← All changes | includes/seo/class-seo-analyzer.php +173 -16 2.14.1 → 2.14.2 View file →

		@@ -67,8 +67,37 @@
67	67	'toggle', // Elementor
68	68	'faq', // Widely used in third-party add-on element names
69	69	];
70	70
	71	+ /**
	72	+ * Builder meta keys that never describe the published page.
	73	+ *
	74	+ * `_fl_builder_draft` holds Beaver Builder changes that were never
	75	+ * published, so a FAQ that exists only there is not on the page (#945).
	76	+ *
	77	+ * @since 2.14.2
	78	+ * @var string[]
	79	+ */
	80	+ private const UNPUBLISHED_BUILDER_META_KEYS = ['_fl_builder_draft'];
	81	+
	82	+ /**
	83	+ * Builder layouts that only count while their builder renders the post.
	84	+ *
	85	+ * Elementor and Beaver Builder both keep their stored layout after the
	86	+ * author switches the page back to the block editor, so the layout alone
	87	+ * does not mean the builder renders it. Each key maps to the builder's own
	88	+ * flag and the value that means "on": Elementor's `_elementor_edit_mode`
	89	+ * is `builder`, Beaver's `_fl_builder_enabled` is any non-empty value
	90	+ * (null here). The same tests as FAQ_Content::builder() (#945).
	91	+ *
	92	+ * @since 2.14.2
	93	+ * @var array<string,array{0:string,1:string|null}>
	94	+ */
	95	+ private const FLAGGED_BUILDER_LAYOUTS = [
	96	+ '_elementor_data' => ['_elementor_edit_mode', 'builder'],
	97	+ '_fl_builder_data' => ['_fl_builder_enabled', null],
	98	+ ];
	99	+
71	100	// Check result statuses.
72	101	public const PASSED = 'passed';
73	102	public const WARNING = 'warning';
74	103	public const FAILED = 'failed';
		@@ -1692,12 +1721,9 @@
1692	1721	}
1693	1722
1694	1723	$collecting = false;
1695	1724
1696		- // `/` is the canonical full block; `/*` is the same instruction
1697		- // written for a wildcard-aware crawler, and every answer engine on
1698		- // the list is one.
1699		- if ('disallow' === $field && ('/' === $value || '/*' === $value)) {
	1725	+ if ('disallow' === $field && self::disallow_blocks_everything($value)) {
1700	1726	foreach ($current as $agent) {
1701	1727	$groups[$agent] = true;
1702	1728	}
1703	1729	}
		@@ -1706,8 +1732,46 @@
1706	1732	return $groups;
1707	1733	}
1708	1734
1709	1735	/**
	1736	+ * Whether a `Disallow:` value matches every URL on the site.
	1737	+ *
	1738	+ * `/` is the canonical full block; `/*` is the same instruction written
	1739	+ * for a wildcard-aware crawler, and every answer engine on the list is
	1740	+ * one. A value is a path pattern in which `*` matches any run of
	1741	+ * characters, so a bare `*` (and `**`, `/**`) matches every path too, and
	1742	+ * Google's parser treats it that way. Recognising only `/` and `/*`
	1743	+ * reported a site that blocked everyone with `Disallow: *` as reachable
	1744	+ * by every answer engine (#890).
	1745	+ *
	1746	+ * A trailing `$` anchors the pattern to the end of the URL. After a `*`
	1747	+ * it changes nothing, since "any run of characters, then the end" still
	1748	+ * matches every path, so `*$` and `/*$` are full blocks as well. Without
	1749	+ * a `*` it narrows the match: `/$` is the home page only, and `/*.pdf$`
	1750	+ * is PDFs only, so neither counts.
	1751	+ *
	1752	+ * An empty value means "allow everything" and must never count as a
	1753	+ * block, so it is rejected here rather than left to the caller.
	1754	+ *
	1755	+ * `Allow:` lines are deliberately not weighed against this: a full
	1756	+ * robots.txt evaluator with longest-match precedence is a separate change.
