| @@ -40,22 +40,18 @@ | ||
| 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 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. | |
| 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. | |
| 52 | 48 | * |
| 53 | 49 | * @var string[] |
| 54 | 50 | */ |
| 55 | 51 | private const BUILDER_META_KEYS = [ |
| 56 | - '_breakdance_data', // Breakdance | |
| 57 | - '_oxygen_data', // Oxygen 6+ (Breakdance engine, Oxygen prefix) | |
| 52 | + '_breakdance_data', // Oxygen 6+ / Breakdance | |
| 53 | + '_oxygen_data', // Oxygen (earlier releases) | |
| 58 | 54 | // Oxygen classic. 4.x writes the tree as JSON to `ct_builder_json` |
| 59 | 55 | // while still keeping `ct_builder_shortcodes`. A post carrying only |
| 60 | 56 | // the JSON key used to match no key at all and fall through to an |
| 61 | 57 | // empty `post_content`, which reads as a one-word page (#776). |
| @@ -135,42 +131,8 @@ | ||
| 135 | 131 | */ |
| 136 | 132 | private const BRICKS_POST_CONTENT_ELEMENT = 'post-content'; |
| 137 | 133 | |
| 138 | 134 | /** |
| 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::$tag`, overridable by the | |
| 144 | - * active theme style's `tag`), and the walker, which only wraps text whose | |
| 145 | - * node names a tag, read it as body copy (#908). | |
| 146 | - * | |
| 147 | - * @since 2.15.0 | |
| 148 | - * @var string | |
| 149 | - */ | |
| 150 | - private const BRICKS_HEADING_ELEMENT = 'heading'; | |
| 151 | - | |
| 152 | - /** | |
| 153 | - * Tag a Bricks Heading renders when neither it nor a theme style sets one. | |
| 154 | - * | |
| 155 | - * @since 2.15.0 | |
| 156 | - * @var string | |
| 157 | - */ | |
| 158 | - private const BRICKS_HEADING_DEFAULT_TAG = 'h3'; | |
| 159 | - | |
| 160 | - /** | |
| 161 | - * Tag an Elementor Heading widget renders when `header_size` is not stored. | |
| 162 | - * | |
| 163 | - * Elementor saves `settings.toJSON({ remove: ['default'] })`, so a heading | |
| 164 | - * left on its default size has no `header_size` in `_elementor_data`, and | |
| 165 | - * that default is `h2` (#908). | |
| 166 | - * | |
| 167 | - * @since 2.15.0 | |
| 168 | - * @var string | |
| 169 | - */ | |
| 170 | - private const ELEMENTOR_HEADING_DEFAULT_TAG = 'h2'; | |
| 171 | - | |
| 172 | - /** | |
| 173 | 135 | * Resolved Bricks trees for this request, keyed by post ID. |
| 174 | 136 | * |
| 175 | 137 | * Rendering one page asks for the tree about twenty times — every |
| 176 | 138 | * description, every schema node, the FAQ guard — and resolving it is not |
| @@ -580,82 +542,12 @@ | ||
| 580 | 542 | if (!is_array($stored) || empty($stored)) { |
| 581 | 543 | return []; |
| 582 | 544 | } |
| 583 | 545 | |
| 584 | - return self::expand_bricks_components(self::bricks_render_order($stored)); | |
| 546 | + return self::expand_bricks_components($stored); | |
| 585 | 547 | } |
| 586 | 548 | |
| 587 | 549 | /** |
| 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 $ordered; | |
| 655 | - } | |
| 656 | - | |
| 657 | - /** | |
| 658 | 550 | * Resolve an arbitrary chunk of editor markup for the given post. |
| 659 | 551 | * |
| 660 | 552 | * The editor sends its live content to the scorer so an author sees their |
| 661 | 553 | * unsaved edits reflected. On a builder page that live string is the raw |
| @@ -884,9 +776,9 @@ | ||
| 884 | 776 | return ''; |
| 885 | 777 | } |
| 886 | 778 | |
| 887 | 779 | return self::strip_bricks_dynamic_tags( |
