| 1 |
<?php |
| 2 |
/** |
| 3 |
* AI Metadata Generator |
| 4 |
* |
| 5 |
* Core functionality for generating SEO metadata using AI |
| 6 |
* |
| 7 |
* @package ThinkRank\AI |
| 8 |
* @since 1.0.0 |
| 9 |
*/ |
| 10 |
|
| 11 |
declare(strict_types=1); |
| 12 |
|
| 13 |
namespace ThinkRank\AI; |
| 14 |
|
| 15 |
use ThinkRank\Core\Settings; |
| 16 |
|
| 17 |
// Prevent direct access |
| 18 |
if (!defined('ABSPATH')) { |
| 19 |
exit; |
| 20 |
} |
| 21 |
|
| 22 |
/** |
| 23 |
* Metadata Generator Class |
| 24 |
* |
| 25 |
* Single Responsibility: Generate SEO metadata using AI |
| 26 |
* |
| 27 |
* @since 1.0.0 |
| 28 |
*/ |
| 29 |
class Metadata_Generator { |
| 30 |
|
| 31 |
/** |
| 32 |
* AI Manager instance |
| 33 |
* |
| 34 |
* @var Manager |
| 35 |
*/ |
| 36 |
private Manager $ai_manager; |
| 37 |
|
| 38 |
/** |
| 39 |
* Settings instance |
| 40 |
* |
| 41 |
* @var Settings |
| 42 |
*/ |
| 43 |
private Settings $settings; |
| 44 |
|
| 45 |
/** |
| 46 |
* Constructor |
| 47 |
* |
| 48 |
* @param Manager|null $ai_manager AI manager instance |
| 49 |
* @param Settings|null $settings Settings instance |
| 50 |
*/ |
| 51 |
public function __construct(?Manager $ai_manager = null, ?Settings $settings = null) { |
| 52 |
$this->ai_manager = $ai_manager ?? new Manager(); |
| 53 |
$this->settings = $settings ?? Settings::instance(); |
| 54 |
} |
| 55 |
|
| 56 |
/** |
| 57 |
* Generate metadata for post content |
| 58 |
* |
| 59 |
* @param string $content Post content |
| 60 |
* @param array $options Generation options |
| 61 |
* @return array Generated metadata |
| 62 |
* @throws \Exception If generation fails |
| 63 |
*/ |
| 64 |
public function generate_for_content(string $content, array $options = []): array { |
| 65 |
// Validate content |
| 66 |
if (empty(trim($content))) { |
| 67 |
throw new \Exception('Content cannot be empty'); |
| 68 |
} |
| 69 |
|
| 70 |
// Prepare content for AI processing |
| 71 |
$processed_content = $this->prepare_content($content); |
| 72 |
|
| 73 |
// Set default options |
| 74 |
$default_options = [ |
| 75 |
'target_keyword' => '', |
| 76 |
'content_type' => 'blog_post', |
| 77 |
'tone' => 'professional', |
| 78 |
'max_title_length' => $this->settings->get('meta_title_length', 60), |
| 79 |
'max_description_length' => $this->settings->get('meta_description_length', 160), |
| 80 |
]; |
| 81 |
|
| 82 |
$options = array_merge($default_options, $options); |
| 83 |
|
| 84 |
// Generate metadata using AI |
| 85 |
$metadata = $this->ai_manager->generate_seo_metadata($processed_content, $options); |
| 86 |
|
| 87 |
// Post-process and validate metadata |
| 88 |
return $this->validate_and_enhance_metadata($metadata, $options); |
| 89 |
} |
| 90 |
|
| 91 |
/** |
| 92 |
* Generate metadata for WordPress post |
| 93 |
* |
| 94 |
* @param int $post_id Post ID |
| 95 |
* @param array $options Generation options |
| 96 |
* @return array Generated metadata |
| 97 |
* @throws \Exception If generation fails |
| 98 |
*/ |
| 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 { |
| 123 |
$post = get_post($post_id); |
| 124 |
|
| 125 |
if (!$post) { |
| 126 |
throw new \Exception('Post not found'); |
| 127 |
} |
| 128 |
|
| 129 |
// Extract content from post |
| 130 |
$content = $this->extract_post_content($post); |
| 131 |
|
| 132 |
// Add post-specific options |
| 133 |
$options = array_merge($options, [ |
| 134 |
'content_type' => $this->determine_content_type($post), |
| 135 |
'post_id' => $post_id, |
| 136 |
]); |
| 137 |
|
| 138 |
return $this->generate_for_content($content, $options); |
| 139 |
} |
| 140 |
|
| 141 |
/** |
| 142 |
* Generate multiple variations of metadata |
| 143 |
* |
| 144 |
* @param string $content Content to analyze |
