x-tools /
xtools
| Total Complexity | 46 |
| Total Lines | 459 |
| Duplicated Lines | 0 % |
| Changes | 0 | ||
Complex classes like ArticleInfoApi often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use ArticleInfoApi, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 20 | class ArticleInfoApi extends Model |
||
| 21 | { |
||
| 22 | /** @var ContainerInterface The application's DI container. */ |
||
| 23 | protected $container; |
||
| 24 | |||
| 25 | /** @var int Number of revisions that belong to the page. */ |
||
| 26 | protected $numRevisions; |
||
| 27 | |||
| 28 | /** @var int Maximum number of revisions to process, as configured. */ |
||
| 29 | protected $maxRevisions; |
||
| 30 | |||
| 31 | /** @var mixed[] Prose stats, with keys 'characters', 'words', 'references', 'unique_references', 'sections'. */ |
||
| 32 | protected $proseStats; |
||
| 33 | |||
| 34 | /** @var array Number of categories, templates and files on the page. */ |
||
| 35 | protected $transclusionData; |
||
| 36 | |||
| 37 | /** @var mixed[] Various statistics about bots that edited the page. */ |
||
| 38 | protected $bots; |
||
| 39 | |||
| 40 | /** @var int Number of edits made to the page by bots. */ |
||
| 41 | protected $botRevisionCount; |
||
| 42 | |||
| 43 | /** @var int[] Number of in and outgoing links and redirects to the page. */ |
||
| 44 | protected $linksAndRedirects; |
||
| 45 | |||
| 46 | /** @var string[] Assessments of the page (see Page::getAssessments). */ |
||
| 47 | protected $assessments; |
||
| 48 | |||
| 49 | /** @var string[] List of Wikidata and Checkwiki errors. */ |
||
| 50 | protected $bugs; |
||
| 51 | |||
| 52 | /** |
||
| 53 | * ArticleInfoApi constructor. |
||
| 54 | * @param Page $page The page to process. |
||
| 55 | * @param ContainerInterface $container The DI container. |
||
| 56 | * @param false|int $start Start date as Unix timestmap. |
||
| 57 | * @param false|int $end End date as Unix timestamp. |
||
| 58 | */ |
||
| 59 | public function __construct(Page $page, ContainerInterface $container, $start = false, $end = false) |
||
| 60 | { |
||
| 61 | $this->page = $page; |
||
| 62 | $this->container = $container; |
||
| 63 | $this->start = $start; |
||
| 64 | $this->end = $end; |
||
| 65 | } |
||
| 66 | |||
| 67 | /** |
||
| 68 | * Get the number of revisions belonging to the page. |
||
| 69 | * @return int |
||
| 70 | */ |
||
| 71 | public function getNumRevisions(): int |
||
| 72 | { |
||
| 73 | if (!isset($this->numRevisions)) { |
||
| 74 | $this->numRevisions = $this->page->getNumRevisions(null, $this->start, $this->end); |
||
| 75 | } |
||
| 76 | return $this->numRevisions; |
||
| 77 | } |
||
| 78 | |||
| 79 | /** |
||
| 80 | * Are there more revisions than we should process, based on the config? |
||
| 81 | * @return bool |
||
| 82 | */ |
||
| 83 | public function tooManyRevisions(): bool |
||
| 84 | { |
||
| 85 | return $this->getMaxRevisions() > 0 && $this->getNumRevisions() > $this->getMaxRevisions(); |
||
| 86 | } |
||
| 87 | |||
| 88 | /** |
||
| 89 | * Get the maximum number of revisions that we should process. |
||
| 90 | * @return int |
||
| 91 | */ |
||
| 92 | public function getMaxRevisions(): int |
||
| 93 | { |
||
| 94 | if (!isset($this->maxRevisions)) { |
||
| 95 | $this->maxRevisions = (int) $this->container->getParameter('app.max_page_revisions'); |
||
| 96 | } |
||
| 97 | return $this->maxRevisions; |
||
| 98 | } |
||
| 99 | |||
| 100 | /** |
||
| 101 | * Get various basic info used in the API, including the number of revisions, unique authors, initial author |
