← All changes | includes/seo/class-seo-analyzer.php +325 -1 2.6.0 → 2.8.0 View file →

		@@ -48,8 +48,27 @@
48	48	*/
49	49	private const ANSWER_FAQ_NAME = 'thinkrank-faq';
50	50	private const ANSWER_HOWTO_NAME = 'thinkrank-howto';
51	51
	52	+ /**
	53	+ * Element names page builders give their own accordion / FAQ widgets.
	54	+ *
	55	+ * Detection used to recognise ThinkRank's surfaces only, so an FAQ built
	56	+ * with the builder's own accordion was invisible — and the check then
	57	+ * advised adding FAQ content to a site that already had it (#686). Used
	58	+ * for the "is there Q&A content here?" question only; whether ThinkRank
	59	+ * *emits* schema for it is a separate question, still answered by
	60	+ * ThinkRank's own element names.
	61	+ *
	62	+ * @since 2.7.0
	63	+ * @var string[]
	64	+ */
	65	+ private const GENERIC_QA_ELEMENT_MARKERS = [
	66	+ 'accordion', // Elementor, Bricks, Beaver Builder, Breakdance, Oxygen
	67	+ 'toggle', // Elementor
	68	+ 'faq', // Widely used in third-party add-on element names
	69	+ ];
	70	+
52	71	// Check result statuses.
53	72	public const PASSED = 'passed';
54	73	public const WARNING = 'warning';
55	74	public const FAILED = 'failed';
		@@ -1184,8 +1203,16 @@
1184	1203	*/
1185	1204	private $sample_post_ids = null;
1186	1205
1187	1206	/**
	1207	+ * Memoised answer to "does this site publish Q&A content?".
	1208	+ *
	1209	+ * @since 2.7.0
	1210	+ * @var bool|null
	1211	+ */
	1212	+ private $site_has_qa_content = null;
	1213	+
	1214	+ /**
1188	1215	* The sampled post ids, fetched once and shared by every check that reads
1189	1216	* the same slice.
1190	1217	*
1191	1218	* @since 2.5.0
		@@ -1658,9 +1685,18 @@
1658	1685	'message' => __('Your site publishes FAQ, How-To or Q&A structured data, which AI answers can quote question-and-answer pairs from directly.', 'thinkrank'),
1659	1686	];
1660	1687	}
1661	1688
1662		- $how_to_fix = __('Add a ThinkRank FAQ or How-To block, widget or element to your key pages, or enable the FAQPage / HowTo schema types under Essential SEO → Schema.', 'thinkrank');
	1689	+ // Only advise FAQ/HowTo markup where there is question-and-answer
	1690	+ // content to mark up. This used to be assigned unconditionally and
	1691	+ // reused by both branches below, so a site of ordinary articles was
	1692	+ // told to enable FAQPage — and a user who follows that gets FAQPage
	1693	+ // schema with an empty or invented mainEntity, which is the failure
	1694	+ // #494 documents. The check is allowed to report the gap; it is not
	1695	+ // allowed to recommend manufacturing the content (#686).
	1696	+ $how_to_fix = $this->site_has_qa_content()
	1697	+ ? __('Add a ThinkRank FAQ or How-To block, widget or element to the pages that answer questions, or enable the FAQPage / HowTo schema types under Essential SEO → Schema.', 'thinkrank')
	1698	+ : __('Only mark up question-and-answer content you already publish. If this site does not answer discrete questions, answer-shaped schema does not apply and there is nothing to fix here.', 'thinkrank');
1663	1699
1664	1700	// Article/WebPage schema still tells an engine what the page is; the
1665	1701	// gap is the answer pairing, not structured data as a whole.
1666	1702	if ($this->schema_is_output()) {
		@@ -1680,8 +1716,147 @@
1680	1716	];
1681	1717	}
1682	1718
1683	1719	/**
	1720	+ * Whether the site plausibly publishes question-and-answer content.
	1721	+ *
	1722	+ * Deliberately a different question from has_answer_schema(). That one asks
	1723	+ * "does ThinkRank emit answer markup?", which is what the check reports on.
	1724	+ * This one asks "is there anything here that answer markup would describe?",
	1725	+ * which is what decides whether recommending it is sound advice — and it has
	1726	+ * to see content ThinkRank did not author, because an FAQ built with a page
	1727	+ * builder's own accordion is still an FAQ (#686).
	1728	+ *
	1729	+ * Any one signal is enough; all of them are bounded.
