x-tools /
xtools
| Total Complexity | 50 |
| Total Lines | 573 |
| Duplicated Lines | 0 % |
| Changes | 0 | ||
Complex classes like PageRepository 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 PageRepository, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 17 | class PageRepository extends Repository |
||
| 18 | { |
||
| 19 | |||
| 20 | /** |
||
| 21 | * Get metadata about a single page from the API. |
||
| 22 | * @param Project $project The project to which the page belongs. |
||
| 23 | * @param string $pageTitle Page title. |
||
| 24 | * @return string[] Array with some of the following keys: pageid, title, missing, displaytitle, |
||
| 25 | * url. |
||
| 26 | */ |
||
| 27 | public function getPageInfo(Project $project, $pageTitle) |
||
| 28 | { |
||
| 29 | $info = $this->getPagesInfo($project, [$pageTitle]); |
||
| 30 | return array_shift($info); |
||
| 31 | } |
||
| 32 | |||
| 33 | /** |
||
| 34 | * Get metadata about a set of pages from the API. |
||
| 35 | * @param Project $project The project to which the pages belong. |
||
| 36 | * @param string[] $pageTitles Array of page titles. |
||
| 37 | * @return string[] Array keyed by the page names, each element with some of the |
||
| 38 | * following keys: pageid, title, missing, displaytitle, url. |
||
| 39 | */ |
||
| 40 | public function getPagesInfo(Project $project, $pageTitles) |
||
| 41 | { |
||
| 42 | // @TODO: Also include 'extlinks' prop when we start checking for dead external links. |
||
| 43 | $params = [ |
||
| 44 | 'prop' => 'info|pageprops', |
||
| 45 | 'inprop' => 'protection|talkid|watched|watchers|notificationtimestamp|subjectid|url|readable|displaytitle', |
||
| 46 | 'converttitles' => '', |
||
| 47 | // 'ellimit' => 20, |
||
| 48 | // 'elexpandurl' => '', |
||
| 49 | 'titles' => join('|', $pageTitles), |
||
| 50 | 'formatversion' => 2 |
||
| 51 | // 'pageids' => $pageIds // FIXME: allow page IDs |
||
| 52 | ]; |
||
| 53 | |||
| 54 | $query = new SimpleRequest('query', $params); |
||
| 55 | $api = $this->getMediawikiApi($project); |
||
| 56 | $res = $api->getRequest($query); |
||
| 57 | $result = []; |
||
| 58 | if (isset($res['query']['pages'])) { |
||
| 59 | foreach ($res['query']['pages'] as $pageInfo) { |
||
| 60 | $result[$pageInfo['title']] = $pageInfo; |
||
| 61 | } |
||
| 62 | } |
||
| 63 | return $result; |
||
| 64 | } |
||
| 65 | |||
| 66 | /** |
||
| 67 | * Get the full page text of a set of pages. |
||
| 68 | * @param Project $project The project to which the pages belong. |
||
| 69 | * @param string[] $pageTitles Array of page titles. |
||
| 70 | * @return string[] Array keyed by the page names, with the page text as the values. |
||
| 71 | */ |
||
| 72 | public function getPagesWikitext(Project $project, $pageTitles) |
||
| 98 | } |
||
| 99 | |||
| 100 | /** |
||
| 101 | * Get revisions of a single page. |
||
| 102 | * @param Page $page The page. |
||
| 103 | * @param User|null $user Specify to get only revisions by the given user. |
||
| 104 | * @param false|int $start |
||
| 105 | * @param false|int $end |
||
| 106 | * @return string[] Each member with keys: id, timestamp, length. |
||
| 107 | */ |
||
| 108 | public function getRevisions(Page $page, User $user = null, $start = false, $end = false) |
||
| 109 | { |
||
| 110 | $cacheKey = $this->getCacheKey(func_get_args(), 'page_revisions'); |
||
| 111 | if ($this->cache->hasItem($cacheKey)) { |
