← All changes | includes/ai/class-seo-score-calculator.php +31 -97 2.14.2 → 2.14.1 View file →

		@@ -1263,16 +1263,14 @@
1263	1263	if (empty($content) || empty($target_keyword)) {
1264	1264	return 0.0;
1265	1265	}
1266	1266
1267		- // Occurrences and the word count both come from the reading text, so
1268		- // shortcode syntax is in neither.
1269		- $content_lower = strtolower(self::reading_text_of($content));
	1267	+ $content_lower = strtolower(wp_strip_all_tags($content));
1270	1268	$keyword_lower = strtolower($target_keyword);
1271	1269
1272	1270	// Calculate keyword and semantic term frequency
1273	1271	$keyword_count = substr_count($content_lower, $keyword_lower);
1274		- $word_count = \ThinkRank\SEO\Word_Count_Index::count_words($content_lower);
	1272	+ $word_count = $this->calculate_word_count_js_style($content_lower);
1275	1273
1276	1274	if ($word_count === 0) {
1277	1275	return 0.0;
1278	1276	}
		@@ -1351,13 +1349,10 @@
1351	1349	private function keyword_density(string $content, string $target_keyword): float {
1352	1350	if (empty($content) || empty($target_keyword)) {
1353	1351	return 0.0;
1354	1352	}
1355		- // Numerator and denominator from the same reading text. Counting the
1356		- // keyword in wp_strip_all_tags() output found it inside shortcode
1357		- // attributes the word count no longer includes.
1358		- $plain = strtolower(self::reading_text_of($content));
1359		- $word_count = \ThinkRank\SEO\Word_Count_Index::count_words($plain);
	1353	+ $plain = strtolower(wp_strip_all_tags($content));
	1354	+ $word_count = $this->calculate_word_count_js_style($plain);
1360	1355	if ($word_count === 0) {
1361	1356	return 0.0;
1362	1357	}
1363	1358	$occurrences = substr_count($plain, strtolower($target_keyword));
		@@ -1392,10 +1387,10 @@
1392	1387	private function keyword_in_first_paragraph(string $content, string $target_keyword): bool {
1393	1388	if (empty($content) || empty($target_keyword)) {
1394	1389	return false;
1395	1390	}
1396		- $plain = strtolower(self::reading_text_of($content));
1397		- $words = preg_split('/\s+/', $plain, -1, PREG_SPLIT_NO_EMPTY) ?: [];
	1391	+ $plain = strtolower(wp_strip_all_tags($content));
	1392	+ $words = preg_split('/\s+/', trim($plain), -1, PREG_SPLIT_NO_EMPTY) ?: [];
1398	1393	$window = array_slice($words, 0, max(50, (int) ceil(count($words) * 0.1)));
1399	1394	return strpos(implode(' ', $window), strtolower($target_keyword)) !== false;
1400	1395	}
1401	1396
		@@ -1655,12 +1650,11 @@
1655	1650
1656	1651	// Extract headings from content
1657	1652	$headings = $this->extract_headings($content);
1658	1653
1659		- // Count the markup, not wp_strip_all_tags() output: stripping deletes
1660		- // tags without a space, so "five</p><p>six" would already be one word
1661		- // before the counter saw it.
1662		- $word_count = $this->calculate_word_count_js_style($content);
	1654	+ // Count words using JavaScript-compatible method
	1655	+ $plain_text = wp_strip_all_tags($content);
	1656	+ $word_count = $this->calculate_word_count_js_style($plain_text);
1663	1657
1664	1658	// Calculate readability
1665	1659	$readability_score = $this->calculate_readability_score($content);
1666	1660
		@@ -1780,9 +1774,9 @@
1780	1774	// 2. Long paragraphs (flag paragraphs over 150 words).
1781	1775	$long_paragraphs = 0;
1782	1776	if (preg_match_all('/<p[^>]*>(.*?)<\/p>/is', $content, $matches)) {
1783	1777	foreach ($matches[1] as $paragraph) {
1784		- if ($this->calculate_word_count_js_style($paragraph) > 150) {
	1778	+ if ($this->calculate_word_count_js_style(wp_strip_all_tags($paragraph)) > 150) {
1785	1779	$long_paragraphs++;
1786	1780	}
1787	1781	}
1788	1782	}
		@@ -1837,22 +1831,22 @@
1837	1831	* @param string $content Content text
1838	1832	* @return float Readability score
1839	1833	*/
1840	1834	private function calculate_readability_score(string $content): float {
1841		- // Sentences, words and syllables all come from one reading text. The
1842		- // word count drops shortcode syntax and punctuation-only tokens; when
1843		- // sentences and syllables were still read off wp_strip_all_tags()
1844		- // output, "[vc_column width="1/2"]" added syllables (and, with a "." in
1845		- // an attribute, sentences) to a word count that did not include it,
1846		- // and a WPBakery page's Flesch fell from 65 to 46.
1847		- $text = self::reading_text_of($content);
	1835	+ $text = wp_strip_all_tags($content);
1848	1836
1849		- if ('' === $text) {
	1837	+ if (empty($text)) {
1850	1838	return 0;
1851	1839	}
1852	1840
1853		- $sentence_count = self::count_sentences($text);
1854		- $word_count = \ThinkRank\SEO\Word_Count_Index::count_words($text);
	1841	+ // Count sentences (approximate)
	1842	+ $sentences = preg_split('/[.!?]+/', $text, -1, PREG_SPLIT_NO_EMPTY);
	1843	+ $sentence_count = count($sentences);
	1844	+
	1845	+ // Count words
	1846	+ $word_count = $this->calculate_word_count_js_style(wp_strip_all_tags($text));
	1847	+
	1848	+ // Count syllables (approximate)
1855	1849	$syllable_count = $this->count_syllables($text);
1856	1850
1857	1851	if ($sentence_count === 0 || $word_count === 0) {
1858	1852	return 0;
		@@ -1864,37 +1858,15 @@
1864	1858	return max(0, min(100, $score));
1865	1859	}
1866	1860
1867	1861	/**
1868		- * Count the sentences in reading text.