| 888 | - self::text_from_tree(self::with_bricks_heading_tags(self::without_bricks_element_labels($tree))) | |
| 780 | + self::text_from_tree(self::without_bricks_element_labels($tree)) | |
| 889 | 781 | ); |
| 890 | 782 | } |
| 891 | 783 | |
| 892 | 784 | /** |
| @@ -1084,9 +976,9 @@ | ||
| 1084 | 976 | continue; |
| 1085 | 977 | } |
| 1086 | 978 | |
| 1087 | 979 | if (isset($component['id']) && $component['id'] === $cid && !empty($component['elements'])) { |
| 1088 | - return is_array($component['elements']) ? self::bricks_render_order($component['elements']) : []; | |
| 980 | + return is_array($component['elements']) ? $component['elements'] : []; | |
| 1089 | 981 | } |
| 1090 | 982 | } |
| 1091 | 983 | |
| 1092 | 984 | return []; |
| @@ -1119,116 +1011,8 @@ | ||
| 1119 | 1011 | return $tree; |
| 1120 | 1012 | } |
| 1121 | 1013 | |
| 1122 | 1014 | /** |
| 1123 | - * Give each Bricks Heading the tag it renders with when none is stored. | |
| 1124 | - * | |
| 1125 | - * Applied on the Bricks path only. Most other builders' text nodes carry no | |
| 1126 | - * tag because they are not headings, so a generic "text without a tag is a | |
| 1127 | - * heading" rule in heading_tag_from() would turn every paragraph into one. | |
| 1128 | - * | |
| 1129 | - * A `tag` of `custom` is left alone: the element then renders its | |
| 1130 | - * `customTag`, which is not necessarily a heading. | |
| 1131 | - * | |
| 1132 | - * @since 2.15.0 | |
| 1133 | - * | |
| 1134 | - * @param array $tree Bricks content area. | |
| 1135 | - * @return array Tree with each untagged Heading's default tag filled in. | |
| 1136 | - */ | |
| 1137 | - private static function with_bricks_heading_tags(array $tree): array { | |
| 1138 | - $default = null; | |
| 1139 | - | |
| 1140 | - foreach ($tree as $index => $element) { | |
| 1141 | - if (!is_array($element) || self::BRICKS_HEADING_ELEMENT !== ($element['name'] ?? null)) { | |
| 1142 | - continue; | |
| 1143 | - } | |
| 1144 | - | |
| 1145 | - $settings = $element['settings'] ?? []; | |
| 1146 | - if (!is_array($settings)) { | |
| 1147 | - continue; | |
| 1148 | - } | |
| 1149 | - | |
| 1150 | - $tag = $settings['tag'] ?? ''; | |
| 1151 | - if (is_string($tag) && '' !== trim($tag)) { | |
| 1152 | - continue; | |
| 1153 | - } | |
| 1154 | - | |
| 1155 | - if (null === $default) { | |
| 1156 | - $default = self::bricks_default_heading_tag(); | |
| 1157 | - } | |
| 1158 | - | |
| 1159 | - $settings['tag'] = $default; | |
| 1160 | - $tree[$index]['settings'] = $settings; | |
| 1161 | - } | |
| 1162 | - | |
| 1163 | - return $tree; | |
| 1164 | - } | |
| 1165 | - | |
| 1166 | - /** | |
| 1167 | - * The tag Bricks gives a Heading that does not set one. | |
| 1168 | - * | |
| 1169 | - * The active theme style can change it. Bricks only loads theme styles for | |
| 1170 | - * a front-end render, so in admin, REST and CLI requests this is the | |
| 1171 | - * element's own default. | |
| 1172 | - * | |
| 1173 | - * @since 2.15.0 | |
| 1174 | - * | |
| 1175 | - * @return string Heading tag, h1 to h6. | |
| 1176 | - */ | |
| 1177 | - private static function bricks_default_heading_tag(): string { | |
| 1178 | - if (class_exists('\\Bricks\\Theme_Styles') | |
| 1179 | - && method_exists('\\Bricks\\Theme_Styles', 'get_setting_by_key') | |
| 1180 | - ) { | |
| 1181 | - try { | |
| 1182 | - $styled = \Bricks\Theme_Styles::get_setting_by_key(self::BRICKS_HEADING_ELEMENT, 'tag'); | |