||
| 112 | return $this->cache->getItem($cacheKey)->get(); |
||
| 113 | } |
||
| 114 | |||
| 115 | $this->stopwatch->start($cacheKey, 'XTools'); |
||
| 116 | |||
| 117 | $stmt = $this->getRevisionsStmt($page, $user, null, null, $start, $end); |
||
| 118 | $result = $stmt->fetchAll(); |
||
| 119 | |||
| 120 | // Cache and return. |
||
| 121 | $this->stopwatch->stop($cacheKey); |
||
| 122 | return $this->setCache($cacheKey, $result); |
||
| 123 | } |
||
| 124 | |||
| 125 | /** |
||
| 126 | * Get the statement for a single revision, so that you can iterate row by row. |
||
| 127 | * @param Page $page The page. |
||
| 128 | * @param User|null $user Specify to get only revisions by the given user. |
||
| 129 | * @param int $limit Max number of revisions to process. |
||
| 130 | * @param int $numRevisions Number of revisions, if known. This is used solely to determine the |
||
| 131 | * OFFSET if we are given a $limit (see below). If $limit is set and $numRevisions is not set, |
||
| 132 | * a separate query is ran to get the nuber of revisions. |
||
| 133 | * @param false|int $start |
||
| 134 | * @param false|int $end |
||
| 135 | * @return Doctrine\DBAL\Driver\PDOStatement |
||
|
Bug
introduced
by
|
|||
| 136 | */ |
||
| 137 | public function getRevisionsStmt( |
||
| 138 | Page $page, |
||
| 139 | User $user = null, |
||
| 140 | $limit = null, |
||
| 141 | $numRevisions = null, |
||
| 142 | $start = false, |
||
| 143 | $end = false |
||
| 144 | ) { |
||
| 145 | $revTable = $this->getTableName($page->getProject()->getDatabaseName(), 'revision'); |
||
| 146 | $userClause = $user ? "revs.rev_user_text = :username AND " : ""; |
||
| 147 | |||
| 148 | // This sorts ascending by rev_timestamp because ArticleInfo must start with the oldest |
||
| 149 | // revision and work its way forward for proper processing. Consequently, if we want to do |
||
| 150 | // a LIMIT we want the most recent revisions, so we also need to know the total count to |
||
| 151 | // supply as the OFFSET. |
||
| 152 | $limitClause = ''; |
||
| 153 | if (intval($limit) > 0 && isset($numRevisions)) { |
||
| 154 | $offset = $numRevisions - $limit; |
||
| 155 | $limitClause = "LIMIT $offset, $limit"; |
||
| 156 | } |
||
| 157 | |||
| 158 | $datesConditions = $this->getDateConditions($start, $end, 'revs.'); |
||
| 159 | |||
| 160 | $sql = "SELECT |
||
| 161 | revs.rev_id AS id, |
||
| 162 | revs.rev_timestamp AS timestamp, |
||
| 163 | revs.rev_minor_edit AS minor, |
||
| 164 | revs.rev_len AS length, |
||
| 165 | (CAST(revs.rev_len AS SIGNED) - IFNULL(parentrevs.rev_len, 0)) AS length_change, |
||
| 166 | revs.rev_user AS user_id, |
||
| 167 | revs.rev_user_text AS username, |
||
| 168 | revs.rev_comment AS comment, |
||
| 169 | revs.rev_sha1 AS sha |
||
| 170 | FROM $revTable AS revs |
||
| 171 | LEFT JOIN $revTable AS parentrevs ON (revs.rev_parent_id = parentrevs.rev_id) |
||
| 172 | WHERE $userClause revs.rev_page = :pageid $datesConditions |
||
| 173 | ORDER BY revs.rev_timestamp ASC |
||
| 174 | $limitClause"; |
||
| 175 | |||
| 176 | $params = ['pageid' => $page->getId()]; |
||
| 177 | if ($user) { |
||
| 178 | $params['username'] = $user->getUsername(); |
||
| 179 | } |
||
| 180 | |||
| 181 | return $this->executeProjectsQuery($sql, $params); |
||
| 182 | } |
||
| 183 | |||
| 184 | /** |
||
| 185 | * Get a count of the number of revisions of a single page |
