← All changes | includes/ai/class-metadata-generator.php +101 -28 1.0.1 → 2.9.0 View file →

		@@ -49,9 +49,9 @@
49	49	* @param Settings|null $settings Settings instance
50	50	*/
51	51	public function __construct(?Manager $ai_manager = null, ?Settings $settings = null) {
52	52	$this->ai_manager = $ai_manager ?? new Manager();
53		- $this->settings = $settings ?? new Settings();
	53	+ $this->settings = $settings ?? Settings::instance();
54	54	}
55	55
56	56	/**
57	57	* Generate metadata for post content
		@@ -96,30 +96,47 @@
96	96	* @return array Generated metadata
97	97	* @throws \Exception If generation fails
98	98	*/
99	99	public function generate_for_post(int $post_id, array $options = []): array {
	100	+ $metadata = $this->suggest_for_post($post_id, $options);
	101	+
	102	+ // Save metadata to post meta
	103	+ $this->save_post_metadata($post_id, $metadata);
	104	+
	105	+ return $metadata;
	106	+ }
	107	+
	108	+ /**
	109	+ * Generate metadata for a post without writing anything.
	110	+ *
	111	+ * The same generation as {@see self::generate_for_post()}, minus the save.
	112	+ * Bulk Snippets shows the result for review and writes only when the user
	113	+ * clicks Save, so it must never persist as a side effect (#727).
	114	+ *
	115	+ * @since 2.8.0
	116	+ *
	117	+ * @param int $post_id Post ID.
	118	+ * @param array $options Generation options.
	119	+ * @return array Generated metadata.
	120	+ * @throws \Exception If the post is missing or generation fails.
	121	+ */
	122	+ public function suggest_for_post(int $post_id, array $options = []): array {
100	123	$post = get_post($post_id);
101		-
	124	+
102	125	if (!$post) {
103	126	throw new \Exception('Post not found');
104	127	}
105		-
	128	+
106	129	// Extract content from post
107	130	$content = $this->extract_post_content($post);
108		-
	131	+
109	132	// Add post-specific options
110	133	$options = array_merge($options, [
111	134	'content_type' => $this->determine_content_type($post),
112	135	'post_id' => $post_id,
113	136	]);
114		-
115		- // Generate metadata
116		- $metadata = $this->generate_for_content($content, $options);
117		-
118		- // Save metadata to post meta
119		- $this->save_post_metadata($post_id, $metadata);
120		-
121		- return $metadata;
	137	+
	138	+ return $this->generate_for_content($content, $options);
122	139	}
123	140
124	141	/**
125	142	* Generate multiple variations of metadata
		@@ -202,11 +219,12 @@
202	219	// Trim and limit length for AI processing
203	220	$content = trim($content);
204	221	$max_length = 4000; // Reasonable limit for AI processing
205	222
206		- if (strlen($content) > $max_length) {
207		- $content = substr($content, 0, $max_length) . '...';
208		- }
	223	+ // strlen()/substr() count BYTES: on Thai or CJK this handed the model a
	224	+ // third of the intended content, and cut the last character in half so
	225	+ // the payload carried an invalid UTF-8 sequence (#687).
	226	+ $content = \ThinkRank\Core\Seo_Text::trim_to_length($content, $max_length);
209	227
210	228	return $content;
211	229	}
212	230
		@@ -252,31 +270,64 @@
252	270	* @param array $options Generation options
253	271	* @return array Enhanced metadata
254	272	*/
255	273	private function validate_and_enhance_metadata(array $metadata, array $options): array {
	274	+ $max_title_length = max(1, (int) $options['max_title_length']);
	275	+ $max_description_length = max(1, (int) $options['max_description_length']);
	276	+
	277	+ $metadata['title'] = trim((string) ($metadata['title'] ?? ''));
	278	+ $metadata['description'] = trim((string) ($metadata['description'] ?? ''));
	279	+
256	280	// Validate title length
257		- if (strlen($metadata['title']) > $options['max_title_length']) {
258		- $metadata['title'] = substr($metadata['title'], 0, $options['max_title_length'] - 3) . '...';
	281	+ if (mb_strlen($metadata['title']) > $max_title_length) {
	282	+ $metadata['title'] = $this->truncate_text($metadata['title'], $max_title_length);
259	283	$metadata['title_truncated'] = true;
260	284	}
261		-
	285	+
262	286	// Validate description length
263		- if (strlen($metadata['description']) > $options['max_description_length']) {
264		- $metadata['description'] = substr($metadata['description'], 0, $options['max_description_length'] - 3) . '...';
	287	+ if (mb_strlen($metadata['description']) > $max_description_length) {
	288	+ $metadata['description'] = $this->truncate_text($metadata['description'], $max_description_length);
265	289	$metadata['description_truncated'] = true;
266	290	}
267		-
	291	+
268	292	// Add character counts
269		- $metadata['title_length'] = strlen($metadata['title']);
270		- $metadata['description_length'] = strlen($metadata['description']);
271		-
	293	+ $metadata['title_length'] = mb_strlen($metadata['title']);
	294	+ $metadata['description_length'] = mb_strlen($metadata['description']);
	295	+
272	296	// Calculate optimization score
273	297	$metadata['optimization_score'] = $this->calculate_optimization_score($metadata, $options);
274		-
	298	+
275	299	return $metadata;
276	300	}
277		-
	301	+
278	302	/**
	303	+ * Truncate text to a hard character limit, preferring a word boundary
	304	+ *
	305	+ * The ellipsis counts towards the limit, so the result is never longer
	306	+ * than $max_length characters.