| 145 |
* @param array $options Generation options |
| 146 |
* @param int $variations Number of variations to generate |
| 147 |
* @return array Array of metadata variations |
| 148 |
*/ |
| 149 |
public function generate_variations(string $content, array $options = [], int $variations = 3): array { |
| 150 |
$results = []; |
| 151 |
|
| 152 |
for ($i = 0; $i < $variations; $i++) { |
| 153 |
// Slightly vary the temperature for different results |
| 154 |
$variation_options = array_merge($options, [ |
| 155 |
'temperature' => 0.3 + ($i * 0.2), // 0.3, 0.5, 0.7 |
| 156 |
'variation' => $i + 1, |
| 157 |
]); |
| 158 |
|
| 159 |
try { |
| 160 |
$metadata = $this->generate_for_content($content, $variation_options); |
| 161 |
$metadata['variation_id'] = $i + 1; |
| 162 |
$results[] = $metadata; |
| 163 |
|
| 164 |
// Small delay to avoid rate limiting |
| 165 |
if ($i < $variations - 1) { |
| 166 |
sleep(1); |
| 167 |
} |
| 168 |
|
| 169 |
} catch (\Exception $e) { |
| 170 |
// Skip failed variations silently |
| 171 |
} |
| 172 |
} |
| 173 |
|
| 174 |
return $results; |
| 175 |
} |
| 176 |
|
| 177 |
/** |
| 178 |
* Analyze content and suggest improvements |
| 179 |
* |
| 180 |
* @param string $content Content to analyze |
| 181 |
* @param array $current_metadata Current metadata |
| 182 |
* @return array Analysis and suggestions |
| 183 |
*/ |
| 184 |
public function analyze_content(string $content, array $current_metadata = []): array { |
| 185 |
$analysis = [ |
| 186 |
'content_length' => str_word_count($content), |
| 187 |
'readability_score' => $this->calculate_readability_score($content), |
| 188 |
'keyword_density' => $this->analyze_keyword_density($content), |
| 189 |
'suggestions' => [], |
| 190 |
]; |
| 191 |
|
| 192 |
// Analyze current metadata if provided |
| 193 |
if (!empty($current_metadata)) { |
| 194 |
$analysis['metadata_analysis'] = $this->analyze_metadata($current_metadata); |
| 195 |
} |
| 196 |
|
| 197 |
// Generate improvement suggestions |
| 198 |
$analysis['suggestions'] = $this->generate_improvement_suggestions($analysis); |
| 199 |
|
| 200 |
return $analysis; |
| 201 |
} |
| 202 |
|
| 203 |
/** |
| 204 |
* Prepare content for AI processing |
| 205 |
* |
| 206 |
* @param string $content Raw content |
| 207 |
* @return string Processed content |
| 208 |
*/ |
| 209 |
private function prepare_content(string $content): string { |
| 210 |
// Remove HTML tags |
| 211 |
$content = wp_strip_all_tags($content); |
| 212 |
|
| 213 |
// Remove shortcodes |
| 214 |
$content = strip_shortcodes($content); |
| 215 |
|
| 216 |
// Normalize whitespace |
| 217 |
$content = preg_replace('/\s+/', ' ', $content); |
| 218 |
|
| 219 |
// Trim and limit length for AI processing |
| 220 |
$content = trim($content); |
| 221 |
$max_length = 4000; // Reasonable limit for AI processing |
| 222 |
|
| 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); |
| 227 |
|
| 228 |
return $content; |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Extract content from WordPress post |
| 233 |
* |
| 234 |
* @param \WP_Post $post Post object |
| 235 |
* @return string Extracted content |
| 236 |
*/ |
| 237 |
private function extract_post_content(\WP_Post $post): string { |
| 238 |
$content = $post->post_title . "\n\n" . $post->post_content; |
| 239 |
|
| 240 |
// Add excerpt if available |
| 241 |
if (!empty($post->post_excerpt)) { |
| 242 |
$content = $post->post_title . "\n\n" . $post->post_excerpt . "\n\n" . $post->post_content; |
| 243 |
} |
| 244 |
|
| 245 |
return $content; |
| 246 |
} |
| 247 |
|
| 248 |
/** |
| 249 |
* Determine content type from post |
| 250 |
* |
| 251 |
* @param \WP_Post $post Post object |
| 252 |
* @return string Content type |
| 253 |
