← All changes | includes/seo/class-sitemap-generator.php +116 -46 2.14.0 → 2.14.2 View file →

		@@ -409,11 +409,105 @@
409	409	add_action($hook, static function (int $term_id, int $tt_id, string $taxonomy): void {
410	410	self::listener()->handle_taxonomy_change($term_id, $tt_id, $taxonomy);
411	411	}, 20, 3);
412	412	}
	413	+
	414	+ // The sitemap leaves out what the robots tag noindexes and what
	415	+ // ThinkRank redirects (#911), so a change to either decides what the
	416	+ // published files should list. Neither is a post or term save, and
	417	+ // without these the files kept the old answer until unrelated content
	418	+ // changed.
	419	+ foreach (self::ROBOTS_SETTINGS_OPTIONS as $option) {
	420	+ add_action('update_option_' . $option, static function ($old_value, $value): void {
	421	+ self::handle_robots_settings_change($old_value, $value);
	422	+ }, 20, 2);
	423	+ add_action('add_option_' . $option, static function ($name, $value): void {
	424	+ self::handle_robots_settings_change([], $value);
	425	+ }, 20, 2);
	426	+ }
	427	+
	428	+ add_action('thinkrank_object_redirect_saved', static function (): void {
	429	+ self::listener()->schedule_regeneration();
	430	+ }, 20, 0);
413	431	}
414	432
415	433	/**
	434	+ * Options holding robots directives the sitemap's item filter reads.
	435	+ *
	436	+ * @since 2.15.0
	437	+ * @var string[]
	438	+ */
	439	+ private const ROBOTS_SETTINGS_OPTIONS = [
	440	+ 'thinkrank_global_seo_settings',
	441	+ 'thinkrank_global_robot_meta_settings',
	442	+ ];
	443	+
	444	+ /**
	445	+ * Queue a rebuild when a saved robots setting changes a noindex decision.
	446	+ *
	447	+ * Both options carry far more than robots directives (titles, schema,
	448	+ * feature switches), so only a change to a noindex outcome rebuilds.
	449	+ *
	450	+ * @since 2.15.0
	451	+ *
	452	+ * @param mixed $old_value Previous option value.
	453	+ * @param mixed $value New option value.
	454	+ * @return void
	455	+ */
	456	+ public static function handle_robots_settings_change($old_value, $value): void {
	457	+ if (self::noindex_fingerprint($old_value) === self::noindex_fingerprint($value)) {
	458	+ return;
	459	+ }
	460	+
	461	+ // The matrix memoises the option for the request, and a rebuild that
	462	+ // runs in this request must read the value just saved.
	463	+ Content_Type_Settings::flush_cache();
	464	+
	465	+ self::listener()->schedule_regeneration();
	466	+ }
	467	+
	468	+ /**
	469	+ * The noindex decisions a robots option makes, keyed by what they apply to.
	470	+ *
	471	+ * Reads both shapes: the flat site-wide directives, and the per-entity
	472	+ * rows, where a row's directives apply only while its robots switch is on.
	473	+ * A row that is on but stores no `noindex` key changes nothing, matching
	474	+ * the array_merge() the robots tag does.
	475	+ *
	476	+ * @since 2.15.0
	477	+ *
	478	+ * @param mixed $value Option value.
	479	+ * @return array<string, bool|null>
	480	+ */
	481	+ private static function noindex_fingerprint($value): array {
	482	+ if (!is_array($value)) {
	483	+ return [];
	484	+ }
	485	+
	486	+ $decisions = [];
	487	+
	488	+ if (array_key_exists('noindex', $value) && !is_array($value['noindex'])) {
	489	+ $decisions['*'] = !empty($value['noindex']);
	490	+ }
	491	+
	492	+ foreach ($value as $key => $row) {
	493	+ if (!is_array($row)) {
	494	+ continue;
	495	+ }
	496	+
	497	+ $robots = $row['robots_meta'] ?? null;
	498	+
	499	+ $decisions[(string) $key] = !empty($row['robots_meta_enabled']) && is_array($robots) && array_key_exists('noindex', $robots)
	500	+ ? !empty($robots['noindex'])
	501	+ : null;
	502	+ }
	503	+
	504	+ ksort($decisions);
	505	+
	506	+ return $decisions;
	507	+ }
	508	+
	509	+ /**
416	510	* The generator the content-change listeners share.
417	511	*
418	512	* @since 2.10.1
419	513	* @return self
		@@ -806,9 +900,12 @@
806	900
807	901	// Add image entries if provided
808	902	foreach ($images as $image) {
809	903	$xml .= " <image:image>\n";
810		- $xml .= " <image:loc>" . esc_url(Url_Scheme::apply((string) $image['url'])) . "</image:loc>\n";
	904	+ // The sitemap protocol wants an escaped URL, and WordPress hands back
	905	+ // attachment URLs with non-ASCII filenames unencoded (esc_url() does
	906	+ // not encode them either), so write the percent-encoded form (#924).