	307	+ *
	308	+ * @param string $text Text to truncate
	309	+ * @param int $max_length Maximum length in characters
	310	+ * @return string Truncated text
	311	+ */
	312	+ private function truncate_text(string $text, int $max_length): string {
	313	+ if (mb_strlen($text) <= $max_length) {
	314	+ return $text;
	315	+ }
	316	+
	317	+ // Reserve one character for the ellipsis.
	318	+ $truncated = mb_substr($text, 0, $max_length - 1);
	319	+
	320	+ // Cut back to the last word boundary, unless that throws away too much.
	321	+ $last_space = mb_strrpos($truncated, ' ');
	322	+ if ($last_space !== false && $last_space > (int) ($max_length * 0.6)) {
	323	+ $truncated = mb_substr($truncated, 0, $last_space);
	324	+ }
	325	+
	326	+ return rtrim($truncated, " \t\n\r\0\x0B,;:-") . '…';
	327	+ }
	328	+
	329	+ /**
279	330	* Save metadata to post meta
280	331	*
281	332	* @param int $post_id Post ID
282	333	* @param array $metadata Metadata to save
		@@ -284,9 +335,13 @@
284	335	*/
285	336	private function save_post_metadata(int $post_id, array $metadata): void {
286	337	update_post_meta($post_id, '_thinkrank_ai_title', $metadata['title']);
287	338	update_post_meta($post_id, '_thinkrank_ai_description', $metadata['description']);
288		- update_post_meta($post_id, '_thinkrank_focus_keyword', $metadata['focus_keyword']);
	339	+ // Persist via Focus_Keywords so the array + legacy meta stay in sync.
	340	+ // Existing keywords are preserved (the AI value seeds only when unset).
	341	+ if (empty(\ThinkRank\SEO\Focus_Keywords::get($post_id))) {
	342	+ \ThinkRank\SEO\Focus_Keywords::save($post_id, $metadata['focus_keyword']);
	343	+ }
289	344	update_post_meta($post_id, '_thinkrank_generated_at', $metadata['generated_at']);
290	345	update_post_meta($post_id, '_thinkrank_metadata_full', $metadata);
291	346	}
292	347
		@@ -321,9 +376,27 @@
321	376	$words = str_word_count($text, 1);
322	377	$syllables = 0;
323	378
324	379	foreach ($words as $word) {
325		- $syllables += max(1, preg_match_all('/[aeiouy]+/i', $word));
	380	+ $word = preg_replace('/[^a-z]/', '', strtolower($word));
	381	+ if ($word === '') {
	382	+ continue;
	383	+ }
	384	+
	385	+ $groups = preg_match_all('/[aeiouy]+/', $word);
	386	+
	387	+ // Standard Flesch heuristic: a trailing silent e does not form a
	388	+ // syllable ("make", "time", "these") — but only when a consonant
	389	+ // precedes it (a vowel+e ending like "movie" already shares its
	390	+ // group) and never for consonant-le ("table"), which does count.
	391	+ // Without this the counter inflated syllables/word by ~0.2-0.3 on
	392	+ // ordinary prose, driving raw Flesch negative and the UI to a
	393	+ // clamped "Very Difficult (0)" (#407).
	394	+ if ($groups > 1 && preg_match('/[^aeiouy]e$/', $word) && !str_ends_with($word, 'le')) {
	395	+ $groups--;
	396	+ }
	397	+
	398	+ $syllables += max(1, $groups);
326	399	}
327	400
328	401	return $syllables;
329	402	}