| @@ -41,8 +41,41 @@ | ||
| 41 | 41 | */ |
| 42 | 42 | private Database $database; |
| 43 | 43 | |
| 44 | 44 | /** |
| 45 | + * Memoised collected performance measurement, and whether it was resolved. | |
| 46 | + * | |
| 47 | + * Two factors read it and both may be asked for on every post in a list, so | |
| 48 | + * the lookup happens once per calculator. `null` is a real answer here — the | |
| 49 | + * separate flag keeps "not looked up yet" distinct from "nothing measured". | |
| 50 | + * | |
| 51 | + * @since 2.3.1 | |
| 52 | + * @var array|null | |
| 53 | + */ | |
| 54 | + private ?array $measured_performance = null; | |
| 55 | + | |
| 56 | + /** | |
| 57 | + * @since 2.3.1 | |
| 58 | + * @var bool | |
| 59 | + */ | |
| 60 | + private bool $measured_performance_resolved = false; | |
| 61 | + | |
| 62 | + /** | |
| 63 | + * Length bands the editor scores against, in characters. | |
| 64 | + * | |
| 65 | + * Public so every surface that judges a title or description — the editor | |
| 66 | + * score and the Bulk Snippets problem filter — reads one set of numbers. | |
| 67 | + * Before these existed the bands were literals inside the scoring methods, | |
| 68 | + * and a second screen would have had to copy them and drift (#727). | |
| 69 | + * | |
| 70 | + * @since 2.8.0 | |
| 71 | + */ | |
| 72 | + public const TITLE_OPTIMAL_MIN = 35; | |
| 73 | + public const TITLE_OPTIMAL_MAX = 60; | |
| 74 | + public const DESCRIPTION_OPTIMAL_MIN = 120; | |
| 75 | + public const DESCRIPTION_OPTIMAL_MAX = 160; | |
| 76 | + | |
| 77 | + /** | |
| 45 | 78 | * 2025 SEO scoring factors (Q1 2025 Google Algorithm) |
| 46 | 79 | * Based on First Page Sage research and Google's latest updates |
| 47 | 80 | * |
| 48 | 81 | * @var array |
| @@ -72,10 +105,17 @@ | ||
| 72 | 105 | $this->database = $database; |
| 73 | 106 | } |
| 74 | 107 | |
| 75 | 108 | /** |
| 76 | - * Calculate comprehensive modern SEO score | |
| 77 | - * | |
| 109 | + * Calculate comprehensive modern SEO score. | |
| 110 | + * | |
| 111 | + * Supports multiple focus keywords: when `$options['target_keywords']` holds | |
| 112 | + * more than one keyword the score is computed independently for each and the | |
| 113 | + * HIGHEST overall score is returned as the final result, with per-keyword | |
| 114 | + * results (`keyword_results`) and OR-combined per-check matches | |
| 115 | + * (`keyword_checks`) attached. A single keyword (or the legacy | |
| 116 | + * `target_keyword` option) falls through to the single-keyword path. | |
| 117 | + * | |
| 78 | 118 | * @param array $content_data Content analysis data |
| 79 | 119 | * @param array $metadata Post metadata |
| 80 | 120 | * @param array $options Additional options |
| 81 | 121 | * @return array Complete scoring result |
| @@ -80,8 +120,492 @@ | ||
| 80 | 120 | * @param array $options Additional options |
| 81 | 121 | * @return array Complete scoring result |
| 82 | 122 | */ |
| 83 | 123 | public function calculate_score(array $content_data, array $metadata, array $options = []): array { |
| 124 | + $keywords = $this->resolve_target_keywords($options, $metadata); | |
| 125 | + | |
| 126 | + if (count($keywords) > 1) { | |
| 127 | + return $this->calculate_score_multi($content_data, $metadata, $keywords, $options); | |
| 128 | + } | |
| 129 | + | |
| 130 | + $options['target_keyword'] = $keywords[0] ?? ''; | |
| 131 | + $result = $this->compute_score($content_data, $metadata, $options); | |
| 132 | + | |
| 133 | + // Expose the keyword surface uniformly so consumers can rely on it | |
| 134 | + // regardless of how many keywords were supplied. | |
| 135 | + $result['target_keywords'] = $keywords; | |
| 136 | + if (!empty($keywords)) { | |
| 137 | + $result['keyword_results'] = [[ | |
| 138 | + 'keyword' => $keywords[0], | |
| 139 | + 'overall_score' => $result['overall_score'], | |
| 140 | + 'grade' => $result['grade'], | |
| 141 | + ]]; | |
| 142 | + $result['keyword_checks'] = $this->analyze_keyword_checks($content_data, $metadata, $keywords); | |
| 143 | + } | |
| 144 | + $result['keywords'] = $this->keyword_placements($content_data, $metadata, $keywords); | |
| 145 | + | |
| 146 | + return $result; | |
| 147 | + } | |
| 148 | + | |
| 149 | + /** | |
| 150 | + * Resolve the target keyword list from the options array, falling back to | |
| 151 | + * the focus keywords carried on the post's metadata. | |
| 152 | + * | |
| 153 | + * Accepts `target_keywords` (array) or the legacy `target_keyword` (string), | |
| 154 | + * trims, drops empties and removes case-insensitive duplicates. | |
| 155 | + * | |
| 156 | + * Options win over metadata on purpose: the editor scores unsaved keyword | |
| 157 | + * edits by passing them explicitly, and that live value must beat whatever | |
| 158 | + * is currently persisted. The metadata fallback applies only when the | |
| 159 | + * caller mentions no keyword option AT ALL — a caller that passes an empty | |
| 160 | + * keyword is deliberately clearing it (the editor does exactly this when | |
| 161 | + * the field is emptied), so the stored value must not resurrect it. | |
| 162 | + * | |
| 163 | + * Without the fallback, every caller that hands over | |
| 164 | + * `Metabox_Manager::get_post_metadata()` (the metabox and the MCP scoring | |
| 165 | + * abilities) silently scored as if no keyword were set. | |
| 166 | + * | |
| 167 | + * @param array $options Scoring options. | |
| 168 | + * @param array $metadata Post metadata (may carry focus_keyword(s)). | |
| 169 | + * @return string[] Normalized keyword list. | |
| 170 | + */ | |
| 171 | + private function resolve_target_keywords(array $options, array $metadata = []): array { | |
| 172 | + $raw = []; | |
| 173 | + if (!empty($options['target_keywords']) && is_array($options['target_keywords'])) { | |
| 174 | + $raw = $options['target_keywords']; | |
| 175 | + } elseif (isset($options['target_keyword']) && $options['target_keyword'] !== '') { | |
| 176 | + $raw = [$options['target_keyword']]; | |
| 177 | + } elseif (!$this->options_mention_keywords($options)) { | |
| 178 | + if (!empty($metadata['focus_keywords']) && is_array($metadata['focus_keywords'])) { | |
| 179 | + $raw = $metadata['focus_keywords']; | |
| 180 | + } elseif (isset($metadata['focus_keyword']) && is_string($metadata['focus_keyword']) && $metadata['focus_keyword'] !== '') { | |
| 181 | + $raw = [$metadata['focus_keyword']]; | |
| 182 | + } | |
| 183 | + } | |
| 184 | + | |
| 185 | + $seen = []; | |
| 186 | + $keywords = []; | |
| 187 | + foreach ($raw as $keyword) { | |
| 188 | + $keyword = trim((string) $keyword); | |
| 189 | + if ($keyword === '') { | |
| 190 | + continue; | |
| 191 | + } | |
| 192 | + $key = strtolower($keyword); | |
| 193 | + if (isset($seen[$key])) { | |
| 194 | + continue; | |
| 195 | + } | |
| 196 | + $seen[$key] = true; | |
| 197 | + $keywords[] = $keyword; | |
| 198 | + } | |
| 199 | + | |
| 200 | + return $keywords; | |
| 201 | + } | |
| 202 | + | |
| 203 | + /** | |
| 204 | + * Whether the caller said anything about keywords — including saying | |
| 205 | + * "none". Distinguishes an intentional clear (score with no keyword) from | |
| 206 | + * silence (fall back to the post's stored focus keywords). | |
| 207 | + * | |
| 208 | + * @param array $options Scoring options. | |
| 209 | + * @return bool True when a keyword option key is present. | |
| 210 | + */ | |
| 211 | + private function options_mention_keywords(array $options): bool { | |
| 212 | + return array_key_exists('target_keywords', $options) | |
| 213 | + || array_key_exists('target_keyword', $options); | |
| 214 | + } | |
| 215 | + | |
| 216 | + /** | |
| 217 | + * Score each keyword independently and return the highest-scoring result. | |
| 218 | + * | |
| 219 | + * @param array $content_data Content analysis data. | |
| 220 | + * @param array $metadata Post metadata. | |
| 221 | + * @param string[] $keywords Target keywords (already normalized, 2+). | |
| 222 | + * @param array $options Additional options. | |
| 223 | + * @return array Best scoring result, with per-keyword data attached. | |
| 224 | + */ | |
| 225 | + private function calculate_score_multi(array $content_data, array $metadata, array $keywords, array $options): array { | |
| 226 | + $per_keyword = []; | |
| 227 | + $best = null; | |
| 228 | + $best_keyword = $keywords[0]; | |
| 229 | + | |
| 230 | + foreach ($keywords as $keyword) { | |
| 231 | + $opts = $options; | |
| 232 | + unset($opts['target_keywords']); | |
| 233 | + $opts['target_keyword'] = $keyword; | |
| 234 | + | |
| 235 | + $result = $this->compute_score($content_data, $metadata, $opts); | |
| 236 | + | |
| 237 | + $per_keyword[] = [ | |
| 238 | + 'keyword' => $keyword, | |
| 239 | + 'overall_score' => $result['overall_score'], | |
| 240 | + 'grade' => $result['grade'], | |
| 241 | + 'score_breakdown' => $result['score_breakdown'], | |
| 242 | + ]; | |
| 243 | + | |
| 244 | + if ($best === null || $result['overall_score'] > $best['overall_score']) { | |
| 245 | + $best = $result; | |
| 246 | + $best_keyword = $keyword; | |
| 247 | + } | |
| 248 | + } | |
| 249 | + | |
| 250 | + // Final score = highest individual keyword score. Retain per-keyword | |
| 251 | + // results and OR-combined checks so the UI can show both. | |
| 252 | + $best['target_keyword'] = $best_keyword; | |
| 253 | + $best['target_keywords'] = $keywords; | |
| 254 | + $best['keyword_results'] = $per_keyword; | |
| 255 | + $best['keyword_checks'] = $this->analyze_keyword_checks($content_data, $metadata, $keywords); | |
| 256 | + $best['keywords'] = $this->keyword_placements($content_data, $metadata, $keywords); | |
| 257 | + | |
| 258 | + return $best; | |
| 259 | + } | |
| 260 | + | |
| 261 | + /** | |
| 262 | + * Evaluate per-location keyword checks across ALL focus keywords. | |
| 263 | + * | |
| 264 | + * Each check (title, meta description, content, image alt, slug) passes when | |
| 265 | + * ANY of the focus keywords matches that location. | |
| 266 | + * | |
| 267 | + * @param array $content_data Content analysis data. | |
| 268 | + * @param array $metadata Post metadata. | |
| 269 | + * @param string[] $keywords Target keywords. | |
| 270 | + * @return array<string,array{passed:bool,matched_keywords:string[]}> | |
| 271 | + */ | |
| 272 | + private function analyze_keyword_checks(array $content_data, array $metadata, array $keywords): array { | |
| 273 | + $title = self::lower((string) ($metadata['title'] ?? $content_data['title'] ?? '')); | |
| 274 | + $description = self::lower((string) ($metadata['description'] ?? '')); | |
| 275 | + $content = self::lower(self::plain_text((string) ($content_data['content'] ?? ''))); | |
| 276 | + | |
| 277 | + $alts = ''; | |
| 278 | + foreach ((array) ($content_data['images'] ?? []) as $image) { | |
| 279 | + $alts .= ' ' . self::lower((string) ($image['alt'] ?? '')); | |
| 280 | + } | |
| 281 | + | |
| 282 | + // Build a searchable slug haystack from the post's OWN slug — never the | |
| 283 | + // full URL path. The path carries ancestors, category bases and date | |
| 284 | + // segments, so a child of /clinical-trials/ reported "keyword in slug" | |
| 285 | + // for a page actually slugged `contact-us`. It also breaks the other | |
| 286 | + // way: an unpublished post has no pretty permalink (get_permalink() | |
| 287 | + // returns ?p=123), so the path held no slug at all and every draft | |
| 288 | + // scored "no match" until it was published. Hyphens/underscores become | |
| 289 | + // spaces so multi-word keywords can match. | |
| 290 | + $slug = self::lower(self::slug_haystack($content_data)); | |
| 291 | + | |
| 292 | + $haystacks = [ | |
| 293 | + 'title' => trim($title), | |
| 294 | + 'meta_description' => trim($description), | |
| 295 | + 'content' => trim($content), | |
| 296 | + 'image_alt' => trim($alts), | |
| 297 | + 'slug' => trim($slug), | |
| 298 | + ]; | |
| 299 | + | |
| 300 | + $checks = []; | |
| 301 | + foreach ($haystacks as $location => $haystack) { | |
| 302 | + $matched = []; | |
| 303 | + foreach ($keywords as $keyword) { | |
| 304 | + if ($this->keyword_matches($haystack, self::lower(trim($keyword)))) { | |
| 305 | + $matched[] = $keyword; | |
| 306 | + } | |
| 307 | + } | |
| 308 | + $checks[$location] = [ | |
| 309 | + 'passed' => !empty($matched), | |
| 310 | + 'matched_keywords' => $matched, | |
| 311 | + ]; | |
| 312 | + } | |
| 313 | + | |
| 314 | + return $checks; | |
| 315 | + } | |
| 316 | + | |
| 317 | + /** | |
| 318 | + * The keyword placements the editor draws one gauge segment for, in the | |
| 319 | + * order a reader meets them (#729). | |
| 320 | + * | |
| 321 | + * @since 2.11.0 | |
| 322 | + * @var string[] | |
| 323 | + */ | |
| 324 | + public const PLACEMENTS = ['title', 'meta_description', 'slug', 'first_paragraph', 'subheading', 'content', 'image_alt']; | |
| 325 | + | |
| 326 | + /** | |
| 327 | + * Characters of plain text read as the opening when the content has no | |
| 328 | + * paragraph tag. | |
| 329 | + * | |
| 330 | + * @since 2.11.0 | |
| 331 | + */ | |
| 332 | + private const OPENING_CHARS = 300; | |
| 333 | + | |
| 334 | + /** | |
| 335 | + * Where each focus keyword is placed, keyword by keyword (#729). | |
| 336 | + * | |
| 337 | + * analyze_keyword_checks() answers "does ANY keyword appear here" for | |
| 338 | + * five places; this answers "where does THIS keyword appear" for seven, | |
| 339 | + * so the editor can show each keyword's own gauge. Same matcher, so a | |
| 340 | + * keyword counts as a word (not a fragment) and a keyword in a script | |
| 341 | + * written without spaces (Thai, Chinese, Japanese) still matches. | |
| 342 | + * | |
| 343 | + * `where` names the heading or alt text that matched, so the editor can | |
| 344 | + * say which one. | |
| 345 | + * | |
| 346 | + * @since 2.11.0 | |
| 347 | + * | |
| 348 | + * @param array $content_data Content analysis data. | |
| 349 | + * @param array $metadata Post metadata (title, description). | |
| 350 | + * @param string[] $keywords Focus keywords. | |
| 351 | + * @return array<int, array{keyword: string, passed: int, total: int, placements: array<string, array{passed: bool, where: string}>}> | |
| 352 | + */ | |
| 353 | + public function keyword_placements(array $content_data, array $metadata, array $keywords): array { | |
| 354 | + $html = (string) ($content_data['content'] ?? ''); | |
| 355 | + $plain = self::lower(self::plain_text($html)); | |
| 356 | + | |
| 357 | + $headings = []; | |
| 358 | + $source = isset($content_data['headings']) && is_array($content_data['headings']) ? $content_data['headings'] : $this->extract_headings($html); | |
| 359 | + foreach ($source as $heading) { | |
| 360 | + $text = trim((string) ($heading['text'] ?? '')); | |
| 361 | + if ((int) ($heading['level'] ?? 0) >= 2 && '' !== $text) { | |
| 362 | + $headings[] = $text; | |
| 363 | + } | |
| 364 | + } | |
| 365 | + | |
| 366 | + $alts = []; | |
| 367 | + foreach ((array) ($content_data['images'] ?? []) as $image) { | |
| 368 | + $alt = trim((string) ($image['alt'] ?? '')); | |
| 369 | + if ('' !== $alt) { | |
| 370 | + $alts[] = $alt; | |
| 371 | + } | |
| 372 | + } | |
| 373 | + | |
| 374 | + $single = [ | |
| 375 | + 'title' => self::lower((string) ($metadata['title'] ?? $content_data['title'] ?? '')), | |
| 376 | + 'meta_description' => self::lower((string) ($metadata['description'] ?? '')), | |
| 377 | + 'slug' => self::lower(self::slug_haystack($content_data)), | |
| 378 | + 'first_paragraph' => self::lower(self::opening($html)), | |
| 379 | + 'content' => $plain, | |
| 380 | + ]; | |
| 381 | + | |
| 382 | + $out = []; | |
| 383 | + foreach ($keywords as $keyword) { | |
| 384 | + $needle = self::lower(trim((string) $keyword)); | |
| 385 | + $placements = []; | |
| 386 | + | |
| 387 | + foreach (self::PLACEMENTS as $placement) { | |
| 388 | + if ('subheading' === $placement || 'image_alt' === $placement) { | |
| 389 | + $where = ''; | |
| 390 | + foreach ('subheading' === $placement ? $headings : $alts as $text) { | |
| 391 | + if ($this->keyword_matches(self::lower($text), $needle)) { | |
| 392 | + $where = $text; | |
| 393 | + break; | |
| 394 | + } | |
| 395 | + } | |
| 396 | + $placements[$placement] = ['passed' => '' !== $where, 'where' => $where]; | |
| 397 | + continue; | |
| 398 | + } | |
| 399 | + | |
| 400 | + $placements[$placement] = ['passed' => $this->keyword_matches($single[$placement], $needle), 'where' => '']; | |
| 401 | + } | |
| 402 | + | |
| 403 | + $out[] = [ | |
| 404 | + 'keyword' => (string) $keyword, | |
| 405 | + 'passed' => count(array_filter(array_column($placements, 'passed'))), | |
| 406 | + 'total' => count(self::PLACEMENTS), | |
| 407 | + 'placements' => $placements, | |
| 408 | + ]; | |
| 409 | + } | |
| 410 | + | |
| 411 | + return $out; | |
| 412 | + } | |
| 413 | + | |
| 414 | + /** | |
| 415 | + * The post's own slug as searchable text: hyphens and underscores become | |
| 416 | + * spaces so a multi-word keyword can match, and a slug WordPress | |
| 417 | + * percent-encoded (Thai, Cyrillic, Chinese) is decoded, or it could never | |
| 418 | + * match a keyword typed in that script. | |
| 419 | + * | |
| 420 | + * Never the full URL path: the path carries ancestors, category bases and | |
| 421 | + * date segments, so a child of /clinical-trials/ reported "keyword in slug" | |
| 422 | + * for a page actually slugged `contact-us`. An unpublished post has no | |
| 423 | + * pretty permalink either, so the path held no slug at all. | |
| 424 | + * | |
| 425 | + * @param array $content_data Content analysis data. | |
| 426 | + * @return string | |
| 427 | + */ | |
| 428 | + private static function slug_haystack(array $content_data): string { | |
| 429 | + $slug = (string) ($content_data['slug'] ?? ''); | |
| 430 | + if ('' === $slug) { | |