*/ |
| 254 |
private function determine_content_type(\WP_Post $post): string { |
| 255 |
switch ($post->post_type) { |
| 256 |
case 'page': |
| 257 |
return 'page'; |
| 258 |
case 'product': |
| 259 |
return 'product'; |
| 260 |
case 'post': |
| 261 |
default: |
| 262 |
return 'blog_post'; |
| 263 |
} |
| 264 |
} |
| 265 |
|
| 266 |
/** |
| 267 |
* Validate and enhance generated metadata |
| 268 |
* |
| 269 |
* @param array $metadata Generated metadata |
| 270 |
* @param array $options Generation options |
| 271 |
* @return array Enhanced metadata |
| 272 |
*/ |
| 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 |
|
| 280 |
// Validate title length |
| 281 |
if (mb_strlen($metadata['title']) > $max_title_length) { |
| 282 |
$metadata['title'] = $this->truncate_text($metadata['title'], $max_title_length); |
| 283 |
$metadata['title_truncated'] = true; |
| 284 |
} |
| 285 |
|
| 286 |
// Validate description length |
| 287 |
if (mb_strlen($metadata['description']) > $max_description_length) { |
| 288 |
$metadata['description'] = $this->truncate_text($metadata['description'], $max_description_length); |
| 289 |
$metadata['description_truncated'] = true; |
| 290 |
} |
| 291 |
|
| 292 |
// Add character counts |
| 293 |
$metadata['title_length'] = mb_strlen($metadata['title']); |
| 294 |
$metadata['description_length'] = mb_strlen($metadata['description']); |
| 295 |
|
| 296 |
// Calculate optimization score |
| 297 |
$metadata['optimization_score'] = $this->calculate_optimization_score($metadata, $options); |
| 298 |
|
| 299 |
return $metadata; |
| 300 |
} |
| 301 |
|
| 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 |
/** |
| 330 |
* Save metadata to post meta |
| 331 |
* |
| 332 |
* @param int $post_id Post ID |
| 333 |
* @param array $metadata Metadata to save |
| 334 |
* @return void |
| 335 |
*/ |
| 336 |
private function save_post_metadata(int $post_id, array $metadata): void { |
| 337 |
update_post_meta($post_id, '_thinkrank_ai_title', $metadata['title']); |
| 338 |
update_post_meta($post_id, '_thinkrank_ai_description', $metadata['description']); |
| 339 |
// Persist via Focus_Keywords so the array + legacy meta stay in sync. |
| 340 |
// Existing keywords are preserved (the AI value seeds only when unset). |
| 341 |
if (empty(\ThinkRank\SEO\Focus_Keywords::get($post_id))) { |
| 342 |
\ThinkRank\SEO\Focus_Keywords::save($post_id, $metadata['focus_keyword']); |
| 343 |
} |
| 344 |
update_post_meta($post_id, '_thinkrank_generated_at', $metadata['generated_at']); |
| 345 |
update_post_meta($post_id, '_thinkrank_metadata_full', $metadata); |
| 346 |
} |
| 347 |
|
| 348 |
/** |
| 349 |
* Calculate basic readability score |
| 350 |
* |
| 351 |
* @param string $content Content to analyze |
| 352 |
* @return float Readability score |
| 353 |
*/ |
| 354 |
private function calculate_readability_score(string $content): float { |
| 355 |
$sentences = preg_split('/[.!?]+/', $content); |
| 356 |
$words = str_word_count($content); |
| 357 |
$syllables = $this->count_syllables($content); |
| 358 |
|
| 359 |
if (count($sentences) === 0 || $words === 0) { |
| 360 |
return 0; |
| 361 |
} |
| 362 |
|
| 363 |
// Simplified Flesch Reading Ease formula |
| 364 |
$score = 206.835 - (1.015 * ($words / count($sentences))) - (84.6 * ($syllables / $words)); |
| 365 |
|
| 366 |
return max(0, min(100, $score)); |
| 367 |
} |
| 368 |
|
| 369 |
/** |
| 370 |
* Count syllables in text (simplified) |
| 371 |
* |
| 372 |
* @param string $text Text to analyze |
| 373 |
* @return int Syllable count |
| 374 |
*/ |
| 375 |
private function count_syllables(string $text): int { |
| 376 |
$words = str_word_count($text, 1); |
| 377 |
$syllables = 0; |
| 378 |
|
| 379 |
