 480 | return PostContentUtils::addMarkerAttributeWithDOMDocument($html, $marker); |
| 481 | } |
| 482 | } |
| 483 | |
| 484 | /** |
| 485 | * Add data-beyondwords-marker attribute to the root elements in a HTML |
| 486 | * string using WP_HTML_Tag_Processor. |
| 487 | * |
| 488 | * @since 4.0.0 |
| 489 | * @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php |
| 490 | * to src/Component/Post/PostContentUtils.php |
| 491 | * @since 4.7.0 Prevent empty data-beyondwords-marker attributes. |
| 492 | * |
| 493 | * @param string $html HTML. |
| 494 | * @param string $marker Marker UUID. |
| 495 | * |
| 496 | * @return string HTML. |
| 497 | */ |
| 498 | public static function addMarkerAttributeWithHTMLTagProcessor(string $html, string $marker): string |
| 499 | { |
| 500 | if (! $marker) { |
| 501 | return $html; |
| 502 | } |
| 503 | |
| 504 | // https://github.com/WordPress/gutenberg/pull/42485 |
| 505 | $tags = new \WP_HTML_Tag_Processor($html); |
| 506 | |
| 507 | if ($tags->next_tag()) { |
| 508 | $tags->set_attribute('data-beyondwords-marker', $marker); |
| 509 | } |
| 510 | |
| 511 | return strval($tags); |
| 512 | } |
| 513 | |
| 514 | /** |
| 515 | * Add data-beyondwords-marker attribute to the root elements in a HTML |
| 516 | * string using DOMDocument. |
| 517 | * |
| 518 | * This is a fallback, since WP_HTML_Tag_Processor was only shipped with |
| 519 | * WordPress 6.2 on 19 April 2023. |
| 520 | * |
| 521 | * https://make.wordpress.org/core/2022/10/13/whats-new-in-gutenberg-14-3-12-october/ |
| 522 | * |
| 523 | * Note: It is not ideal to do all the $bodyElement/$fullHtml processing |
| 524 | * in this method, but without it DOMDocument does not work as expected if |
| 525 | * there is more than 1 root element. The approach here has been taken from |
| 526 | * some historic Gutenberg code before they implemented WP_HTML_Tag_Processor: |
| 527 | * |
| 528 | * https://github.com/WordPress/gutenberg/blob/6671cef1179412a2bbd4969cbbc82705c7f69bac/lib/block-supports/index.php |
| 529 | * |
| 530 | * @since 4.0.0 |
| 531 | * @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php |
| 532 | * to src/Component/Post/PostContentUtils.php |
| 533 | * @since 4.7.0 Prevent empty data-beyondwords-marker attributes. |
| 534 | * |
| 535 | * @param string $html HTML. |
| 536 | * @param string $marker Marker UUID. |
| 537 | * |
| 538 | * @return string HTML. |
| 539 | */ |
| 540 | public static function addMarkerAttributeWithDOMDocument(string $html, string $marker): string |
| 541 | { |
| 542 | if (! $marker) { |
| 543 | return $html; |
| 544 | } |
| 545 | |
| 546 | $dom = new \DOMDocument('1.0', 'utf-8'); |
| 547 | |
| 548 | $wrappedHtml = |
| 549 | '<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body>' |
| 550 | . $html |
| 551 | . '</body></html>'; |
| 552 | |
| 553 | $success = $dom->loadHTML($wrappedHtml, LIBXML_HTML_NODEFDTD | LIBXML_COMPACT); |
| 554 | |
| 555 | if (! $success) { |
| 556 | return $html; |
| 557 | } |
| 558 | |
| 559 | // Structure is like `<html><head/><body/></html>`, so body is the `lastChild` of our document. |
| 560 | $bodyElement = $dom->documentElement->lastChild; |
| 561 | |
| 562 | $xpath = new \DOMXPath($dom); |
| 563 | $blockRoot = $xpath->query('./*', $bodyElement)[0]; |
| 564 | |
| 565 | if (empty($blockRoot)) { |
| 566 | return $html; |
| 567 | } |
| 568 | |
| 569 | $blockRoot->setAttribute('data-beyondwords-marker', $marker); |
| 570 | |
| 571 | // Avoid using `$dom->saveHtml( $node )` because the node results may not produce consistent |
| 572 | // whitespace. Saving the root HTML `$dom->saveHtml()` prevents this behavior. |
| 573 | $fullHtml = $dom->saveHtml(); |
| 574 | |
| 575 | // Find the <body> open/close tags. The open tag needs to be adjusted so we get inside the tag |
| 576 | // and not the tag itself. |
| 577 | $start = strpos($fullHtml, '<body>', 0) + strlen('<body>'); |
| 578 | $end = strpos($fullHtml, '</body>', $start); |
| 579 | |
| 580 | return trim(substr($fullHtml, $start, $end - $start)); |
| 581 | } |
| 582 | } |