| 431 | + // Draft with no slug assigned yet: score what WordPress would | |
| 432 | + // generate from the title, which is what the editor shows as the | |
| 433 | + // proposed URL — so the check reads the same before and after | |
| 434 | + // publishing instead of flipping. | |
| 435 | + $slug = sanitize_title((string) ($content_data['title'] ?? '')); | |
| 436 | + } | |
| 437 | + | |
| 438 | + return trim(str_replace(['-', '_'], ' ', rawurldecode($slug))); | |
| 439 | + } | |
| 440 | + | |
| 441 | + /** | |
| 442 | + * The opening of the content: its first paragraph with text, or the first | |
| 443 | + * few hundred characters when it has none. Counted in characters, not | |
| 444 | + * words, so a language written without spaces is not read as one word. | |
| 445 | + * | |
| 446 | + * @param string $html Content HTML. | |
| 447 | + * @return string Plain text. | |
| 448 | + */ | |
| 449 | + private static function opening(string $html): string { | |
| 450 | + if (preg_match_all('/<p\b[^>]*>(.*?)<\/p>/isu', $html, $matches)) { | |
| 451 | + foreach ($matches[1] as $paragraph) { | |
| 452 | + $text = self::collapse_whitespace(wp_strip_all_tags($paragraph)); | |
| 453 | + if ('' !== $text) { | |
| 454 | + return $text; | |
| 455 | + } | |
| 456 | + } | |
| 457 | + } | |
| 458 | + | |
| 459 | + $text = self::plain_text($html); | |
| 460 | + | |
| 461 | + return function_exists('mb_substr') ? mb_substr($text, 0, self::OPENING_CHARS) : substr($text, 0, self::OPENING_CHARS); | |
| 462 | + } | |
| 463 | + | |
| 464 | + /** | |
| 465 | + * Content as plain text, with a space where each tag was. wp_strip_all_tags() | |
| 466 | + * alone joins neighbouring blocks — "…coffee grinder</h3><p>A good…" became | |
| 467 | + * "coffee grinderA good" — so a keyword at the end of a heading or a | |
| 468 | + * paragraph was no longer a word and did not match. | |
| 469 | + * | |
| 470 | + * @param string $html Content HTML. | |
| 471 | + * @return string | |
| 472 | + */ | |
| 473 | + private static function plain_text(string $html): string { | |
| 474 | + $spaced = preg_replace('/<[^>]+>/', ' $0 ', $html); | |
| 475 | + | |
| 476 | + return self::collapse_whitespace(wp_strip_all_tags(null === $spaced ? $html : (string) $spaced)); | |
| 477 | + } | |
| 478 | + | |
| 479 | + /** | |
| 480 | + * Runs of whitespace down to one space. | |
| 481 | + * | |
| 482 | + * The `/u` pass is the one that understands a multibyte space, but | |
| 483 | + * preg_replace() answers null on bytes that are not valid UTF-8 rather than | |
| 484 | + * throwing — and casting that null to a string blanked the haystack, so a | |
| 485 | + * post carrying one mojibake byte (a Latin-1 paste, an old import) reported | |
| 486 | + * every keyword as missing from its body, its opening, and every | |
| 487 | + * subheading. The gauge said 0/7 and told the author to add a keyword that | |
| 488 | + * was already there. | |
| 489 | + * | |
| 490 | + * Falls back to the byte-wise collapse, which is what this did before the | |
| 491 | + * multibyte work added the modifier. Same reasoning keyword_matches() | |
| 492 | + * already records for its own PCRE failure: a pattern PCRE refuses must not | |
| 493 | + * be reported as a confident "no match". | |
| 494 | + * | |
| 495 | + * @param string $text Text to collapse. | |
| 496 | + * @return string | |
| 497 | + */ | |
| 498 | + private static function collapse_whitespace(string $text): string { | |
| 499 | + $collapsed = preg_replace('/\s+/u', ' ', $text); | |
| 500 | + | |
| 501 | + if (null === $collapsed) { | |
| 502 | + $collapsed = preg_replace('/\s+/', ' ', $text); | |
| 503 | + } | |
| 504 | + | |
| 505 | + return trim(null === $collapsed ? $text : (string) $collapsed); | |
| 506 | + } | |
| 507 | + | |
| 508 | + /** | |
| 509 | + * Lowercase in any script. strtolower() only folds ASCII, so "Кофе" never | |
| 510 | + * matched "кофе". | |
| 511 | + * | |
| 512 | + * @param string $text Text. | |
| 513 | + * @return string | |
| 514 | + */ | |
| 515 | + private static function lower(string $text): string { | |
| 516 | + return function_exists('mb_strtolower') ? mb_strtolower($text, 'UTF-8') : strtolower($text); | |
| 517 | + } | |
| 518 | + | |
| 519 | + /** | |
| 520 | + * Scripts written without spaces between words. | |
| 521 | + * | |
| 522 | + * @since 2.1.0 | |
| 523 | + * @var string | |
| 524 | + */ | |
| 525 | + private const SCRIPTIO_CONTINUA = '/[\p{Han}\p{Hiragana}\p{Katakana}\p{Thai}\p{Lao}\p{Khmer}\p{Myanmar}]/u'; | |
| 526 | + | |
| 527 | + /** | |
| 528 | + * Whether a keyword appears in a haystack as a word rather than as a | |
| 529 | + * fragment of a longer one. | |
| 530 | + * | |
| 531 | + * The five keyword checks used a plain strpos(), so any substring hit | |
| 532 | + * counted: "test coronavirus" matched "la|test coronavirus|news", "art" | |
| 533 | + * matched "start", "cat" matched "category". The panel then confidently | |
| 534 | + * reported a keyword placement that does not exist (#416). Same class of | |
| 535 | + * problem #71 fixed in the Image SEO rewriter, and the same remedy. | |
| 536 | + * | |
| 537 | + * Both arguments are expected lowercased already. | |
| 538 | + * | |
| 539 | + * @since 2.1.0 | |
| 540 | + * | |
| 541 | + * @param string $haystack Text to search. | |
| 542 | + * @param string $needle Keyword, lowercased and trimmed. | |
| 543 | + * @return bool | |
| 544 | + */ | |
| 545 | + private function keyword_matches(string $haystack, string $needle): bool { | |
| 546 | + if ($needle === '' || $haystack === '') { | |
| 547 | + return false; | |
| 548 | + } | |
| 549 | + | |
| 550 | + if (!$this->supports_word_boundaries($needle)) { | |
| 551 | + return strpos($haystack, $needle) !== false; | |
| 552 | + } | |
| 553 | + | |
| 554 | + $matched = preg_match('/\b' . preg_quote($needle, '/') . '\b/u', $haystack); | |
| 555 | + | |
| 556 | + // PCRE refusing the pattern — invalid UTF-8 in the keyword, a | |
| 557 | + // backtrack limit — must not be reported as a confident "no match". | |
| 558 | + // Fall back to the behaviour this replaced rather than invent a | |
| 559 | + // negative the user cannot explain. | |
| 560 | + if ($matched === false) { | |
| 561 | + return strpos($haystack, $needle) !== false; | |
| 562 | + } | |
| 563 | + | |
| 564 | + return $matched === 1; | |
| 565 | + } | |
| 566 | + | |
| 567 | + /** | |
| 568 | + * Whether \b can express "this keyword, as a word" for this keyword. | |
| 569 | + * | |
| 570 | + * It asserts a transition between a word and a non-word character, which | |
| 571 | + * only means something where words are separated. Two cases where it is | |
| 572 | + * not, both verified against PCRE rather than assumed: | |
| 573 | + * | |
| 574 | + * - the keyword's own edges are not word characters ("c++", "#seo"), so | |
| 575 | + * no boundary can assert there and a real match is lost; | |
| 576 | + * - scripts written without spaces, where the neighbouring characters | |
| 577 | + * are word characters too — "冠状病毒" inside "最新冠状病毒新闻" is a | |
| 578 | + * legitimate match that \b never sees. | |
| 579 | + * | |
| 580 | + * Accented Latin and Cyrillic need no special handling: PHP's /u modifier | |
| 581 | + * turns on Unicode character properties, so "café" correctly does not | |
| 582 | + * match "cafés" and "коронавирус" does not match "коронавирусный". | |
| 583 | + * | |
| 584 | + * @since 2.1.0 | |
| 585 | + * | |
| 586 | + * @param string $needle Keyword, lowercased and trimmed. | |
| 587 | + * @return bool | |
| 588 | + */ | |
| 589 | + private function supports_word_boundaries(string $needle): bool { | |
| 590 | + if (preg_match(self::SCRIPTIO_CONTINUA, $needle)) { | |
| 591 | + return false; | |
| 592 | + } | |
| 593 | + | |
| 594 | + return preg_match('/^\w/u', $needle) === 1 && preg_match('/\w$/u', $needle) === 1; | |
| 595 | + } | |
| 596 | + | |
| 597 | + /** | |
| 598 | + * Compute the SEO score for a single target keyword. | |
| 599 | + * | |
| 600 | + * @param array $content_data Content analysis data | |
| 601 | + * @param array $metadata Post metadata | |
| 602 | + * @param array $options Additional options (expects scalar target_keyword) | |
| 603 | + * @return array Complete scoring result | |
| 604 | + * | |
| 605 | + * @throws \Exception On failure. | |
| 606 | + */ | |
| 607 | + private function compute_score(array $content_data, array $metadata, array $options = []): array { | |
| 84 | 608 | $scores = []; |
| 85 | 609 | $suggestions = []; |
| 86 | 610 | $total_score = 0; |
| 87 | 611 | |
| @@ -86,9 +610,9 @@ | ||
| 86 | 610 | $total_score = 0; |
| 87 | 611 | |
| 88 | 612 | try { |
| 89 | 613 | // 1. Satisfying Content (23 points) - #1 factor in 2025 |
| 90 | - $satisfying_result = $this->score_satisfying_content($content_data, $options['target_keyword'] ?? ''); | |
| 614 | + $satisfying_result = $this->score_satisfying_content($content_data, $options['target_keyword'] ?? '', $metadata); | |
| 91 | 615 | $scores['satisfying_content'] = $satisfying_result; |
| 92 | 616 | $total_score += $satisfying_result['score']; |
| 93 | 617 | $suggestions = array_merge($suggestions, $satisfying_result['suggestions']); |
| 94 | 618 | } catch (\Exception $e) { |
| @@ -183,9 +707,9 @@ | ||
| 183 | 707 | $total_score += $technical_result['score']; |
| 184 | 708 | $suggestions = array_merge($suggestions, $technical_result['suggestions']); |
| 185 | 709 | |
| 186 | 710 | try { |
| 187 | - $prioritized_suggestions = $this->prioritize_suggestions($suggestions); | |
| 711 | + $prioritized_suggestions = $this->prioritize_suggestions($suggestions, $scores); | |
| 188 | 712 | $grade = $this->get_grade_from_score($total_score); |
| 189 | 713 | |
| 190 | 714 | return [ |
| 191 | 715 | 'overall_score' => min(100, $total_score), |
| @@ -191,10 +715,16 @@ | ||
| 191 | 715 | 'overall_score' => min(100, $total_score), |
| 192 | 716 | 'score_breakdown' => $scores, |
| 193 | 717 | 'suggestions' => $prioritized_suggestions, |
| 194 | 718 | 'grade' => $grade, |
| 719 | + // Readability + content-quality labels so persisted scores | |
| 720 | + // (e.g. bulk-analyzed on import) populate the post-list columns | |
| 721 | + // without a manual re-analyze. The REST endpoint still overrides | |
| 722 | + // these with live-editor values when the metabox provides them. | |
| 723 | + 'readability_score' => $this->format_readability_label($content_data), | |
| 724 | + 'content_quality' => $this->derive_content_quality($content_data), | |
| 195 | 725 | 'calculated_at' => current_time('mysql'), |
| 196 | - 'algorithm_version' => '2025.1', | |
| 726 | + 'algorithm_version' => '2025.2', | |
| 197 | 727 | 'algorithm_source' => 'First Page Sage Q1 2025 Research', |
| 198 | 728 | 'factors_count' => count($scores), |
| 199 | 729 | ]; |
| 200 | 730 | } catch (\Exception $e) { |
| @@ -207,11 +737,12 @@ | ||
| 207 | 737 | * Google tests content to see if it satisfies search intent |
| 208 | 738 | * |
| 209 | 739 | * @param array $content_data Content analysis data |
| 210 | 740 | * @param string $target_keyword Target keyword |
| 741 | + * @param array $metadata Post metadata (title, description) | |
| 211 | 742 | * @return array Scoring result |
| 212 | 743 | */ |
| 213 | - private function score_satisfying_content(array $content_data, string $target_keyword): array { | |
| 744 | + private function score_satisfying_content(array $content_data, string $target_keyword, array $metadata = []): array { | |
| 214 | 745 | $score = 0; |
| 215 | 746 | $max_score = $this->scoring_factors['satisfying_content']; |
| 216 | 747 | $suggestions = []; |
| 217 | 748 | |
| @@ -216,37 +747,63 @@ | ||
| 216 | 747 | $suggestions = []; |
| 217 | 748 | |
| 218 | 749 | $content = $content_data['content'] ?? ''; |
| 219 | 750 | $word_count = $content_data['word_count'] ?? 0; |
| 751 | + $meta_description = (string) ($metadata['description'] ?? ''); | |
| 220 | 752 | |
| 221 | - // Content depth and comprehensiveness (8 points) | |
| 222 | - if ($word_count >= 2000) { | |
| 753 | + // Content depth and comprehensiveness (8 points). Tiers softened so a | |
| 754 | + // genuinely useful post is not capped the way the old 2000-word gate did | |
| 755 | + // (Rank Math awards full content credit well below 2000 words). | |
| 756 | + if ($word_count >= 1500) { | |
| 223 | 757 | $score += 8; |
| 224 | - } elseif ($word_count >= 1200) { | |
| 225 | - $score += 6; | |
| 226 | - $suggestions[] = 'Consider expanding content to 2000+ words for comprehensive topic coverage'; | |
| 227 | - } elseif ($word_count >= 800) { | |
| 228 | - $score += 4; | |
| 229 | - $suggestions[] = 'Content needs more depth - aim for 1200+ words for better satisfaction signals'; | |
| 758 | + } elseif ($word_count >= 1000) { | |
| 759 | + $score += 7; | |
| 760 | + $suggestions[] = 'Consider expanding content to 1500+ words for more comprehensive coverage'; | |
| 761 | + } elseif ($word_count >= 600) { | |
| 762 | + $score += 5; | |
| 763 | + $suggestions[] = 'Content is adequate - aim for 1000+ words for stronger topic depth'; | |
| 764 | + } elseif ($word_count >= 300) { | |
| 765 | + $score += 3; | |
| 766 | + $suggestions[] = 'Content is thin - aim for 600+ words minimum'; | |
| 230 | 767 | } else { |
| 231 | - $score += 1; | |
| 232 | - $suggestions[] = 'Content too shallow - Google prioritizes comprehensive, satisfying content (800+ words minimum)'; | |
| 768 | + $score++; | |
| 769 | + $suggestions[] = 'Content too shallow - Google prioritizes comprehensive, satisfying content'; | |
| 233 | 770 | } |
| 234 | 771 | |
| 235 | - // Search intent satisfaction (8 points) | |
| 236 | - $intent_satisfaction = $this->calculate_intent_satisfaction($content, $target_keyword); | |
| 237 | - $score += round($intent_satisfaction * 8); | |
| 238 | - | |
| 239 | - if ($intent_satisfaction < 0.7) { | |
| 240 | - $suggestions[] = 'Improve content to better satisfy search intent - focus on answering user questions completely'; | |
| 772 | + // Keyword presence & placement (8 points) - deterministic, replaces the | |
| 773 | + // old literal-phrase intent heuristic ("what is"/"because"). Measures | |
| 774 | + // signals the editor actually controls: body (3), first paragraph (3), | |
| 775 | + // meta description (2) - the last mirrors Rank Math's "keyword in meta | |
| 776 | + // description" basic-SEO check. | |
| 777 | + if (empty($target_keyword)) { | |
| 778 | + $score += 4; // Benefit of the doubt when no focus keyword is set. | |
| 779 | + $suggestions[] = 'Set a focus keyword so content relevance can be measured'; | |
| 780 | + } else { | |
| 781 | + if ($this->keyword_in_content($content, $target_keyword)) { | |
| 782 | + $score += 3; | |
| 783 | + } else { | |
| 784 | + $suggestions[] = "Use the focus keyword '{$target_keyword}' in the body content"; | |
| 785 | + } | |
| 786 | + if ($this->keyword_in_first_paragraph($content, $target_keyword)) { | |
| 787 | + $score += 3; | |
| 788 | + } else { | |
| 789 | + $suggestions[] = "Mention '{$target_keyword}' near the start of the content (first paragraph)"; | |
| 790 | + } | |
| 791 | + if ($this->keyword_in_meta($meta_description, $target_keyword)) { | |
| 792 | + $score += 2; | |
| 793 | + } else { | |
| 794 | + $suggestions[] = "Include the focus keyword '{$target_keyword}' in the meta description"; | |
| 795 | + } | |
| 241 | 796 | } |
| 242 | 797 | |
| 243 | - // Content uniqueness and value (7 points) | |
| 244 | - $uniqueness_score = $this->assess_content_uniqueness($content); | |
| 245 | - $score += round($uniqueness_score * 7); | |
| 798 | + // Content structure & value (7 points) - reuses the deterministic | |
| 799 | + // content-quality signal (length, paragraph length, subheading | |
| 800 | + // distribution) instead of the noisy sentence-length variety heuristic. | |
| 801 | + $quality = $this->derive_content_quality_score($content_data); // 0-100 | |
| 802 | + $score += (int) round(($quality / 100) * 7); | |
| 246 | 803 | |
| 247 | - if ($uniqueness_score < 0.6) { | |
| 248 | - $suggestions[] = 'Add unique insights, examples, or perspectives to stand out from competitors'; | |
| 804 | + if ($quality < 60) { | |
| 805 | + $suggestions[] = 'Improve content structure - break up long paragraphs and add subheadings'; | |
| 249 | 806 | } |
| 250 | 807 | |
| 251 | 808 | return [ |
| 252 | 809 | 'score' => $score, |
| @@ -253,10 +810,12 @@ | ||
| 253 | 810 | 'max_score' => $max_score, |
| 254 | 811 | 'suggestions' => $suggestions, |
| 255 | 812 | 'details' => [ |
| 256 | 813 | 'word_count' => $word_count, |
| 257 | - 'intent_satisfaction' => round($intent_satisfaction * 100, 1), | |
| 258 | - 'uniqueness_score' => round($uniqueness_score * 100, 1), | |
| 814 | + 'keyword_in_content' => !empty($target_keyword) && $this->keyword_in_content($content, $target_keyword), | |
| 815 | + 'keyword_in_first_paragraph' => !empty($target_keyword) && $this->keyword_in_first_paragraph($content, $target_keyword), | |
| 816 | + 'keyword_in_meta_description' => !empty($target_keyword) && $this->keyword_in_meta($meta_description, $target_keyword), | |
| 817 | + 'content_quality_score' => $quality, | |
| 259 | 818 | 'content_depth' => $this->assess_content_depth_2025($word_count), |
| 260 | 819 | ] |
| 261 | 820 | ]; |
| 262 | 821 | } |
| @@ -261,105 +820,22 @@ | ||
| 261 | 820 | ]; |
| 262 | 821 | } |
| 263 | 822 | |
| 264 | 823 | /** |
| 265 | - * Calculate intent satisfaction score | |
| 824 | + * Assess content depth for 2025 standards | |
| 266 | 825 | * |
| 267 | - * @param string $content Content text | |
| 268 | - * @param string $target_keyword Target keyword | |
| 269 | - * @return float Satisfaction score (0-1) | |
| 826 | + * @param int $word_count Word count | |
| 827 | + * @return string Depth assessment | |
| 270 | 828 | */ |
