| @@ -40,18 +40,22 @@ | ||
| 40 | 40 | |
| 41 | 41 | /** |
| 42 | 42 | * Post meta keys that hold builder data, in priority order. |
| 43 | 43 | * |
| 44 | - * Several generations of the same builder are listed on purpose: Oxygen 6 | |
| 45 | - * is Breakdance under the hood (`_breakdance_data`), while earlier Oxygen | |
| 46 | - * releases used `_oxygen_data` or the shortcode-based | |
| 47 | - * `ct_builder_shortcodes`. A site can only have one of them. | |
| 44 | + * Several generations of the same builder are listed on purpose. Oxygen 6 | |
| 45 | + * is Breakdance under the hood and writes the same tree, under its own | |
| 46 | + * prefix: Breakdance keeps it in `_breakdance_data`, Oxygen 6 in | |
| 47 | + * `_oxygen_data` (the key is `__bdox('_meta_prefix') . 'data'`, and the | |
| 48 | + * prefix is `_oxygen_` under Oxygen). Both store it inside a | |
| 49 | + * `tree_json_string` envelope, see unwrap_tree_envelope(). Earlier Oxygen | |
| 50 | + * releases used the shortcode-based `ct_builder_shortcodes` and its JSON | |
| 51 | + * sibling. A site can only have one of them. | |
| 48 | 52 | * |
| 49 | 53 | * @var string[] |
| 50 | 54 | */ |
| 51 | 55 | private const BUILDER_META_KEYS = [ |
| 52 | - '_breakdance_data', // Oxygen 6+ / Breakdance | |
| 53 | - '_oxygen_data', // Oxygen (earlier releases) | |
| 56 | + '_breakdance_data', // Breakdance | |
| 57 | + '_oxygen_data', // Oxygen 6+ (Breakdance engine, Oxygen prefix) | |
| 54 | 58 | // Oxygen classic. 4.x writes the tree as JSON to `ct_builder_json` |
| 55 | 59 | // while still keeping `ct_builder_shortcodes`. A post carrying only |
| 56 | 60 | // the JSON key used to match no key at all and fall through to an |
| 57 | 61 | // empty `post_content`, which reads as a one-word page (#776). |
| @@ -131,8 +135,42 @@ | ||
| 131 | 135 | */ |
| 132 | 136 | private const BRICKS_POST_CONTENT_ELEMENT = 'post-content'; |
| 133 | 137 | |
| 134 | 138 | /** |
| 139 | + * Bricks' Heading element, and the tag it renders when none is stored. | |
| 140 | + * | |
| 141 | + * Bricks leaves a setting out of storage while it equals its default, so a | |
| 142 | + * Heading left on its default tag is stored with no `tag` at all. Bricks | |
| 143 | + * 2.4.1 renders it as `h3` (`Element_Heading::$tag`, overridable by the | |
| 144 | + * active theme style's `tag`), and the walker, which only wraps text whose | |
| 145 | + * node names a tag, read it as body copy (#908). | |
| 146 | + * | |
| 147 | + * @since 2.15.0 | |
| 148 | + * @var string | |
| 149 | + */ | |
| 150 | + private const BRICKS_HEADING_ELEMENT = 'heading'; | |
| 151 | + | |
| 152 | + /** | |
| 153 | + * Tag a Bricks Heading renders when neither it nor a theme style sets one. | |
| 154 | + * | |
| 155 | + * @since 2.15.0 | |
| 156 | + * @var string | |
| 157 | + */ | |
| 158 | + private const BRICKS_HEADING_DEFAULT_TAG = 'h3'; | |
| 159 | + | |
| 160 | + /** | |
| 161 | + * Tag an Elementor Heading widget renders when `header_size` is not stored. | |
| 162 | + * | |
| 163 | + * Elementor saves `settings.toJSON({ remove: ['default'] })`, so a heading | |
| 164 | + * left on its default size has no `header_size` in `_elementor_data`, and | |
| 165 | + * that default is `h2` (#908). | |
| 166 | + * | |
| 167 | + * @since 2.15.0 | |
| 168 | + * @var string | |
| 169 | + */ | |
| 170 | + private const ELEMENTOR_HEADING_DEFAULT_TAG = 'h2'; | |
| 171 | + | |
| 172 | + /** | |
| 135 | 173 | * Resolved Bricks trees for this request, keyed by post ID. |
