← All changes | includes/ai/class-metadata-generator.php +23 -4 1.27.0 → 2.7.0 View file →

		@@ -202,11 +202,12 @@
202	202	// Trim and limit length for AI processing
203	203	$content = trim($content);
204	204	$max_length = 4000; // Reasonable limit for AI processing
205	205
206		- if (strlen($content) > $max_length) {
207		- $content = substr($content, 0, $max_length) . '...';
208		- }
	206	+ // strlen()/substr() count BYTES: on Thai or CJK this handed the model a
	207	+ // third of the intended content, and cut the last character in half so
	208	+ // the payload carried an invalid UTF-8 sequence (#687).
	209	+ $content = \ThinkRank\Core\Seo_Text::trim_to_length($content, $max_length);
209	210
210	211	return $content;
211	212	}
212	213
		@@ -358,9 +359,27 @@
358	359	$words = str_word_count($text, 1);
359	360	$syllables = 0;
360	361
361	362	foreach ($words as $word) {
362		- $syllables += max(1, preg_match_all('/[aeiouy]+/i', $word));
	363	+ $word = preg_replace('/[^a-z]/', '', strtolower($word));
	364	+ if ($word === '') {
	365	+ continue;
	366	+ }
	367	+
	368	+ $groups = preg_match_all('/[aeiouy]+/', $word);
	369	+
	370	+ // Standard Flesch heuristic: a trailing silent e does not form a
	371	+ // syllable ("make", "time", "these") — but only when a consonant
	372	+ // precedes it (a vowel+e ending like "movie" already shares its
	373	+ // group) and never for consonant-le ("table"), which does count.
	374	+ // Without this the counter inflated syllables/word by ~0.2-0.3 on
	375	+ // ordinary prose, driving raw Flesch negative and the UI to a
	376	+ // clamped "Very Difficult (0)" (#407).
	377	+ if ($groups > 1 && preg_match('/[^aeiouy]e$/', $word) && !str_ends_with($word, 'le')) {
	378	+ $groups--;
	379	+ }
	380	+
	381	+ $syllables += max(1, $groups);
363	382	}
364	383
365	384	return $syllables;
366	385	}