| 271 | - private function calculate_intent_satisfaction(string $content, string $target_keyword): float { | |
| 272 | - if (empty($content) || empty($target_keyword)) { | |
| 273 | - return 0.3; // Partial credit for content without keyword focus | |
| 829 | + private function assess_content_depth_2025(int $word_count): string { | |
| 830 | + if ($word_count >= 3000) { return 'Comprehensive'; | |
| 274 | 831 | } |
| 275 | - | |
| 276 | - $content_lower = strtolower(wp_strip_all_tags($content)); | |
| 277 | - $keyword_lower = strtolower($target_keyword); | |
| 278 | - | |
| 279 | - // Check for question answering patterns | |
| 280 | - $question_patterns = ['what is', 'how to', 'why', 'when', 'where', 'who']; | |
| 281 | - $answer_patterns = ['because', 'therefore', 'as a result', 'in conclusion', 'to summarize']; | |
| 282 | - | |
| 283 | - $question_score = 0; | |
| 284 | - $answer_score = 0; | |
| 285 | - | |
| 286 | - foreach ($question_patterns as $pattern) { | |
| 287 | - if (strpos($content_lower, $pattern) !== false) { | |
| 288 | - $question_score += 0.1; | |
| 289 | - } | |
| 832 | + if ($word_count >= 2000) { return 'Detailed'; | |
| 290 | 833 | } |
| 291 | - | |
| 292 | - foreach ($answer_patterns as $pattern) { | |
| 293 | - if (strpos($content_lower, $pattern) !== false) { | |
| 294 | - $answer_score += 0.1; | |
| 295 | - } | |
| 834 | + if ($word_count >= 1200) { return 'Adequate'; | |
| 296 | 835 | } |
| 297 | - | |
| 298 | - // Base satisfaction from keyword relevance | |
| 299 | - $keyword_relevance = $this->calculate_topic_relevance($content, $target_keyword); | |
| 300 | - | |
| 301 | - // Combine factors | |
| 302 | - $satisfaction = ($keyword_relevance * 0.6) + (min(1.0, $question_score) * 0.2) + (min(1.0, $answer_score) * 0.2); | |
| 303 | - | |
| 304 | - return min(1.0, $satisfaction); | |
| 305 | - } | |
| 306 | - | |
| 307 | - /** | |
| 308 | - * Assess content uniqueness | |
| 309 | - * | |
| 310 | - * @param string $content Content text | |
| 311 | - * @return float Uniqueness score (0-1) | |
| 312 | - */ | |
| 313 | - private function assess_content_uniqueness(string $content): float { | |
| 314 | - if (empty($content)) { | |
| 315 | - return 0.0; | |
| 836 | + if ($word_count >= 800) { return 'Basic'; | |
| 316 | 837 | } |
| 317 | - | |
| 318 | - // Simple heuristics for uniqueness assessment | |
| 319 | - $sentences = preg_split('/[.!?]+/', wp_strip_all_tags($content), -1, PREG_SPLIT_NO_EMPTY); | |
| 320 | - $sentence_count = count($sentences); | |
| 321 | - | |
| 322 | - if ($sentence_count === 0) { | |
| 323 | - return 0.0; | |
| 324 | - } | |
| 325 | - | |
| 326 | - // Check for varied sentence structures | |
| 327 | - $short_sentences = 0; | |
| 328 | - $medium_sentences = 0; | |
| 329 | - $long_sentences = 0; | |
| 330 | - | |
| 331 | - foreach ($sentences as $sentence) { | |
| 332 | - $word_count = $this->calculate_word_count_js_style(trim($sentence)); | |
| 333 | - if ($word_count <= 10) { | |
| 334 | - $short_sentences++; | |
| 335 | - } elseif ($word_count <= 20) { | |
| 336 | - $medium_sentences++; | |
| 337 | - } else { | |
| 338 | - $long_sentences++; | |
| 339 | - } | |
| 340 | - } | |
| 341 | - | |
| 342 | - // Variety in sentence length indicates more natural, unique content | |
| 343 | - $variety_score = 0; | |
| 344 | - if ($short_sentences > 0) $variety_score += 0.3; | |
| 345 | - if ($medium_sentences > 0) $variety_score += 0.4; | |
| 346 | - if ($long_sentences > 0) $variety_score += 0.3; | |
| 347 | - | |
| 348 | - return min(1.0, $variety_score); | |
| 349 | - } | |
| 350 | - | |
| 351 | - /** | |
| 352 | - * Assess content depth for 2025 standards | |
| 353 | - * | |
| 354 | - * @param int $word_count Word count | |
| 355 | - * @return string Depth assessment | |
| 356 | - */ | |
| 357 | - private function assess_content_depth_2025(int $word_count): string { | |
| 358 | - if ($word_count >= 3000) return 'Comprehensive'; | |
| 359 | - if ($word_count >= 2000) return 'Detailed'; | |
| 360 | - if ($word_count >= 1200) return 'Adequate'; | |
| 361 | - if ($word_count >= 800) return 'Basic'; | |
| 362 | 838 | return 'Insufficient'; |
| 363 | 839 | } |
| 364 | 840 | |
| 365 | 841 | /** |
| @@ -378,24 +854,25 @@ | ||
| 378 | 854 | $suggestions[] = 'Add a compelling, click-worthy title that matches search intent'; |
| 379 | 855 | return ['score' => 0, 'max_score' => $max_score, 'suggestions' => $suggestions]; |
| 380 | 856 | } |
| 381 | 857 | |
| 382 | - $title_length = strlen($title); | |
| 858 | + $title_length = mb_strlen($title); | |
| 383 | 859 | |
| 384 | - // 2025 length optimization (7 points) - Slightly more flexible | |
| 385 | - if ($title_length >= 35 && $title_length <= 65) { | |
| 386 | - $score += 7; | |
| 860 | + // 2025 length optimization (6 points). 60 characters is the recommended | |
| 861 | + // maximum for best SERP visibility before Google truncates the title. | |
| 862 | + if ($title_length >= self::TITLE_OPTIMAL_MIN && $title_length <= self::TITLE_OPTIMAL_MAX) { | |
| 863 | + $score += 6; | |
| 387 | 864 | } elseif ($title_length >= 25 && $title_length <= 75) { |
| 388 | - $score += 5; | |
| 389 | - $suggestions[] = 'Optimize title length to 35-65 characters for better SERP visibility'; | |
| 865 | + $score += 4; | |
| 866 | + $suggestions[] = 'Optimize title length to 35-60 characters for better SERP visibility'; | |
| 390 | 867 | } else { |
| 391 | - $score += 2; | |
| 868 | + $score++; | |
| 392 | 869 | $suggestions[] = $title_length < 25 ? |
| 393 | - 'Title too short - aim for 35-65 characters' : | |
| 870 | + 'Title too short - aim for 35-60 characters' : | |
| 394 | 871 | 'Title too long - risk truncation in search results'; |
| 395 | 872 | } |
| 396 | 873 | |
| 397 | - // Looser keyword matching (7 points) - 2025 update | |
| 874 | + // Looser keyword matching (6 points) - 2025 update | |
| 398 | 875 | if (!empty($target_keyword)) { |
| 399 | 876 | $title_lower = strtolower($title); |
| 400 | 877 | $keyword_lower = strtolower($target_keyword); |
| 401 | 878 | |
| @@ -400,14 +877,14 @@ | ||
| 400 | 877 | $keyword_lower = strtolower($target_keyword); |
| 401 | 878 | |
| 402 | 879 | // Exact match |
| 403 | 880 | if (strpos($title_lower, $keyword_lower) !== false) { |
| 404 | - $score += 7; | |
| 881 | + $score += 6; | |
| 405 | 882 | } else { |
| 406 | 883 | // Check for semantic variations (2025 improvement) |
| 407 | 884 | $semantic_match = $this->check_semantic_keyword_match($title, $target_keyword); |
| 408 | 885 | if ($semantic_match) { |
| 409 | - $score += 6; // Almost full credit for semantic match | |
| 886 | + $score += 5; // Almost full credit for semantic match | |
| 410 | 887 | $suggestions[] = 'Good semantic keyword usage - Google now recognizes keyword variations'; |
| 411 | 888 | } else { |
| 412 | 889 | // Check for partial keyword match |
| 413 | 890 | $keyword_parts = explode(' ', $keyword_lower); |
| @@ -418,21 +895,37 @@ | ||
| 418 | 895 | } |
| 419 | 896 | } |
| 420 | 897 | |
| 421 | 898 | if ($partial_matches > 0) { |
| 422 | - $score += round(($partial_matches / count($keyword_parts)) * 5); | |
| 899 | + $score += round(($partial_matches / count($keyword_parts)) * 4); | |
| 423 | 900 | $suggestions[] = "Include more parts of target keyword '{$target_keyword}' in title"; |
| 424 | 901 | } else { |
| 425 | - $score += 1; | |
| 426 | 902 | $suggestions[] = "Include target keyword '{$target_keyword}' or related terms in title"; |
| 427 | 903 | } |
| 428 | 904 | } |
| 429 | 905 | } |
| 430 | 906 | } else { |
| 431 | - $score += 3; // Partial credit | |
| 907 | + $score += 2; // Partial credit | |
| 432 | 908 | $suggestions[] = 'Set a target keyword to optimize title effectiveness'; |
| 433 | 909 | } |
| 434 | 910 | |
| 911 | + // Title readability (2 points) - mirrors Rank Math's title checks for a | |
| 912 | + // number/power word (drives CTR) and emotional sentiment. | |
| 913 | + $has_number = (bool) preg_match('/\d/', $title); | |
| 914 | + $has_power_word = $this->title_has_power_word($title); | |
| 915 | + $has_sentiment = $this->title_has_sentiment_word($title); | |
| 916 | + | |
| 917 | + if ($has_number || $has_power_word) { | |
| 918 | + $score++; | |
| 919 | + } else { | |
| 920 | + $suggestions[] = 'Add a number or a power word to the title to boost click-through rate'; | |
| 921 | + } | |
| 922 | + if ($has_sentiment) { | |
| 923 | + $score++; | |
| 924 | + } else { | |
| 925 | + $suggestions[] = 'Use an emotional/sentiment word in the title to make it more compelling'; | |
| 926 | + } | |
| 927 | + | |
| 435 | 928 | return [ |
| 436 | 929 | 'score' => $score, |
| 437 | 930 | 'max_score' => $max_score, |
| 438 | 931 | 'suggestions' => $suggestions, |
| @@ -437,11 +930,13 @@ | ||
| 437 | 930 | 'max_score' => $max_score, |
| 438 | 931 | 'suggestions' => $suggestions, |
| 439 | 932 | 'details' => [ |
| 440 | 933 | 'title_length' => $title_length, |
| 441 | - 'optimal_range' => '35-65 characters', | |
| 934 | + 'optimal_range' => '35-60 characters', | |
| 442 | 935 | 'keyword_present' => !empty($target_keyword) && strpos(strtolower($title), strtolower($target_keyword)) !== false, |
| 443 | 936 | 'semantic_match' => !empty($target_keyword) ? $this->check_semantic_keyword_match($title, $target_keyword) : false, |
| 937 | + 'has_number_or_power_word' => $has_number || $has_power_word, | |
| 938 | + 'has_sentiment_word' => $has_sentiment, | |
| 444 | 939 | ] |
| 445 | 940 | ]; |
| 446 | 941 | } |
| 447 | 942 | |
| @@ -447,295 +942,185 @@ | ||
| 447 | 942 | |
| 448 | 943 | // Placeholder methods for remaining 2025 factors |
| 449 | 944 | |
| 450 | 945 | private function score_niche_expertise(array $content_data, string $target_keyword): array { |
| 946 | + $max_score = $this->scoring_factors['niche_expertise']; | |
| 947 | + $suggestions = []; | |
| 948 | + | |
| 949 | + // Without a focus keyword we cannot measure topical coverage; award | |
| 950 | + // partial credit rather than capping the ceiling with a placeholder. | |
| 951 | + if (empty($target_keyword)) { | |
| 952 | + return [ | |
| 953 | + 'score' => 7, | |
| 954 | + 'max_score' => $max_score, | |
| 955 | + 'suggestions' => ['Set a focus keyword and use it in subheadings and the URL for stronger topical signals'], | |
| 956 | + 'details' => ['expertise_level' => 'Unmeasured (no focus keyword)'], | |
| 957 | + ]; | |
| 958 | + } | |
| 959 | + | |
| 960 | + $score = 0; | |
| 961 | + $content = (string) ($content_data['content'] ?? ''); | |
| 962 | + $headings = (array) ($content_data['headings'] ?? []); | |
| 963 | + $images = (array) ($content_data['images'] ?? []); | |
| 964 | + $slug = strtolower((string) ($content_data['slug'] ?? '')); | |
| 965 | + | |
| 966 | + // Keyword in a subheading (4 points). | |
| 967 | + if ($this->keyword_in_subheadings($headings, $target_keyword)) { | |
| 968 | + $score += 4; | |
| 969 | + } else { | |
| 970 | + $suggestions[] = "Include '{$target_keyword}' in at least one subheading (H2-H6)"; | |
| 971 | + } | |
| 972 | + | |
| 973 | + // Keyword density in a healthy band (4 points). Rank Math treats | |
| 974 | + // ~0.5%-2.5% as optimal; reward in-band, partial when present but thin. | |
| 975 | + $density = $this->keyword_density($content, $target_keyword); | |
| 976 | + if ($density >= 0.5 && $density <= 2.5) { | |
| 977 | + $score += 4; | |
| 978 | + } elseif ($density > 0) { | |
| 979 | + $score += 2; | |
| 980 | + $suggestions[] = $density > 2.5 | |
| 981 | + ? 'Keyword density is high - reduce repetition to avoid over-optimization' | |
| 982 | + : 'Keyword density is low - use the focus keyword a little more often'; | |
| 983 | + } else { | |
| 984 | + $suggestions[] = "Use the focus keyword '{$target_keyword}' in the content"; | |
| 985 | + } | |
| 986 | + | |
| 987 | + // URL optimization (3 points): keyword in slug (2) + a reasonably short | |
| 988 | + // URL (1). Rank Math flags overly long URLs, so reward concise slugs. | |
| 989 | + $keyword_slug = str_replace(' ', '-', strtolower($target_keyword)); | |
| 990 | + $keyword_in_slug = $slug !== '' && (strpos($slug, $keyword_slug) !== false || strpos(str_replace('-', '', $slug), str_replace('-', '', $keyword_slug)) !== false); | |
| 991 | + if ($keyword_in_slug) { | |
| 992 | + $score += 2; | |
| 993 | + } else { | |
| 994 | + $suggestions[] = 'Include the focus keyword in the URL slug'; | |
| 995 | + } | |
| 996 | + | |
| 997 | + // A slug under ~75 chars keeps the URL clean and fully visible in SERPs. | |
| 998 | + $slug_length = strlen($slug); | |
| 999 | + if ($slug === '' || $slug_length <= 75) { | |
| 1000 | + $score++; | |
| 1001 | + } else { | |
| 1002 | + $suggestions[] = 'Shorten the URL slug - long URLs are harder to read and share'; | |
| 1003 | + } | |
| 1004 | + | |
| 1005 | + // Keyword in image alt text (2 points) - mirrors Rank Math's | |
| 1006 | + // "keyword in image alt" check. When the post has no images the check | |
| 1007 | + // does not apply, so award the points (benefit of the doubt) rather | |
| 1008 | + // than capping the ceiling for legitimately image-less posts. | |
| 1009 | + if (empty($images)) { | |
| 1010 | + $score += 2; | |
| 1011 | + $suggestions[] = 'Add a relevant image with the focus keyword in its alt text'; | |
| 1012 | + } elseif ($this->keyword_in_alt($images, $target_keyword)) { | |
| 1013 | + $score += 2; | |
| 1014 | + } else { | |
| 1015 | + $suggestions[] = 'Include the focus keyword in at least one image alt attribute'; | |
| 1016 | + } | |
| 1017 | + | |
| 451 | 1018 | return [ |
| 452 | - 'score' => 8, // Partial score for basic implementation | |
| 453 | - 'max_score' => $this->scoring_factors['niche_expertise'], | |
| 454 | - 'suggestions' => ['Develop hub and spoke content clusters around your main topics'], | |
| 455 | - 'details' => ['expertise_level' => 'Developing'] | |
| 1019 | + 'score' => $score, | |
| 1020 | + 'max_score' => $max_score, | |
| 1021 | + 'suggestions' => $suggestions, | |
| 1022 | + 'details' => [ | |
| 1023 | + 'keyword_in_subheading' => $this->keyword_in_subheadings($headings, $target_keyword), | |
| 1024 | + 'keyword_density' => round($density, 2), | |
| 1025 | + 'keyword_in_slug' => $keyword_in_slug, | |
| 1026 | + 'slug_length' => $slug_length, | |
| 1027 | + 'keyword_in_image_alt' => $this->keyword_in_alt($images, $target_keyword), | |
| 1028 | + ], | |
| 456 | 1029 | ]; |
| 457 | 1030 | } |
| 458 | 1031 | |
| 459 | 1032 | private function score_searcher_engagement(array $content_data): array { |
| 460 | - $readability = $content_data['readability_score'] ?? 50; | |
| 461 | - $score = round(($readability / 100) * $this->scoring_factors['searcher_engagement']); | |
| 1033 | + $max_score = $this->scoring_factors['searcher_engagement']; | |
| 462 | 1034 | |
| 1035 | + // Engagement (dwell time / bounce) is off-page, so estimate it from the | |
| 1036 | + // on-page signals that drive it: readability (half) + content structure | |
| 1037 | + // (half). This raises the old readability-only floor. | |
| 1038 | + $readability = (float) ($content_data['readability_score'] ?? 50); | |
| 1039 | + $structure = (float) $this->derive_content_quality_score($content_data); | |
| 1040 | + | |
| 1041 | + $readability_pts = ($readability / 100) * ($max_score / 2); | |
| 1042 | + $structure_pts = ($structure / 100) * ($max_score / 2); | |
| 1043 | + $score = (int) round($readability_pts + $structure_pts); | |
| 1044 | + | |
| 463 | 1045 | return [ |
| 464 | 1046 | 'score' => $score, |
| 465 | - 'max_score' => $this->scoring_factors['searcher_engagement'], | |
| 466 | - 'suggestions' => $score < 8 ? ['Improve content engagement with better readability and structure'] : [], | |
| 467 | - 'details' => ['engagement_estimate' => round(($score / $this->scoring_factors['searcher_engagement']) * 100, 1) . '%'] | |
| 1047 | + 'max_score' => $max_score, | |
| 1048 | + 'suggestions' => $score < ($max_score * 0.7) | |
| 1049 | + ? ['Improve readability and structure (shorter sentences, subheadings, shorter paragraphs)'] | |
| 1050 | + : [], | |
| 1051 | + 'details' => ['engagement_estimate' => round(($score / $max_score) * 100, 1) . '%'], | |
| 468 | 1052 | ]; |
| 469 | 1053 | } |
| 470 | 1054 | |
| 471 | 1055 | private function score_backlink_authority(array $content_data): array { |
| 472 | - $external_links = $content_data['external_links'] ?? 0; | |
| 473 | - $score = min($this->scoring_factors['backlink_authority'], $external_links * 3); | |
| 1056 | + $max_score = $this->scoring_factors['backlink_authority']; | |
| 1057 | + $external_links = (int) ($content_data['external_links'] ?? 0); | |
| 1058 | + $dofollow_links = (int) ($content_data['external_dofollow_links'] ?? 0); | |
| 474 | 1059 | |
| 1060 | + // Backlinks are off-page and cannot be read from post content. We use | |
| 1061 | + // the only on-page proxy available - whether the content cites external | |
| 1062 | + // sources - and avoid hard-penalizing posts for something outside the | |
| 1063 | + // editor's control (the old external_links*3 formula needed 5 outbound | |
| 1064 | + // links just to reach full marks, dragging nearly every post down). | |
| 1065 | + // Dofollow links pass equity, so they earn full credit (Rank Math's | |
| 1066 | + // "external dofollow link" check); nofollow-only citations earn less. | |
| 1067 | + $suggestions = []; | |
| 1068 | + if ($dofollow_links >= 2) { | |
| 1069 | + $score = $max_score; | |
| 1070 | + } elseif ($dofollow_links === 1) { | |
| 1071 | + $score = (int) round($max_score * 0.85); | |
| 1072 | + } elseif ($external_links > 0) { | |
| 1073 | + // Cites sources but every external link is nofollow. | |
| 1074 | + $score = (int) round($max_score * 0.75); | |
| 1075 | + $suggestions[] = 'Add at least one dofollow link to an authoritative external source'; | |