||
| 186 | * @param Page $page The page. |
||
| 187 | * @param User|null $user Specify to only count revisions by the given user. |
||
| 188 | * @param false|int $start |
||
| 189 | * @param false|int $end |
||
| 190 | * @return int |
||
| 191 | */ |
||
| 192 | public function getNumRevisions(Page $page, User $user = null, $start = false, $end = false) |
||
| 193 | { |
||
| 194 | $cacheKey = $this->getCacheKey(func_get_args(), 'page_numrevisions'); |
||
| 195 | if ($this->cache->hasItem($cacheKey)) { |
||
| 196 | return $this->cache->getItem($cacheKey)->get(); |
||
| 197 | } |
||
| 198 | |||
| 199 | $revTable = $page->getProject()->getTableName('revision'); |
||
| 200 | $userClause = $user ? "rev_user_text = :username AND " : ""; |
||
| 201 | |||
| 202 | $datesConditions = $this->getDateConditions($start, $end); |
||
| 203 | |||
| 204 | $sql = "SELECT COUNT(*) |
||
| 205 | FROM $revTable |
||
| 206 | WHERE $userClause rev_page = :pageid $datesConditions"; |
||
| 207 | $params = ['pageid' => $page->getId()]; |
||
| 208 | if ($user) { |
||
| 209 | $params['username'] = $user->getUsername(); |
||
| 210 | } |
||
| 211 | |||
| 212 | $result = $this->executeProjectsQuery($sql, $params)->fetchColumn(0); |
||
| 213 | |||
| 214 | // Cache and return. |
||
| 215 | return $this->setCache($cacheKey, $result); |
||
| 216 | } |
||
| 217 | |||
| 218 | /** |
||
| 219 | * Get various basic info used in the API, including the |
||
| 220 | * number of revisions, unique authors, initial author |
||
| 221 | * and edit count of the initial author. |
||
| 222 | * This is combined into one query for better performance. |
||
| 223 | * Caching is only applied if it took considerable time to process, |
||
| 224 | * because using the gadget, this will get hit for a different page |
||
| 225 | * constantly, where the likelihood of cache benefiting us is slim. |
||
| 226 | * @param Page $page The page. |
||
| 227 | * @return string[] |
||
| 228 | */ |
||
| 229 | public function getBasicEditingInfo(Page $page) |
||
| 230 | { |
||
| 231 | $cacheKey = $this->getCacheKey(func_get_args(), 'page_basicinfo'); |
||
| 232 | if ($this->cache->hasItem($cacheKey)) { |
||
| 233 | return $this->cache->getItem($cacheKey)->get(); |
||
| 234 | } |
||
| 235 | |||
| 236 | $revTable = $this->getTableName($page->getProject()->getDatabaseName(), 'revision'); |
||
| 237 | $userTable = $this->getTableName($page->getProject()->getDatabaseName(), 'user'); |
||
| 238 | $pageTable = $this->getTableName($page->getProject()->getDatabaseName(), 'page'); |
||
| 239 | |||
| 240 | $sql = "SELECT *, ( |
||
| 241 | SELECT user_editcount |
||
| 242 | FROM $userTable |
||
| 243 | WHERE user_name = author |
||
| 244 | ) AS author_editcount |
||
| 245 | FROM ( |
||
| 246 | ( |
||
| 247 | SELECT COUNT(*) AS num_edits, |
||
| 248 | COUNT(DISTINCT(rev_user_text)) AS num_editors |
||
| 249 | FROM $revTable |
||
| 250 | WHERE rev_page = :pageid |
||
| 251 | ) a, |
||
| 252 | ( |
||
| 253 | # With really old pages, the rev_timestamp may need to be sorted ASC, |
||
| 254 | # and the lowest rev_id may not be the first revision. |
||
| 255 | SELECT rev_user_text AS author, |
||
| 256 | rev_timestamp AS created_at, |
||
| 257 | rev_id AS created_rev_id |
||
| 258 | FROM $revTable |
||
| 259 | WHERE rev_page = :pageid |
||
| 260 | ORDER BY rev_timestamp ASC |
||
| 261 | LIMIT 1 |
||
| 262 | ) b, |
||
| 263 | ( |
||