foreach ($words as $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); |
| 399 |
} |
| 400 |
|
| 401 |
return $syllables; |
| 402 |
} |
| 403 |
|
| 404 |
/** |
| 405 |
* Analyze keyword density |
| 406 |
* |
| 407 |
* @param string $content Content to analyze |
| 408 |
* @return array Keyword density analysis |
| 409 |
*/ |
| 410 |
private function analyze_keyword_density(string $content): array { |
| 411 |
$words = str_word_count(strtolower($content), 1); |
| 412 |
$total_words = count($words); |
| 413 |
$word_counts = array_count_values($words); |
| 414 |
|
| 415 |
// Remove common stop words |
| 416 |
$stop_words = ['the', 'and', 'or', 'but', 'in', 'on', 'at', 'to', 'for', 'of', 'with', 'by']; |
| 417 |
$word_counts = array_diff_key($word_counts, array_flip($stop_words)); |
| 418 |
|
| 419 |
// Calculate density for top keywords |
| 420 |
arsort($word_counts); |
| 421 |
$top_keywords = array_slice($word_counts, 0, 10, true); |
| 422 |
|
| 423 |
$density = []; |
| 424 |
foreach ($top_keywords as $word => $count) { |
| 425 |
$density[$word] = round(($count / $total_words) * 100, 2); |
| 426 |
} |
| 427 |
|
| 428 |
return $density; |
| 429 |
} |
| 430 |
|
| 431 |
/** |
| 432 |
* Analyze existing metadata |
| 433 |
* |
| 434 |
* @param array $metadata Metadata to analyze |
| 435 |
* @return array Analysis results |
| 436 |
*/ |
| 437 |
private function analyze_metadata(array $metadata): array { |
| 438 |
return [ |
| 439 |
'title_length_ok' => strlen($metadata['title'] ?? '') <= 60, |
| 440 |
'description_length_ok' => strlen($metadata['description'] ?? '') <= 160, |
| 441 |
'has_focus_keyword' => !empty($metadata['focus_keyword'] ?? ''), |
| 442 |
]; |
| 443 |
} |
| 444 |
|
| 445 |
/** |
| 446 |
* Generate improvement suggestions |
| 447 |
* |
| 448 |
* @param array $analysis Content analysis |
| 449 |
* @return array Suggestions |
| 450 |
*/ |
| 451 |
private function generate_improvement_suggestions(array $analysis): array { |
| 452 |
$suggestions = []; |
| 453 |
|
| 454 |
if ($analysis['content_length'] < 300) { |
| 455 |
$suggestions[] = 'Consider adding more content. Longer content typically performs better in search results.'; |
| 456 |
} |
| 457 |
|
| 458 |
if ($analysis['readability_score'] < 60) { |
| 459 |
$suggestions[] = 'Content readability could be improved. Try using shorter sentences and simpler words.'; |
| 460 |
} |
| 461 |
|
| 462 |
return $suggestions; |
| 463 |
} |
| 464 |
|
| 465 |
/** |
| 466 |
* Calculate optimization score |
| 467 |
* |
| 468 |
* @param array $metadata Generated metadata |
| 469 |
* @param array $options Generation options |
| 470 |
* @return int Optimization score (0-100) |
| 471 |
*/ |
| 472 |
private function calculate_optimization_score(array $metadata, array $options): int { |
| 473 |
$score = 0; |
| 474 |
|
| 475 |
// Title optimization (30 points) |
| 476 |
if (!empty($metadata['title'])) { |
| 477 |
$score += 15; |
| 478 |
if (strlen($metadata['title']) >= 30 && strlen($metadata['title']) <= 60) { |
| 479 |
$score += 15; |
| 480 |
} |
| 481 |
} |
| 482 |
|
| 483 |
// Description optimization (30 points) |
| 484 |
if (!empty($metadata['description'])) { |
| 485 |
$score += 15; |
| 486 |
if (strlen($metadata['description']) >= 120 && strlen($metadata['description']) <= 160) { |
| 487 |
$score += 15; |
| 488 |
} |
| 489 |
} |
| 490 |
|
| 491 |
// Keyword optimization (40 points) |
| 492 |
if (!empty($metadata['focus_keyword'])) { |
| 493 |
$score += 20; |
| 494 |
if (!empty($options['target_keyword']) && |
| 495 |
stripos($metadata['title'], $options['target_keyword']) !== false) { |
| 496 |
$score += 20; |
| 497 |
} |
| 498 |
} |
| 499 |
|
| 500 |
return min(100, $score); |
| 501 |
} |
| 502 |
} |
| 503 |
|