| 1076 | + } else { | |
| 1077 | + $score = (int) round($max_score * 0.6); | |
| 1078 | + $suggestions[] = 'Cite authoritative external sources, and build quality backlinks to this page'; | |
| 1079 | + } | |
| 1080 | + | |
| 475 | 1081 | return [ |
| 476 | 1082 | 'score' => $score, |
| 477 | - 'max_score' => $this->scoring_factors['backlink_authority'], | |
| 478 | - 'suggestions' => $score < 8 ? ['Build high-quality backlinks from authoritative sources'] : [], | |
| 479 | - 'details' => ['authority_estimate' => 'Moderate'] | |
| 1083 | + 'max_score' => $max_score, | |
| 1084 | + 'suggestions' => $suggestions, | |
| 1085 | + 'details' => [ | |
| 1086 | + 'external_links' => $external_links, | |
| 1087 | + 'external_dofollow_links' => $dofollow_links, | |
| 1088 | + ], | |
| 480 | 1089 | ]; |
| 481 | 1090 | } |
| 482 | - | |
| 483 | - private function score_2025_content_freshness(array $content_data): array { | |
| 484 | - return [ | |
| 485 | - 'score' => 6, // Full score for new content | |
| 486 | - 'max_score' => $this->scoring_factors['content_freshness'], | |
| 487 | - 'suggestions' => ['Update content quarterly to maintain freshness signals'], | |
| 488 | - 'details' => ['freshness_status' => 'Current', 'last_updated' => current_time('mysql')] | |
| 489 | - ]; | |
| 490 | - } | |
| 491 | - | |
| 492 | - private function score_mobile_experience_score(array $content_data): array { | |
| 493 | - return [ | |
| 494 | - 'score' => 4, // Good default score | |
| 495 | - 'max_score' => $this->scoring_factors['mobile_experience'], | |
| 496 | - 'suggestions' => ['Ensure optimal mobile experience with fast loading and easy navigation'], | |
| 497 | - 'details' => ['mes_score' => 'Good', 'mobile_friendly' => true] | |
| 498 | - ]; | |
| 499 | - } | |
| 500 | - | |
| 501 | 1091 | private function score_trustworthiness(array $content_data, array $metadata): array { |
| 1092 | + // E-E-A-T is an off-page/site-wide signal we cannot reliably measure | |
| 1093 | + // from a single post. Award full credit (benefit of the doubt) rather | |
| 1094 | + // than a fixed partial that silently caps every post's ceiling. | |
| 502 | 1095 | return [ |
| 503 | - 'score' => 3, // Moderate trust score | |
| 1096 | + 'score' => $this->scoring_factors['trustworthiness'], | |
| 504 | 1097 | 'max_score' => $this->scoring_factors['trustworthiness'], |
| 505 | - 'suggestions' => ['Add author credentials, citations, and contact information to improve trustworthiness'], | |
| 506 | - 'details' => ['trust_level' => 'Moderate'] | |
| 1098 | + 'suggestions' => ['Add author credentials, citations, and contact information to reinforce trustworthiness'], | |
| 1099 | + 'details' => ['trust_level' => 'Assumed adequate'], | |
| 507 | 1100 | ]; |
| 508 | 1101 | } |
| 509 | 1102 | |
| 510 | 1103 | private function score_link_diversity(array $content_data): array { |
| 1104 | + // Off-page link distribution; not measurable per post. Full credit. | |
| 511 | 1105 | return [ |
| 512 | - 'score' => 2, // Basic score | |
| 1106 | + 'score' => $this->scoring_factors['link_diversity'], | |
| 513 | 1107 | 'max_score' => $this->scoring_factors['link_diversity'], |
| 514 | - 'suggestions' => ['Build backlinks to multiple pages across your site'], | |
| 515 | - 'details' => ['diversity_level' => 'Basic'] | |
| 1108 | + 'suggestions' => [], | |
| 1109 | + 'details' => ['diversity_level' => 'Assumed adequate'], | |
| 516 | 1110 | ]; |
| 517 | 1111 | } |
| 518 | - | |
| 519 | - private function score_core_web_vitals_2025(array $content_data): array { | |
| 520 | - return [ | |
| 521 | - 'score' => 2, // Assume decent performance | |
| 522 | - 'max_score' => $this->scoring_factors['core_web_vitals'], | |
| 523 | - 'suggestions' => ['Optimize for Interaction Readiness and Cumulative Layout Shift 2.0'], | |
| 524 | - 'details' => ['vitals_status' => 'Good', 'interaction_readiness' => 'Optimized'] | |
| 525 | - ]; | |
| 526 | - } | |
| 527 | - | |
| 528 | 1112 | private function score_site_security(array $content_data): array { |
| 529 | - return [ | |
| 530 | - 'score' => 2, // Full score for SSL | |
| 531 | - 'max_score' => $this->scoring_factors['site_security'], | |
| 532 | - 'suggestions' => [], | |
| 533 | - 'details' => ['ssl_enabled' => true, 'security_level' => 'Good'] | |
| 534 | - ]; | |
| 535 | - } | |
| 1113 | + $max_score = $this->scoring_factors['site_security']; | |
| 1114 | + $ssl = function_exists('is_ssl') ? is_ssl() : true; | |
| 536 | 1115 | |
| 537 | - private function score_internal_linking_2025(array $content_data): array { | |
| 538 | - $internal_links = $content_data['internal_links'] ?? 0; | |
| 539 | - $score = min($this->scoring_factors['internal_linking'], $internal_links > 0 ? 1 : 0); | |
| 540 | - | |
| 541 | 1116 | return [ |
| 542 | - 'score' => $score, | |
| 543 | - 'max_score' => $this->scoring_factors['internal_linking'], | |
| 544 | - 'suggestions' => $score < 1 ? ['Add internal links to related content'] : [], | |
| 545 | - 'details' => ['internal_links' => $internal_links] | |
| 546 | - ]; | |
| 547 | - } | |
| 548 | - | |
| 549 | - private function score_technical_factors_2025(array $content_data, array $metadata, array $options): array { | |
| 550 | - $meta_desc = $metadata['description'] ?? ''; | |
| 551 | - $score = !empty($meta_desc) ? 1 : 0; | |
| 552 | - | |
| 553 | - return [ | |
| 554 | - 'score' => $score, | |
| 555 | - 'max_score' => $this->scoring_factors['technical_factors'], | |
| 556 | - 'suggestions' => $score < 1 ? ['Add meta description and schema markup'] : [], | |
| 557 | - 'details' => ['meta_description' => !empty($meta_desc), 'schema_markup' => false] | |
| 558 | - ]; | |
| 559 | - } | |
| 560 | - | |
| 561 | - /** | |
| 562 | - * Prioritize suggestions for 2025 | |
| 563 | - * | |
| 564 | - * @param array $suggestions Raw suggestions | |
| 565 | - * @return array Prioritized suggestions | |
| 566 | - */ | |
| 567 | - private function prioritize_suggestions_2025(array $suggestions): array { | |
| 568 | - $prioritized = []; | |
| 569 | - | |
| 570 | - foreach ($suggestions as $suggestion) { | |
| 571 | - $priority = $this->determine_2025_priority($suggestion); | |
| 572 | - $prioritized[] = [ | |
| 573 | - 'text' => $suggestion, | |
| 574 | - 'priority' => $priority, | |
| 575 | - 'impact' => $this->estimate_2025_impact($suggestion), | |
| 576 | - 'effort' => $this->estimate_effort($suggestion), | |
| 577 | - ]; | |
| 578 | - } | |
| 579 | - | |
| 580 | - // Sort by 2025 priorities | |
| 581 | - usort($prioritized, function($a, $b) { | |
| 582 | - $priority_order = ['Critical' => 4, 'High' => 3, 'Medium' => 2, 'Low' => 1]; | |
| 583 | - return $priority_order[$b['priority']] - $priority_order[$a['priority']]; | |
| 584 | - }); | |
| 585 | - | |
| 586 | - return $prioritized; | |
| 587 | - } | |
| 588 | - | |
| 589 | - /** | |
| 590 | - * Determine 2025 suggestion priority | |
| 591 | - * | |
| 592 | - * @param string $suggestion Suggestion text | |
| 593 | - * @return string Priority level | |
| 594 | - */ | |
| 595 | - private function determine_2025_priority(string $suggestion): string { | |
| 596 | - $critical_keywords = ['satisfying content', 'search intent', 'comprehensive']; | |
| 597 | - $high_priority_keywords = ['title', 'engagement', 'freshness']; | |
| 598 | - $medium_priority_keywords = ['mobile', 'trust', 'backlink']; | |
| 599 | - | |
| 600 | - $suggestion_lower = strtolower($suggestion); | |
| 601 | - | |
| 602 | - foreach ($critical_keywords as $keyword) { | |
| 603 | - if (strpos($suggestion_lower, $keyword) !== false) { | |
| 604 | - return 'Critical'; | |
| 605 | - } | |
| 606 | - } | |
| 607 | - | |
| 608 | - foreach ($high_priority_keywords as $keyword) { | |
| 609 | - if (strpos($suggestion_lower, $keyword) !== false) { | |
| 610 | - return 'High'; | |
| 611 | - } | |
| 612 | - } | |
| 613 | - | |
| 614 | - foreach ($medium_priority_keywords as $keyword) { | |
| 615 | - if (strpos($suggestion_lower, $keyword) !== false) { | |
| 616 | - return 'Medium'; | |
| 617 | - } | |
| 618 | - } | |
| 619 | - | |
| 620 | - return 'Low'; | |
| 621 | - } | |
| 622 | - | |
| 623 | - /** | |
| 624 | - * Estimate 2025 impact | |
| 625 | - * | |
| 626 | - * @param string $suggestion Suggestion text | |
| 627 | - * @return string Impact level | |
| 628 | - */ | |
| 629 | - private function estimate_2025_impact(string $suggestion): string { | |
| 630 | - if (strpos(strtolower($suggestion), 'satisfying') !== false) return 'Critical'; | |
| 631 | - if (strpos(strtolower($suggestion), 'engagement') !== false) return 'High'; | |
| 632 | - if (strpos(strtolower($suggestion), 'title') !== false) return 'High'; | |
| 633 | - if (strpos(strtolower($suggestion), 'freshness') !== false) return 'Medium'; | |
| 634 | - return 'Low'; | |
| 635 | - } | |
| 636 | - | |
| 637 | - /** | |
| 638 | - * Get 2025 grade from score | |
| 639 | - * | |
| 640 | - * @param int $score Numeric score | |
| 641 | - * @return string Letter grade | |
| 642 | - */ | |
| 643 | - private function get_2025_grade(int $score): string { | |
| 644 | - if ($score >= 95) return 'A+'; | |
| 645 | - if ($score >= 90) return 'A'; | |
| 646 | - if ($score >= 85) return 'A-'; | |
| 647 | - if ($score >= 80) return 'B+'; | |
| 648 | - if ($score >= 75) return 'B'; | |
| 649 | - if ($score >= 70) return 'B-'; | |
| 650 | - if ($score >= 65) return 'C+'; | |
| 651 | - if ($score >= 60) return 'C'; | |
| 652 | - if ($score >= 55) return 'C-'; | |
| 653 | - if ($score >= 45) return 'D+'; | |
| 654 | - if ($score >= 35) return 'D'; | |
| 655 | - return 'F'; | |
| 656 | - } | |
| 657 | - | |
| 658 | - /** | |
| 659 | - * Score title optimization - #2 factor in 2025 (14 points) | |
| 660 | - * Looser keyword matching, focus on click-through rate and intent satisfaction | |
| 661 | - * | |
| 662 | - * @param string $title Page title | |
| 663 | - * @param string $target_keyword Target keyword | |
| 664 | - * @return array Scoring result | |
| 665 | - */ | |
| 666 | - private function score_title_optimization(string $title, string $target_keyword): array { | |
| 667 | - $score = 0; | |
| 668 | - $max_score = $this->scoring_factors['title_optimization']; | |
| 669 | - $suggestions = []; | |
| 670 | - | |
| 671 | - if (empty($title)) { | |
| 672 | - $suggestions[] = 'Add a compelling, click-worthy title that matches search intent'; | |
| 673 | - return ['score' => 0, 'max_score' => $max_score, 'suggestions' => $suggestions]; | |
| 674 | - } | |
| 675 | - | |
| 676 | - $title_length = strlen($title); | |
| 677 | - | |
| 678 | - // Modern length optimization (8 points) - Updated for 2024 | |
| 679 | - if ($title_length >= 40 && $title_length <= 60) { | |
| 680 | - $score += 8; | |
| 681 | - } elseif ($title_length >= 30 && $title_length <= 70) { | |
| 682 | - $score += 6; | |
| 683 | - $suggestions[] = 'Optimize title length to 40-60 characters for better SERP visibility'; | |
| 684 | - } else { | |
| 685 | - $score += 2; | |
| 686 | - $suggestions[] = $title_length < 30 ? | |
| 687 | - 'Title too short - aim for 40-60 characters to maximize SERP real estate' : | |
| 688 | - 'Title too long - risk truncation in search results'; | |
| 689 | - } | |
| 690 | - | |
| 691 | - // Semantic keyword presence (10 points) - Modern approach | |
| 692 | - if (!empty($target_keyword)) { | |
| 693 | - $title_lower = strtolower($title); | |
| 694 | - $keyword_lower = strtolower($target_keyword); | |
| 695 | - | |
| 696 | - if (strpos($title_lower, $keyword_lower) !== false) { | |
| 697 | - $score += 10; | |
| 698 | - | |
| 699 | - // Bonus for keyword placement | |
| 700 | - if (strpos($title_lower, $keyword_lower) === 0) { | |
| 701 | - $suggestions[] = 'Excellent: Target keyword at beginning of title for maximum impact'; | |
| 702 | - } | |
| 703 | - } else { | |
| 704 | - // Check for semantic variations | |
| 705 | - $semantic_match = $this->check_semantic_keyword_match($title, $target_keyword); | |
| 706 | - if ($semantic_match) { | |
| 707 | - $score += 7; | |
| 708 | - $suggestions[] = 'Good semantic keyword usage - consider including exact keyword for clarity'; | |
| 709 | - } else { | |
| 710 | - $score += 2; | |
| 711 | - $suggestions[] = "Include target keyword '{$target_keyword}' or semantic variations in title"; | |
| 712 | - } | |
| 713 | - } | |
| 714 | - } else { | |
| 715 | - $score += 5; // Partial credit | |
| 716 | - $suggestions[] = 'Set a target keyword to optimize title effectiveness'; | |
| 717 | - } | |
| 718 | - | |
| 719 | - return [ | |
| 720 | - 'score' => $score, | |
| 1117 | + 'score' => $ssl ? $max_score : 0, | |
| 721 | 1118 | 'max_score' => $max_score, |
| 722 | - 'suggestions' => $suggestions, | |
| 723 | - 'details' => [ | |
| 724 | - 'title_length' => $title_length, | |
| 725 | - 'optimal_range' => '40-60 characters', | |
| 726 | - 'keyword_present' => !empty($target_keyword) && strpos(strtolower($title), strtolower($target_keyword)) !== false, | |
| 727 | - ] | |
| 1119 | + 'suggestions' => $ssl ? [] : ['Serve the site over HTTPS (install an SSL certificate)'], | |
| 1120 | + 'details' => ['ssl_enabled' => $ssl, 'security_level' => $ssl ? 'Good' : 'Insecure'], | |
| 728 | 1121 | ]; |
| 729 | 1122 | } |
| 730 | - | |
| 731 | - /** | |
| 732 | - * Check for semantic keyword matching | |
| 733 | - * | |
| 734 | - * @param string $text Text to analyze | |
| 735 | - * @param string $keyword Target keyword | |
| 736 | - * @return bool True if semantic match found | |
| 737 | - */ | |
| 738 | 1123 | private function check_semantic_keyword_match(string $text, string $keyword): bool { |
| 739 | 1124 | // Simple semantic matching - can be enhanced with AI/NLP |
| 740 | 1125 | $keyword_parts = explode(' ', strtolower($keyword)); |
| 741 | 1126 | $text_lower = strtolower($text); |
| @@ -749,55 +1134,69 @@ | ||
| 749 | 1134 | |
| 750 | 1135 | // Consider it a semantic match if 70% of keyword parts are present |
| 751 | 1136 | return ($matches / count($keyword_parts)) >= 0.7; |
| 752 | 1137 | } |
| 753 | - | |
| 754 | - /** | |
| 755 | - * Get modern grade from score (updated scale) | |
| 756 | - * | |
| 757 | - * @param int $score Numeric score | |
| 758 | - * @return string Letter grade | |
| 759 | - */ | |
| 760 | - private function get_modern_grade(int $score): string { | |
| 761 | - if ($score >= 95) return 'A+'; | |
| 762 | - if ($score >= 90) return 'A'; | |
| 763 | - if ($score >= 85) return 'A-'; | |
| 764 | - if ($score >= 80) return 'B+'; | |
| 765 | - if ($score >= 75) return 'B'; | |
| 766 | - if ($score >= 70) return 'B-'; | |
| 767 | - if ($score >= 65) return 'C+'; | |
| 768 | - if ($score >= 60) return 'C'; | |
| 769 | - if ($score >= 55) return 'C-'; | |
| 770 | - if ($score >= 45) return 'D+'; | |
| 771 | - if ($score >= 35) return 'D'; | |
| 772 | - return 'F'; | |
| 773 | - } | |
| 774 | - | |
| 775 | - /** | |
| 776 | - * Prioritize suggestions by impact and effort | |
| 777 | - * | |
| 778 | - * @param array $suggestions Raw suggestions | |
| 779 | - * @return array Prioritized suggestions with metadata | |
| 780 | - */ | |
| 781 | - private function prioritize_suggestions(array $suggestions): array { | |
| 1138 | + private function prioritize_suggestions(array $suggestions, array $scores = []): array { | |
| 1139 | + // Map each suggestion back to the factor that emitted it, so priority | |
| 1140 | + // can rank by the points the factor actually lost instead of keyword- | |
| 1141 | + // matching the advice text — which sorted a 2-point title tweak above | |
| 1142 | + // a 6-point thin-content loss and contradicted the row's own impact | |
| 1143 | + // tag (#408). | |
| 1144 | + $by_text = []; | |
| 1145 | + foreach ($scores as $factor => $result) { | |
| 1146 | + if (!is_array($result) || empty($result['suggestions']) || !is_array($result['suggestions'])) { | |
| 1147 | + continue; | |
| 1148 | + } | |
| 1149 | + $lost = max(0, (float) ($result['max_score'] ?? 0) - (float) ($result['score'] ?? 0)); | |
| 1150 | + foreach ($result['suggestions'] as $text) { | |
| 1151 | + if (is_string($text) && !isset($by_text[$text])) { | |
| 1152 | + $by_text[$text] = ['factor' => (string) $factor, 'lost' => $lost]; | |
| 1153 | + } | |
| 1154 | + } | |
| 1155 | + } | |
| 1156 | + | |
| 782 | 1157 | $prioritized = []; |
| 783 | - | |
| 1158 | + | |
| 784 | 1159 | foreach ($suggestions as $suggestion) { |
| 785 | - $priority = $this->determine_suggestion_priority($suggestion); | |
| 1160 | + $origin = $by_text[$suggestion] ?? null; | |
| 1161 | + | |
| 1162 | + // A factor already at full marks loses nothing to this advice — | |
| 1163 | + // it was occupying list positions (sometimes at "High") while | |
| 1164 | + // recovering zero points. Dropped rather than sorted last. | |
| 1165 | + if (null !== $origin && $origin['lost'] <= 0) { | |
| 1166 | + continue; | |
| 1167 | + } | |
| 1168 | + | |
| 1169 | + if (null !== $origin) { | |
| 1170 | + $priority = $origin['lost'] >= 4 ? 'High' : ($origin['lost'] >= 2 ? 'Medium' : 'Low'); | |
| 1171 | + } else { | |
| 1172 | + // No factor attached (defensive: a filter-added or legacy | |
| 1173 | + // suggestion) — the old keyword map is the fallback. | |
| 1174 | + $priority = $this->determine_suggestion_priority($suggestion); | |
| 1175 | + } | |
| 1176 | + | |
| 786 | 1177 | $prioritized[] = [ |
| 787 | 1178 | 'text' => $suggestion, |
| 788 | 1179 | 'priority' => $priority, |
| 789 | 1180 | 'impact' => $this->estimate_impact($suggestion), |
| 790 | 1181 | 'effort' => $this->estimate_effort($suggestion), |
| 1182 | + 'factor' => $origin['factor'] ?? null, | |
| 1183 | + 'points_recoverable' => $origin['lost'] ?? null, | |
| 791 | 1184 | ]; |
| 792 | 1185 | } |
| 793 | - | |
| 794 | - // Sort by priority (High > Medium > Low) | |
| 1186 | + | |
| 1187 | + // Biggest recoverable loss first; keyword-mapped stragglers (no | |