	1730	+ *
	1731	+ * @since 2.7.0
	1732	+ * @return bool
	1733	+ */
	1734	+ private function site_has_qa_content(): bool {
	1735	+ if (null !== $this->site_has_qa_content) {
	1736	+ return $this->site_has_qa_content;
	1737	+ }
	1738	+
	1739	+ $this->site_has_qa_content = $this->qa_content_markers_exist()
	1740	+ || $this->sample_has_question_headings();
	1741	+
	1742	+ return $this->site_has_qa_content;
	1743	+ }
	1744	+
	1745	+ /**
	1746	+ * Q&A markers in stored content or builder trees, site-wide.
	1747	+ *
	1748	+ * Two bounded queries rather than a walk over the content sample: the
	1749	+ * sample is the 100 most recent posts and pages, so a site whose only FAQ
	1750	+ * lives in a custom post type, or further back than that, answered "no
	1751	+ * answer content" while publishing exactly that (#686).
	1752	+ *
	1753	+ * @since 2.7.0
	1754	+ * @return bool
	1755	+ */
	1756	+ private function qa_content_markers_exist(): bool {
	1757	+ global $wpdb;
	1758	+
	1759	+ // Block markup and core's details/summary block, in any public type.
	1760	+ $content_markers = ['wp:thinkrank/faq', 'wp:thinkrank/howto', 'wp:details'];
	1761	+ $clauses = [];
	1762	+ $values = [];
	1763	+
	1764	+ foreach ($content_markers as $marker) {
	1765	+ $clauses[] = 'p.post_content LIKE %s';
	1766	+ $values[] = '%' . $wpdb->esc_like($marker) . '%';
	1767	+ }
	1768	+
	1769	+ // The OR list is one '%s' per entry in a fixed class-level marker list,
	1770	+ // so its length varies but its content never comes from input; every
	1771	+ // value goes through prepare(). phpcs cannot see that, and this is the
	1772	+ // usual variable-length-IN exemption.
	1773	+ // phpcs:disable WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
	1774	+ $found = (int) $wpdb->get_var(
	1775	+ $wpdb->prepare(
	1776	+ "SELECT 1 FROM {$wpdb->posts} p
	1777	+ WHERE p.post_status = 'publish'
	1778	+ AND (" . implode(' OR ', $clauses) . ')
	1779	+ LIMIT 1',
	1780	+ ...$values
	1781	+ )
	1782	+ );
	1783	+ // phpcs:enable
	1784	+
	1785	+ if (1 === $found) {
	1786	+ return true;
	1787	+ }
	1788	+
	1789	+ // Builder trees: ThinkRank's own elements, and the builders' generic
	1790	+ // accordion/toggle/FAQ elements, which are what a non-ThinkRank FAQ
	1791	+ // is actually built from.
	1792	+ $meta_keys = self::builder_meta_keys();
	1793	+
	1794	+ if (empty($meta_keys)) {
	1795	+ return false;
	1796	+ }
	1797	+
	1798	+ $markers = array_merge(
	1799	+ [self::ANSWER_FAQ_NAME, self::ANSWER_HOWTO_NAME],
	1800	+ self::GENERIC_QA_ELEMENT_MARKERS
	1801	+ );
	1802	+
	1803	+ $key_placeholders = implode(', ', array_fill(0, count($meta_keys), '%s'));
	1804	+ $marker_clauses = [];
	1805	+ $marker_values = [];
	1806	+
	1807	+ foreach ($markers as $marker) {
	1808	+ $marker_clauses[] = 'pm.meta_value LIKE %s';
	1809	+ $marker_values[] = '%' . $wpdb->esc_like($marker) . '%';
	1810	+ }
	1811	+
	1812	+ // Same exemption: both placeholder runs are sized from fixed lists —
	1813	+ // the builder meta keys and the marker list — and every value is
	1814	+ // prepared.
	1815	+ // phpcs:disable WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
	1816	+ $found = (int) $wpdb->get_var(
	1817	+ $wpdb->prepare(
	1818	+ "SELECT 1 FROM {$wpdb->postmeta} pm
	1819	+ INNER JOIN {$wpdb->posts} p ON p.ID = pm.post_id
	1820	+ WHERE p.post_status = 'publish'
	1821	+ AND pm.meta_key IN ({$key_placeholders})
	1822	+ AND (" . implode(' OR ', $marker_clauses) . ')
	1823	+ LIMIT 1',
	1824	+ ...array_merge($meta_keys, $marker_values)
	1825	+ )
	1826	+ );
	1827	+ // phpcs:enable
	1828	+
	1829	+ return 1 === $found;
	1830	+ }
	1831	+
	1832	+ /**
	1833	+ * Whether the sampled content asks questions in its headings.