| 1183 | - } catch (\Throwable $e) { | |
| 1184 | - $styled = null; | |
| 1185 | - } | |
| 1186 | - | |
| 1187 | - if (is_string($styled) && preg_match('/^h[1-6]$/i', trim($styled))) { | |
| 1188 | - return strtolower(trim($styled)); | |
| 1189 | - } | |
| 1190 | - } | |
| 1191 | - | |
| 1192 | - return self::BRICKS_HEADING_DEFAULT_TAG; | |
| 1193 | - } | |
| 1194 | - | |
| 1195 | - /** | |
| 1196 | - * Give each Elementor Heading widget its default `header_size` if unstored. | |
| 1197 | - * | |
| 1198 | - * @since 2.15.0 | |
| 1199 | - * | |
| 1200 | - * @param array $elements Decoded `_elementor_data`. | |
| 1201 | - * @return array The same tree, with untagged Heading widgets tagged. | |
| 1202 | - */ | |
| 1203 | - private static function with_elementor_heading_tags(array $elements): array { | |
| 1204 | - foreach ($elements as $index => $element) { | |
| 1205 | - if (!is_array($element)) { | |
| 1206 | - continue; | |
| 1207 | - } | |
| 1208 | - | |
| 1209 | - if ('heading' === ($element['widgetType'] ?? null)) { | |
| 1210 | - $settings = $element['settings'] ?? []; | |
| 1211 | - if (is_array($settings)) { | |
| 1212 | - $size = $settings['header_size'] ?? ''; | |
| 1213 | - if (!is_string($size) || '' === trim($size)) { | |
| 1214 | - $settings['header_size'] = self::ELEMENTOR_HEADING_DEFAULT_TAG; | |
| 1215 | - $element['settings'] = $settings; | |
| 1216 | - } | |
| 1217 | - } | |
| 1218 | - } | |
| 1219 | - | |
| 1220 | - if (!empty($element['elements']) && is_array($element['elements'])) { | |
| 1221 | - $element['elements'] = self::with_elementor_heading_tags($element['elements']); | |
| 1222 | - } | |
| 1223 | - | |
| 1224 | - $elements[$index] = $element; | |
| 1225 | - } | |
| 1226 | - | |
| 1227 | - return $elements; | |
| 1228 | - } | |
| 1229 | - | |
| 1230 | - /** | |
| 1231 | 1015 | * Remove Bricks dynamic-data tags from extracted text. |
| 1232 | 1016 | * |
| 1233 | 1017 | * Bricks stores `{post_title}`, `{post_meta:price}`, `{echo:my_fn}` and the |
| 1234 | 1018 | * like verbatim and resolves them when it renders. Extraction reads the |
| @@ -1336,15 +1120,12 @@ | ||
| 1336 | 1120 | $stored = get_post_meta($post_id, $key, true); |
| 1337 | 1121 | |
| 1338 | 1122 | if (is_string($stored) && '' !== trim($stored)) { |
| 1339 | 1123 | $decoded = json_decode($stored, true); |
| 1340 | - if ('_elementor_data' === $key && is_array($decoded)) { | |
| 1341 | - $decoded = self::with_elementor_heading_tags($decoded); | |
| 1342 | - } | |
| 1343 | 1124 | |
| 1344 | 1125 | // JSON node tree (Breakdance/Oxygen 6, Elementor). |
| 1345 | 1126 | if (is_array($decoded)) { |
| 1346 | - $text = self::text_from_tree(self::unwrap_tree_envelope($decoded)); | |
| 1127 | + $text = self::text_from_tree($decoded); | |
| 1347 | 1128 | if (!self::is_blank($text)) { |
| 1348 | 1129 | return $text; |
| 1349 | 1130 | } |
| 1350 | 1131 | continue; |
| @@ -1356,9 +1137,8 @@ | ||
| 1356 | 1137 | // Some builders store an already-decoded tree — an array for most, |
| 1357 | 1138 | // an array of objects for Beaver Builder (#449). |
| 1358 | 1139 | $tree = self::as_children($stored); |
| 1359 | 1140 | if (null !== $tree) { |
| 1360 | - $tree = self::unwrap_tree_envelope($tree); | |
| 1361 | 1141 | $text = self::text_from_tree($tree); |
| 1362 | 1142 | if (!self::is_blank($text)) { |
| 1363 | 1143 | return $text; |
| 1364 | 1144 | } |