||
| 102 | * and edit count of the initial author. This is combined into one query for better performance. Caching is |
||
| 103 | * intentionally disabled, because using the gadget, this will get hit for a different page constantly, where |
||
| 104 | * the likelihood of cache benefiting us is slim. |
||
| 105 | * @return string[]|false false if the page was not found. |
||
| 106 | */ |
||
| 107 | public function getBasicEditingInfo() |
||
| 108 | { |
||
| 109 | return $this->getRepository()->getBasicEditingInfo($this->page); |
||
|
Bug
introduced
by
|
|||
| 110 | } |
||
| 111 | |||
| 112 | /** |
||
| 113 | * Get the top editors to the page by edit count. |
||
| 114 | * @param int $limit Default 20, maximum 1,000. |
||
| 115 | * @param bool $noBots Set to non-false to exclude bots from the result. |
||
| 116 | * @return array |
||
| 117 | */ |
||
| 118 | public function getTopEditorsByEditCount(int $limit = 20, bool $noBots = false): array |
||
| 119 | { |
||
| 120 | // Quick cache, valid only for the same request. |
||
| 121 | static $topEditors = null; |
||
| 122 | if (null !== $topEditors) { |
||
| 123 | return $topEditors; |
||
| 124 | } |
||
| 125 | |||
| 126 | $rows = $this->getRepository()->getTopEditorsByEditCount( |
||
| 127 | $this->page, |
||
| 128 | $this->start, |
||
| 129 | $this->end, |
||
| 130 | min($limit, 1000), |
||
| 131 | $noBots |
||
| 132 | ); |
||
| 133 | |||
| 134 | $topEditors = []; |
||
| 135 | $rank = 0; |
||
| 136 | foreach ($rows as $row) { |
||
| 137 | $topEditors[] = [ |
||
| 138 | 'rank' => ++$rank, |
||
| 139 | 'username' => $row['username'], |
||
| 140 | 'count' => $row['count'], |
||
| 141 | 'minor' => $row['minor'], |
||
| 142 | 'first_edit' => [ |
||
| 143 | 'id' => $row['first_revid'], |
||
| 144 | 'timestamp' => $row['first_timestamp'], |
||
| 145 | ], |
||
| 146 | 'latest_edit' => [ |
||
| 147 | 'id' => $row['latest_revid'], |
||
| 148 | 'timestamp' => $row['latest_timestamp'], |
||
| 149 | ], |
||
| 150 | ]; |
||
| 151 | } |
||
| 152 | |||
| 153 | return $topEditors; |
||
| 154 | } |
||
| 155 | |||
| 156 | /** |
||
| 157 | * Get prose and reference information. |
||
| 158 | * @return array With keys 'characters', 'words', 'references', 'unique_references' |
||
| 159 | */ |
||
| 160 | public function getProseStats(): array |
||
| 161 | { |
||
| 162 | if (isset($this->proseStats)) { |
||
| 163 | return $this->proseStats; |
||
| 164 | } |
||
| 165 | |||
| 166 | $datetime = is_int($this->end) ? new DateTime("@$this->end") : null; |
||
| 167 | $html = $this->page->getHTMLContent($datetime); |
||
| 168 | |||
| 169 | $crawler = new Crawler($html); |
||
| 170 | |||
| 171 | [$chars, $words] = $this->countCharsAndWords($crawler, '#mw-content-text p'); |
||
| 172 | |||
| 173 | $refs = $crawler->filter('#mw-content-text .reference'); |
||
| 174 | $refContent = []; |
||
| 175 | $refs->each(function ($ref) use (&$refContent): void { |
||
| 176 | $refContent[] = $ref->text(); |
||
| 177 | }); |
||
| 178 | $uniqueRefs = count(array_unique($refContent)); |
||
| 179 | |||
| 180 | $sections = count($crawler->filter('#mw-content-text .mw-headline')); |
||
| 181 | |||
| 182 | $this->proseStats = [ |
||
| 183 | 'characters' => $chars, |
||
| 184 | 'words' => $words, |
||
| 185 | 'references' => $refs->count(), |
||
| 186 | 'unique_references' => $uniqueRefs, |
||
| 187 | 'sections' => $sections, |
||
| 188 | ]; |
||
| 189 | return $this->proseStats; |
||
| 190 | } |
||
| 191 | |||
| 192 | /** |
||
| 193 | * Count the number of characters and words of the plain text within the DOM element matched by the given selector. |