| 1188 | + // factor) sort within their priority band after the measured rows. | |
| 795 | 1189 | usort($prioritized, function($a, $b) { |
| 1190 | + $al = $a['points_recoverable'] ?? -1; | |
| 1191 | + $bl = $b['points_recoverable'] ?? -1; | |
| 1192 | + if ($al !== $bl) { | |
| 1193 | + return $bl <=> $al; | |
| 1194 | + } | |
| 796 | 1195 | $priority_order = ['High' => 3, 'Medium' => 2, 'Low' => 1]; |
| 797 | 1196 | return $priority_order[$b['priority']] - $priority_order[$a['priority']]; |
| 798 | 1197 | }); |
| 799 | - | |
| 1198 | + | |
| 800 | 1199 | return $prioritized; |
| 801 | 1200 | } |
| 802 | 1201 | |
| 803 | 1202 | /** |
| @@ -834,11 +1233,14 @@ | ||
| 834 | 1233 | * @return string Impact level |
| 835 | 1234 | */ |
| 836 | 1235 | private function estimate_impact(string $suggestion): string { |
| 837 | 1236 | // Simple heuristic - can be enhanced with ML |
| 838 | - if (strpos(strtolower($suggestion), 'title') !== false) return 'High'; | |
| 839 | - if (strpos(strtolower($suggestion), 'content') !== false) return 'High'; | |
| 840 | - if (strpos(strtolower($suggestion), 'keyword') !== false) return 'Medium'; | |
| 1237 | + if (strpos(strtolower($suggestion), 'title') !== false) { return 'High'; | |
| 1238 | + } | |
| 1239 | + if (strpos(strtolower($suggestion), 'content') !== false) { return 'High'; | |
| 1240 | + } | |
| 1241 | + if (strpos(strtolower($suggestion), 'keyword') !== false) { return 'Medium'; | |
| 1242 | + } | |
| 841 | 1243 | return 'Low'; |
| 842 | 1244 | } |
| 843 | 1245 | |
| 844 | 1246 | /** |
| @@ -848,86 +1250,16 @@ | ||
| 848 | 1250 | * @return string Effort level |
| 849 | 1251 | */ |
| 850 | 1252 | private function estimate_effort(string $suggestion): string { |
| 851 | 1253 | // Simple heuristic - can be enhanced with ML |
| 852 | - if (strpos(strtolower($suggestion), 'rewrite') !== false) return 'High'; | |
| 853 | - if (strpos(strtolower($suggestion), 'add') !== false) return 'Medium'; | |
| 854 | - if (strpos(strtolower($suggestion), 'optimize') !== false) return 'Medium'; | |
| 855 | - return 'Low'; | |
| 856 | - } | |
| 857 | - | |
| 858 | - /** | |
| 859 | - * Score content quality using modern E-A-T principles | |
| 860 | - * | |
| 861 | - * @param array $content_data Content analysis data | |
| 862 | - * @param string $target_keyword Target keyword | |
| 863 | - * @return array Scoring result | |
| 864 | - */ | |
| 865 | - private function score_content_quality(array $content_data, string $target_keyword): array { | |
| 866 | - $score = 0; | |
| 867 | - $max_score = $this->scoring_factors['content_quality']; | |
| 868 | - $suggestions = []; | |
| 869 | - | |
| 870 | - $word_count = $content_data['word_count'] ?? 0; | |
| 871 | - $content = $content_data['content'] ?? ''; | |
| 872 | - | |
| 873 | - // Content depth and comprehensiveness (8 points) | |
| 874 | - if ($word_count >= 1500) { | |
| 875 | - $score += 8; | |
| 876 | - } elseif ($word_count >= 800) { | |
| 877 | - $score += 6; | |
| 878 | - $suggestions[] = 'Consider expanding content for more comprehensive coverage (1500+ words ideal)'; | |
| 879 | - } elseif ($word_count >= 300) { | |
| 880 | - $score += 4; | |
| 881 | - $suggestions[] = 'Content needs more depth - aim for 800+ words for better topic coverage'; | |
| 882 | - } else { | |
| 883 | - $suggestions[] = 'Content too thin - add substantial value with detailed information (minimum 300 words)'; | |
| 1254 | + if (strpos(strtolower($suggestion), 'rewrite') !== false) { return 'High'; | |
| 884 | 1255 | } |
| 885 | - | |
| 886 | - // Content structure and organization (6 points) | |
| 887 | - $headings = $content_data['headings'] ?? []; | |
| 888 | - if (count($headings) >= 3) { | |
| 889 | - $score += 6; | |
| 890 | - } elseif (count($headings) >= 1) { | |
| 891 | - $score += 3; | |
| 892 | - $suggestions[] = 'Add more headings to improve content structure and scannability'; | |
| 893 | - } else { | |
| 894 | - $suggestions[] = 'Add headings (H2, H3) to organize content and improve user experience'; | |
| 1256 | + if (strpos(strtolower($suggestion), 'add') !== false) { return 'Medium'; | |
| 895 | 1257 | } |
| 896 | - | |
| 897 | - // Topic relevance and focus (6 points) | |
| 898 | - if (!empty($target_keyword) && !empty($content)) { | |
| 899 | - $relevance_score = $this->calculate_topic_relevance($content, $target_keyword); | |
| 900 | - $score += round($relevance_score * 6); | |
| 901 | - | |
| 902 | - if ($relevance_score < 0.5) { | |
| 903 | - $suggestions[] = "Improve content relevance to target topic '{$target_keyword}'"; | |
| 904 | - } | |
| 905 | - } else { | |
| 906 | - $score += 3; // Partial credit | |
| 907 | - $suggestions[] = 'Define target keyword to optimize content focus and relevance'; | |
| 1258 | + if (strpos(strtolower($suggestion), 'optimize') !== false) { return 'Medium'; | |
| 908 | 1259 | } |
| 909 | - | |
| 910 | - return [ | |
| 911 | - 'score' => $score, | |
| 912 | - 'max_score' => $max_score, | |
| 913 | - 'suggestions' => $suggestions, | |
| 914 | - 'details' => [ | |
| 915 | - 'word_count' => $word_count, | |
| 916 | - 'heading_count' => count($headings), | |
| 917 | - 'content_depth' => $this->assess_content_depth($word_count), | |
| 918 | - 'topic_coverage' => $this->assess_topic_coverage($content, $target_keyword), | |
| 919 | - ] | |
| 920 | - ]; | |
| 1260 | + return 'Low'; | |
| 921 | 1261 | } |
| 922 | - | |
| 923 | - /** | |
| 924 | - * Calculate topic relevance score | |
| 925 | - * | |
| 926 | - * @param string $content Content text | |
| 927 | - * @param string $target_keyword Target keyword | |
| 928 | - * @return float Relevance score (0-1) | |
| 929 | - */ | |
| 930 | 1262 | private function calculate_topic_relevance(string $content, string $target_keyword): float { |
| 931 | 1263 | if (empty($content) || empty($target_keyword)) { |
| 932 | 1264 | return 0.0; |
| 933 | 1265 | } |
| @@ -999,9 +1331,9 @@ | ||
| 999 | 1331 | $terms = array_merge($terms, ['optimization', 'search engine', 'ranking', 'visibility']); |
| 1000 | 1332 | } |
| 1001 | 1333 | |
| 1002 | 1334 | // WordPress-related terms |
| 1003 | - if (strpos($keyword_lower, 'wordpress') !== false) { | |
| 1335 | + if (strpos($keyword_lower, 'wordpress') !== false) { // phpcs:ignore WordPress.WP.CapitalPDangit.MisspelledInText -- lowercase on purpose: the haystack is strtolower()ed. | |
| 1004 | 1336 | $terms = array_merge($terms, ['wp', 'plugin', 'theme', 'cms']); |
| 1005 | 1337 | } |
| 1006 | 1338 | |
| 1007 | 1339 | return $terms; |
| @@ -1007,494 +1339,292 @@ | ||
| 1007 | 1339 | return $terms; |
| 1008 | 1340 | } |
| 1009 | 1341 | |
| 1010 | 1342 | /** |
| 1011 | - * Assess content depth based on word count | |
| 1343 | + * Keyword density (%) of the target keyword across the plain-text body. | |
| 1012 | 1344 | * |
| 1013 | - * @param int $word_count Word count | |
| 1014 | - * @return string Depth assessment | |
| 1345 | + * @param string $content Raw/HTML content. | |
| 1346 | + * @param string $target_keyword Target keyword. | |
| 1347 | + * @return float Density percentage (0 when no keyword/content). | |
| 1015 | 1348 | */ |
| 1016 | - private function assess_content_depth(int $word_count): string { | |
| 1017 | - if ($word_count >= 2000) return 'Comprehensive'; | |
| 1018 | - if ($word_count >= 1000) return 'Detailed'; | |
| 1019 | - if ($word_count >= 500) return 'Moderate'; | |
| 1020 | - if ($word_count >= 300) return 'Basic'; | |
| 1021 | - return 'Insufficient'; | |
| 1349 | + private function keyword_density(string $content, string $target_keyword): float { | |
| 1350 | + if (empty($content) || empty($target_keyword)) { | |
| 1351 | + return 0.0; | |
| 1352 | + } | |
| 1353 | + $plain = strtolower(wp_strip_all_tags($content)); | |
| 1354 | + $word_count = $this->calculate_word_count_js_style($plain); | |
| 1355 | + if ($word_count === 0) { | |
| 1356 | + return 0.0; | |
| 1357 | + } | |
| 1358 | + $occurrences = substr_count($plain, strtolower($target_keyword)); | |
| 1359 | + return ($occurrences / $word_count) * 100; | |
| 1022 | 1360 | } |
| 1023 | 1361 | |
| 1024 | 1362 | /** |
| 1025 | - * Assess topic coverage | |
| 1363 | + * Whether the target keyword (or a semantic variation) appears in the body. | |
| 1026 | 1364 | * |
| 1027 | - * @param string $content Content text | |
| 1028 | - * @param string $target_keyword Target keyword | |
| 1029 | - * @return string Coverage assessment | |
| 1365 | + * @param string $content Raw/HTML content. | |
| 1366 | + * @param string $target_keyword Target keyword. | |
| 1367 | + * @return bool | |
| 1030 | 1368 | */ |
| 1031 | - private function assess_topic_coverage(string $content, string $target_keyword): string { | |
| 1032 | - $relevance = $this->calculate_topic_relevance($content, $target_keyword); | |
| 1033 | - | |
| 1034 | - if ($relevance >= 0.8) return 'Excellent'; | |
| 1035 | - if ($relevance >= 0.6) return 'Good'; | |
| 1036 | - if ($relevance >= 0.4) return 'Fair'; | |
| 1037 | - if ($relevance >= 0.2) return 'Poor'; | |
| 1038 | - return 'Off-topic'; | |
| 1039 | - } | |
| 1040 | - | |
| 1041 | - /** | |
| 1042 | - * Score modern heading structure | |
| 1043 | - * | |
| 1044 | - * @param array $headings Array of headings | |
| 1045 | - * @return array Scoring result | |
| 1046 | - */ | |
| 1047 | - private function score_modern_headings(array $headings): array { | |
| 1048 | - $score = 0; | |
| 1049 | - $max_score = $this->scoring_factors['heading_structure']; | |
| 1050 | - $suggestions = []; | |
| 1051 | - | |
| 1052 | - if (empty($headings)) { | |
| 1053 | - $suggestions[] = 'Add headings (H1, H2, H3) to structure content and improve readability'; | |
| 1054 | - return ['score' => 0, 'max_score' => $max_score, 'suggestions' => $suggestions]; | |
| 1369 | + private function keyword_in_content(string $content, string $target_keyword): bool { | |
| 1370 | + if (empty($content) || empty($target_keyword)) { | |
| 1371 | + return false; | |
| 1055 | 1372 | } |
| 1056 | - | |
| 1057 | - $h1_count = 0; | |
| 1058 | - $h2_count = 0; | |
| 1059 | - $h3_count = 0; | |
| 1060 | - | |
| 1061 | - foreach ($headings as $heading) { | |
| 1062 | - $level = $heading['level'] ?? 1; | |
| 1063 | - switch ($level) { | |
| 1064 | - case 1: $h1_count++; break; | |
| 1065 | - case 2: $h2_count++; break; | |
| 1066 | - case 3: $h3_count++; break; | |
| 1067 | - } | |
| 1373 | + $plain = strtolower(wp_strip_all_tags($content)); | |
| 1374 | + if (strpos($plain, strtolower($target_keyword)) !== false) { | |
| 1375 | + return true; | |
| 1068 | 1376 | } |
| 1069 | - | |
| 1070 | - // H1 optimization (4 points) | |
| 1071 | - if ($h1_count === 1) { | |
| 1072 | - $score += 4; | |
| 1073 | - } elseif ($h1_count === 0) { | |
| 1074 | - $suggestions[] = 'Add exactly one H1 heading for your main topic'; | |
| 1075 | - } else { | |
| 1076 | - $suggestions[] = 'Use only one H1 heading per page for better SEO'; | |
| 1077 | - } | |
| 1078 | - | |
| 1079 | - // H2 structure (4 points) | |
| 1080 | - if ($h2_count >= 2 && $h2_count <= 8) { | |
| 1081 | - $score += 4; | |
| 1082 | - } elseif ($h2_count === 1) { | |
| 1083 | - $score += 2; | |
| 1084 | - $suggestions[] = 'Add more H2 headings to break content into logical sections'; | |
| 1085 | - } elseif ($h2_count === 0) { | |
| 1086 | - $suggestions[] = 'Add H2 headings to organize content into main sections'; | |
| 1087 | - } else { | |
| 1088 | - $suggestions[] = 'Too many H2 headings - consider consolidating some sections'; | |
| 1089 | - } | |
| 1090 | - | |
| 1091 | - // H3 usage (4 points) | |
| 1092 | - if ($h3_count >= 1 && $h3_count <= 12) { | |
| 1093 | - $score += 4; | |
| 1094 | - } elseif ($h3_count === 0 && $h2_count >= 2) { | |
| 1095 | - $score += 2; | |
| 1096 | - $suggestions[] = 'Consider adding H3 subheadings for better content organization'; | |
| 1097 | - } | |
| 1098 | - | |
| 1099 | - return [ | |
| 1100 | - 'score' => $score, | |
| 1101 | - 'max_score' => $max_score, | |
| 1102 | - 'suggestions' => $suggestions, | |
| 1103 | - 'details' => [ | |
| 1104 | - 'h1_count' => $h1_count, | |
| 1105 | - 'h2_count' => $h2_count, | |
| 1106 | - 'h3_count' => $h3_count, | |
| 1107 | - 'total_headings' => count($headings), | |
| 1108 | - 'structure_quality' => $this->assess_heading_structure($h1_count, $h2_count, $h3_count), | |
| 1109 | - ] | |
| 1110 | - ]; | |
| 1377 | + return $this->check_semantic_keyword_match($plain, $target_keyword); | |
| 1111 | 1378 | } |
| 1112 | 1379 | |
| 1113 | 1380 | /** |
| 1114 | - * Assess heading structure quality | |
| 1381 | + * Whether the keyword appears early (first paragraph / first ~10% of words). | |
| 1115 | 1382 | * |
| 1116 | - * @param int $h1_count H1 count | |
| 1117 | - * @param int $h2_count H2 count | |
| 1118 | - * @param int $h3_count H3 count | |
| 1119 | - * @return string Structure quality | |
| 1383 | + * @param string $content Raw/HTML content. | |
| 1384 | + * @param string $target_keyword Target keyword. | |
| 1385 | + * @return bool | |
| 1120 | 1386 | */ |
| 1121 | - private function assess_heading_structure(int $h1_count, int $h2_count, int $h3_count): string { | |
| 1122 | - if ($h1_count === 1 && $h2_count >= 2 && $h3_count >= 1) return 'Excellent'; | |
| 1123 | - if ($h1_count === 1 && $h2_count >= 2) return 'Good'; | |
| 1124 | - if ($h1_count === 1 && $h2_count >= 1) return 'Fair'; | |
| 1125 | - if ($h1_count === 1) return 'Basic'; | |
| 1126 | - return 'Poor'; | |
| 1387 | + private function keyword_in_first_paragraph(string $content, string $target_keyword): bool { | |
| 1388 | + if (empty($content) || empty($target_keyword)) { | |
| 1389 | + return false; | |
| 1390 | + } | |
| 1391 | + $plain = strtolower(wp_strip_all_tags($content)); | |
| 1392 | + $words = preg_split('/\s+/', trim($plain), -1, PREG_SPLIT_NO_EMPTY) ?: []; | |
| 1393 | + $window = array_slice($words, 0, max(50, (int) ceil(count($words) * 0.1))); | |
| 1394 | + return strpos(implode(' ', $window), strtolower($target_keyword)) !== false; | |
| 1127 | 1395 | } |
| 1128 | 1396 | |
| 1129 | 1397 | /** |
| 1130 | - * Score semantic relevance using modern NLP principles | |
| 1398 | + * Whether the keyword appears in any subheading (H2–H6). | |
| 1131 | 1399 | * |
| 1132 | - * @param string $content Content text | |
| 1133 | - * @param string $target_keyword Target keyword | |
| 1134 | - * @return array Scoring result | |
| 1400 | + * @param array $headings Extracted headings (each with 'level' + 'text'). | |
| 1401 | + * @param string $target_keyword Target keyword. | |
| 1402 | + * @return bool | |
| 1135 | 1403 | */ |
| 1136 | - private function score_semantic_relevance(string $content, string $target_keyword): array { | |
| 1137 | - $score = 0; | |
| 1138 | - $max_score = $this->scoring_factors['semantic_relevance']; | |
| 1139 | - $suggestions = []; | |
| 1140 | - | |
| 1141 | - if (empty($content)) { | |
| 1142 | - $suggestions[] = 'Add content to analyze semantic relevance'; | |
| 1143 | - return ['score' => 0, 'max_score' => $max_score, 'suggestions' => $suggestions]; | |
| 1144 | - } | |
| 1145 | - | |
| 1404 | + private function keyword_in_subheadings(array $headings, string $target_keyword): bool { | |
| 1146 | 1405 | if (empty($target_keyword)) { |
| 1147 | - $score += 7; // Partial credit | |
| 1148 | - $suggestions[] = 'Set a target keyword to optimize semantic relevance'; | |
| 1149 | - return ['score' => $score, 'max_score' => $max_score, 'suggestions' => $suggestions]; | |
| 1406 | + return false; | |
| 1150 | 1407 | } |
| 1151 | - | |
| 1152 | - // Calculate semantic relevance | |
| 1153 | - $relevance = $this->calculate_topic_relevance($content, $target_keyword); | |
| 1154 | - $score = round($relevance * $max_score); | |
| 1155 | - | |
| 1156 | - // Provide specific suggestions based on relevance score | |
| 1157 | - if ($relevance >= 0.8) { | |
| 1158 | - $suggestions[] = 'Excellent semantic relevance - content strongly matches target topic'; | |
| 1159 | - } elseif ($relevance >= 0.6) { | |
| 1160 | - $suggestions[] = 'Good semantic relevance - consider adding more related terms'; | |
| 1161 | - } elseif ($relevance >= 0.4) { | |
| 1162 | - $suggestions[] = "Improve content relevance to '{$target_keyword}' with related terms and concepts"; | |
| 1163 | - } else { | |
| 1164 | - $suggestions[] = "Content lacks focus on '{$target_keyword}' - add more relevant information"; | |
| 1408 | + $keyword_lower = strtolower($target_keyword); | |
| 1409 | + foreach ($headings as $heading) { | |
| 1410 | + if ((int) ($heading['level'] ?? 0) < 2) { | |
| 1411 | + continue; | |
| 1412 | + } | |
| 1413 | + $text = strtolower((string) ($heading['text'] ?? '')); | |
| 1414 | + if ($text !== '' && strpos($text, $keyword_lower) !== false) { | |
| 1415 | + return true; | |
| 1416 | + } | |
| 1165 | 1417 | } |
| 1166 | - | |
| 1167 | - return [ | |
| 1168 | - 'score' => $score, | |
| 1169 | - 'max_score' => $max_score, | |
| 1170 | - 'suggestions' => $suggestions, | |
| 1171 | - 'details' => [ | |
| 1172 | - 'relevance_score' => round($relevance * 100, 1), | |
| 1173 | - 'keyword_density' => $this->calculate_keyword_density($content, $target_keyword), | |
| 1174 | - 'semantic_terms_found' => $this->count_semantic_terms($content, $target_keyword), | |
| 1175 | - ] | |
| 1176 | - ]; | |
| 1418 | + return false; | |
| 1177 | 1419 | } |
| 1178 | 1420 | |
| 1179 | 1421 | /** |
| 1180 | - * Calculate keyword density | |
| 1422 | + * Days elapsed since a MySQL datetime string, or null when unparseable. | |
| 1181 | 1423 | * |
| 1182 | - * @param string $content Content text | |
| 1183 | - * @param string $keyword Target keyword | |
| 1184 | - * @return float Keyword density percentage | |
| 1424 | + * @param string $datetime MySQL datetime (e.g. post_modified). | |