| @@ -1368,48 +1148,8 @@ | ||
| 1368 | 1148 | return ''; |
| 1369 | 1149 | } |
| 1370 | 1150 | |
| 1371 | 1151 | /** |
| 1372 | - * The node tree inside a Breakdance / Oxygen 6 storage envelope. | |
| 1373 | - * | |
| 1374 | - * Neither builder stores its tree directly. The meta value is | |
| 1375 | - * `{"tree_json_string": "<the tree, JSON-encoded again>"}`, so one | |
| 1376 | - * json_decode() yields the envelope, not the tree. Walked as a tree, the | |
| 1377 | - * envelope is a single string leaf: kept whole as "content" when any | |
| 1378 | - * element held rich text (the encoded JSON then reached scoring, the | |
| 1379 | - * get-post-content ability and Markdown for AI), dropped when none did, | |
| 1380 | - * leaving the page empty (#905). | |
| 1381 | - * | |
| 1382 | - * An envelope whose inner string does not decode returns an empty tree, | |
| 1383 | - * never the string: handing the raw JSON back to the walker would bring | |
| 1384 | - * the JSON-as-content failure back on corrupt data. Anything that is not | |
| 1385 | - * an envelope is returned unchanged, so a bare tree still resolves. | |
| 1386 | - * | |
| 1387 | - * @since 2.15.0 | |
| 1388 | - * | |
| 1389 | - * @param array $decoded Decoded meta value. | |
| 1390 | - * @return array The node tree. | |
| 1391 | - */ | |
| 1392 | - private static function unwrap_tree_envelope(array $decoded): array { | |
| 1393 | - if (!array_key_exists('tree_json_string', $decoded)) { | |
| 1394 | - return $decoded; | |
| 1395 | - } | |
| 1396 | - | |
| 1397 | - $inner = is_string($decoded['tree_json_string']) | |
| 1398 | - ? json_decode($decoded['tree_json_string'], true) | |
| 1399 | - : $decoded['tree_json_string']; | |
| 1400 | - | |
| 1401 | - if (is_array($inner)) { | |
| 1402 | - return $inner; | |
| 1403 | - } | |
| 1404 | - | |
| 1405 | - // Re-serialised envelopes can carry the tree as an object. | |
| 1406 | - $inner = self::as_children($inner); | |
| 1407 | - | |
| 1408 | - return null !== $inner ? $inner : []; | |
| 1409 | - } | |
| 1410 | - | |
| 1411 | - /** | |
| 1412 | 1152 | * Whether a meta key is one of Oxygen classic's storage keys. |
| 1413 | 1153 | * |
| 1414 | 1154 | * @since 2.10.0 |
| 1415 | 1155 | * |
| @@ -1698,117 +1438,90 @@ | ||
| 1698 | 1438 | * |
| 1699 | 1439 | * Values are joined with block-level markup so downstream heading, link and |
| 1700 | 1440 | * image detection keeps working on the result. |
| 1701 | 1441 | * |
| 1702 | - * One depth-first walk, so the output follows the tree's own order, which | |
| 1703 | - * is the order the builders read here render in. This used to be two | |
| 1704 | - * passes over the whole tree, one for the reconstructed headings, links and | |
| 1705 | - * images and one for the remaining text, and the output followed pass | |
| 1706 | - * order: every heading and button on the page first, every paragraph after | |
| 1707 | - * them. That order became the meta description, og:description, the schema | |
| 1708 | - * description and Pro's Markdown for AI document (#907). | |
| 1709 | - * | |
| 1710 | 1442 | * @param array $tree Decoded builder tree. |
| 1711 | 1443 | * @return string Collected HTML. |
| 1712 | 1444 | */ |
| 1713 | 1445 | private static function text_from_tree(array $tree): string { |
| 1714 | - // Each entry is [value, is_markup], in tree order. | |
| 1715 | - $entries = []; | |
| 1446 | + $collected = []; | |