	1757	+ *
	1758	+ * @since 2.14.2
	1759	+ * @param string $value Trimmed `Disallow:` value, comment already removed.
	1760	+ * @return bool
	1761	+ */
	1762	+ private static function disallow_blocks_everything(string $value): bool {
	1763	+ if ('' === $value) {
	1764	+ return false;
	1765	+ }
	1766	+
	1767	+ // Optional leading slash, then either nothing (`/`) or a run of `*`
	1768	+ // with an optional end anchor (`*`, `/*`, `/*$`, `**$`). A `$` with
	1769	+ // no `*` before it never matches here, so `/$` stays a partial block.
	1770	+ return 1 === preg_match('#^/?(?:\*+\$?)?$#', $value);
	1771	+ }
	1772	+
	1773	+ /**
1710	1774	* /llms.txt should be published — it is the one file whose entire purpose
1711	1775	* is telling an AI assistant what this site is and what to read.
1712	1776	*
1713	1777	* @since 2.5.0
		@@ -1862,10 +1926,10 @@
1862	1926	}
1863	1927
1864	1928	// Builder trees: ThinkRank's own elements, and the builders' generic
1865	1929	// accordion/toggle/FAQ elements, which are what a non-ThinkRank FAQ
1866		- // is actually built from.
1867		- $meta_keys = self::builder_meta_keys();
	1930	+ // is actually built from. Only layouts the visitor is served count.
	1931	+ $meta_keys = self::rendered_builder_meta_keys();
1868	1932
1869	1933	if (empty($meta_keys)) {
1870	1934	return false;
1871	1935	}
		@@ -1883,8 +1947,10 @@
1883	1947	$marker_clauses[] = 'pm.meta_value LIKE %s';
1884	1948	$marker_values[] = '%' . $wpdb->esc_like($marker) . '%';
1885	1949	}
1886	1950
	1951	+ [$gate_sql, $gate_values] = self::builder_layout_renders_sql('pm');
	1952	+
1887	1953	// Same exemption: both placeholder runs are sized from fixed lists —
1888	1954	// the builder meta keys and the marker list — and every value is
1889	1955	// prepared.
1890	1956	// phpcs:disable WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
		@@ -1893,11 +1959,12 @@
1893	1959	"SELECT 1 FROM {$wpdb->postmeta} pm
1894	1960	INNER JOIN {$wpdb->posts} p ON p.ID = pm.post_id
1895	1961	WHERE p.post_status = 'publish'
1896	1962	AND pm.meta_key IN ({$key_placeholders})
1897		- AND (" . implode(' OR ', $marker_clauses) . ')
1898		- LIMIT 1',
1899		- ...array_merge($meta_keys, $marker_values)
	1963	+ AND (" . implode(' OR ', $marker_clauses) . ")
	1964	+ {$gate_sql}
	1965	+ LIMIT 1",
	1966	+ ...array_merge($meta_keys, $marker_values, $gate_values)
1900	1967	)
1901	1968	);
1902	1969	// phpcs:enable
1903	1970
		@@ -2014,9 +2081,9 @@
2014	2081	$content_clauses[] = 'p.post_content LIKE %s';
2015	2082	$content_values[] = '%' . $wpdb->esc_like($marker) . '%';
2016	2083	}
2017	2084
2018		- $meta_keys = self::builder_meta_keys();
	2085	+ $meta_keys = self::rendered_builder_meta_keys();
2019	2086	$meta_sql = '';
2020	2087	$values = $content_values;
2021	2088
2022	2089	if (!empty($meta_keys)) {
		@@ -2026,16 +2093,19 @@
2026	2093	$meta_clauses[] = 'pm.meta_value LIKE %s';
2027	2094	$meta_values[] = '%' . $wpdb->esc_like($marker) . '%';
2028	2095	}
2029	2096
	2097	+ [$gate_sql, $gate_values] = self::builder_layout_renders_sql('pm');
	2098	+
2030	2099	$key_placeholders = implode(', ', array_fill(0, count($meta_keys), '%s'));