| 1425 | + * @return int|null | |
| 1185 | 1426 | */ |
| 1186 | - private function calculate_keyword_density(string $content, string $keyword): float { | |
| 1187 | - if (empty($content) || empty($keyword)) { | |
| 1188 | - return 0.0; | |
| 1427 | + private function days_since(string $datetime): ?int { | |
| 1428 | + $datetime = trim($datetime); | |
| 1429 | + if ($datetime === '' || strpos($datetime, '0000-00-00') === 0) { | |
| 1430 | + return null; | |
| 1189 | 1431 | } |
| 1190 | - | |
| 1191 | - $content_lower = strtolower(wp_strip_all_tags($content)); | |
| 1192 | - $keyword_lower = strtolower($keyword); | |
| 1193 | - | |
| 1194 | - $keyword_count = substr_count($content_lower, $keyword_lower); | |
| 1195 | - $word_count = $this->calculate_word_count_js_style($content_lower); | |
| 1196 | - | |
| 1197 | - return $word_count > 0 ? round(($keyword_count / $word_count) * 100, 2) : 0.0; | |
| 1432 | + $ts = strtotime($datetime); | |
| 1433 | + if ($ts === false) { | |
| 1434 | + return null; | |
| 1435 | + } | |
| 1436 | + // strtotime() returns a real unix timestamp, so this must compare against | |
| 1437 | + // one: current_time('timestamp') is offset by the site timezone and made | |
| 1438 | + // every "days ago" figure wrong by that offset. | |
| 1439 | + $now = time(); | |
| 1440 | + return (int) floor(($now - $ts) / 86400); | |
| 1198 | 1441 | } |
| 1199 | 1442 | |
| 1200 | 1443 | /** |
| 1201 | - * Count semantic terms found in content | |
| 1444 | + * Whether the keyword appears in the meta description. | |
| 1202 | 1445 | * |
| 1203 | - * @param string $content Content text | |
| 1204 | - * @param string $keyword Target keyword | |
| 1205 | - * @return int Number of semantic terms found | |
| 1446 | + * @param string $meta_description Meta description text. | |
| 1447 | + * @param string $target_keyword Target keyword. | |
| 1448 | + * @return bool | |
| 1206 | 1449 | */ |
| 1207 | - private function count_semantic_terms(string $content, string $keyword): int { | |
| 1208 | - $semantic_terms = $this->get_semantic_terms($keyword); | |
| 1209 | - $content_lower = strtolower($content); | |
| 1210 | - $found = 0; | |
| 1211 | - | |
| 1212 | - foreach ($semantic_terms as $term) { | |
| 1213 | - if (strpos($content_lower, strtolower($term)) !== false) { | |
| 1214 | - $found++; | |
| 1215 | - } | |
| 1450 | + private function keyword_in_meta(string $meta_description, string $target_keyword): bool { | |
| 1451 | + if ($meta_description === '' || $target_keyword === '') { | |
| 1452 | + return false; | |
| 1216 | 1453 | } |
| 1217 | - | |
| 1218 | - return $found; | |
| 1454 | + return strpos(strtolower($meta_description), strtolower($target_keyword)) !== false; | |
| 1219 | 1455 | } |
| 1220 | 1456 | |
| 1221 | 1457 | /** |
| 1222 | - * Score user experience factors | |
| 1458 | + * Whether the keyword appears in any image alt text. | |
| 1223 | 1459 | * |
| 1224 | - * @param array $content_data Content analysis data | |
| 1225 | - * @return array Scoring result | |
| 1460 | + * @param array $images Images (each with an 'alt' key). | |
| 1461 | + * @param string $target_keyword Target keyword. | |
| 1462 | + * @return bool | |
| 1226 | 1463 | */ |
| 1227 | - private function score_user_experience(array $content_data): array { | |
| 1228 | - $score = 0; | |
| 1229 | - $max_score = $this->scoring_factors['user_experience']; | |
| 1230 | - $suggestions = []; | |
| 1231 | - | |
| 1232 | - // Readability score (6 points) | |
| 1233 | - $readability = $content_data['readability_score'] ?? 0; | |
| 1234 | - if ($readability >= 60) { | |
| 1235 | - $score += 6; | |
| 1236 | - } elseif ($readability >= 40) { | |
| 1237 | - $score += 4; | |
| 1238 | - $suggestions[] = 'Improve readability with shorter sentences and simpler words'; | |
| 1239 | - } else { | |
| 1240 | - $score += 2; | |
| 1241 | - $suggestions[] = 'Content is difficult to read - use shorter sentences and common words'; | |
| 1464 | + private function keyword_in_alt(array $images, string $target_keyword): bool { | |
| 1465 | + if ($target_keyword === '') { | |
| 1466 | + return false; | |
| 1242 | 1467 | } |
| 1243 | - | |
| 1244 | - // Content scannability (4 points) | |
| 1245 | - $headings = $content_data['headings'] ?? []; | |
| 1246 | - $word_count = $content_data['word_count'] ?? 0; | |
| 1247 | - | |
| 1248 | - if (count($headings) >= 3 && $word_count > 500) { | |
| 1249 | - $score += 4; | |
| 1250 | - } elseif (count($headings) >= 1) { | |
| 1251 | - $score += 2; | |
| 1252 | - $suggestions[] = 'Add more headings to make content easier to scan'; | |
| 1253 | - } else { | |
| 1254 | - $suggestions[] = 'Add headings to break up text and improve scannability'; | |
| 1468 | + $keyword_lower = strtolower($target_keyword); | |
| 1469 | + foreach ($images as $image) { | |
| 1470 | + $alt = strtolower((string) ($image['alt'] ?? '')); | |
| 1471 | + if ($alt !== '' && strpos($alt, $keyword_lower) !== false) { | |
| 1472 | + return true; | |
| 1473 | + } | |
| 1255 | 1474 | } |
| 1256 | - | |
| 1257 | - return [ | |
| 1258 | - 'score' => $score, | |
| 1259 | - 'max_score' => $max_score, | |
| 1260 | - 'suggestions' => $suggestions, | |
| 1261 | - 'details' => [ | |
| 1262 | - 'readability_score' => $readability, | |
| 1263 | - 'readability_level' => $this->get_readability_level($readability), | |
| 1264 | - 'scannability' => count($headings) >= 3 ? 'Good' : 'Needs improvement', | |
| 1265 | - ] | |
| 1266 | - ]; | |
| 1475 | + return false; | |
| 1267 | 1476 | } |
| 1268 | 1477 | |
| 1269 | 1478 | /** |
| 1270 | - * Get readability level description | |
| 1479 | + * Whether the title contains a common power word (CTR booster). | |
| 1271 | 1480 | * |
| 1272 | - * @param float $score Readability score | |
| 1273 | - * @return string Readability level | |
| 1481 | + * @param string $title Post title. | |
| 1482 | + * @return bool | |
| 1274 | 1483 | */ |
| 1275 | - private function get_readability_level(float $score): string { | |
| 1276 | - if ($score >= 90) return 'Very Easy'; | |
| 1277 | - if ($score >= 80) return 'Easy'; | |
| 1278 | - if ($score >= 70) return 'Fairly Easy'; | |
| 1279 | - if ($score >= 60) return 'Standard'; | |
| 1280 | - if ($score >= 50) return 'Fairly Difficult'; | |
| 1281 | - if ($score >= 30) return 'Difficult'; | |
| 1282 | - return 'Very Difficult'; | |
| 1484 | + private function title_has_power_word(string $title): bool { | |
| 1485 | + $title_lower = strtolower($title); | |
| 1486 | + foreach (self::get_title_power_words() as $word) { | |
| 1487 | + if (strpos($title_lower, $word) !== false) { | |
| 1488 | + return true; | |
| 1489 | + } | |
| 1490 | + } | |
| 1491 | + return false; | |
| 1283 | 1492 | } |
| 1284 | 1493 | |
| 1285 | 1494 | /** |
| 1286 | - * Score strategic internal linking | |
| 1495 | + * The power words rewarded by the title check. Single source of truth so the | |
| 1496 | + * AI title improver can require the generated title to actually contain one. | |
| 1287 | 1497 | * |
| 1288 | - * @param array $content_data Content analysis data | |
| 1289 | - * @return array Scoring result | |
| 1498 | + * @return string[] | |
| 1290 | 1499 | */ |
| 1291 | - private function score_strategic_linking(array $content_data): array { | |
| 1292 | - $score = 0; | |
| 1293 | - $max_score = $this->scoring_factors['internal_linking']; | |
| 1294 | - $suggestions = []; | |
| 1295 | - | |
| 1296 | - $internal_links = $content_data['internal_links'] ?? 0; | |
| 1297 | - $word_count = $content_data['word_count'] ?? 0; | |
| 1298 | - | |
| 1299 | - // Calculate optimal link ratio (1 link per 150-200 words) | |
| 1300 | - $optimal_links = max(1, round($word_count / 175)); | |
| 1301 | - | |
| 1302 | - if ($internal_links >= $optimal_links) { | |
| 1303 | - $score += 8; | |
| 1304 | - } elseif ($internal_links >= max(1, $optimal_links - 1)) { | |
| 1305 | - $score += 6; | |
| 1306 | - $suggestions[] = 'Consider adding 1-2 more internal links to related content'; | |
| 1307 | - } elseif ($internal_links >= 1) { | |
| 1308 | - $score += 4; | |
| 1309 | - $suggestions[] = 'Add more internal links to improve site navigation and SEO'; | |
| 1310 | - } else { | |
| 1311 | - $suggestions[] = 'Add internal links to related pages to improve site structure'; | |
| 1312 | - } | |
| 1313 | - | |
| 1500 | + public static function get_title_power_words(): array { | |
| 1314 | 1501 | return [ |
| 1315 | - 'score' => $score, | |
| 1316 | - 'max_score' => $max_score, | |
| 1317 | - 'suggestions' => $suggestions, | |
| 1318 | - 'details' => [ | |
| 1319 | - 'current_links' => $internal_links, | |
| 1320 | - 'optimal_links' => $optimal_links, | |
| 1321 | - 'link_ratio' => $word_count > 0 ? round($word_count / max(1, $internal_links)) : 0, | |
| 1322 | - ] | |
| 1502 | + 'ultimate', 'essential', 'complete', 'proven', 'guide', 'best', 'top', | |
| 1503 | + 'free', 'easy', 'simple', 'quick', 'fast', 'powerful', 'secret', 'expert', | |
| 1504 | + 'effective', 'amazing', 'incredible', 'exclusive', 'definitive', 'step-by-step', | |
| 1323 | 1505 | ]; |
| 1324 | 1506 | } |
| 1325 | 1507 | |
| 1326 | 1508 | /** |
| 1327 | - * Score external authority signals | |
| 1509 | + * The emotion/sentiment words rewarded by the title check. Single source of | |
| 1510 | + * truth shared with the AI title improver so a generated title satisfies the | |
| 1511 | + * same validation. | |
| 1328 | 1512 | * |
| 1329 | - * @param array $content_data Content analysis data | |
| 1330 | - * @return array Scoring result | |
| 1513 | + * @return string[] | |
| 1331 | 1514 | */ |
| 1332 | - private function score_external_authority(array $content_data): array { | |
| 1333 | - $score = 0; | |
| 1334 | - $max_score = $this->scoring_factors['external_authority']; | |
| 1335 | - $suggestions = []; | |
| 1336 | - | |
| 1337 | - $external_links = $content_data['external_links'] ?? 0; | |
| 1338 | - | |
| 1339 | - if ($external_links >= 2 && $external_links <= 5) { | |
| 1340 | - $score += 7; | |
| 1341 | - } elseif ($external_links === 1) { | |
| 1342 | - $score += 5; | |
| 1343 | - $suggestions[] = 'Consider adding 1-2 more high-quality external links to authoritative sources'; | |
| 1344 | - } elseif ($external_links === 0) { | |
| 1345 | - $score += 2; | |
| 1346 | - $suggestions[] = 'Add external links to authoritative sources to support your content'; | |
| 1347 | - } else { | |
| 1348 | - $score += 4; | |
| 1349 | - $suggestions[] = 'Too many external links - focus on 2-5 high-quality authoritative sources'; | |
| 1350 | - } | |
| 1351 | - | |
| 1515 | + public static function get_title_sentiment_words(): array { | |
| 1352 | 1516 | return [ |
| 1353 | - 'score' => $score, | |
| 1354 | - 'max_score' => $max_score, | |
| 1355 | - 'suggestions' => $suggestions, | |
| 1356 | - 'details' => [ | |
| 1357 | - 'external_links' => $external_links, | |
| 1358 | - 'optimal_range' => '2-5 links', | |
| 1359 | - 'authority_signal' => $external_links > 0 ? 'Present' : 'Missing', | |
| 1360 | - ] | |
| 1517 | + // Positive | |
| 1518 | + 'great', 'good', 'better', 'awesome', 'love', 'win', 'boost', 'improve', | |
| 1519 | + 'success', 'smart', 'brilliant', 'perfect', 'happy', 'beautiful', | |
| 1520 | + // Negative (drives clicks too) | |
| 1521 | + 'avoid', 'mistake', 'worst', 'stop', 'never', 'bad', 'wrong', 'fail', | |
| 1522 | + 'danger', 'warning', 'painful', 'ugly', | |
| 1361 | 1523 | ]; |
| 1362 | 1524 | } |
| 1363 | 1525 | |
| 1364 | 1526 | /** |
| 1365 | - * Score technical SEO factors | |
| 1527 | + * Whether the title carries an emotional/sentiment word (positive or | |
| 1528 | + * negative), which Rank Math rewards for higher engagement. | |
| 1366 | 1529 | * |
| 1367 | - * @param array $content_data Content analysis data | |
| 1368 | - * @param array $metadata Post metadata | |
| 1369 | - * @param array $options Additional options | |
| 1370 | - * @return array Scoring result | |
| 1530 | + * @param string $title Post title. | |
| 1531 | + * @return bool | |
| 1371 | 1532 | */ |
| 1372 | - private function score_technical_seo(array $content_data, array $metadata, array $options): array { | |
| 1373 | - $score = 0; | |
| 1374 | - $max_score = $this->scoring_factors['technical_seo']; | |
| 1375 | - $suggestions = []; | |
| 1376 | - | |
| 1377 | - // Meta description (2 points) | |
| 1378 | - $meta_desc = $metadata['description'] ?? ''; | |
| 1379 | - $meta_length = strlen($meta_desc); | |
| 1380 | - | |
| 1381 | - if ($meta_length >= 120 && $meta_length <= 160) { | |
| 1382 | - $score += 2; | |
| 1383 | - } elseif ($meta_length >= 100 && $meta_length <= 180) { | |
| 1384 | - $score += 1; | |
| 1385 | - $suggestions[] = 'Optimize meta description length to 120-160 characters'; | |
| 1386 | - } else { | |
| 1387 | - $suggestions[] = empty($meta_desc) ? | |
| 1388 | - 'Add a compelling meta description (120-160 characters)' : | |
| 1389 | - 'Adjust meta description length to 120-160 characters'; | |
| 1390 | - } | |
| 1391 | - | |
| 1392 | - // Image optimization (2 points) | |
| 1393 | - $images = $content_data['images'] ?? []; | |
| 1394 | - if (!empty($images)) { | |
| 1395 | - $images_with_alt = array_filter($images, function($img) { | |
| 1396 | - return !empty($img['alt']); | |
| 1397 | - }); | |
| 1398 | - | |
| 1399 | - $alt_ratio = count($images_with_alt) / count($images); | |
| 1400 | - if ($alt_ratio >= 0.9) { | |
| 1401 | - $score += 2; | |
| 1402 | - } elseif ($alt_ratio >= 0.7) { | |
| 1403 | - $score += 1; | |
| 1404 | - $suggestions[] = 'Add alt text to remaining images for better accessibility'; | |
| 1405 | - } else { | |
| 1406 | - $suggestions[] = 'Add descriptive alt text to all images'; | |
| 1533 | + private function title_has_sentiment_word(string $title): bool { | |
| 1534 | + $title_lower = strtolower($title); | |
| 1535 | + foreach (self::get_title_sentiment_words() as $word) { | |
| 1536 | + if (strpos($title_lower, $word) !== false) { | |
| 1537 | + return true; | |
| 1407 | 1538 | } |
| 1408 | - } else { | |
| 1409 | - $score += 1; // Partial credit if no images | |
| 1410 | 1539 | } |
| 1411 | - | |
| 1412 | - // URL structure (1 point) | |
| 1413 | - $url = $content_data['url'] ?? ''; | |
| 1414 | - if (!empty($url)) { | |
| 1415 | - $slug = basename(wp_parse_url($url, PHP_URL_PATH)); | |
| 1416 | - if (strlen($slug) <= 60 && !empty($options['target_keyword'])) { | |
| 1417 | - if (strpos(strtolower($slug), strtolower($options['target_keyword'])) !== false) { | |
| 1418 | - $score += 1; | |
| 1419 | - } else { | |
| 1420 | - $suggestions[] = 'Include target keyword in URL slug for better SEO'; | |
| 1421 | - } | |
| 1422 | - } else { | |
| 1423 | - $suggestions[] = 'Optimize URL structure - keep it short and descriptive'; | |
| 1424 | - } | |
| 1425 | - } | |
| 1426 | - | |
| 1427 | - return [ | |
| 1428 | - 'score' => $score, | |
| 1429 | - 'max_score' => $max_score, | |
| 1430 | - 'suggestions' => $suggestions, | |
| 1431 | - 'details' => [ | |
| 1432 | - 'meta_description_length' => $meta_length, | |
| 1433 | - 'images_with_alt' => !empty($images) ? count(array_filter($images, function($img) { return !empty($img['alt']); })) : 0, | |
| 1434 | - 'total_images' => count($images), | |
| 1435 | - 'url_optimized' => !empty($url) && !empty($options['target_keyword']) ? | |
| 1436 | - strpos(strtolower(basename(wp_parse_url($url, PHP_URL_PATH))), strtolower($options['target_keyword'])) !== false : false, | |
| 1437 | - ] | |
| 1438 | - ]; | |
| 1540 | + return false; | |
| 1439 | 1541 | } |
| 1440 | 1542 | |
| 1441 | 1543 | /** |
| 1442 | - * Score content freshness | |
| 1544 | + * Assess content depth based on word count | |
| 1443 | 1545 | * |
| 1444 | - * @param array $content_data Content analysis data | |
| 1445 | - * @return array Scoring result | |
| 1546 | + * @param int $word_count Word count | |
| 1547 | + * @return string Depth assessment | |
| 1446 | 1548 | */ |
| 1549 | + private function assess_content_depth(int $word_count): string { | |
| 1550 | + if ($word_count >= 2000) { return 'Comprehensive'; | |
| 1551 | + } | |
| 1552 | + if ($word_count >= 1000) { return 'Detailed'; | |
| 1553 | + } | |
| 1554 | + if ($word_count >= 500) { return 'Moderate'; | |
| 1555 | + } | |
| 1556 | + if ($word_count >= 300) { return 'Basic'; | |
| 1557 | + } | |
| 1558 | + return 'Insufficient'; | |
| 1559 | + } | |
| 1447 | 1560 | private function score_content_freshness(array $content_data): array { |
| 1448 | - $score = 3; // Default full score for new content | |
| 1449 | 1561 | $max_score = $this->scoring_factors['content_freshness']; |
| 1450 | 1562 | $suggestions = []; |
| 1451 | 1563 | |
| 1452 | - // For now, give full score - can be enhanced with post date analysis | |
| 1453 | - $suggestions[] = 'Keep content updated regularly for better freshness signals'; | |
| 1564 | + // Derive freshness from the real last-modified date when available | |
| 1565 | + // (live editing has no stored date yet -> treat as fresh). | |
| 1566 | + $days = $this->days_since((string) ($content_data['post_modified'] ?? '')); | |
| 1454 | 1567 | |
| 1568 | + if ($days === null || $days <= 180) { | |
| 1569 | + $score = $max_score; | |
| 1570 | + $status = 'Current'; | |
| 1571 | + } elseif ($days <= 365) { | |
| 1572 | + $score = (int) round($max_score * 0.66); | |
| 1573 | + $status = 'Aging'; | |
| 1574 | + $suggestions[] = 'Content is 6-12 months old - review and refresh it for better freshness signals'; | |
| 1575 | + } else { | |
| 1576 | + $score = (int) round($max_score * 0.33); | |
| 1577 | + $status = 'Stale'; | |
| 1578 | + $suggestions[] = 'Content is over a year old - update it to maintain freshness signals'; | |
| 1579 | + } | |
| 1580 | + | |
| 1455 | 1581 | return [ |
| 1456 | 1582 | 'score' => $score, |