	1834	+ *
	1835	+ * The weakest signal and the last one tried: prose that poses questions is
	1836	+ * content answer markup could describe, even where nothing has been marked
	1837	+ * up yet. Runs over the existing sample, so it costs nothing extra.
	1838	+ *
	1839	+ * @since 2.7.0
	1840	+ * @return bool
	1841	+ */
	1842	+ private function sample_has_question_headings(): bool {
	1843	+ foreach ($this->get_content_sample() as $row) {
	1844	+ $content = (string) ($row['content'] ?? '');
	1845	+
	1846	+ if ('' === $content) {
	1847	+ continue;
	1848	+ }
	1849	+
	1850	+ if (preg_match('/<h[2-4][^>]*>\s*[^<]*\?\s*<\/h[2-4]>/i', $content)) {
	1851	+ return true;
	1852	+ }
	1853	+ }
	1854	+
	1855	+ return false;
	1856	+ }
	1857	+
	1858	+ /**
1684	1859	* Whether the site publishes FAQ / HowTo / Q&A structured data.
1685	1860	*
1686	1861	* Three sources, because three things emit it: the Schema Management
1687	1862	* System's enabled types, a per-post-type schema_type in Global SEO, and
		@@ -1726,12 +1901,97 @@
1726	1901	return true;
1727	1902	}
1728	1903	}
1729	1904
	1905	+ // ...and again beyond the sample, which is the 100 most recent posts
	1906	+ // and pages. A site whose only FAQ lives in a custom post type, or
	1907	+ // simply further back than that, reported no answer schema while
	1908	+ // ThinkRank was publishing exactly that (#686). Added alongside the
	1909	+ // walk above rather than replacing it: the sample is already loaded and
	1910	+ // resolves Bricks trees, so it stays the primary source and this only
	1911	+ // extends the reach.
	1912	+ foreach ($this->answer_content_candidates() as $post_id => $content) {
	1913	+ if ($this->has_answer_content($post_id, $content)) {
	1914	+ return true;
	1915	+ }
	1916	+ }
	1917	+
1730	1918	return false;
1731	1919	}
1732	1920
1733	1921	/**
	1922	+ * Posts that might carry a ThinkRank FAQ / How-To, from anywhere on the site.
	1923	+ *
	1924	+ * Narrowed in SQL to posts whose content or builder meta names one of
	1925	+ * ThinkRank's answer surfaces, so the per-post confirmation below runs over
	1926	+ * a handful of rows rather than the whole site.
	1927	+ *
	1928	+ * @since 2.7.0
	1929	+ * @return array<int,string> Post id => raw content.
	1930	+ */
	1931	+ private function answer_content_candidates(): array {
	1932	+ global $wpdb;
	1933	+
	1934	+ $markers = [self::ANSWER_FAQ_NAME, self::ANSWER_HOWTO_NAME, 'wp:thinkrank/faq', 'wp:thinkrank/howto'];
	1935	+
	1936	+ $content_clauses = [];
	1937	+ $content_values = [];
	1938	+ foreach ($markers as $marker) {
	1939	+ $content_clauses[] = 'p.post_content LIKE %s';
	1940	+ $content_values[] = '%' . $wpdb->esc_like($marker) . '%';
	1941	+ }
	1942	+
	1943	+ $meta_keys = self::builder_meta_keys();
	1944	+ $meta_sql = '';
	1945	+ $values = $content_values;
	1946	+
	1947	+ if (!empty($meta_keys)) {
	1948	+ $meta_clauses = [];
	1949	+ $meta_values = [];
	1950	+ foreach ($markers as $marker) {
	1951	+ $meta_clauses[] = 'pm.meta_value LIKE %s';
	1952	+ $meta_values[] = '%' . $wpdb->esc_like($marker) . '%';
	1953	+ }
	1954	+
	1955	+ $key_placeholders = implode(', ', array_fill(0, count($meta_keys), '%s'));
	1956	+ $meta_sql = " OR EXISTS (
	1957	+ SELECT 1 FROM {$wpdb->postmeta} pm
	1958	+ WHERE pm.post_id = p.ID
	1959	+ AND pm.meta_key IN ({$key_placeholders})
	1960	+ AND (" . implode(' OR ', $meta_clauses) . ')
	1961	+ )';
	1962	+ $values = array_merge($content_values, $meta_keys, $meta_values);
	1963	+ }
	1964	+
	1965	+ $values[] = self::CONTENT_SAMPLE_SIZE;
	1966	+
	1967	+ // Same exemption as qa_content_markers_exist(): the clause lists are
	1968	+ // sized from fixed marker and meta-key lists, and every value is
	1969	+ // prepared.