| 136 | 174 | * |
| 137 | 175 | * Rendering one page asks for the tree about twenty times — every |
| 138 | 176 | * description, every schema node, the FAQ guard — and resolving it is not |
| @@ -542,12 +580,82 @@ | ||
| 542 | 580 | if (!is_array($stored) || empty($stored)) { |
| 543 | 581 | return []; |
| 544 | 582 | } |
| 545 | 583 | |
| 546 | - return self::expand_bricks_components($stored); | |
| 584 | + return self::expand_bricks_components(self::bricks_render_order($stored)); | |
| 547 | 585 | } |
| 548 | 586 | |
| 549 | 587 | /** |
| 588 | + * A flat Bricks element list, in the order Bricks renders it. | |
| 589 | + * | |
| 590 | + * Bricks stores one flat list and links it with `parent` and `children` | |
| 591 | + * ids. `Frontend::render_data()` renders the root elements in list order | |
| 592 | + * and each element's children in the order of its `children` array, so a | |
| 593 | + * child's position in the list says nothing about where it appears on the | |
| 594 | + * page. Walking the list as stored put a section's contents wherever they | |
| 595 | + * happened to be saved (#907). | |
| 596 | + * | |
| 597 | + * Anything the walk does not reach (an orphan, a cycle) keeps its stored | |
| 598 | + * position after the rest, so no copy is dropped. | |
| 599 | + * | |
| 600 | + * @since 2.15.0 | |
| 601 | + * | |
| 602 | + * @param array $elements Flat Bricks element list. | |
| 603 | + * @return array The same elements, in render order. | |
| 604 | + */ | |
| 605 | + private static function bricks_render_order(array $elements): array { | |
| 606 | + $by_id = []; | |
| 607 | + foreach ($elements as $index => $element) { | |
| 608 | + $id = is_array($element) ? ($element['id'] ?? null) : null; | |
| 609 | + if (is_scalar($id) && '' !== (string) $id && !isset($by_id[(string) $id])) { | |
| 610 | + $by_id[(string) $id] = $index; | |
| 611 | + } | |
| 612 | + } | |
| 613 | + | |
| 614 | + if (empty($by_id)) { | |
| 615 | + return $elements; | |
| 616 | + } | |
| 617 | + | |
| 618 | + $ordered = []; | |
| 619 | + $placed = []; | |
| 620 | + | |
| 621 | + $place = static function ($index) use (&$place, &$ordered, &$placed, $elements, $by_id): void { | |
| 622 | + if (isset($placed[$index])) { | |
| 623 | + return; | |
| 624 | + } | |
| 625 | + | |
| 626 | + $placed[$index] = true; | |
| 627 | + $ordered[] = $elements[$index]; | |
| 628 | + | |
| 629 | + $children = is_array($elements[$index]) ? ($elements[$index]['children'] ?? []) : []; | |
| 630 | + if (!is_array($children)) { | |
| 631 | + return; | |
| 632 | + } | |
| 633 | + | |
| 634 | + foreach ($children as $child_id) { | |
| 635 | + if (is_scalar($child_id) && isset($by_id[(string) $child_id])) { | |
| 636 | + $place($by_id[(string) $child_id]); | |
| 637 | + } | |
| 638 | + } | |
| 639 | + }; | |
| 640 | + | |
| 641 | + foreach ($elements as $index => $element) { | |
| 642 | + $parent = is_array($element) ? ($element['parent'] ?? null) : null; | |
| 643 | + if (empty($parent) || !is_scalar($parent) || !isset($by_id[(string) $parent])) { | |
| 644 | + $place($index); | |
| 645 | + } | |
| 646 | + } | |
| 647 | + | |
| 648 | + foreach ($elements as $index => $element) { | |
| 649 | + if (!isset($placed[$index])) { | |
| 650 | + $ordered[] = $element; | |
| 651 | + } | |
| 652 | + } | |
| 653 | + | |
| 654 | + return $ordered; | |
| 655 | + } | |
| 656 | + | |
| 657 | + /** | |
| 550 | 658 | * Resolve an arbitrary chunk of editor markup for the given post. |
| 551 | 659 | * |