2031	2100	$meta_sql = " OR EXISTS (
2032	2101	SELECT 1 FROM {$wpdb->postmeta} pm
2033	2102	WHERE pm.post_id = p.ID
2034	2103	AND pm.meta_key IN ({$key_placeholders})
2035		- AND (" . implode(' OR ', $meta_clauses) . ')
2036		- )';
2037		- $values = array_merge($content_values, $meta_keys, $meta_values);
	2104	+ AND (" . implode(' OR ', $meta_clauses) . ")
	2105	+ {$gate_sql}
	2106	+ )";
	2107	+ $values = array_merge($content_values, $meta_keys, $meta_values, $gate_values);
2038	2108	}
2039	2109
2040	2110	$values[] = self::CONTENT_SAMPLE_SIZE;
2041	2111
		@@ -2089,9 +2159,13 @@
2089	2159	return false;
2090	2160	}
2091	2161
2092	2162	// Elementor stores its tree as JSON, so the widget name appears verbatim.
2093		- $elementor = get_post_meta($post_id, '_elementor_data', true);
	2163	+ // Read only while Elementor renders the post: the tree survives
	2164	+ // switching the page back to the block editor (#945).
	2165	+ $elementor = self::builder_layout_renders($post_id, '_elementor_data')
	2166	+ ? get_post_meta($post_id, '_elementor_data', true)
	2167	+ : '';
2094	2168	if (is_string($elementor)
2095	2169	&& (false !== strpos($elementor, '"' . self::ANSWER_FAQ_NAME . '"')
2096	2170	|| false !== strpos($elementor, '"' . self::ANSWER_HOWTO_NAME . '"'))) {
2097	2171	return true;
		@@ -2103,9 +2177,12 @@
2103	2177	return true;
2104	2178	}
2105	2179
2106	2180	// Beaver Builder keeps its layout in postmeta as a map of node objects.
2107		- $layout = get_post_meta($post_id, '_fl_builder_data', true);
	2181	+ // Read only while Beaver renders the post, for the same reason (#945).
	2182	+ $layout = self::builder_layout_renders($post_id, '_fl_builder_data')
	2183	+ ? get_post_meta($post_id, '_fl_builder_data', true)
	2184	+ : [];
2108	2185	if (is_array($layout)) {
2109	2186	foreach ($layout as $node) {
2110	2187	$settings = is_object($node) ? ($node->settings ?? null) : ($node['settings'] ?? null);
2111	2188	$settings = is_object($settings) ? get_object_vars($settings) : $settings;
		@@ -2122,9 +2199,9 @@
2122	2199	// while the page was publishing exactly that (#686). Builder_Content
2123	2200	// already knows every key involved — Oxygen 6 is Breakdance under the
2124	2201	// hood, and older releases used two other keys — so ask it rather than
2125	2202	// keeping a second list that can drift.
2126		- foreach (self::builder_meta_keys() as $meta_key) {
	2203	+ foreach (self::rendered_builder_meta_keys() as $meta_key) {
2127	2204	if ('_fl_builder_data' === $meta_key || '_elementor_data' === $meta_key) {
2128	2205	continue; // Handled above, in their own storage shapes.
2129	2206	}
2130	2207
		@@ -2157,8 +2234,88 @@
2157	2234	require_once $file;
2158	2235	}
2159	2236
2160	2237	return Builder_Content::builder_meta_keys();
	2238	+ }
	2239	+
	2240	+ /**
	2241	+ * Builder meta keys that can hold the published page.
	2242	+ *
	2243	+ * {@see builder_meta_keys()} without the unpublished draft keys, which
	2244	+ * Builder_Content lists because the word-count index watches them, not
	2245	+ * because they are ever served (#945).
