x-tools /
xtools
| Total Complexity | 65 |
| Total Lines | 508 |
| Duplicated Lines | 0 % |
| Changes | 0 | ||
Complex classes like Page 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 Page, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 16 | class Page extends Model |
||
| 17 | { |
||
| 18 | /** @var string The page name as provided at instantiation. */ |
||
| 19 | protected $unnormalizedPageName; |
||
| 20 | |||
| 21 | /** @var string[] Metadata about this page. */ |
||
| 22 | protected $pageInfo = []; |
||
| 23 | |||
| 24 | /** @var string[] Revision history of this page. */ |
||
| 25 | protected $revisions; |
||
| 26 | |||
| 27 | /** @var int Number of revisions for this page. */ |
||
| 28 | protected $numRevisions; |
||
| 29 | |||
| 30 | /** @var string[] List of Wikidata sitelinks for this page. */ |
||
| 31 | protected $wikidataItems; |
||
| 32 | |||
| 33 | /** @var int Number of Wikidata sitelinks for this page. */ |
||
| 34 | protected $numWikidataItems; |
||
| 35 | |||
| 36 | /** @var string Title of the page. */ |
||
| 37 | protected $pageTitle; |
||
| 38 | |||
| 39 | /** @var int Length of the page in bytes. */ |
||
| 40 | protected $length; |
||
| 41 | |||
| 42 | /** |
||
| 43 | * Page constructor. |
||
| 44 | * @param Project $project |
||
| 45 | * @param string $pageName |
||
| 46 | */ |
||
| 47 | public function __construct(Project $project, string $pageName) |
||
| 48 | { |
||
| 49 | $this->project = $project; |
||
| 50 | $this->unnormalizedPageName = $pageName; |
||
| 51 | } |
||
| 52 | |||
| 53 | /** |
||
| 54 | * Get a Page instance given a database row (either from or JOINed on the page table). |
||
| 55 | * @param Project $project |
||
| 56 | * @param array $row Must contain 'page_title' and 'page_namespace'. May contain 'page_len'. |
||
| 57 | * @return static |
||
| 58 | */ |
||
| 59 | public static function newFromRow(Project $project, array $row): self |
||
| 60 | { |
||
| 61 | $pageTitle = $row['page_title']; |
||
| 62 | |||
| 63 | if (0 === (int)$row['page_namespace']) { |
||
| 64 | $fullPageTitle = $pageTitle; |
||
| 65 | } else { |
||
| 66 | $namespaces = $project->getNamespaces(); |
||
| 67 | $fullPageTitle = $namespaces[$row['page_namespace']].":$pageTitle"; |
||
| 68 | } |
||
| 69 | |||
| 70 | $page = new self($project, $fullPageTitle); |
||
| 71 | $page->pageInfo['ns'] = $row['page_namespace']; |
||
| 72 | if (isset($row['page_len'])) { |
||
| 73 | $page->length = (int)$row['page_len']; |
||
| 74 | } |
||
| 75 | |||
| 76 | return $page; |
||
| 77 | } |
||
| 78 | |||
| 79 | /** |
||
| 80 | * Unique identifier for this Page, to be used in cache keys. |
||
| 81 | * Use of md5 ensures the cache key does not contain reserved characters. |
||
| 82 | * @see Repository::getCacheKey() |
||
| 83 | * @return string |
||
| 84 | * @codeCoverageIgnore |
||
| 85 | */ |
||
| 86 | public function getCacheKey(): string |
||
| 87 | { |
||
| 88 | return md5((string)$this->getId()); |
||
| 89 | } |
||
| 90 | |||
| 91 | /** |
||
| 92 | * Get basic information about this page from the repository. |
||
| 93 | * @return array|null |
||
| 94 | */ |
||
| 95 | protected function getPageInfo(): ?array |
||
| 96 | { |
||
| 97 | if (empty($this->pageInfo)) { |
||
| 98 | $this->pageInfo = $this->getRepository() |
||
| 99 | ->getPageInfo($this->project, $this->unnormalizedPageName); |
||
|
Bug
introduced
by
|
|||
| 100 | } |
||
| 101 | return $this->pageInfo; |
||
| 102 | } |
||
| 103 | |||
| 104 | /** |
||
| 105 | * Get the page's title. |
||
| 106 | * @param bool $useUnnormalized Use the unnormalized page title to avoid an API call. This should be used only if |
||
| 107 | * you fetched the page title via other means (SQL query), and is not from user input alone. |
||
| 108 | * @return string |
||
| 109 | */ |
||
| 110 | public function getTitle(bool $useUnnormalized = false): string |
||
| 111 | { |
||
| 112 | if ($useUnnormalized) { |
||
| 113 | return $this->unnormalizedPageName; |
||
| 114 | } |
||
| 115 | $info = $this->getPageInfo(); |
||
| 116 | return $info['title'] ?? $this->unnormalizedPageName; |
||
| 117 | } |
||
| 118 | |||
| 119 | /** |