| 264 | SELECT MAX(rev_timestamp) AS modified_at |
||
| 265 | FROM $revTable |
||
| 266 | WHERE rev_page = :pageid |
||
| 267 | ) c, |
||
| 268 | ( |
||
| 269 | SELECT page_latest AS modified_rev_id |
||
| 270 | FROM $pageTable |
||
| 271 | WHERE page_id = :pageid |
||
| 272 | ) d |
||
| 273 | );"; |
||
| 274 | $params = ['pageid' => $page->getId()]; |
||
| 275 | |||
| 276 | // Get current time so we can compare timestamps |
||
| 277 | // and decide whether or to cache the result. |
||
| 278 | $time1 = time(); |
||
| 279 | |||
| 280 | /** |
||
| 281 | * This query can sometimes take too long to run for pages with tens of thousands |
||
| 282 | * of revisions. This query is used by the ArticleInfo gadget, which shows basic |
||
| 283 | * data in real-time, so if it takes too long than the user probably didn't even |
||
| 284 | * wait to see the result. We'll pass 60 as the last parameter to executeProjectsQuery, |
||
| 285 | * which will set the max_statement_time to 60 seconds. |
||
| 286 | */ |
||
| 287 | $result = $this->executeProjectsQuery($sql, $params, 60)->fetch(); |
||
| 288 | |||
| 289 | $time2 = time(); |
||
| 290 | |||
| 291 | // If it took over 5 seconds, cache the result for 20 minutes. |
||
| 292 | if ($time2 - $time1 > 5) { |
||
| 293 | $this->setCache($cacheKey, $result, 'PT20M'); |
||
| 294 | } |
||
| 295 | |||
| 296 | return $result; |
||
| 297 | } |
||
| 298 | |||
| 299 | /** |
||
| 300 | * Get assessment data for the given pages |
||
| 301 | * @param Project $project The project to which the pages belong. |
||
| 302 | * @param int[] $pageIds Page IDs |
||
| 303 | * @return string[] Assessment data as retrieved from the database. |
||
| 304 | */ |
||
| 305 | public function getAssessments(Project $project, $pageIds) |
||
| 306 | { |
||
| 307 | $cacheKey = $this->getCacheKey(func_get_args(), 'page_assessments'); |
||
| 308 | if ($this->cache->hasItem($cacheKey)) { |
||
| 309 | return $this->cache->getItem($cacheKey)->get(); |
||
| 310 | } |
||
| 311 | |||
| 312 | if (!$project->hasPageAssessments()) { |
||
| 313 | return []; |
||
| 314 | } |
||
| 315 | $paTable = $this->getTableName($project->getDatabaseName(), 'page_assessments'); |
||
| 316 | $papTable = $this->getTableName($project->getDatabaseName(), 'page_assessments_projects'); |
||
| 317 | $pageIds = implode($pageIds, ','); |
||
| 318 | |||
| 319 | $sql = "SELECT pap_project_title AS wikiproject, pa_class AS class, pa_importance AS importance |
||
| 320 | FROM $paTable |
||
| 321 | LEFT JOIN $papTable ON pa_project_id = pap_project_id |
||
| 322 | WHERE pa_page_id IN ($pageIds)"; |
||
| 323 | |||
| 324 | $result = $this->executeProjectsQuery($sql)->fetchAll(); |
||
| 325 | |||
| 326 | // Cache and return. |
||
| 327 | return $this->setCache($cacheKey, $result); |
||
| 328 | } |
||
| 329 | |||
| 330 | /** |
||
| 331 | * Get any CheckWiki errors of a single page |
||
| 332 | * @param Page $page |
||
| 333 | * @return array Results from query |
||
| 334 | */ |
||
| 335 | public function getCheckWikiErrors(Page $page) |
||
| 336 | { |
||
| 337 | // Only support mainspace on Labs installations |
||
| 338 | if ($page->getNamespace() !== 0 || !$this->isLabs()) { |
||
| 339 | return []; |
||
| 340 | } |
||
| 341 | |||
| 342 | $sql = "SELECT error, notice, found, name_trans AS name, prio, text_trans AS explanation |
||
| 343 | FROM s51080__checkwiki_p.cw_error a |
||