| 552 | 660 | * The editor sends its live content to the scorer so an author sees their |
| 553 | 661 | * unsaved edits reflected. On a builder page that live string is the raw |
| @@ -627,8 +735,19 @@ | ||
| 627 | 735 | * runs after the form prints its hidden `post_ID` and before the title and |
| 628 | 736 | * editor, which then showed the related post, and Update saved it over the |
| 629 | 737 | * original (#860). |
| 630 | 738 | * |
| 739 | + * Secondary queries also run with front-end statuses. In wp-admin core | |
| 740 | + * marks every `WP_Query` as an admin query and, when no `post_status` is | |
| 741 | + * set, adds the statuses the admin post list shows, draft among them, so | |
| 742 | + * a related-posts shortcode listed drafts the front end never shows and | |
| 743 | + * the editor-load analysis disagreed with REST and the page (#902). | |
| 744 | + * | |
| 745 | + * The main query points at the post too. Restoring the globals afterwards | |
| 746 | + * (#860) did not reach between shortcodes: a related-posts loop's own | |
| 747 | + * `wp_reset_postdata()` still found no post on the main query, so every | |
| 748 | + * later shortcode in the same render saw the last looped post (#903). | |
| 749 | + * | |
| 631 | 750 | * @param string $raw Raw post content. |
| 632 | 751 | * @param \WP_Post $post Post the content belongs to. |
| 633 | 752 | * @return string Rendered content. |
| 634 | 753 | */ |
| @@ -639,11 +758,29 @@ | ||
| 639 | 758 | |
| 640 | 759 | $content = $raw; |
| 641 | 760 | $previous = self::snapshot_post_globals(); |
| 642 | 761 | |
| 762 | + // After pre_get_posts core reads `is_admin` only to add the admin | |
| 763 | + // list's statuses when none were asked for, so queries that set | |
| 764 | + // `post_status`, and the main query, are untouched. | |
| 765 | + $front_end_statuses = static function ($query): void { | |
| 766 | + if ($query instanceof \WP_Query && !$query->is_main_query()) { | |
| 767 | + $query->is_admin = false; | |
| 768 | + } | |
| 769 | + }; | |
| 770 | + | |
| 771 | + // `wp_reset_postdata()` returns to the main query's post, which admin | |
| 772 | + // and REST requests do not have. Restored in finally, null included. | |
| 773 | + $main_query = (isset($GLOBALS['wp_query']) && $GLOBALS['wp_query'] instanceof \WP_Query) ? $GLOBALS['wp_query'] : null; | |
| 774 | + $main_query_post = $main_query ? $main_query->post : null; | |
| 775 | + | |
| 643 | 776 | try { |
| 644 | 777 | // phpcs:ignore WordPress.WP.GlobalVariablesOverride.Prohibited -- Made current for the render, restored in finally. |
| 645 | 778 | $GLOBALS['post'] = $post; |
| 779 | + add_action('pre_get_posts', $front_end_statuses, PHP_INT_MIN); | |
| 780 | + if ($main_query) { | |
| 781 | + $main_query->post = $post; | |
| 782 | + } | |
| 646 | 783 | |
| 647 | 784 | // Fires `the_post`, which these paths never fired before: admin, |
| 648 | 785 | // REST and cron analysis had no current post at all. That is the |
| 649 | 786 | // same signal the front-end loop sends and it is what makes |
| @@ -666,8 +803,12 @@ | ||
| 666 | 803 | } |
| 667 | 804 | } catch (\Throwable $e) { |
| 668 | 805 | return $raw; |
| 669 | 806 | } finally { |
| 807 | + if ($main_query) { | |
| 808 | + $main_query->post = $main_query_post; | |
| 809 | + } | |
| 810 | + remove_action('pre_get_posts', $front_end_statuses, PHP_INT_MIN); | |
| 670 | 811 | self::restore_post_globals($previous); |
| 671 | 812 | } |
| 672 | 813 | |