1869		- *
1870		- * A sentence is a run of text between terminal punctuation that holds at
1871		- * least one letter or digit. A fragment with no word in it (the space
1872		- * after the final full stop, a stray "!" between two "?") is not a
1873		- * sentence. contentAnalysis.js countSentences() applies the same rule.
1874		- *
1875		- * @since 2.14.2
1876		- *
1877		- * @param string $text Text as returned by {@see self::reading_text_of()}.
1878		- * @return int
1879		- */
1880		- private static function count_sentences(string $text): int {
1881		- $fragments = preg_split('/[.!?]+/', $text, -1, PREG_SPLIT_NO_EMPTY) ?: [];
1882		-
1883		- return count(preg_grep('/[\p{L}\p{N}]/u', $fragments) ?: []);
1884		- }
1885		-
1886		- /**
1887	1862	* Count syllables in text (approximate)
1888	1863	*
1889		- * Expects reading text ({@see self::reading_text_of()}); tags are not
1890		- * stripped here, so a decoded "<" in prose is not read as a tag.
1891		- *
1892	1864	* @param string $text Text to analyze
1893	1865	* @return int Syllable count
1894	1866	*/
1895	1867	private function count_syllables(string $text): int {
1896		- $words = preg_split('/\s+/', trim(strtolower($text)), -1, PREG_SPLIT_NO_EMPTY) ?: [];
	1868	+ $words = preg_split('/\s+/', trim(strtolower(wp_strip_all_tags($text))), -1, PREG_SPLIT_NO_EMPTY);
1897	1869	$syllables = 0;
1898	1870
1899	1871	foreach ($words as $word) {
1900	1872	$word = preg_replace('/[^a-z]/', '', $word);
		@@ -2504,12 +2476,11 @@
2504	2476
2505	2477	// Extract headings from content
2506	2478	$headings = $this->extract_headings($content);
2507	2479
2508		- // Count the markup, not wp_strip_all_tags() output: stripping deletes
2509		- // tags without a space, so "five</p><p>six" would already be one word
2510		- // before the counter saw it.
2511		- $word_count = $this->calculate_word_count_js_style($content);
	2480	+ // Count words using JavaScript-compatible method
	2481	+ $plain_text = wp_strip_all_tags($content);
	2482	+ $word_count = $this->calculate_word_count_js_style($plain_text);
2512	2483
2513	2484	// Calculate readability
2514	2485	$readability_score = $this->calculate_readability_score($content);
2515	2486
		@@ -2543,59 +2514,22 @@
2543	2514	];
2544	2515	}
2545	2516
2546	2517	/**
2547		- * Count the words a reader reads in HTML or text.
	2518	+ * Calculate word count using JavaScript-compatible method
	2519	+ * Matches the logic in contentAnalysis.js for consistency
2548	2520	*
2549		- * The same extraction and counting as the thin content report
2550		- * ({@see \ThinkRank\SEO\Word_Count_Index::reading_text()} and
2551		- * {@see \ThinkRank\SEO\Word_Count_Index::count_words()}), so the editor
2552		- * score and the report give one number for one page. Splitting on
2553		- * whitespace counted shortcode syntax as words: a WPBakery page with 216
2554		- * words of prose scored a word count of 325 while the report said 216
2555		- * (#893 fixed the report only). It also counted tokens of punctuation
2556		- * alone, such as a full stop after a link.
2557		- *
2558		- * Always words, whatever the locale's unit, because every threshold that
2559		- * reads this value (content length, long paragraphs, headings per 300
2560		- * words, readability, keyword density) is in words.
2561		- *
2562		- * contentAnalysis.js calculateWordCount() applies the same two rules in
2563		- * the editor.
2564		- *
2565		- * @param string $text HTML or text to count words in.
	2521	+ * @param string $text Text to count words in
2566	2522	* @return int Word count
2567	2523	*/
2568	2524	private function calculate_word_count_js_style(string $text): int {
2569		- if ('' === $text) {
	2525	+ if (empty($text)) {
2570	2526	return 0;
2571	2527	}
2572	2528
2573		- return \ThinkRank\SEO\Word_Count_Index::count_words(self::reading_text_of($text));
2574		- }
2575		-
2576		- /**
2577		- * The text a reader reads in HTML or text, as the word count sees it.
2578		- *
2579		- * Every check that divides by the word count (readability, keyword
2580		- * density, topic relevance) reads its numerator from this same text, so
2581		- * shortcode syntax is never on one side of a ratio and not the other.
2582		- *
2583		- * @since 2.14.2
2584		- *
2585		- * @param string $content HTML or text.
2586		- * @return string Plain text, whitespace collapsed.
2587		- */
2588		- private static function reading_text_of(string $content): string {
2589		- if ('' === $content) {
2590		- return '';
2591		- }
2592		-
2593		- if (!class_exists('\ThinkRank\SEO\Word_Count_Index')) {
2594		- require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-word-count-index.php';
2595		- }
2596		-
2597		- return \ThinkRank\SEO\Word_Count_Index::reading_text($content);
	2529	+ // Match JavaScript: trim, split by whitespace, filter empty
	2530	+ $words = preg_split('/\s+/', trim($text), -1, PREG_SPLIT_NO_EMPTY);
	2531	+ return count($words);
2598	2532	}
2599	2533
2600	2534	/**
2601	2535	* Get existing score data for a post from database