| 1716 | 1447 | |
| 1717 | 1448 | // Strings already represented inside reconstructed markup, so the plain |
| 1718 | - // text doesn't emit a link label or heading a second time and double it | |
| 1719 | - // in the word count. Applied after the walk, against the whole tree: | |
| 1720 | - // a string folded into markup anywhere is dropped everywhere, exactly | |
| 1721 | - // as it was when the markup pass ran over the whole tree first. Checking | |
| 1722 | - // it during the walk instead would let a bare copy that appears before | |
| 1723 | - // its heading through. | |
| 1449 | + // sweep below doesn't emit a link label or heading a second time and | |
| 1450 | + // double it in the word count. | |
| 1724 | 1451 | $consumed = []; |
| 1725 | 1452 | |
| 1726 | - // Markup is content wherever it appears; bare strings only count when | |
| 1727 | - // their key says they are content, so slugs and class names stay out of | |
| 1728 | - // the word count. | |
| 1729 | - $leaf = static function ($value, $key) use (&$entries): void { | |
| 1730 | - if (!is_string($value) || '' === trim($value)) { | |
| 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) { | |
| 1731 | 1460 | return; |
| 1732 | 1461 | } |
| 1733 | 1462 | |
| 1734 | - $is_content_key = is_string($key) | |
| 1735 | - && in_array(strtolower($key), self::CONTENT_KEYS, true); | |
| 1463 | + $markup = self::markup_for_node($node, $consumed); | |
| 1464 | + if ('' !== $markup) { | |
| 1465 | + $collected[] = $markup; | |
| 1466 | + } | |
| 1736 | 1467 | |
| 1737 | - if ($is_content_key || strpos($value, '<') !== false) { | |
| 1738 | - $entries[] = [$value, false]; | |
| 1468 | + foreach ($node as $child_key => $child) { | |
| 1469 | + // A `link` / `image` sub-object is a destination descriptor the | |
| 1470 | + // parent has already folded into its markup. Descending into it | |
| 1471 | + // would emit the same URL a second time as a bare link, and | |
| 1472 | + // would turn an image's own `url` field into a spurious <a>. | |
| 1473 | + if (is_string($child_key) | |
| 1474 | + && (in_array(strtolower($child_key), self::URL_KEYS, true) | |
| 1475 | + || in_array(strtolower($child_key), self::IMAGE_KEYS, true) | |
| 1476 | + || in_array(strtolower($child_key), self::VIDEO_KEYS, true)) | |
| 1477 | + ) { | |
| 1478 | + continue; | |
| 1479 | + } | |
| 1480 | + | |
| 1481 | + $reconstruct($child); | |
| 1739 | 1482 | } |
| 1740 | 1483 | }; |
| 1484 | + $reconstruct($tree); | |
| 1741 | 1485 | |
| 1742 | - // Text only, no reconstruction. Used for a `link` / `image` / video | |
| 1743 | - // sub-object: a destination descriptor the parent has already folded | |
| 1744 | - // into its markup. Rebuilding inside it would emit the same URL a second | |
| 1745 | - // time as a bare link and turn an image's own `url` field into a | |
| 1746 | - // spurious <a>, but any copy it carries still counts. | |
| 1747 | - $sweep = static function ($node, $key) use (&$sweep, $leaf): void { | |
| 1486 | + // Pass 2 — remaining visible text. | |
| 1487 | + $walk = static function ($node, $key = null) use (&$walk, &$collected, &$consumed): void { | |
| 1748 | 1488 | $children = self::as_children($node); |
| 1749 | - if (null === $children) { | |
| 1750 | - $leaf($node, $key); | |
| 1489 | + if (null !== $children) { | |
| 1490 | + foreach ($children as $child_key => $child) { | |
| 1491 | + $walk($child, is_string($child_key) ? $child_key : $key); | |