| 673 | 814 | return self::is_blank($content) ? $raw : $content; |
| @@ -776,9 +917,9 @@ | ||
| 776 | 917 | return ''; |
| 777 | 918 | } |
| 778 | 919 | |
| 779 | 920 | return self::strip_bricks_dynamic_tags( |
| 780 | - self::text_from_tree(self::without_bricks_element_labels($tree)) | |
| 921 | + self::text_from_tree(self::with_bricks_heading_tags(self::without_bricks_element_labels($tree))) | |
| 781 | 922 | ); |
| 782 | 923 | } |
| 783 | 924 | |
| 784 | 925 | /** |
| @@ -976,9 +1117,9 @@ | ||
| 976 | 1117 | continue; |
| 977 | 1118 | } |
| 978 | 1119 | |
| 979 | 1120 | if (isset($component['id']) && $component['id'] === $cid && !empty($component['elements'])) { |
| 980 | - return is_array($component['elements']) ? $component['elements'] : []; | |
| 1121 | + return is_array($component['elements']) ? self::bricks_render_order($component['elements']) : []; | |
| 981 | 1122 | } |
| 982 | 1123 | } |
| 983 | 1124 | |
| 984 | 1125 | return []; |
| @@ -1011,8 +1152,116 @@ | ||
| 1011 | 1152 | return $tree; |
| 1012 | 1153 | } |
| 1013 | 1154 | |
| 1014 | 1155 | /** |
| 1156 | + * Give each Bricks Heading the tag it renders with when none is stored. | |
| 1157 | + * | |
| 1158 | + * Applied on the Bricks path only. Most other builders' text nodes carry no | |
| 1159 | + * tag because they are not headings, so a generic "text without a tag is a | |
| 1160 | + * heading" rule in heading_tag_from() would turn every paragraph into one. | |
| 1161 | + * | |
| 1162 | + * A `tag` of `custom` is left alone: the element then renders its | |
| 1163 | + * `customTag`, which is not necessarily a heading. | |
| 1164 | + * | |
| 1165 | + * @since 2.15.0 | |
| 1166 | + * | |
| 1167 | + * @param array $tree Bricks content area. | |
| 1168 | + * @return array Tree with each untagged Heading's default tag filled in. | |
| 1169 | + */ | |
| 1170 | + private static function with_bricks_heading_tags(array $tree): array { | |
| 1171 | + $default = null; | |
| 1172 | + | |
| 1173 | + foreach ($tree as $index => $element) { | |
| 1174 | + if (!is_array($element) || self::BRICKS_HEADING_ELEMENT !== ($element['name'] ?? null)) { | |
| 1175 | + continue; | |
| 1176 | + } | |
| 1177 | + | |
| 1178 | + $settings = $element['settings'] ?? []; | |
| 1179 | + if (!is_array($settings)) { | |
| 1180 | + continue; | |
| 1181 | + } | |
| 1182 | + | |
| 1183 | + $tag = $settings['tag'] ?? ''; | |
| 1184 | + if (is_string($tag) && '' !== trim($tag)) { | |
| 1185 | + continue; | |
| 1186 | + } | |
| 1187 | + | |
| 1188 | + if (null === $default) { | |
| 1189 | + $default = self::bricks_default_heading_tag(); | |
| 1190 | + } | |
| 1191 | + | |
| 1192 | + $settings['tag'] = $default; | |
| 1193 | + $tree[$index]['settings'] = $settings; | |
| 1194 | + } | |
| 1195 | + | |
| 1196 | + return $tree; | |
| 1197 | + } | |
| 1198 | + | |
| 1199 | + /** | |
| 1200 | + * The tag Bricks gives a Heading that does not set one. | |
| 1201 | + * | |
| 1202 | + * The active theme style can change it. Bricks only loads theme styles for | |
| 1203 | + * a front-end render, so in admin, REST and CLI requests this is the | |
| 1204 | + * element's own default. | |
| 1205 | + * | |
| 1206 | + * @since 2.15.0 | |
| 1207 | + * | |
| 1208 | + * @return string Heading tag, h1 to h6. | |
| 1209 | + */ | |
| 1210 | + private static function bricks_default_heading_tag(): string { | |
| 1211 | + if (class_exists('\\Bricks\\Theme_Styles') | |
| 1212 | + && method_exists('\\Bricks\\Theme_Styles', 'get_setting_by_key') | |