| 344 | JOIN s51080__checkwiki_p.cw_overview_errors b |
||
| 345 | WHERE a.project = b.project |
||
| 346 | AND a.project = :dbName |
||
| 347 | AND a.title = :title |
||
| 348 | AND a.error = b.id |
||
| 349 | AND a.ok = 0"; |
||
| 350 | |||
| 351 | // remove _p if present |
||
| 352 | $dbName = preg_replace('/_p$/', '', $page->getProject()->getDatabaseName()); |
||
| 353 | |||
| 354 | // Page title without underscores (str_replace just to be sure) |
||
| 355 | $pageTitle = str_replace('_', ' ', $page->getTitle()); |
||
| 356 | |||
| 357 | $resultQuery = $this->getToolsConnection()->prepare($sql); |
||
| 358 | $resultQuery->bindParam(':dbName', $dbName); |
||
| 359 | $resultQuery->bindParam(':title', $pageTitle); |
||
| 360 | $resultQuery->execute(); |
||
| 361 | |||
| 362 | return $resultQuery->fetchAll(); |
||
| 363 | } |
||
| 364 | |||
| 365 | /** |
||
| 366 | * Get basic wikidata on the page: label and description. |
||
| 367 | * @param Page $page |
||
| 368 | * @return string[] In the format: |
||
| 369 | * [[ |
||
| 370 | * 'term' => string such as 'label', |
||
| 371 | * 'term_text' => string (value for 'label'), |
||
| 372 | * ], ... ] |
||
| 373 | */ |
||
| 374 | public function getWikidataInfo(Page $page) |
||
| 375 | { |
||
| 376 | if (empty($page->getWikidataId())) { |
||
| 377 | return []; |
||
| 378 | } |
||
| 379 | |||
| 380 | $wikidataId = ltrim($page->getWikidataId(), 'Q'); |
||
| 381 | $lang = $page->getProject()->getLang(); |
||
| 382 | |||
| 383 | $sql = "SELECT term_type AS term, term_text |
||
| 384 | FROM wikidatawiki_p.wb_terms |
||
| 385 | WHERE term_entity_id = :wikidataId |
||
| 386 | AND term_type IN ('label', 'description') |
||
| 387 | AND term_language = :lang"; |
||
| 388 | |||
| 389 | return $this->executeProjectsQuery($sql, [ |
||
| 390 | 'lang' => $lang, |
||
| 391 | 'wikidataId' => $wikidataId, |
||
| 392 | ])->fetchAll(); |
||
| 393 | } |
||
| 394 | |||
| 395 | /** |
||
| 396 | * Get or count all wikidata items for the given page, |
||
| 397 | * not just languages of sister projects |
||
| 398 | * @param Page $page |
||
| 399 | * @param bool $count Set to true to get only a COUNT |
||
| 400 | * @return string[]|int Records as returend by the DB, |
||
| 401 | * or raw COUNT of the records. |
||
| 402 | */ |
||
| 403 | public function getWikidataItems(Page $page, $count = false) |
||
| 404 | { |
||
| 405 | if (!$page->getWikidataId()) { |
||
| 406 | return $count ? 0 : []; |
||
| 407 | } |
||
| 408 | |||
| 409 | $wikidataId = ltrim($page->getWikidataId(), 'Q'); |
||
| 410 | |||
| 411 | $sql = "SELECT " . ($count ? 'COUNT(*) AS count' : '*') . " |
||
| 412 | FROM wikidatawiki_p.wb_items_per_site |
||
| 413 | WHERE ips_item_id = :wikidataId"; |
||
| 414 | |||
| 415 | $result = $this->executeProjectsQuery($sql, [ |
||
| 416 | 'wikidataId' => $wikidataId, |
||
| 417 | ])->fetchAll(); |
||
| 418 | |||
| 419 | return $count ? (int) $result[0]['count'] : $result; |
||
| 420 | } |
||
| 421 | |||
| 422 | /** |
||
| 423 | * Get number of in and outgoing links and redirects to the given page. |
||
| 424 | * @param Page $page |
||
| 425 | * @return string[] Counts with the keys 'links_ext_count', 'links_out_count', |
||
| 426 | * 'links_in_count' and 'redirects_count' |
||
| 427 | */ |
||
| 428 | public function countLinksAndRedirects(Page $page) |
||
| 429 | { |
||
