| @@ -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 |
| @@ -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 | |
| @@ -252,31 +253,64 @@ | ||
| 252 | 253 | * @param array $options Generation options |
| 253 | 254 | * @return array Enhanced metadata |
| 254 | 255 | */ |
| 255 | 256 | private function validate_and_enhance_metadata(array $metadata, array $options): array { |
| 257 | + $max_title_length = max(1, (int) $options['max_title_length']); | |
| 258 | + $max_description_length = max(1, (int) $options['max_description_length']); | |
| 259 | + | |
| 260 | + $metadata['title'] = trim((string) ($metadata['title'] ?? '')); | |
| 261 | + $metadata['description'] = trim((string) ($metadata['description'] ?? '')); | |
| 262 | + | |
| 256 | 263 | // Validate title length |
| 257 | - if (strlen($metadata['title']) > $options['max_title_length']) { | |
| 258 | - $metadata['title'] = substr($metadata['title'], 0, $options['max_title_length'] - 3) . '...'; | |
| 264 | + if (mb_strlen($metadata['title']) > $max_title_length) { | |
| 265 | + $metadata['title'] = $this->truncate_text($metadata['title'], $max_title_length); | |
| 259 | 266 | $metadata['title_truncated'] = true; |
| 260 | 267 | } |
| 261 | - | |
| 268 | + | |
| 262 | 269 | // Validate description length |
| 263 | - if (strlen($metadata['description']) > $options['max_description_length']) { | |
| 264 | - $metadata['description'] = substr($metadata['description'], 0, $options['max_description_length'] - 3) . '...'; | |
| 270 | + if (mb_strlen($metadata['description']) > $max_description_length) { | |
| 271 | + $metadata['description'] = $this->truncate_text($metadata['description'], $max_description_length); | |
| 265 | 272 | $metadata['description_truncated'] = true; |
| 266 | 273 | } |
| 267 | - | |
| 274 | + | |
| 268 | 275 | // Add character counts |
| 269 | - $metadata['title_length'] = strlen($metadata['title']); | |
| 270 | - $metadata['description_length'] = strlen($metadata['description']); | |
| 271 | - | |
| 276 | + $metadata['title_length'] = mb_strlen($metadata['title']); | |
| 277 | + $metadata['description_length'] = mb_strlen($metadata['description']); | |
| 278 | + | |
| 272 | 279 | // Calculate optimization score |
| 273 | 280 | $metadata['optimization_score'] = $this->calculate_optimization_score($metadata, $options); |
| 274 | - | |
| 281 | + | |
| 275 | 282 | return $metadata; |
| 276 | 283 | } |
| 277 | - | |
| 284 | + | |
| 278 | 285 | /** |
| 286 | + * Truncate text to a hard character limit, preferring a word boundary | |
| 287 | + * | |
| 288 | + * The ellipsis counts towards the limit, so the result is never longer | |
| 289 | + * than $max_length characters. | |
| 290 | + * | |
| 291 | + * @param string $text Text to truncate | |
| 292 | + * @param int $max_length Maximum length in characters | |
| 293 | + * @return string Truncated text | |
| 294 | + */ | |
| 295 | + private function truncate_text(string $text, int $max_length): string { | |
| 296 | + if (mb_strlen($text) <= $max_length) { | |
| 297 | + return $text; | |
| 298 | + } | |
| 299 | + | |
| 300 | + // Reserve one character for the ellipsis. | |
| 301 | + $truncated = mb_substr($text, 0, $max_length - 1); | |
| 302 | + | |
| 303 | + // Cut back to the last word boundary, unless that throws away too much. | |
| 304 | + $last_space = mb_strrpos($truncated, ' '); | |
| 305 | + if ($last_space !== false && $last_space > (int) ($max_length * 0.6)) { | |
| 306 | + $truncated = mb_substr($truncated, 0, $last_space); | |
| 307 | + } | |
| 308 | + | |
| 309 | + return rtrim($truncated, " \t\n\r\0\x0B,;:-") . '…'; | |
| 310 | + } | |
| 311 | + | |
| 312 | + /** | |
| 279 | 313 | * Save metadata to post meta |
| 280 | 314 | * |
| 281 | 315 | * @param int $post_id Post ID |
| 282 | 316 | * @param array $metadata Metadata to save |
| @@ -284,9 +318,13 @@ | ||
| 284 | 318 | */ |
| 285 | 319 | private function save_post_metadata(int $post_id, array $metadata): void { |
| 286 | 320 | update_post_meta($post_id, '_thinkrank_ai_title', $metadata['title']); |
| 287 | 321 | update_post_meta($post_id, '_thinkrank_ai_description', $metadata['description']); |
| 288 | - update_post_meta($post_id, '_thinkrank_focus_keyword', $metadata['focus_keyword']); | |
| 322 | + // Persist via Focus_Keywords so the array + legacy meta stay in sync. | |
| 323 | + // Existing keywords are preserved (the AI value seeds only when unset). | |
| 324 | + if (empty(\ThinkRank\SEO\Focus_Keywords::get($post_id))) { | |
| 325 | + \ThinkRank\SEO\Focus_Keywords::save($post_id, $metadata['focus_keyword']); | |
| 326 | + } | |
| 289 | 327 | update_post_meta($post_id, '_thinkrank_generated_at', $metadata['generated_at']); |
| 290 | 328 | update_post_meta($post_id, '_thinkrank_metadata_full', $metadata); |
| 291 | 329 | } |
| 292 | 330 | |
| @@ -321,9 +359,27 @@ | ||
| 321 | 359 | $words = str_word_count($text, 1); |
| 322 | 360 | $syllables = 0; |
| 323 | 361 | |
| 324 | 362 | foreach ($words as $word) { |
| 325 | - $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); | |
| 326 | 382 | } |
| 327 | 383 | |
| 328 | 384 | return $syllables; |
| 329 | 385 | } |