	907	+ $xml .= " <image:loc>" . esc_url(\ThinkRank\Core\Url_Validator::to_ascii(Url_Scheme::apply((string) $image['url']))) . "</image:loc>\n";
811	908
812	909	if (!empty($image['title'])) {
813	910	$xml .= " <image:title>" . esc_html($image['title']) . "</image:title>\n";
814	911	}
		@@ -1063,12 +1160,12 @@
1063	1160	continue;
1064	1161	}
1065	1162
1066	1163	foreach ($terms as $term) {
1067		- // A term the user marked noindex must not be advertised in the
1068		- // sitemap: the robots tag now honours term meta, so listing it
1069		- // here would have the sitemap contradict the page's own tag.
1070		- if ($this->term_is_noindexed((int) $term->term_id)) {
	1164	+ // A term whose archive says noindex (its own override or its
	1165	+ // taxonomy's), or that redirects, must not be advertised: the
	1166	+ // sitemap would contradict the page's own signal (#911).
	1167	+ if (!Indexability::is_indexable_term($term)) {
1071	1168	continue;
1072	1169	}
1073	1170
1074	1171	$url = get_term_link($term);
		@@ -1256,9 +1353,9 @@
1256	1353	if (preg_match('/src=["\']([^"\']+)["\']/', $img_tag, $src_match)) {
1257	1354	$image_url = $src_match[1];
1258	1355
1259	1356	// Skip if not a valid URL or external image
1260		- if (!filter_var($image_url, FILTER_VALIDATE_URL)) {
	1357	+ if (!\ThinkRank\Core\Url_Validator::is_valid($image_url)) {
1261	1358	continue;
1262	1359	}
1263	1360
1264	1361	// Extract title and alt attributes
		@@ -1373,14 +1470,15 @@
1373	1470	if (is_wp_error($all_terms)) {
1374	1471	return [];
1375	1472	}
1376	1473
1377		- // Drop terms the user marked noindex. This path feeds the single general
1378		- // sitemap while collect_taxonomy_entries_iter() feeds the segmented ones,
1379		- // so both need the filter or the two disagree about the same term.
	1474	+ // Drop terms that are not indexable destinations. This path feeds the
	1475	+ // single general sitemap while collect_taxonomy_entries_iter() feeds
	1476	+ // the segmented ones, so both need the filter or the two disagree about
	1477	+ // the same term.
1380	1478	$all_terms = array_values(array_filter(
1381	1479	$all_terms,
1382		- fn($term) => !$this->term_is_noindexed((int) $term->term_id)
	1480	+ static fn($term) => $term instanceof \WP_Term && Indexability::is_indexable_term($term)
1383	1481	));
1384	1482
1385	1483	// Group terms by taxonomy
1386	1484	return $this->group_terms_by_taxonomy($all_terms);
		@@ -1562,17 +1660,16 @@
1562	1660	if (!in_array($post->post_status, ['publish', 'private'], true)) {
1563	1661	return false;
1564	1662	}
1565	1663
1566		- // Check if post overrides robots and sets noindex.
1567		- if ((bool) get_post_meta($post->ID, '_thinkrank_robots_meta_enabled', true)) {
1568		- $raw = get_post_meta($post->ID, '_thinkrank_robots_meta', true);
1569		- if (is_string($raw) && $raw !== '') {
1570		- $robots = json_decode($raw, true);
1571		- if (is_array($robots) && !empty($robots['noindex'])) {
1572		- return false;
1573		- }
1574		- }
	1664	+ // A URL whose page says noindex, or that ThinkRank redirects, is not a
	1665	+ // destination. Only the per-post noindex used to be read here, so a
	1666	+ // post type set to No-index still had every item listed, and redirected
	1667	+ // posts were submitted as "Page with redirect" (#911). The password
	1668	+ // check stays with the setting above: listing protected posts is a
	1669	+ // choice this sitemap has always offered.
	1670	+ if (Indexability::is_post_noindexed($post) || Indexability::is_post_redirected($post)) {
	1671	+ return false;
1575	1672	}
1576	1673
1577	1674	return true;
1578	1675	}
		@@ -1608,35 +1705,8 @@
1608	1705
1609	1706	$this->woocommerce_excluded_page_ids = $ids;
1610	1707
1611	1708	return $ids;
1612		- }
1613		-
1614		- /**
1615		- * Whether a term carries an explicit noindex override.
1616		- *
1617		- * Mirrors the post-side check in should_include_post(); terms store the same
1618		- * `_thinkrank_robots_meta_enabled` / `_thinkrank_robots_meta` keys, written
1619		- * by the update-term-seo ability and by the SEO importer.
1620		- *
1621		- * @since 1.31.0
1622		- *
1623		- * @param int $term_id Term to test.
1624		- * @return bool True when the term is marked noindex.
1625		- */
1626		- private function term_is_noindexed(int $term_id): bool {
1627		- if (!(bool) get_term_meta($term_id, '_thinkrank_robots_meta_enabled', true)) {
1628		- return false;
1629		- }
1630		-
1631		- $raw = get_term_meta($term_id, '_thinkrank_robots_meta', true);
1632		- if (!is_string($raw) || $raw === '') {
1633		- return false;
1634		- }
1635		-
1636		- $robots = json_decode($raw, true);
1637		-
1638		- return is_array($robots) && !empty($robots['noindex']);
1639	1709	}
1640	1710
1641	1711	/**
1642	1712	* Count total URLs in sitemap