| 1213 | + ) { | |
| 1214 | + try { | |
| 1215 | + $styled = \Bricks\Theme_Styles::get_setting_by_key(self::BRICKS_HEADING_ELEMENT, 'tag'); | |
| 1216 | + } catch (\Throwable $e) { | |
| 1217 | + $styled = null; | |
| 1218 | + } | |
| 1219 | + | |
| 1220 | + if (is_string($styled) && preg_match('/^h[1-6]$/i', trim($styled))) { | |
| 1221 | + return strtolower(trim($styled)); | |
| 1222 | + } | |
| 1223 | + } | |
| 1224 | + | |
| 1225 | + return self::BRICKS_HEADING_DEFAULT_TAG; | |
| 1226 | + } | |
| 1227 | + | |
| 1228 | + /** | |
| 1229 | + * Give each Elementor Heading widget its default `header_size` if unstored. | |
| 1230 | + * | |
| 1231 | + * @since 2.15.0 | |
| 1232 | + * | |
| 1233 | + * @param array $elements Decoded `_elementor_data`. | |
| 1234 | + * @return array The same tree, with untagged Heading widgets tagged. | |
| 1235 | + */ | |
| 1236 | + private static function with_elementor_heading_tags(array $elements): array { | |
| 1237 | + foreach ($elements as $index => $element) { | |
| 1238 | + if (!is_array($element)) { | |
| 1239 | + continue; | |
| 1240 | + } | |
| 1241 | + | |
| 1242 | + if ('heading' === ($element['widgetType'] ?? null)) { | |
| 1243 | + $settings = $element['settings'] ?? []; | |
| 1244 | + if (is_array($settings)) { | |
| 1245 | + $size = $settings['header_size'] ?? ''; | |
| 1246 | + if (!is_string($size) || '' === trim($size)) { | |
| 1247 | + $settings['header_size'] = self::ELEMENTOR_HEADING_DEFAULT_TAG; | |
| 1248 | + $element['settings'] = $settings; | |
| 1249 | + } | |
| 1250 | + } | |
| 1251 | + } | |
| 1252 | + | |
| 1253 | + if (!empty($element['elements']) && is_array($element['elements'])) { | |
| 1254 | + $element['elements'] = self::with_elementor_heading_tags($element['elements']); | |
| 1255 | + } | |
| 1256 | + | |
| 1257 | + $elements[$index] = $element; | |
| 1258 | + } | |
| 1259 | + | |
| 1260 | + return $elements; | |
| 1261 | + } | |
| 1262 | + | |
| 1263 | + /** | |
| 1015 | 1264 | * Remove Bricks dynamic-data tags from extracted text. |
| 1016 | 1265 | * |
| 1017 | 1266 | * Bricks stores `{post_title}`, `{post_meta:price}`, `{echo:my_fn}` and the |
| 1018 | 1267 | * like verbatim and resolves them when it renders. Extraction reads the |
| @@ -1120,12 +1369,15 @@ | ||
| 1120 | 1369 | $stored = get_post_meta($post_id, $key, true); |
| 1121 | 1370 | |
| 1122 | 1371 | if (is_string($stored) && '' !== trim($stored)) { |
| 1123 | 1372 | $decoded = json_decode($stored, true); |
| 1373 | + if ('_elementor_data' === $key && is_array($decoded)) { | |
| 1374 | + $decoded = self::with_elementor_heading_tags($decoded); | |
| 1375 | + } | |
| 1124 | 1376 | |
| 1125 | 1377 | // JSON node tree (Breakdance/Oxygen 6, Elementor). |
| 1126 | 1378 | if (is_array($decoded)) { |
| 1127 | - $text = self::text_from_tree($decoded); | |
| 1379 | + $text = self::text_from_tree(self::unwrap_tree_envelope($decoded)); | |
| 1128 | 1380 | if (!self::is_blank($text)) { |
| 1129 | 1381 | return $text; |
| 1130 | 1382 | } |
| 1131 | 1383 | continue; |
| @@ -1137,8 +1389,9 @@ | ||
| 1137 | 1389 | // Some builders store an already-decoded tree — an array for most, |
| 1138 | 1390 | // an array of objects for Beaver Builder (#449). |
| 1139 | 1391 | $tree = self::as_children($stored); |
| 1140 | 1392 | if (null !== $tree) { |
| 1393 | + $tree = self::unwrap_tree_envelope($tree); | |
| 1141 | 1394 | $text = self::text_from_tree($tree); |
| 1142 | 1395 | if (!self::is_blank($text)) { |
| 1143 | 1396 | return $text; |