||
| 194 | * @param Crawler $crawler |
||
| 195 | * @param string $selector HTML selector. |
||
| 196 | * @return array [num chars, num words] |
||
| 197 | */ |
||
| 198 | private function countCharsAndWords(Crawler $crawler, string $selector): array |
||
| 199 | { |
||
| 200 | $totalChars = 0; |
||
| 201 | $totalWords = 0; |
||
| 202 | $paragraphs = $crawler->filter($selector); |
||
| 203 | $paragraphs->each(function ($node) use (&$totalChars, &$totalWords): void { |
||
| 204 | /** @var Crawler $node */ |
||
| 205 | $text = preg_replace('/\[\d+]/', '', trim($node->text(null, true))); |
||
| 206 | $totalChars += strlen($text); |
||
| 207 | $totalWords += count(explode(' ', $text)); |
||
| 208 | }); |
||
| 209 | |||
| 210 | return [$totalChars, $totalWords]; |
||
| 211 | } |
||
| 212 | |||
| 213 | /** |
||
| 214 | * Get the page assessments of the page. |
||
| 215 | * @see https://www.mediawiki.org/wiki/Extension:PageAssessments |
||
| 216 | * @return string[]|false False if unsupported. |
||
| 217 | * @codeCoverageIgnore |
||
| 218 | */ |
||
| 219 | public function getAssessments() |
||
| 220 | { |
||
| 221 | if (!is_array($this->assessments)) { |
||
| 222 | $this->assessments = $this->page |
||
| 223 | ->getProject() |
||
| 224 | ->getPageAssessments() |
||
| 225 | ->getAssessments($this->page); |
||
| 226 | } |
||
| 227 | return $this->assessments; |
||
| 228 | } |
||
| 229 | |||
| 230 | /** |
||
| 231 | * Get the list of page's wikidata and Checkwiki errors. |
||
| 232 | * @see Page::getErrors() |
||
| 233 | * @return string[] |
||
| 234 | */ |
||
| 235 | public function getBugs(): array |
||
| 236 | { |
||
| 237 | if (!is_array($this->bugs)) { |
||
| 238 | $this->bugs = $this->page->getErrors(); |
||
| 239 | } |
||
| 240 | return $this->bugs; |
||
| 241 | } |
||
| 242 | |||
| 243 | /** |
||
| 244 | * Get the number of wikidata nad CheckWiki errors. |
||
| 245 | * @return int |
||
| 246 | */ |
||
| 247 | public function numBugs(): int |
||
| 248 | { |
||
| 249 | return count($this->getBugs()); |
||
| 250 | } |
||
| 251 | |||
| 252 | /** |
||
| 253 | * Generate the data structure that will used in the ArticleInfo API response. |
||
| 254 | * @param Project $project |
||
| 255 | * @param Page $page |
||
| 256 | * @return array |
||
| 257 | * @codeCoverageIgnore |
||
| 258 | */ |
||
| 259 | public function getArticleInfoApiData(Project $project, Page $page): array |
||
| 260 | { |
||
| 261 | /** @var int $pageviewsOffset Number of days to query for pageviews */ |
||
| 262 | $pageviewsOffset = 30; |
||
| 263 | |||
| 264 | $data = [ |
||
| 265 | 'project' => $project->getDomain(), |
||
| 266 | 'page' => $page->getTitle(), |
||
| 267 | 'watchers' => (int) $page->getWatchers(), |
||
| 268 | 'pageviews' => $page->getLastPageviews($pageviewsOffset), |
||
| 269 | 'pageviews_offset' => $pageviewsOffset, |
||
| 270 | ]; |
||
| 271 | |||
| 272 | $info = false; |
||
| 273 | |||
| 274 | try { |
||
| 275 | $articleInfoRepo = new ArticleInfoRepository(); |
||
| 276 | $articleInfoRepo->setContainer($this->container); |
||
| 277 | $info = $articleInfoRepo->getBasicEditingInfo($page); |
||
| 278 | } catch (ServiceUnavailableHttpException $e) { |
||
| 279 | // No more open database connections. |
||
| 280 | $data['error'] = 'Unable to fetch revision data. Please try again later.'; |
||
| 281 | } catch (HttpException $e) { |
||
| 282 | /** |
||
| 283 | * The query most likely exceeded the maximum query time, |
||
| 284 | * so we'll abort and give only info retrieved by the API. |
||
| 285 | */ |
