| @@ -260,31 +260,72 @@ | ||
| 260 | 260 | * @return array The snippets, and how many times the text says it. |
| 261 | 261 | */ |
| 262 | 262 | private static function snippets($text, $query, $context) |
| 263 | 263 | { |
| 264 | - $length = mb_strlen($query); | |
| 265 | - $end = mb_strlen($text); | |
| 264 | + // Turkish İ lower cases to two characters, so a folded copy's offsets miss the match. | |
| 265 | + $pattern = '/' . preg_quote($query, '/') . '/iu'; | |
| 266 | + if (!preg_match_all($pattern, $text, $found, PREG_OFFSET_CAPTURE)) { | |
| 267 | + return ['matches' => [], 'count' => 0]; | |
| 268 | + } | |
| 266 | 269 | |
| 270 | + $end = strlen($text); | |
| 267 | 271 | $matches = []; |
| 268 | - $count = 0; | |
| 269 | 272 | $quoted = 0; |
| 270 | - $offset = 0; | |
| 271 | - // Turkish İ lower cases to two characters, so a folded copy's offsets miss the match. | |
| 272 | - while (($at = mb_stripos($text, $query, $offset)) !== false) { | |
| 273 | - $count++; | |
| 274 | - $offset = $at + $length; | |
| 273 | + foreach ($found[0] as $occurrence) { | |
| 274 | + list($match, $at) = $occurrence; | |
| 275 | 275 | if (count($matches) >= self::MAX_MATCHES || $at < $quoted) { |
| 276 | 276 | continue; |
| 277 | 277 | } |
| 278 | 278 | |
| 279 | - $from = max(0, $at - $context); | |
| 280 | - $quoted = min($end, $at + $length + $context); | |
| 281 | - $matches[] = ($from > 0 ? '…' : '') | |
| 282 | - . mb_substr($text, $from, $quoted - $from) | |
| 279 | + $before = self::quoteBefore($text, $at, $context); | |
| 280 | + $after = self::quoteAfter($text, $at + strlen($match), $context); | |
| 281 | + $quoted = $at + strlen($match) + strlen($after); | |
| 282 | + $matches[] = ($at > strlen($before) ? '…' : '') | |
| 283 | + . $before . $match . $after | |
| 283 | 284 | . ($quoted < $end ? '…' : ''); |
| 284 | 285 | } |
| 285 | 286 | |
| 286 | - return ['matches' => $matches, 'count' => $count]; | |
| 287 | + return ['matches' => $matches, 'count' => count($found[0])]; | |
| 288 | + } | |
| 289 | + | |
| 290 | + /** | |
| 291 | + * Up to $context characters of $text ending at byte offset $at. | |
| 292 | + * | |
| 293 | + * @param string $text The text a match was found in. | |
| 294 | + * @param integer $at Where the match begins, in bytes. | |
| 295 | + * @param integer $context How many characters to quote. | |
| 296 | + * @return string | |
| 297 | + */ | |
| 298 | + private static function quoteBefore($text, $at, $context) | |
| 299 | + { | |
| 300 | + // A character is at most four bytes, so the window always holds enough. | |
| 301 | + $slice = substr($text, max(0, $at - (4 * $context)), min($at, 4 * $context)); | |
| 302 | + // A slice starting mid-character would fail the /u match and empty the quote. | |
| 303 | + $slice = (string) preg_replace('/^[\x80-\xBF]+/', '', $slice); | |
| 304 | + preg_match('/.{0,' . $context . '}$/su', $slice, $match); | |
| 305 | + | |
| 306 | + return $match[0] ?? ''; | |
| 307 | + } | |
| 308 | + | |
| 309 | + /** | |
| 310 | + * Up to $context characters of $text starting at byte offset $at. | |
| 311 | + * | |
| 312 | + * @param string $text The text a match was found in. | |
| 313 | + * @param integer $at Where the match ends, in bytes. | |
| 314 | + * @param integer $context How many characters to quote. | |
| 315 | + * @return string | |
| 316 | + */ | |
| 317 | + private static function quoteAfter($text, $at, $context) | |
| 318 | + { | |
| 319 | + // A slice ending mid-character would fail the /u match and empty the quote. | |
| 320 | + $slice = (string) preg_replace( | |
| 321 | + '/(?:[\xC2-\xDF]|[\xE0-\xEF][\x80-\xBF]?|[\xF0-\xF4][\x80-\xBF]{0,2})$/D', | |
| 322 | + '', | |
| 323 | + substr($text, $at, 4 * $context) | |
| 324 | + ); | |
| 325 | + preg_match('/^.{0,' . $context . '}/su', $slice, $match); | |
| 326 | + | |
| 327 | + return $match[0] ?? ''; | |
| 287 | 328 | } |
| 288 | 329 | |
| 289 | 330 | /** |
| 290 | 331 | * @param integer $id A post id. |