| 1144 | 1397 | } |
| @@ -1148,8 +1401,48 @@ | ||
| 1148 | 1401 | return ''; |
| 1149 | 1402 | } |
| 1150 | 1403 | |
| 1151 | 1404 | /** |
| 1405 | + * The node tree inside a Breakdance / Oxygen 6 storage envelope. | |
| 1406 | + * | |
| 1407 | + * Neither builder stores its tree directly. The meta value is | |
| 1408 | + * `{"tree_json_string": "<the tree, JSON-encoded again>"}`, so one | |
| 1409 | + * json_decode() yields the envelope, not the tree. Walked as a tree, the | |
| 1410 | + * envelope is a single string leaf: kept whole as "content" when any | |
| 1411 | + * element held rich text (the encoded JSON then reached scoring, the | |
| 1412 | + * get-post-content ability and Markdown for AI), dropped when none did, | |
| 1413 | + * leaving the page empty (#905). | |
| 1414 | + * | |
| 1415 | + * An envelope whose inner string does not decode returns an empty tree, | |
| 1416 | + * never the string: handing the raw JSON back to the walker would bring | |
| 1417 | + * the JSON-as-content failure back on corrupt data. Anything that is not | |
| 1418 | + * an envelope is returned unchanged, so a bare tree still resolves. | |
| 1419 | + * | |
| 1420 | + * @since 2.15.0 | |
| 1421 | + * | |
| 1422 | + * @param array $decoded Decoded meta value. | |
| 1423 | + * @return array The node tree. | |
| 1424 | + */ | |
| 1425 | + private static function unwrap_tree_envelope(array $decoded): array { | |
| 1426 | + if (!array_key_exists('tree_json_string', $decoded)) { | |
| 1427 | + return $decoded; | |
| 1428 | + } | |
| 1429 | + | |
| 1430 | + $inner = is_string($decoded['tree_json_string']) | |
| 1431 | + ? json_decode($decoded['tree_json_string'], true) | |
| 1432 | + : $decoded['tree_json_string']; | |
| 1433 | + | |
| 1434 | + if (is_array($inner)) { | |
| 1435 | + return $inner; | |
| 1436 | + } | |
| 1437 | + | |
| 1438 | + // Re-serialised envelopes can carry the tree as an object. | |
| 1439 | + $inner = self::as_children($inner); | |
| 1440 | + | |
| 1441 | + return null !== $inner ? $inner : []; | |
| 1442 | + } | |
| 1443 | + | |
| 1444 | + /** | |
| 1152 | 1445 | * Whether a meta key is one of Oxygen classic's storage keys. |
| 1153 | 1446 | * |
| 1154 | 1447 | * @since 2.10.0 |
| 1155 | 1448 | * |
| @@ -1438,90 +1731,117 @@ | ||
| 1438 | 1731 | * |
| 1439 | 1732 | * Values are joined with block-level markup so downstream heading, link and |
| 1440 | 1733 | * image detection keeps working on the result. |
| 1441 | 1734 | * |
| 1735 | + * One depth-first walk, so the output follows the tree's own order, which | |
| 1736 | + * is the order the builders read here render in. This used to be two | |
| 1737 | + * passes over the whole tree, one for the reconstructed headings, links and | |
| 1738 | + * images and one for the remaining text, and the output followed pass | |
| 1739 | + * order: every heading and button on the page first, every paragraph after | |
| 1740 | + * them. That order became the meta description, og:description, the schema | |
| 1741 | + * description and Pro's Markdown for AI document (#907). | |
| 1742 | + * | |
| 1442 | 1743 | * @param array $tree Decoded builder tree. |
| 1443 | 1744 | * @return string Collected HTML. |
| 1444 | 1745 | */ |
| 1445 | 1746 | private static function text_from_tree(array $tree): string { |
| 1446 | - $collected = []; | |
| 1747 | + // Each entry is [value, is_markup], in tree order. | |
| 1748 | + $entries = []; | |
| 1447 | 1749 | |
| 1448 | 1750 | // Strings already represented inside reconstructed markup, so the plain |