||
| 286 | $data['error'] = 'Unable to fetch revision data. The query may have timed out.'; |
||
| 287 | } |
||
| 288 | |||
| 289 | if (false !== $info) { |
||
| 290 | $creationDateTime = DateTime::createFromFormat('YmdHis', $info['created_at']); |
||
| 291 | $modifiedDateTime = DateTime::createFromFormat('YmdHis', $info['modified_at']); |
||
| 292 | $secsSinceLastEdit = (new DateTime)->getTimestamp() - $modifiedDateTime->getTimestamp(); |
||
| 293 | |||
| 294 | // Some wikis (such foundation.wikimedia.org) may be missing the creation date. |
||
| 295 | $creationDateTime = false === $creationDateTime |
||
| 296 | ? null |
||
| 297 | : $creationDateTime->format('Y-m-d'); |
||
| 298 | |||
| 299 | $assessment = $page->getProject() |
||
| 300 | ->getPageAssessments() |
||
| 301 | ->getAssessment($page); |
||
| 302 | |||
| 303 | $data = array_merge($data, [ |
||
| 304 | 'revisions' => (int) $info['num_edits'], |
||
| 305 | 'editors' => (int) $info['num_editors'], |
||
| 306 | 'minor_edits' => (int) $info['minor_edits'], |
||
| 307 | 'author' => $info['author'], |
||
| 308 | 'author_editcount' => null === $info['author_editcount'] ? null : (int) $info['author_editcount'], |
||
| 309 | 'created_at' => $creationDateTime, |
||
| 310 | 'created_rev_id' => $info['created_rev_id'], |
||
| 311 | 'modified_at' => $modifiedDateTime->format('Y-m-d H:i'), |
||
| 312 | 'secs_since_last_edit' => $secsSinceLastEdit, |
||
| 313 | 'last_edit_id' => (int) $info['modified_rev_id'], |
||
| 314 | 'assessment' => $assessment, |
||
| 315 | ]); |
||
| 316 | } |
||
| 317 | |||
| 318 | return $data; |
||
| 319 | } |
||
| 320 | |||
| 321 | /************************ Link statistics ************************/ |
||
| 322 | |||
| 323 | /** |
||
| 324 | * Get the number of external links on the page. |
||
| 325 | * @return int |
||
| 326 | */ |
||
| 327 | public function linksExtCount(): int |
||
| 328 | { |
||
| 329 | return $this->getLinksAndRedirects()['links_ext_count']; |
||
| 330 | } |
||
| 331 | |||
| 332 | /** |
||
| 333 | * Get the number of incoming links to the page. |
||
| 334 | * @return int |
||
| 335 | */ |
||
| 336 | public function linksInCount(): int |
||
| 337 | { |
||
| 338 | return $this->getLinksAndRedirects()['links_in_count']; |
||
| 339 | } |
||
| 340 | |||
| 341 | /** |
||
| 342 | * Get the number of outgoing links from the page. |
||
| 343 | * @return int |
||
| 344 | */ |
||
| 345 | public function linksOutCount(): int |
||
| 346 | { |
||
| 347 | return $this->getLinksAndRedirects()['links_out_count']; |
||
| 348 | } |
||
| 349 | |||
| 350 | /** |
||
| 351 | * Get the number of redirects to the page. |
||
| 352 | * @return int |
||
| 353 | */ |
||
| 354 | public function redirectsCount(): int |
||
| 355 | { |
||
| 356 | return $this->getLinksAndRedirects()['redirects_count']; |
||
| 357 | } |
||
| 358 | |||
| 359 | /** |
||
| 360 | * Get the number of external, incoming and outgoing links, along with the number of redirects to the page. |
||
| 361 | * @return int[] |
||
| 362 | * @codeCoverageIgnore |
||
| 363 | */ |
||
| 364 | private function getLinksAndRedirects(): array |
||
| 365 | { |
||
| 366 | if (!is_array($this->linksAndRedirects)) { |
||
| 367 | $this->linksAndRedirects = $this->page->countLinksAndRedirects(); |
||
| 368 | } |
||
| 369 | return $this->linksAndRedirects; |
||
| 370 | } |
||
| 371 | |||
| 372 | /** |
||
| 373 | * Fetch transclusion data (categories, templates and files) that are on the page. |
||
| 374 | * @return array With keys 'categories', 'templates' and 'files'. |
||
| 375 | */ |