| 1457 | 1583 | 'max_score' => $max_score, |
| 1458 | 1584 | 'suggestions' => $suggestions, |
| 1459 | 1585 | 'details' => [ |
| 1460 | - 'freshness_status' => 'Current', | |
| 1461 | - 'last_updated' => current_time('mysql'), | |
| 1462 | - ] | |
| 1586 | + 'freshness_status' => $status, | |
| 1587 | + 'days_since_modified' => $days, | |
| 1588 | + ], | |
| 1463 | 1589 | ]; |
| 1464 | 1590 | } |
| 1465 | - | |
| 1466 | 1591 | /** |
| 1467 | - * Score mobile optimization | |
| 1592 | + * Resolve a post's content into something worth analyzing. | |
| 1468 | 1593 | * |
| 1469 | - * @param array $content_data Content analysis data | |
| 1470 | - * @return array Scoring result | |
| 1594 | + * Delegates to Builder_Content, which knows where each page builder keeps | |
| 1595 | + * its text. Kept as the historical entry point for existing callers. | |
| 1596 | + * | |
| 1597 | + * @since 1.23.0 | |
| 1598 | + * | |
| 1599 | + * @param \WP_Post $post Post being analyzed. | |
| 1600 | + * @return string Content to analyze. | |
| 1471 | 1601 | */ |
| 1472 | - private function score_mobile_optimization(array $content_data): array { | |
| 1473 | - $score = 2; // Default full score - basic mobile checks | |
| 1474 | - $max_score = $this->scoring_factors['mobile_optimization']; | |
| 1475 | - $suggestions = []; | |
| 1602 | + public static function resolve_analyzable_content(\WP_Post $post): string { | |
| 1603 | + if (!class_exists('\ThinkRank\SEO\Builder_Content')) { | |
| 1604 | + require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-builder-content.php'; | |
| 1605 | + } | |
| 1476 | 1606 | |
| 1477 | - // For now, give full score - can be enhanced with actual mobile testing | |
| 1478 | - $suggestions[] = 'Ensure content displays well on mobile devices'; | |
| 1479 | - | |
| 1480 | - return [ | |
| 1481 | - 'score' => $score, | |
| 1482 | - 'max_score' => $max_score, | |
| 1483 | - 'suggestions' => $suggestions, | |
| 1484 | - 'details' => [ | |
| 1485 | - 'mobile_friendly' => 'Assumed', | |
| 1486 | - 'responsive_design' => 'Theme dependent', | |
| 1487 | - ] | |
| 1488 | - ]; | |
| 1607 | + return \ThinkRank\SEO\Builder_Content::resolve($post); | |
| 1489 | 1608 | } |
| 1490 | 1609 | |
| 1491 | 1610 | /** |
| 1492 | - * Analyze post content and extract data for scoring | |
| 1611 | + * Resolve editor-supplied live content into something worth analyzing. | |
| 1493 | 1612 | * |
| 1494 | - * @param int $post_id Post ID | |
| 1495 | - * @return array Content analysis data | |
| 1613 | + * @since 1.23.0 | |
| 1614 | + * | |
| 1615 | + * @param string $live_content Markup supplied by the editor. | |
| 1616 | + * @param \WP_Post $post Post the markup belongs to. | |
| 1617 | + * @return string Content to analyze. | |
| 1496 | 1618 | */ |
| 1619 | + public static function resolve_live_content(string $live_content, \WP_Post $post): string { | |
| 1620 | + if (!class_exists('\ThinkRank\SEO\Builder_Content')) { | |
| 1621 | + require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-builder-content.php'; | |
| 1622 | + } | |
| 1623 | + | |
| 1624 | + return \ThinkRank\SEO\Builder_Content::resolve_markup($live_content, $post); | |
| 1625 | + } | |
| 1626 | + | |
| 1497 | 1627 | public function analyze_post_content(int $post_id): array { |
| 1498 | 1628 | $post = get_post($post_id); |
| 1499 | 1629 | if (!$post) { |
| 1500 | 1630 | return []; |
| @@ -1499,9 +1629,9 @@ | ||
| 1499 | 1629 | if (!$post) { |
| 1500 | 1630 | return []; |
| 1501 | 1631 | } |
| 1502 | 1632 | |
| 1503 | - $content = $post->post_content; | |
| 1633 | + $content = self::resolve_analyzable_content($post); | |
| 1504 | 1634 | $title = $post->post_title; |
| 1505 | 1635 | |
| 1506 | 1636 | // Extract headings from content |
| 1507 | 1637 | $headings = $this->extract_headings($content); |
| @@ -1515,8 +1645,9 @@ | ||
| 1515 | 1645 | |
| 1516 | 1646 | // Count links |
| 1517 | 1647 | $internal_links = $this->count_internal_links($content); |
| 1518 | 1648 | $external_links = $this->count_external_links($content); |
| 1649 | + $external_dofollow_links = $this->count_external_dofollow_links($content); | |
| 1519 | 1650 | |
| 1520 | 1651 | // Analyze images |
| 1521 | 1652 | $images = $this->analyze_images($content); |
| 1522 | 1653 | |
| @@ -1530,14 +1661,135 @@ | ||
| 1530 | 1661 | 'word_count' => $word_count, |
| 1531 | 1662 | 'readability_score' => $readability_score, |
| 1532 | 1663 | 'internal_links' => $internal_links, |
| 1533 | 1664 | 'external_links' => $external_links, |
| 1665 | + 'external_dofollow_links' => $external_dofollow_links, | |
| 1534 | 1666 | 'images' => $images, |
| 1535 | 1667 | 'url' => $url, |
| 1668 | + 'slug' => $post->post_name, | |
| 1669 | + 'post_modified' => $post->post_modified, | |
| 1670 | + 'schema_present' => $this->detect_schema_present($content) | |
| 1671 | + || $this->thinkrank_global_schema_active($post->post_type) | |
| 1672 | + || $this->thinkrank_deployed_schema_active($post), | |
| 1536 | 1673 | ]; |
| 1537 | 1674 | } |
| 1538 | 1675 | |
| 1539 | 1676 | /** |
| 1677 | + * Build the human-readable readability label (mirrors the editor's | |
| 1678 | + * calculateReadabilityScore: "<level> (<flesch>)") from analyzed content. | |
| 1679 | + * | |
| 1680 | + * @param array $content_data Output of analyze_post_content()/analyze_live_content() | |
| 1681 | + * @return string Readability label, e.g. "Standard (62)" | |
| 1682 | + */ | |
| 1683 | + private function format_readability_label(array $content_data): string { | |
| 1684 | + if ((int) ($content_data['word_count'] ?? 0) === 0) { | |
| 1685 | + return 'No content'; | |
| 1686 | + } | |
| 1687 | + | |
| 1688 | + $rounded = (int) round((float) ($content_data['readability_score'] ?? 0)); | |
| 1689 | + | |
| 1690 | + if ($rounded >= 90) { | |
| 1691 | + $level = 'Very Easy'; | |
| 1692 | + } elseif ($rounded >= 80) { | |
| 1693 | + $level = 'Easy'; | |
| 1694 | + } elseif ($rounded >= 70) { | |
| 1695 | + $level = 'Fairly Easy'; | |
| 1696 | + } elseif ($rounded >= 60) { | |
| 1697 | + $level = 'Standard'; | |
| 1698 | + } elseif ($rounded >= 50) { | |
| 1699 | + $level = 'Fairly Difficult'; | |
| 1700 | + } elseif ($rounded >= 30) { | |
| 1701 | + $level = 'Difficult'; | |
| 1702 | + } else { | |
| 1703 | + $level = 'Very Difficult'; | |
| 1704 | + } | |
| 1705 | + | |
| 1706 | + return "{$level} ({$rounded})"; | |
| 1707 | + } | |
| 1708 | + | |
| 1709 | + /** | |
| 1710 | + * Derive the content-quality label (mirrors the editor's | |
| 1711 | + * calculateContentQuality: word count + long-paragraph + subheading scoring) | |
| 1712 | + * so persisted scores carry a non-null quality value. | |
| 1713 | + * | |
| 1714 | + * @param array $content_data Output of analyze_post_content()/analyze_live_content() | |
| 1715 | + * @return string One of: No content, Good, OK, Needs improvement | |
| 1716 | + */ | |
| 1717 | + private function derive_content_quality(array $content_data): string { | |
| 1718 | + $word_count = (int) ($content_data['word_count'] ?? 0); | |
| 1719 | + if ($word_count === 0) { | |
| 1720 | + return 'No content'; | |
| 1721 | + } | |
| 1722 | + | |
| 1723 | + $final = $this->derive_content_quality_score($content_data); | |
| 1724 | + | |
| 1725 | + if ($final >= 80) { | |
| 1726 | + return 'Good'; | |
| 1727 | + } | |
| 1728 | + if ($final >= 50) { | |
| 1729 | + return 'OK'; | |
| 1730 | + } | |
| 1731 | + | |
| 1732 | + return 'Needs improvement'; | |
| 1733 | + } | |
| 1734 | + | |
| 1735 | + /** | |
| 1736 | + * Numeric content-quality score (0-100): word count + long-paragraph + | |
| 1737 | + * subheading distribution. Shared by the quality label and the | |
| 1738 | + * satisfying-content factor so both stay in sync. | |
| 1739 | + * | |
| 1740 | + * @param array $content_data Output of analyze_post_content()/analyze_live_content() | |
| 1741 | + * @return int Quality score 0-100. | |
| 1742 | + */ | |
| 1743 | + private function derive_content_quality_score(array $content_data): int { | |
| 1744 | + $word_count = (int) ($content_data['word_count'] ?? 0); | |
| 1745 | + if ($word_count === 0) { | |
| 1746 | + return 0; | |
| 1747 | + } | |
| 1748 | + | |
| 1749 | + $content = (string) ($content_data['content'] ?? ''); | |
| 1750 | + $score = 0; | |
| 1751 | + | |
| 1752 | + // 1. Word count (industry standard: 300+ words). | |
| 1753 | + if ($word_count >= 600) { | |
| 1754 | + $score += 100; | |
| 1755 | + } elseif ($word_count >= 300) { | |
| 1756 | + $score += 50; | |
| 1757 | + } | |
| 1758 | + | |
| 1759 | + // 2. Long paragraphs (flag paragraphs over 150 words). | |
| 1760 | + $long_paragraphs = 0; | |
| 1761 | + if (preg_match_all('/<p[^>]*>(.*?)<\/p>/is', $content, $matches)) { | |
| 1762 | + foreach ($matches[1] as $paragraph) { | |
| 1763 | + if ($this->calculate_word_count_js_style(wp_strip_all_tags($paragraph)) > 150) { | |
| 1764 | + $long_paragraphs++; | |
| 1765 | + } | |
| 1766 | + } | |
| 1767 | + } | |
| 1768 | + if ($long_paragraphs === 0) { | |
| 1769 | + $score += 100; | |
| 1770 | + } elseif ($long_paragraphs <= 2) { | |
| 1771 | + $score += 50; | |
| 1772 | + } | |
| 1773 | + | |
| 1774 | + // 3. Subheading distribution (H2–H6, ~one per 300 words). | |
| 1775 | + $subheadings = 0; | |
| 1776 | + foreach ((array) ($content_data['headings'] ?? []) as $heading) { | |
| 1777 | + if ((int) ($heading['level'] ?? 0) >= 2) { | |
| 1778 | + $subheadings++; | |
| 1779 | + } | |
| 1780 | + } | |
| 1781 | + $expected = (int) floor($word_count / 300); | |
| 1782 | + if ($subheadings > 0 && $subheadings >= $expected) { | |
| 1783 | + $score += 100; | |
| 1784 | + } elseif ($subheadings > 0) { | |
| 1785 | + $score += 50; | |
| 1786 | + } | |
| 1787 | + | |
| 1788 | + return (int) round($score / 3); | |
| 1789 | + } | |
| 1790 | + | |
| 1791 | + /** | |
| 1540 | 1792 | * Extract headings from content |
| 1541 | 1793 | * |
| 1542 | 1794 | * @param string $content Content HTML |
| 1543 | 1795 | * @return array Array of headings with levels |
| @@ -1601,9 +1853,27 @@ | ||
| 1601 | 1853 | $words = preg_split('/\s+/', trim(strtolower(wp_strip_all_tags($text))), -1, PREG_SPLIT_NO_EMPTY); |
| 1602 | 1854 | $syllables = 0; |
| 1603 | 1855 | |
| 1604 | 1856 | foreach ($words as $word) { |
| 1605 | - $syllables += max(1, preg_match_all('/[aeiouy]+/', $word)); | |
| 1857 | + $word = preg_replace('/[^a-z]/', '', $word); | |
| 1858 | + if ($word === '') { | |
| 1859 | + continue; | |
| 1860 | + } | |
| 1861 | + | |
| 1862 | + $groups = preg_match_all('/[aeiouy]+/', $word); | |
| 1863 | + | |
| 1864 | + // Standard Flesch heuristic: a trailing silent e does not form a | |
| 1865 | + // syllable ("make", "time", "these") — but only when a consonant | |
| 1866 | + // precedes it (a vowel+e ending like "movie" already shares its | |
| 1867 | + // group) and never for consonant-le ("table"), which does count. | |
| 1868 | + // Without this the counter inflated syllables/word by ~0.2-0.3 on | |
| 1869 | + // ordinary prose, driving raw Flesch negative and the UI to a | |
| 1870 | + // clamped "Very Difficult (0)" (#407). | |
| 1871 | + if ($groups > 1 && preg_match('/[^aeiouy]e$/', $word) && !str_ends_with($word, 'le')) { | |
| 1872 | + $groups--; | |
| 1873 | + } | |
| 1874 | + | |
| 1875 | + $syllables += max(1, $groups); | |
| 1606 | 1876 | } |
| 1607 | 1877 | |
| 1608 | 1878 | return $syllables; |
| 1609 | 1879 | } |
| @@ -1650,13 +1920,117 @@ | ||
| 1650 | 1920 | return $count; |
| 1651 | 1921 | } |
| 1652 | 1922 | |
| 1653 | 1923 | /** |
| 1924 | + * Count external links that pass link equity (not rel="nofollow"). | |
| 1925 | + * Mirrors Rank Math's "external dofollow link" check. | |
| 1926 | + * | |
| 1927 | + * @param string $content Content HTML | |
| 1928 | + * @return int External dofollow link count | |
| 1929 | + */ | |
| 1930 | + private function count_external_dofollow_links(string $content): int { | |
| 1931 | + $site_url = get_site_url(); | |
| 1932 | + $count = 0; | |
| 1933 | + | |
| 1934 | + if (preg_match_all('/<a\b[^>]*>/i', $content, $matches)) { | |
| 1935 | + foreach ($matches[0] as $tag) { | |
| 1936 | + if (!preg_match('/href=["\']([^"\']+)["\']/i', $tag, $href)) { | |
| 1937 | + continue; | |
| 1938 | + } | |
| 1939 | + $url = $href[1]; | |
| 1940 | + $is_external = strpos($url, 'http') === 0 && strpos($url, $site_url) === false; | |
| 1941 | + if (!$is_external) { | |
| 1942 | + continue; | |
| 1943 | + } | |
| 1944 | + if (preg_match('/rel=["\'][^"\']*\bnofollow\b[^"\']*["\']/i', $tag)) { | |
| 1945 | + continue; | |
| 1946 | + } | |
| 1947 | + $count++; | |
| 1948 | + } | |
| 1949 | + } | |
| 1950 | + | |
| 1951 | + return $count; | |
| 1952 | + } | |
| 1953 | + | |
| 1954 | + /** | |
| 1654 | 1955 | * Analyze images in content |
| 1655 | 1956 | * |
| 1656 | 1957 | * @param string $content Content HTML |
| 1657 | 1958 | * @return array Image analysis data |
| 1658 | 1959 | */ |
| 1960 | + /** | |
| 1961 | + * Detect structured data embedded directly in the content (JSON-LD script | |
| 1962 | + * blocks or microdata attributes). Site-wide schema injected at render time | |
| 1963 | + * is a separate feature and intentionally out of scope here. | |
| 1964 | + * | |
| 1965 | + * @param string $content Raw/HTML content. | |
| 1966 | + * @return bool | |
| 1967 | + */ | |
| 1968 | + private function detect_schema_present(string $content): bool { | |
| 1969 | + if ($content === '') { | |
| 1970 | + return false; | |
| 1971 | + } | |
| 1972 | + return stripos($content, 'application/ld+json') !== false | |
| 1973 | + || stripos($content, 'itemscope') !== false | |
| 1974 | + || stripos($content, 'itemtype') !== false; | |
| 1975 | + } | |
| 1976 | + | |
| 1977 | + /** | |
| 1978 | + * Whether ThinkRank's Global SEO schema output is active for a post type. | |
| 1979 | + * | |
| 1980 | + * ThinkRank injects JSON-LD at render time (wp_head) when a schema type is | |
| 1981 | + * configured for the post type, so a post can have valid structured data | |
| 1982 | + * even when none is embedded in the post body. The score credits this so the | |
| 1983 | + * "add structured data" suggestion reflects ThinkRank's own schema engine. | |
| 1984 | + * | |
| 1985 | + * @param string $post_type Post type slug. | |
| 1986 | + * @return bool | |
| 1987 | + */ | |
| 1988 | + private function thinkrank_global_schema_active(string $post_type): bool { | |
| 1989 | + if ($post_type === '') { | |
| 1990 | + return false; | |
| 1991 | + } | |
| 1992 | + $all_settings = get_option('thinkrank_global_seo_settings', []); | |
| 1993 | + return !empty($all_settings[$post_type]['schema_type']); | |
| 1994 | + } | |
| 1995 | + | |
| 1996 | + /** | |
| 1997 | + * Whether the Schema Manager has an active deployed schema for this post. | |
| 1998 | + * | |
| 1999 | + * Per-post schema deployed from the editor's Schema tab is stored in the | |
| 2000 | + * Schema Manager's own table and emitted at wp_head by | |
| 2001 | + * Frontend\SEO_Manager::output_site_schema_markup(). Neither | |
| 2002 | + * detect_schema_present() (body scan) nor thinkrank_global_schema_active() | |
| 2003 | + * (post-type option) sees it, so without this the score reported "no | |
| 2004 | + * structured data" for posts that do emit it. | |
| 2005 | + * | |
| 2006 | + * Mirrors the context_type whitelist the emitter and the metabox both use, so | |
| 2007 | + * the lookup targets the same row the front end reads. | |
| 2008 | + * | |
| 2009 | + * @param \WP_Post $post Post being scored. | |
| 2010 | + * @return bool | |
| 2011 | + */ | |
| 2012 | + private function thinkrank_deployed_schema_active(\WP_Post $post): bool { | |
| 2013 | + if (!class_exists('ThinkRank\\SEO\\Schema_Management_System')) { | |
| 2014 | + $manager_file = THINKRANK_PLUGIN_DIR . 'includes/seo/class-schema-management-system.php'; | |
| 2015 | + if (!file_exists($manager_file)) { | |
| 2016 | + return false; | |
| 2017 | + } | |
| 2018 | + require_once $manager_file; | |
| 2019 | + } | |
| 2020 | + | |
| 2021 | + $context_type = in_array($post->post_type, ['site', 'post', 'page', 'product'], true) | |
| 2022 | + ? $post->post_type | |
| 2023 | + : 'post'; | |
| 2024 | + | |
| 2025 | + try { | |
| 2026 | + $manager = new \ThinkRank\SEO\Schema_Management_System(); | |
| 2027 | + return !empty($manager->get_deployed_schemas($context_type, (int) $post->ID)); | |
| 2028 | + } catch (\Throwable $e) { | |
| 2029 | + return false; | |
| 2030 | + } | |
| 2031 | + } | |
| 2032 | + | |
| 1659 | 2033 | private function analyze_images(string $content): array { |
| 1660 | 2034 | $images = []; |
| 1661 | 2035 | |
| 1662 | 2036 | if (preg_match_all('/<img[^>]+>/i', $content, $matches)) { |
| @@ -1693,25 +2067,42 @@ | ||
| 1693 | 2067 | global $wpdb; |
| 1694 | 2068 | |
| 1695 | 2069 | $table_name = $wpdb->prefix . 'thinkrank_seo_scores'; |
| 1696 | 2070 | |
| 1697 | - // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- SEO score storage requires direct database access | |
| 2071 | + // Prepare data for insertion | |
| 2072 | + $insert_data = [ | |
| 2073 | + 'post_id' => $post_id, | |
| 2074 | + 'user_id' => $user_id, | |
| 2075 | + 'overall_score' => $score_data['overall_score'], | |
| 2076 | + 'score_breakdown' => wp_json_encode($score_data['score_breakdown']), | |
| 2077 | + 'suggestions' => wp_json_encode($score_data['suggestions']), | |
| 2078 | + 'grade' => $score_data['grade'], | |
| 2079 | + 'algorithm_version' => $score_data['algorithm_version'] ?? '2024.1', | |
| 2080 | + 'calculated_at' => $score_data['calculated_at'], | |
| 2081 | + 'created_at' => current_time('mysql'), | |
| 2082 | + ]; | |
| 2083 | + | |
| 2084 | + $format = [ | |
| 2085 | + '%d', '%d', '%d', '%s', '%s', '%s', '%s', '%s', '%s' | |
| 2086 | + ]; | |
| 2087 | + | |
| 2088 | + // Add readability_score if provided | |