| 1449 | - // sweep below doesn't emit a link label or heading a second time and | |
| 1450 | - // double it in the word count. | |
| 1751 | + // text doesn't emit a link label or heading a second time and double it | |
| 1752 | + // in the word count. Applied after the walk, against the whole tree: | |
| 1753 | + // a string folded into markup anywhere is dropped everywhere, exactly | |
| 1754 | + // as it was when the markup pass ran over the whole tree first. Checking | |
| 1755 | + // it during the walk instead would let a bare copy that appears before | |
| 1756 | + // its heading through. | |
| 1451 | 1757 | $consumed = []; |
| 1452 | 1758 | |
| 1453 | - // Pass 1 — rebuild <a>, <img> and <hN> from node *shape*. This has to | |
| 1454 | - // happen per node rather than per leaf: a link's label and its | |
| 1455 | - // destination are separate sibling fields, so once the tree is | |
| 1456 | - // flattened to leaves the pairing is gone. | |
| 1457 | - $reconstruct = static function ($node) use (&$reconstruct, &$collected, &$consumed): void { | |
| 1458 | - $node = self::as_children($node); | |
| 1459 | - if (null === $node) { | |
| 1759 | + // Markup is content wherever it appears; bare strings only count when | |
| 1760 | + // their key says they are content, so slugs and class names stay out of | |
| 1761 | + // the word count. | |
| 1762 | + $leaf = static function ($value, $key) use (&$entries): void { | |
| 1763 | + if (!is_string($value) || '' === trim($value)) { | |
| 1460 | 1764 | return; |
| 1461 | 1765 | } |
| 1462 | 1766 | |
| 1463 | - $markup = self::markup_for_node($node, $consumed); | |
| 1767 | + $is_content_key = is_string($key) | |
| 1768 | + && in_array(strtolower($key), self::CONTENT_KEYS, true); | |
| 1769 | + | |
| 1770 | + if ($is_content_key || strpos($value, '<') !== false) { | |
| 1771 | + $entries[] = [$value, false]; | |
| 1772 | + } | |
| 1773 | + }; | |
| 1774 | + | |
| 1775 | + // Text only, no reconstruction. Used for a `link` / `image` / video | |
| 1776 | + // sub-object: a destination descriptor the parent has already folded | |
| 1777 | + // into its markup. Rebuilding inside it would emit the same URL a second | |
| 1778 | + // time as a bare link and turn an image's own `url` field into a | |
| 1779 | + // spurious <a>, but any copy it carries still counts. | |
| 1780 | + $sweep = static function ($node, $key) use (&$sweep, $leaf): void { | |
| 1781 | + $children = self::as_children($node); | |
| 1782 | + if (null === $children) { | |
| 1783 | + $leaf($node, $key); | |
| 1784 | + return; | |
| 1785 | + } | |
| 1786 | + | |
| 1787 | + foreach ($children as $child_key => $child) { | |
| 1788 | + $sweep($child, is_string($child_key) ? $child_key : $key); | |
| 1789 | + } | |
| 1790 | + }; | |
| 1791 | + | |
| 1792 | + // Rebuild <a>, <img> and <hN> from node *shape*, then carry on through | |
| 1793 | + // the node's own fields in order. This has to happen per node rather | |
| 1794 | + // than per leaf: a link's label and its destination are separate | |
| 1795 | + // sibling fields, so once the tree is flattened to leaves the pairing | |
| 1796 | + // is gone. | |
| 1797 | + $walk = static function ($node, $key = null) use (&$walk, $sweep, $leaf, &$entries, &$consumed): void { | |
| 1798 | + $children = self::as_children($node); | |
| 1799 | + if (null === $children) { | |
| 1800 | + $leaf($node, $key); | |
| 1801 | + return; | |
| 1802 | + } | |
| 1803 | + | |