||
| 376 | public function getTransclusionData(): array |
||
| 377 | { |
||
| 378 | if (!is_array($this->transclusionData)) { |
||
| 379 | $this->transclusionData = $this->getRepository() |
||
| 380 | ->getTransclusionData($this->page); |
||
| 381 | } |
||
| 382 | return $this->transclusionData; |
||
| 383 | } |
||
| 384 | |||
| 385 | /** |
||
| 386 | * Get the number of categories that are on the page. |
||
| 387 | * @return int |
||
| 388 | */ |
||
| 389 | public function getNumCategories(): int |
||
| 390 | { |
||
| 391 | return $this->getTransclusionData()['categories']; |
||
| 392 | } |
||
| 393 | |||
| 394 | /** |
||
| 395 | * Get the number of templates that are on the page. |
||
| 396 | * @return int |
||
| 397 | */ |
||
| 398 | public function getNumTemplates(): int |
||
| 399 | { |
||
| 400 | return $this->getTransclusionData()['templates']; |
||
| 401 | } |
||
| 402 | |||
| 403 | /** |
||
| 404 | * Get the number of files that are on the page. |
||
| 405 | * @return int |
||
| 406 | */ |
||
| 407 | public function getNumFiles(): int |
||
| 408 | { |
||
| 409 | return $this->getTransclusionData()['files']; |
||
| 410 | } |
||
| 411 | |||
| 412 | /************************ Bot statistics ************************/ |
||
| 413 | |||
| 414 | /** |
||
| 415 | * Number of edits made to the page by current or former bots. |
||
| 416 | * @param string[] $bots Used only in unit tests, where we supply mock data for the bots that will get processed. |
||
| 417 | * @return int |
||
| 418 | */ |
||
| 419 | public function getBotRevisionCount(?array $bots = null): int |
||
| 420 | { |
||
| 421 | if (isset($this->botRevisionCount)) { |
||
| 422 | return $this->botRevisionCount; |
||
| 423 | } |
||
| 424 | |||
| 425 | if (null === $bots) { |
||
| 426 | $bots = $this->getBots(); |
||
| 427 | } |
||
| 428 | |||
| 429 | $count = 0; |
||
| 430 | |||
| 431 | foreach (array_values($bots) as $data) { |
||
| 432 | $count += $data['count']; |
||
| 433 | } |
||
| 434 | |||
| 435 | $this->botRevisionCount = $count; |
||
| 436 | return $count; |
||
| 437 | } |
||
| 438 | |||
| 439 | /** |
||
| 440 | * Get and set $this->bots about bots that edited the page. This is done separately from the main query because |
||
| 441 | * we use this information when computing the top 10 editors in ArticleInfo, where we don't want to include bots. |
||
| 442 | * @return mixed[] |
||
| 443 | */ |
||
| 444 | public function getBots(): array |
||
| 445 | { |
||
| 446 | if (isset($this->bots)) { |
||
| 447 | return $this->bots; |
||
| 448 | } |
||
| 449 | |||
| 450 | // Parse the bot edits. |
||
| 451 | $this->bots = []; |
||
| 452 | |||
| 453 | $limit = $this->tooManyRevisions() ? $this->getMaxRevisions() : null; |
||
| 454 | |||
| 455 | /** @var Statement $botData */ |
||
| 456 | $botData = $this->getRepository()->getBotData($this->page, $this->start, $this->end, $limit); |
||
| 457 | while ($bot = $botData->fetch()) { |
||
| 458 | $this->bots[$bot['username']] = [ |
||
| 459 | 'count' => (int)$bot['count'], |
||
| 460 | 'current' => '1' === $bot['current'], |
||
| 461 | ]; |
||
| 462 | } |
||
| 463 | |||
| 464 | // Sort by edit count. |
||
| 465 | uasort($this->bots, function ($a, $b) { |
||
| 466 | return $b['count'] - $a['count']; |
||
| 467 | }); |
||
| 468 | |||
| 469 | return $this->bots; |
||
| 470 | } |
||
| 471 | |||
| 472 | /** |
||
| 473 | * Get the number of bots that edited the page. |
||
| 474 | * @return int |
||
| 475 | */ |
||
| 476 | public function getNumBots(): int |
||
| 479 | } |
||
| 480 | } |
||
| 481 |
Loading history...