| 1492 | + } | |
| 1751 | 1493 | return; |
| 1752 | 1494 | } |
| 1753 | 1495 | |
| 1754 | - foreach ($children as $child_key => $child) { | |
| 1755 | - $sweep($child, is_string($child_key) ? $child_key : $key); | |
| 1496 | + if (!is_string($node) || '' === trim($node)) { | |
| 1497 | + return; | |
| 1756 | 1498 | } |
| 1757 | - }; | |
| 1758 | 1499 | |
| 1759 | - // Rebuild <a>, <img> and <hN> from node *shape*, then carry on through | |
| 1760 | - // the node's own fields in order. This has to happen per node rather | |
| 1761 | - // than per leaf: a link's label and its destination are separate | |
| 1762 | - // sibling fields, so once the tree is flattened to leaves the pairing | |
| 1763 | - // is gone. | |
| 1764 | - $walk = static function ($node, $key = null) use (&$walk, $sweep, $leaf, &$entries, &$consumed): void { | |
| 1765 | - $children = self::as_children($node); | |
| 1766 | - if (null === $children) { | |
| 1767 | - $leaf($node, $key); | |
| 1500 | + // Already inside a reconstructed tag. | |
| 1501 | + if (in_array($node, $consumed, true)) { | |
| 1768 | 1502 | return; |
| 1769 | 1503 | } |
| 1770 | 1504 | |
| 1771 | - $markup = self::markup_for_node($children, $consumed); | |
| 1772 | - if ('' !== $markup) { | |
| 1773 | - $entries[] = [$markup, true]; | |
| 1774 | - } | |
| 1505 | + $is_content_key = is_string($key) | |
| 1506 | + && in_array(strtolower($key), self::CONTENT_KEYS, true); | |
| 1775 | 1507 | |
| 1776 | - foreach ($children as $child_key => $child) { | |
| 1777 | - $next_key = is_string($child_key) ? $child_key : $key; | |
| 1778 | - | |
| 1779 | - if (is_string($child_key) | |
| 1780 | - && (in_array(strtolower($child_key), self::URL_KEYS, true) | |
| 1781 | - || in_array(strtolower($child_key), self::IMAGE_KEYS, true) | |
| 1782 | - || in_array(strtolower($child_key), self::VIDEO_KEYS, true)) | |
| 1783 | - ) { | |
| 1784 | - $sweep($child, $next_key); | |
| 1785 | - continue; | |
| 1786 | - } | |
| 1787 | - | |
| 1788 | - $walk($child, $next_key); | |
| 1508 | + // Markup is content wherever it appears; bare strings only count | |
| 1509 | + // when their key says they are content, so slugs and class names | |
| 1510 | + // stay out of the word count. | |
| 1511 | + if ($is_content_key || strpos($node, '<') !== false) { | |
| 1512 | + $collected[] = $node; | |
| 1789 | 1513 | } |
| 1790 | 1514 | }; |
| 1791 | 1515 | |
| 1792 | 1516 | $walk($tree); |
| 1793 | 1517 | |
| 1794 | - $collected = []; | |
| 1795 | - foreach ($entries as [$value, $is_markup]) { | |
| 1796 | - // Already inside a reconstructed tag. | |
| 1797 | - if (!$is_markup && in_array($value, $consumed, true)) { | |
| 1798 | - continue; | |
| 1799 | - } | |
| 1800 | - | |
| 1801 | - $collected[] = $value; | |
| 1802 | - } | |
| 1803 | - | |
| 1804 | 1518 | if (empty($collected)) { |
| 1805 | 1519 | return ''; |
| 1806 | 1520 | } |
| 1807 | 1521 | |
| 1808 | 1522 | // De-duplicate: builder trees often repeat a value across responsive |
| 1809 | - // breakpoints, which would otherwise multiply the word count. Keeps the | |
| 1810 | - // first occurrence, so a repeat never moves a value later in the page. | |
| 1523 | + // breakpoints, which would otherwise multiply the word count. | |
| 1811 | 1524 | $collected = array_unique($collected); |
| 1812 | 1525 | |
| 1813 | 1526 | return implode("\n", $collected); |
| 1814 | 1527 | } |