| 430 | $externalLinksTable = $this->getTableName($page->getProject()->getDatabaseName(), 'externallinks'); |
||
| 431 | $pageLinksTable = $this->getTableName($page->getProject()->getDatabaseName(), 'pagelinks'); |
||
| 432 | $redirectTable = $this->getTableName($page->getProject()->getDatabaseName(), 'redirect'); |
||
| 433 | |||
| 434 | $sql = "SELECT COUNT(*) AS value, 'links_ext' AS type |
||
| 435 | FROM $externalLinksTable WHERE el_from = :id |
||
| 436 | UNION |
||
| 437 | SELECT COUNT(*) AS value, 'links_out' AS type |
||
| 438 | FROM $pageLinksTable WHERE pl_from = :id |
||
| 439 | UNION |
||
| 440 | SELECT COUNT(*) AS value, 'links_in' AS type |
||
| 441 | FROM $pageLinksTable WHERE pl_namespace = :namespace AND pl_title = :title |
||
| 442 | UNION |
||
| 443 | SELECT COUNT(*) AS value, 'redirects' AS type |
||
| 444 | FROM $redirectTable WHERE rd_namespace = :namespace AND rd_title = :title"; |
||
| 445 | |||
| 446 | $params = [ |
||
| 447 | 'id' => $page->getId(), |
||
| 448 | 'title' => str_replace(' ', '_', $page->getTitleWithoutNamespace()), |
||
| 449 | 'namespace' => $page->getNamespace(), |
||
| 450 | ]; |
||
| 451 | |||
| 452 | $res = $this->executeProjectsQuery($sql, $params); |
||
| 453 | $data = []; |
||
| 454 | |||
| 455 | // Transform to associative array by 'type' |
||
| 456 | foreach ($res as $row) { |
||
| 457 | $data[$row['type'] . '_count'] = $row['value']; |
||
| 458 | } |
||
| 459 | |||
| 460 | return $data; |
||
| 461 | } |
||
| 462 | |||
| 463 | /** |
||
| 464 | * Count wikidata items for the given page, not just languages of sister projects |
||
| 465 | * @param Page $page |
||
| 466 | * @return int Number of records. |
||
| 467 | */ |
||
| 468 | public function countWikidataItems(Page $page) |
||
| 471 | } |
||
| 472 | |||
| 473 | /** |
||
| 474 | * Get page views for the given page and timeframe. |
||
| 475 | * @FIXME use Symfony Guzzle package. |
||
| 476 | * @param Page $page |
||
| 477 | * @param string|DateTime $start In the format YYYYMMDD |
||
| 478 | * @param string|DateTime $end In the format YYYYMMDD |
||
| 479 | * @return string[] |
||
| 480 | */ |
||
| 481 | public function getPageviews(Page $page, $start, $end) |
||
| 504 | } |
||
| 505 | |||
| 506 | /** |
||
| 507 | * Get the full HTML content of the the page. |
||
| 508 | * @param Page $page |
||
| 509 | * @param int $revId What revision to query for. |
||
| 510 | * @return string |
||
| 511 | */ |
||
| 512 | public function getHTMLContent(Page $page, $revId = null) |
||
| 522 | } |
||
| 523 | |||
| 524 | /** |
||
| 525 | * Get the ID of the revision of a page at the time of the given DateTime. |
||
| 526 | * @param Page $page |
||
| 527 | * @param DateTime $date |
||
| 528 | * @return int |
||
| 529 | */ |
||
| 530 | public function getRevisionIdAtDate(Page $page, DateTime $date) |
||
| 541 | } |
||
| 542 | |||
| 543 | /** |
||
| 544 | * Get HTML display titles of a set of pages (or the normal title if there's no display title). |
||
| 545 | * This will send t/50 API requests where t is the number of titles supplied. |
||
| 546 | * @param Project $project The project. |
||
| 547 | * @param string[] $pageTitles The titles to fetch. |
||
| 548 | * @return string[] Keys are the original supplied title, and values are the display titles. |
||
| 549 | * @static |
||
| 550 | */ |
||
| 551 | public static function displayTitles(Project $project, $pageTitles) |
||
| 592 |
Loading history...