	2246	+ *
	2247	+ * @since 2.14.2
	2248	+ * @return string[]
	2249	+ */
	2250	+ private static function rendered_builder_meta_keys(): array {
	2251	+ return array_values(array_diff(self::builder_meta_keys(), self::UNPUBLISHED_BUILDER_META_KEYS));
	2252	+ }
	2253	+
	2254	+ /**
	2255	+ * Whether a builder layout key holds what the visitor is served.
	2256	+ *
	2257	+ * The answer-content readers' guard for #945. Builder_Content gains the
	2258	+ * same guards for its resolver in #910, and once that lands this can ask
	2259	+ * it instead; until then it reads the builders' own flags, exactly as
	2260	+ * FAQ_Content::builder() does.
	2261	+ *
	2262	+ * @since 2.14.2
	2263	+ * @param int $post_id Post being inspected.
	2264	+ * @param string $meta_key Builder meta key about to be read.
	2265	+ * @return bool
	2266	+ */
	2267	+ private static function builder_layout_renders(int $post_id, string $meta_key): bool {
	2268	+ if (in_array($meta_key, self::UNPUBLISHED_BUILDER_META_KEYS, true)) {
	2269	+ return false;
	2270	+ }
	2271	+
	2272	+ if (!isset(self::FLAGGED_BUILDER_LAYOUTS[$meta_key])) {
	2273	+ return true;
	2274	+ }
	2275	+
	2276	+ [$flag, $on] = self::FLAGGED_BUILDER_LAYOUTS[$meta_key];
	2277	+ $value = get_post_meta($post_id, $flag, true);
	2278	+
	2279	+ return null === $on ? !empty($value) : $on === (string) $value;
	2280	+ }
	2281	+
	2282	+ /**
	2283	+ * {@see builder_layout_renders()} as an SQL condition on a postmeta row.
	2284	+ *
	2285	+ * The site-wide queries match markers in SQL with nothing confirming each
	2286	+ * row in PHP afterwards, so the flag test has to happen there too. Rows
	2287	+ * under any other key pass untouched. `!empty()` in PHP is false for '' and
	2288	+ * '0', which is what the "any non-empty value" branch excludes.
	2289	+ *
	2290	+ * @since 2.14.2
	2291	+ * @param string $alias Alias of the postmeta row being tested.
	2292	+ * @return array{0:string,1:array<int,string>} ` AND (...)` clause and its values.
	2293	+ */
	2294	+ private static function builder_layout_renders_sql(string $alias): array {
	2295	+ global $wpdb;
	2296	+
	2297	+ $flagged = array_keys(self::FLAGGED_BUILDER_LAYOUTS);
	2298	+ $clauses = [$alias . '.meta_key NOT IN (' . implode(', ', array_fill(0, count($flagged), '%s')) . ')'];
	2299	+ $values = $flagged;
	2300	+
	2301	+ foreach (self::FLAGGED_BUILDER_LAYOUTS as $meta_key => [$flag, $on]) {
	2302	+ $test = null === $on ? "flag.meta_value NOT IN ('', '0')" : 'flag.meta_value = %s';
	2303	+ $clauses[] = "({$alias}.meta_key = %s AND EXISTS (
	2304	+ SELECT 1 FROM {$wpdb->postmeta} flag
	2305	+ WHERE flag.post_id = {$alias}.post_id
	2306	+ AND flag.meta_key = %s
	2307	+ AND {$test}
	2308	+ ))";
	2309	+ $values[] = $meta_key;
	2310	+ $values[] = $flag;
	2311	+
	2312	+ if (null !== $on) {
	2313	+ $values[] = $on;
	2314	+ }
	2315	+ }
	2316	+
	2317	+ return [' AND (' . implode(' OR ', $clauses) . ')', $values];
2161	2318	}
2162	2319
2163	2320	/**
2164	2321	* Whether a stored builder blob names a ThinkRank FAQ or How-To.