| 1804 | + $markup = self::markup_for_node($children, $consumed); | |
| 1464 | 1805 | if ('' !== $markup) { |
| 1465 | - $collected[] = $markup; | |
| 1806 | + $entries[] = [$markup, true]; | |
| 1466 | 1807 | } |
| 1467 | 1808 | |
| 1468 | - foreach ($node as $child_key => $child) { | |
| 1469 | - // A `link` / `image` sub-object is a destination descriptor the | |
| 1470 | - // parent has already folded into its markup. Descending into it | |
| 1471 | - // would emit the same URL a second time as a bare link, and | |
| 1472 | - // would turn an image's own `url` field into a spurious <a>. | |
| 1809 | + foreach ($children as $child_key => $child) { | |
| 1810 | + $next_key = is_string($child_key) ? $child_key : $key; | |
| 1811 | + | |
| 1473 | 1812 | if (is_string($child_key) |
| 1474 | 1813 | && (in_array(strtolower($child_key), self::URL_KEYS, true) |
| 1475 | 1814 | || in_array(strtolower($child_key), self::IMAGE_KEYS, true) |
| 1476 | 1815 | || in_array(strtolower($child_key), self::VIDEO_KEYS, true)) |
| 1477 | 1816 | ) { |
| 1817 | + $sweep($child, $next_key); | |
| 1478 | 1818 | continue; |
| 1479 | 1819 | } |
| 1480 | 1820 | |
| 1481 | - $reconstruct($child); | |
| 1821 | + $walk($child, $next_key); | |
| 1482 | 1822 | } |
| 1483 | 1823 | }; |
| 1484 | - $reconstruct($tree); | |
| 1485 | 1824 | |
| 1486 | - // Pass 2 — remaining visible text. | |
| 1487 | - $walk = static function ($node, $key = null) use (&$walk, &$collected, &$consumed): void { | |
| 1488 | - $children = self::as_children($node); | |
| 1489 | - if (null !== $children) { | |
| 1490 | - foreach ($children as $child_key => $child) { | |
| 1491 | - $walk($child, is_string($child_key) ? $child_key : $key); | |
| 1492 | - } | |
| 1493 | - return; | |
| 1494 | - } | |
| 1825 | + $walk($tree); | |
| 1495 | 1826 | |
| 1496 | - if (!is_string($node) || '' === trim($node)) { | |
| 1497 | - return; | |
| 1498 | - } | |
| 1499 | - | |
| 1827 | + $collected = []; | |
| 1828 | + foreach ($entries as [$value, $is_markup]) { | |
| 1500 | 1829 | // Already inside a reconstructed tag. |
| 1501 | - if (in_array($node, $consumed, true)) { | |
| 1502 | - return; | |
| 1830 | + if (!$is_markup && in_array($value, $consumed, true)) { | |
| 1831 | + continue; | |
| 1503 | 1832 | } |
| 1504 | 1833 | |
| 1505 | - $is_content_key = is_string($key) | |
| 1506 | - && in_array(strtolower($key), self::CONTENT_KEYS, true); | |
| 1834 | + $collected[] = $value; | |
| 1835 | + } | |
| 1507 | 1836 | |
| 1508 | - // Markup is content wherever it appears; bare strings only count | |
| 1509 | - // when their key says they are content, so slugs and class names | |
| 1510 | - // stay out of the word count. | |
| 1511 | - if ($is_content_key || strpos($node, '<') !== false) { | |
| 1512 | - $collected[] = $node; | |
| 1513 | - } | |
| 1514 | - }; | |
| 1515 | - | |
| 1516 | - $walk($tree); | |
| 1517 | - | |
| 1518 | 1837 | if (empty($collected)) { |
| 1519 | 1838 | return ''; |
| 1520 | 1839 | } |
| 1521 | 1840 | |
| 1522 | 1841 | // De-duplicate: builder trees often repeat a value across responsive |
| 1523 | - // breakpoints, which would otherwise multiply the word count. | |
| 1842 | + // breakpoints, which would otherwise multiply the word count. Keeps the | |
| 1843 | + // first occurrence, so a repeat never moves a value later in the page. | |
| 1524 | 1844 | $collected = array_unique($collected); |
| 1525 | 1845 | |
| 1526 | 1846 | return implode("\n", $collected); |
| 1527 | 1847 | } |