| 2089 | + if (isset($score_data['readability_score'])) { | |
| 2090 | + $insert_data['readability_score'] = $score_data['readability_score']; | |
| 2091 | + $format[] = '%s'; | |
| 2092 | + } | |
| 2093 | + | |
| 2094 | + // Add content_quality if provided | |
| 2095 | + if (isset($score_data['content_quality'])) { | |
| 2096 | + $insert_data['content_quality'] = $score_data['content_quality']; | |
| 2097 | + $format[] = '%s'; | |
| 2098 | + } | |
| 2099 | + | |
| 2100 | + // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- SEO score storage requires direct database access | |
| 1698 | 2101 | $result = $wpdb->insert( |
| 1699 | 2102 | $table_name, |
| 1700 | - [ | |
| 1701 | - 'post_id' => $post_id, | |
| 1702 | - 'user_id' => $user_id, | |
| 1703 | - 'overall_score' => $score_data['overall_score'], | |
| 1704 | - 'score_breakdown' => json_encode($score_data['score_breakdown']), | |
| 1705 | - 'suggestions' => json_encode($score_data['suggestions']), | |
| 1706 | - 'grade' => $score_data['grade'], | |
| 1707 | - 'algorithm_version' => $score_data['algorithm_version'] ?? '2024.1', | |
| 1708 | - 'calculated_at' => $score_data['calculated_at'], | |
| 1709 | - 'created_at' => current_time('mysql'), | |
| 1710 | - ], | |
| 1711 | - [ | |
| 1712 | - '%d', '%d', '%d', '%s', '%s', '%s', '%s', '%s', '%s' | |
| 1713 | - ] | |
| 2103 | + $insert_data, | |
| 2104 | + $format | |
| 1714 | 2105 | ); |
| 1715 | 2106 | |
| 1716 | 2107 | return $result ? $wpdb->insert_id : false; |
| 1717 | 2108 | } |
| @@ -1728,12 +2119,12 @@ | ||
| 1728 | 2119 | |
| 1729 | 2120 | // Get table name and escape it properly (table names cannot be parameterized) |
| 1730 | 2121 | $table_name = esc_sql($wpdb->prefix . 'thinkrank_seo_scores'); |
| 1731 | 2122 | |
| 1732 | - // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- SEO score history requires direct database access | |
| 2123 | + // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- SEO score history requires direct database access | |
| 1733 | 2124 | $results = $wpdb->get_results( |
| 1734 | 2125 | $wpdb->prepare( |
| 1735 | - // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Table name is properly escaped using esc_sql() | |
| 2126 | + // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Table name is properly escaped using esc_sql() | |
| 1736 | 2127 | "SELECT * FROM `{$table_name}` WHERE post_id = %d ORDER BY created_at DESC LIMIT %d", |
| 1737 | 2128 | $post_id, |
| 1738 | 2129 | $limit |
| 1739 | 2130 | ), |
| @@ -1766,17 +2157,101 @@ | ||
| 1766 | 2157 | * @param array $content_data Content analysis data |
| 1767 | 2158 | * @return array Scoring result |
| 1768 | 2159 | */ |
| 1769 | 2160 | private function score_mobile_experience(array $content_data): array { |
| 2161 | + $max = $this->scoring_factors['mobile_experience']; | |
| 2162 | + | |
| 2163 | + // Mobile experience is theme/site-level, not controlled by post content | |
| 2164 | + // — but the plugin already measures it. When a mobile Lighthouse score | |
| 2165 | + // has been collected, score against it; the "benefit of the doubt" below | |
| 2166 | + // is for sites nobody has measured, not for sites measured as slow. | |
| 2167 | + $performance_score = $this->measured_performance_score(); | |
| 2168 | + | |
| 2169 | + if ($performance_score === null) { | |
| 2170 | + return [ | |
| 2171 | + 'score' => $max, | |
| 2172 | + 'max_score' => $max, | |
| 2173 | + 'suggestions' => ['Ensure mobile-first design and fast loading on mobile devices'], | |
| 2174 | + 'details' => ['mobile_score' => 'Assumed adequate', 'measured' => false], | |
| 2175 | + ]; | |
| 2176 | + } | |
| 2177 | + | |
| 2178 | + $score = (int) round($max * $performance_score / 100); | |
| 2179 | + | |
| 1770 | 2180 | return [ |
| 1771 | - 'score' => 4, // Default good score for mobile | |
| 1772 | - 'max_score' => $this->scoring_factors['mobile_experience'], | |
| 1773 | - 'suggestions' => ['Ensure mobile-first design and fast loading on mobile devices'], | |
| 1774 | - 'details' => ['mobile_score' => 'Good'] | |
| 2181 | + 'score' => $score, | |
| 2182 | + 'max_score' => $max, | |
| 2183 | + 'suggestions' => $score < $max | |
| 2184 | + ? ['Improve mobile page speed: the last PageSpeed run scored ' . $performance_score . '/100 on mobile'] | |
| 2185 | + : [], | |
| 2186 | + 'details' => [ | |
| 2187 | + 'mobile_score' => $performance_score, | |
| 2188 | + 'measured' => true, | |
| 2189 | + 'source' => 'pagespeed_mobile', | |
| 2190 | + ], | |
| 1775 | 2191 | ]; |
| 1776 | 2192 | } |
| 1777 | 2193 | |
| 1778 | 2194 | /** |
| 2195 | + * The last collected mobile Lighthouse score, or null when unmeasured. | |
| 2196 | + * | |
| 2197 | + * Memoised per instance: compute_score() asks twice, and a post-list screen | |
| 2198 | + * scores a page of posts at a time. | |
| 2199 | + * | |
| 2200 | + * Every failure — no performance module, no collected row, an unreadable | |
| 2201 | + * table — resolves to null, which the callers read as "not measured" and | |
| 2202 | + * answer with the full-credit fallback. A site is never penalised for | |
| 2203 | + * ThinkRank being unable to look. | |
| 2204 | + * | |
| 2205 | + * @since 2.3.1 | |
| 2206 | + * @return int|null Score 0-100, or null when nothing has been collected. | |
| 2207 | + */ | |
| 2208 | + private function measured_performance_score(): ?int { | |
| 2209 | + $measurement = $this->measured_performance(); | |
| 2210 | + | |
| 2211 | + if ($measurement === null || !isset($measurement['performance_score'])) { | |
| 2212 | + return null; | |
| 2213 | + } | |
| 2214 | + | |
| 2215 | + $score = $measurement['performance_score']; | |
| 2216 | + | |
| 2217 | + if (!is_numeric($score)) { | |
| 2218 | + return null; | |
| 2219 | + } | |
| 2220 | + | |
| 2221 | + return (int) round(max(0, min(100, (float) $score))); | |
| 2222 | + } | |
| 2223 | + | |
| 2224 | + /** | |
| 2225 | + * The last collected mobile measurement, or null when there is none. | |
| 2226 | + * | |
| 2227 | + * @since 2.3.1 | |
| 2228 | + * @return array|null { core_web_vitals: array, performance_score: float|null } | |
| 2229 | + */ | |
| 2230 | + private function measured_performance(): ?array { | |
| 2231 | + if ($this->measured_performance_resolved) { | |
| 2232 | + return $this->measured_performance; | |
| 2233 | + } | |
| 2234 | + | |
| 2235 | + $this->measured_performance_resolved = true; | |
| 2236 | + | |
| 2237 | + if (!class_exists('ThinkRank\\SEO\\Performance_Monitoring_Manager')) { | |
| 2238 | + return null; | |
| 2239 | + } | |
| 2240 | + | |
| 2241 | + try { | |
| 2242 | + $manager = new \ThinkRank\SEO\Performance_Monitoring_Manager(); | |
| 2243 | + // Mobile deliberately: Google indexes mobile-first, and it is the | |
| 2244 | + // device the mobile_experience factor is named after. | |
| 2245 | + $this->measured_performance = $manager->get_stored_performance_measurement('mobile'); | |
| 2246 | + } catch (\Throwable $e) { | |
| 2247 | + $this->measured_performance = null; | |
| 2248 | + } | |
| 2249 | + | |
| 2250 | + return $this->measured_performance; | |
| 2251 | + } | |
| 2252 | + | |
| 2253 | + /** | |
| 1779 | 2254 | * Score core web vitals - 2025 version (3 points) |
| 1780 | 2255 | * |
| 1781 | 2256 | * @param array $content_data Content analysis data |
| 1782 | 2257 | * @return array Scoring result |
| @@ -1781,17 +2256,98 @@ | ||
| 1781 | 2256 | * @param array $content_data Content analysis data |
| 1782 | 2257 | * @return array Scoring result |
| 1783 | 2258 | */ |
| 1784 | 2259 | private function score_core_web_vitals(array $content_data): array { |
| 2260 | + $max = $this->scoring_factors['core_web_vitals']; | |
| 2261 | + | |
| 2262 | + // Not derivable from post content — but it is measured, and the audit | |
| 2263 | + // stores LCP, INP and CLS with a rating each. Score against those when | |
| 2264 | + // they exist; fall back to the benefit of the doubt when they do not. | |
| 2265 | + $rated = $this->measured_vitals_score(); | |
| 2266 | + | |
| 2267 | + if ($rated === null) { | |
| 2268 | + return [ | |
| 2269 | + 'score' => $max, | |
| 2270 | + 'max_score' => $max, | |
| 2271 | + 'suggestions' => ['Optimize Core Web Vitals: LCP, INP, and CLS for better user experience'], | |
| 2272 | + 'details' => ['vitals_status' => 'Assumed adequate', 'measured' => false], | |
| 2273 | + ]; | |
| 2274 | + } | |
| 2275 | + | |
| 2276 | + $score = (int) round($max * $rated['average'] / 100); | |
| 2277 | + | |
| 1785 | 2278 | return [ |
| 1786 | - 'score' => 2, // Default partial score | |
| 1787 | - 'max_score' => $this->scoring_factors['core_web_vitals'], | |
| 1788 | - 'suggestions' => ['Optimize Core Web Vitals: LCP, FID, and CLS for better user experience'], | |
| 1789 | - 'details' => ['vitals_status' => 'Needs improvement'] | |
| 2279 | + 'score' => $score, | |
| 2280 | + 'max_score' => $max, | |
| 2281 | + // Gated on the measurement, not the rounded score: two good metrics | |
| 2282 | + // and one needing improvement averages 88.33, which rounds to the | |
| 2283 | + // full 3 of 3 and used to swallow the suggestion naming the metric | |
| 2284 | + // that is actually failing. | |
| 2285 | + 'suggestions' => !empty($rated['failing']) | |
| 2286 | + ? ['Optimize Core Web Vitals: ' . implode(', ', $rated['failing']) . ' below target on mobile'] | |
| 2287 | + : [], | |
| 2288 | + 'details' => [ | |
| 2289 | + 'vitals_status' => $rated['statuses'], | |
| 2290 | + 'measured' => true, | |
| 2291 | + 'source' => 'pagespeed_mobile', | |
| 2292 | + ], | |
| 1790 | 2293 | ]; |
| 1791 | 2294 | } |
| 1792 | 2295 | |
| 1793 | 2296 | /** |
| 2297 | + * Rate the collected Core Web Vitals, or null when none were measured. | |
| 2298 | + * | |
| 2299 | + * Reuses the per-metric score the performance module already assigns | |
| 2300 | + * (good 100, needs improvement 65, poor 30) rather than inventing a second | |
| 2301 | + * scale, so the SEO score and the performance card cannot disagree about | |
| 2302 | + * whether a metric is healthy. | |
| 2303 | + * | |
| 2304 | + * Metrics with no stored value — fcp is not always collected — are skipped | |
| 2305 | + * rather than counted as failures. | |
| 2306 | + * | |
| 2307 | + * @since 2.3.1 | |
| 2308 | + * @return array|null { average: float, statuses: array, failing: string[] } | |
| 2309 | + */ | |
| 2310 | + private function measured_vitals_score(): ?array { | |
| 2311 | + $measurement = $this->measured_performance(); | |
| 2312 | + $vitals = $measurement['core_web_vitals'] ?? null; | |
| 2313 | + | |
| 2314 | + if (!is_array($vitals)) { | |
| 2315 | + return null; | |
| 2316 | + } | |
| 2317 | + | |
| 2318 | + $scores = []; | |
| 2319 | + $statuses = []; | |
| 2320 | + $failing = []; | |
| 2321 | + | |
| 2322 | + // The three Google ranks on. fcp is diagnostic and not a Core Web Vital. | |
| 2323 | + foreach (['lcp', 'inp', 'cls'] as $metric) { | |
| 2324 | + $data = $vitals[$metric] ?? null; | |
| 2325 | + | |
| 2326 | + if (!is_array($data) || !isset($data['value'], $data['score']) || $data['value'] === null) { | |
| 2327 | + continue; | |
| 2328 | + } | |
| 2329 | + | |
| 2330 | + $scores[] = (float) $data['score']; | |
| 2331 | + $statuses[$metric] = $data['status'] ?? 'unknown'; | |
| 2332 | + | |
| 2333 | + if (($data['status'] ?? '') !== 'good') { | |
| 2334 | + $failing[] = strtoupper($metric); | |
| 2335 | + } | |
| 2336 | + } | |
| 2337 | + | |
| 2338 | + if (empty($scores)) { | |
| 2339 | + return null; | |
| 2340 | + } | |
| 2341 | + | |
| 2342 | + return [ | |
| 2343 | + 'average' => array_sum($scores) / count($scores), | |
| 2344 | + 'statuses' => $statuses, | |
| 2345 | + 'failing' => $failing, | |
| 2346 | + ]; | |
| 2347 | + } | |
| 2348 | + | |
| 2349 | + /** | |
| 1794 | 2350 | * Score internal linking - declining importance (1 point) |
| 1795 | 2351 | * |
| 1796 | 2352 | * @param array $content_data Content analysis data |
| 1797 | 2353 | * @return array Scoring result |
| @@ -1821,9 +2377,9 @@ | ||
| 1821 | 2377 | $suggestions = []; |
| 1822 | 2378 | |
| 1823 | 2379 | // Meta description check |
| 1824 | 2380 | $meta_desc = $metadata['description'] ?? ''; |
| 1825 | - if (!empty($meta_desc) && strlen($meta_desc) >= 120 && strlen($meta_desc) <= 160) { | |
| 2381 | + if (!empty($meta_desc) && mb_strlen($meta_desc) >= self::DESCRIPTION_OPTIMAL_MIN && mb_strlen($meta_desc) <= self::DESCRIPTION_OPTIMAL_MAX) { | |
| 1826 | 2382 | $score += 0.5; |
| 1827 | 2383 | } else { |
| 1828 | 2384 | $suggestions[] = 'Add a compelling meta description (120-160 characters)'; |
| 1829 | 2385 | } |
| @@ -1839,9 +2395,9 @@ | ||
| 1839 | 2395 | 'score' => $score, |
| 1840 | 2396 | 'max_score' => $this->scoring_factors['technical_factors'], |
| 1841 | 2397 | 'suggestions' => $suggestions, |
| 1842 | 2398 | 'details' => [ |
| 1843 | - 'meta_description_length' => strlen($meta_desc), | |
| 2399 | + 'meta_description_length' => mb_strlen($meta_desc), | |
| 1844 | 2400 | 'schema_present' => !empty($content_data['schema_present']) |
| 1845 | 2401 | ] |
| 1846 | 2402 | ]; |
| 1847 | 2403 | } |
| @@ -1854,19 +2410,30 @@ | ||
| 1854 | 2410 | */ |
| 1855 | 2411 | private function get_grade_from_score($score): string { |
| 1856 | 2412 | $score = (int) $score; // Ensure it's an integer |
| 1857 | 2413 | |
| 1858 | - if ($score >= 95) return 'A+'; | |
| 1859 | - if ($score >= 90) return 'A'; | |
| 1860 | - if ($score >= 85) return 'A-'; | |
| 1861 | - if ($score >= 80) return 'B+'; | |
| 1862 | - if ($score >= 75) return 'B'; | |
| 1863 | - if ($score >= 70) return 'B-'; | |
| 1864 | - if ($score >= 65) return 'C+'; | |
| 1865 | - if ($score >= 60) return 'C'; | |
| 1866 | - if ($score >= 55) return 'C-'; | |
| 1867 | - if ($score >= 45) return 'D+'; | |
| 1868 | - if ($score >= 35) return 'D'; | |
| 2414 | + if ($score >= 95) { return 'A+'; | |
| 2415 | + } | |
| 2416 | + if ($score >= 90) { return 'A'; | |
| 2417 | + } | |
| 2418 | + if ($score >= 85) { return 'A-'; | |
| 2419 | + } | |
| 2420 | + if ($score >= 80) { return 'B+'; | |
| 2421 | + } | |
| 2422 | + if ($score >= 75) { return 'B'; | |
| 2423 | + } | |
| 2424 | + if ($score >= 70) { return 'B-'; | |
| 2425 | + } | |
| 2426 | + if ($score >= 65) { return 'C+'; | |
| 2427 | + } | |
| 2428 | + if ($score >= 60) { return 'C'; | |
| 2429 | + } | |
| 2430 | + if ($score >= 55) { return 'C-'; | |
| 2431 | + } | |
| 2432 | + if ($score >= 45) { return 'D+'; | |
| 2433 | + } | |
| 2434 | + if ($score >= 35) { return 'D'; | |
| 2435 | + } | |
| 1869 | 2436 | return 'F'; |
| 1870 | 2437 | } |
| 1871 | 2438 | |
| 1872 | 2439 | /** |
| @@ -1882,9 +2449,15 @@ | ||
| 1882 | 2449 | if (!$post) { |
| 1883 | 2450 | return []; |
| 1884 | 2451 | } |
| 1885 | 2452 | |
| 1886 | - $content = $live_content; // Use live content instead of $post->post_content | |
| 2453 | + // Resolve the live string the same way stored content is resolved. On a | |
| 2454 | + // builder page the editor hands over raw builder markup (the block | |
| 2455 | + // editor cannot render blocks it has no client-side registration for), | |
| 2456 | + // which analyzed as-is reads as zero words — the reason a Divi page | |
| 2457 | + // could show a correct saved score beside a live panel still claiming | |
| 2458 | + // "No content". | |
| 2459 | + $content = self::resolve_live_content($live_content, $post); | |
| 1887 | 2460 | $title = $post->post_title; |
| 1888 | 2461 | |
| 1889 | 2462 | // Extract headings from content |
| 1890 | 2463 | $headings = $this->extract_headings($content); |
| @@ -1898,8 +2471,9 @@ | ||
| 1898 | 2471 | |
| 1899 | 2472 | // Count links |
| 1900 | 2473 | $internal_links = $this->count_internal_links($content); |
| 1901 | 2474 | $external_links = $this->count_external_links($content); |
| 2475 | + $external_dofollow_links = $this->count_external_dofollow_links($content); | |
| 1902 | 2476 | |
| 1903 | 2477 | // Analyze images |
| 1904 | 2478 | $images = $this->analyze_images($content); |
| 1905 | 2479 | |
| @@ -1913,10 +2487,16 @@ | ||
| 1913 | 2487 | 'word_count' => $word_count, |
| 1914 | 2488 | 'readability_score' => $readability_score, |
| 1915 | 2489 | 'internal_links' => $internal_links, |
| 1916 | 2490 | 'external_links' => $external_links, |
| 2491 | + 'external_dofollow_links' => $external_dofollow_links, | |
| 1917 | 2492 | 'images' => $images, |
| 1918 | 2493 | 'url' => $url, |
| 2494 | + 'slug' => $post->post_name, | |
| 2495 | + 'post_modified' => $post->post_modified, | |
| 2496 | + 'schema_present' => $this->detect_schema_present($content) | |
| 2497 | + || $this->thinkrank_global_schema_active($post->post_type) | |
| 2498 | + || $this->thinkrank_deployed_schema_active($post), | |
| 1919 | 2499 | ]; |
| 1920 | 2500 | } |
| 1921 | 2501 | |
| 1922 | 2502 | /** |
| @@ -1948,11 +2528,11 @@ | ||
| 1948 | 2528 | // Get table name and escape it properly (table names cannot be parameterized) |
| 1949 | 2529 | $table_name = esc_sql($wpdb->prefix . 'thinkrank_seo_scores'); |
| 1950 | 2530 | |
| 1951 | 2531 | // Get the most recent score for this post |
| 1952 | - // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- SEO score retrieval requires direct database access | |
| 2532 | + // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching, PluginCheck.Security.DirectDB.UnescapedDBParameter -- SEO score retrieval requires direct database access | |
| 1953 | 2533 | $result = $wpdb->get_row($wpdb->prepare( |
| 1954 | - // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Table name is properly escaped using esc_sql() | |
| 2534 | + // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- Table name is properly escaped using esc_sql() | |
| 1955 | 2535 | "SELECT * FROM `{$table_name}` |
| 1956 | 2536 | WHERE post_id = %d |
| 1957 | 2537 | ORDER BY calculated_at DESC |
| 1958 | 2538 | LIMIT 1", |