	1970	+ // phpcs:disable WordPress.DB.PreparedSQL.NotPrepared, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
	1971	+ $rows = $wpdb->get_results(
	1972	+ $wpdb->prepare(
	1973	+ "SELECT p.ID, p.post_content FROM {$wpdb->posts} p
	1974	+ WHERE p.post_status = 'publish'
	1975	+ AND ((" . implode(' OR ', $content_clauses) . ")
	1976	+ {$meta_sql})
	1977	+ ORDER BY p.post_date DESC
	1978	+ LIMIT %d",
	1979	+ ...$values
	1980	+ ),
	1981	+ ARRAY_A
	1982	+ );
	1983	+ // phpcs:enable
	1984	+
	1985	+ $candidates = [];
	1986	+ foreach ((array) $rows as $row) {
	1987	+ $candidates[(int) $row['ID']] = (string) $row['post_content'];
	1988	+ }
	1989	+
	1990	+ return $candidates;
	1991	+ }
	1992	+
	1993	+ /**
1734	1994	* Whether one post carries a ThinkRank FAQ or How-To that emits schema.
1735	1995	*
1736	1996	* Four surfaces, because `Schema_Graph` collects from four: the Gutenberg
1737	1997	* block in `post_content`, the Elementor widget, the Bricks element and the
		@@ -1781,9 +2041,73 @@
1781	2041	}
1782	2042	}
1783	2043	}
1784	2044
	2045	+ // Oxygen and Breakdance were missing entirely, so a ThinkRank FAQ
	2046	+ // element placed inside one of those pages reported no answer schema
	2047	+ // while the page was publishing exactly that (#686). Builder_Content
	2048	+ // already knows every key involved — Oxygen 6 is Breakdance under the
	2049	+ // hood, and older releases used two other keys — so ask it rather than
	2050	+ // keeping a second list that can drift.
	2051	+ foreach (self::builder_meta_keys() as $meta_key) {
	2052	+ if ('_fl_builder_data' === $meta_key || '_elementor_data' === $meta_key) {
	2053	+ continue; // Handled above, in their own storage shapes.
	2054	+ }
	2055	+
	2056	+ $stored = get_post_meta($post_id, $meta_key, true);
	2057	+
	2058	+ if (self::blob_names_answer_element($stored)) {
	2059	+ return true;
	2060	+ }
	2061	+ }
	2062	+
1785	2063	return false;
	2064	+ }
	2065	+
	2066	+ /**
	2067	+ * Builder meta keys, or [] when Builder_Content is unavailable.
	2068	+ *
	2069	+ * Mirrors the defensive load in bricks_has_answer_element(): the analyzer
	2070	+ * must degrade to "no answer content found" on a partial checkout rather
	2071	+ * than fatal mid-audit.
	2072	+ *
	2073	+ * @since 2.7.0
	2074	+ * @return string[]
	2075	+ */
	2076	+ private static function builder_meta_keys(): array {
	2077	+ if (!class_exists('ThinkRank\\SEO\\Builder_Content')) {
	2078	+ $file = THINKRANK_PLUGIN_DIR . 'includes/seo/class-builder-content.php';
	2079	+ if (!file_exists($file)) {
	2080	+ return [];
	2081	+ }
	2082	+ require_once $file;
	2083	+ }
	2084	+
	2085	+ return Builder_Content::builder_meta_keys();
	2086	+ }
	2087	+
	2088	+ /**
	2089	+ * Whether a stored builder blob names a ThinkRank FAQ or How-To.
	2090	+ *
	2091	+ * The blob is a JSON tree for Breakdance/Oxygen 6 and a shortcode string
	2092	+ * for Oxygen classic, so this matches on the element name appearing in the
	2093	+ * serialized form rather than parsing each dialect.
	2094	+ *
	2095	+ * @since 2.7.0
	2096	+ * @param mixed $stored Raw meta value.
	2097	+ * @return bool
	2098	+ */
	2099	+ private static function blob_names_answer_element($stored): bool {
	2100	+ if (is_array($stored)) {
	2101	+ $stored = wp_json_encode($stored);
	2102	+ }
	2103	+
	2104	+ if (!is_string($stored) || '' === $stored) {
	2105	+ return false;
	2106	+ }
	2107	+
	2108	+ return false !== strpos($stored, self::ANSWER_FAQ_NAME)
	2109	+ || false !== strpos($stored, self::ANSWER_HOWTO_NAME);
1786	2110	}
1787	2111
1788	2112	/**
1789	2113	* Whether a Bricks-rendered post holds a ThinkRank FAQ or How-To element.