||
| 120 | * Get the page's title without the namespace. |
||
| 121 | * @return string |
||
| 122 | */ |
||
| 123 | public function getTitleWithoutNamespace(): string |
||
| 124 | { |
||
| 125 | $info = $this->getPageInfo(); |
||
| 126 | $title = $info['title'] ?? $this->unnormalizedPageName; |
||
| 127 | $nsName = $this->getNamespaceName(); |
||
| 128 | return $nsName |
||
| 129 | ? str_replace($nsName . ':', '', $title) |
||
| 130 | : $title; |
||
| 131 | } |
||
| 132 | |||
| 133 | /** |
||
| 134 | * Get this page's database ID. |
||
| 135 | * @return int|null Null if nonexistent. |
||
| 136 | */ |
||
| 137 | public function getId(): ?int |
||
| 138 | { |
||
| 139 | $info = $this->getPageInfo(); |
||
| 140 | return isset($info['pageid']) ? (int)$info['pageid'] : null; |
||
| 141 | } |
||
| 142 | |||
| 143 | /** |
||
| 144 | * Get this page's length in bytes. |
||
| 145 | * @return int|null Null if nonexistent. |
||
| 146 | */ |
||
| 147 | public function getLength(): ?int |
||
| 148 | { |
||
| 149 | if (isset($this->length)) { |
||
| 150 | return $this->length; |
||
| 151 | } |
||
| 152 | $info = $this->getPageInfo(); |
||
| 153 | $this->length = isset($info['length']) ? (int)$info['length'] : null; |
||
| 154 | return $this->length; |
||
| 155 | } |
||
| 156 | |||
| 157 | /** |
||
| 158 | * Get HTML for the stylized display of the title. |
||
| 159 | * The text will be the same as Page::getTitle(). |
||
| 160 | * @return string |
||
| 161 | */ |
||
| 162 | public function getDisplayTitle(): string |
||
| 163 | { |
||
| 164 | $info = $this->getPageInfo(); |
||
| 165 | if (isset($info['displaytitle'])) { |
||
| 166 | return $info['displaytitle']; |
||
| 167 | } |
||
| 168 | return $this->getTitle(); |
||
| 169 | } |
||
| 170 | |||
| 171 | /** |
||
| 172 | * Get the full URL of this page. |
||
| 173 | * @return string|null Null if nonexistent. |
||
| 174 | */ |
||
| 175 | public function getUrl(): ?string |
||
| 179 | } |
||
| 180 | |||
| 181 | /** |
||
| 182 | * Get the numerical ID of the namespace of this page. |
||
| 183 | * @return int|null Null if page doesn't exist. |
||
| 184 | */ |
||
| 185 | public function getNamespace(): ?int |
||
| 186 | { |
||
| 187 | if (isset($this->pageInfo['ns']) && is_numeric($this->pageInfo['ns'])) { |
||
| 188 | return (int)$this->pageInfo['ns']; |
||
| 189 | } |
||
| 190 | $info = $this->getPageInfo(); |
||
| 191 | return isset($info['ns']) ? (int)$info['ns'] : null; |
||
| 192 | } |
||
| 193 | |||
| 194 | /** |
||
| 195 | * Get the name of the namespace of this page. |
||
| 196 | * @return string|null Null if could not be determined. |
||
| 197 | */ |
||
| 198 | public function getNamespaceName(): ?string |
||
| 199 | { |
||
| 200 | $info = $this->getPageInfo(); |
||
| 201 | return isset($info['ns']) |
||
| 202 | ? ($this->getProject()->getNamespaces()[$info['ns']] ?? null) |
||
| 203 | : null; |
||
| 204 | } |
||
| 205 | |||
| 206 | /** |
||
| 207 | * Get the number of page watchers. |
||
| 208 | * @return int|null Null if unknown. |
||
| 209 | */ |
||
| 210 | public function getWatchers(): ?int |
||
| 211 | { |
||
| 212 | $info = $this->getPageInfo(); |
||
| 213 | return isset($info['watchers']) ? (int)$info['watchers'] : null; |
||
| 214 | } |
||
| 215 | |||
| 216 | /** |
||
| 217 | * Get the HTML content of the body of the page. |
||
| 218 | * @param DateTime|int $target If a DateTime object, the |
||
| 219 | * revision at that time will be returned. If an integer, it is |
||
| 220 | * assumed to be the actual revision ID. |
||
| 221 | * @return string |
||
| 222 | */ |
||
| 223 | public function getHTMLContent($target = null): string |
||
| 224 | { |
||
| 225 | if (is_a($target, 'DateTime')) { |
||
| 226 | $target = $this->getRepository()->getRevisionIdAtDate($this, $target); |
||
| 227 | } |
||
| 228 | return $this->getRepository()->getHTMLContent($this, $target); |
||
| 229 | } |
||
| 230 | |||
| 231 | /** |
||
| 232 | * Whether or not this page exists. |
||
| 233 | * @return bool |
||
| 234 | */ |
||
| 235 | public function exists(): bool |
||
| 236 | { |
||
| 237 | $info = $this->getPageInfo(); |
||
| 238 | return null !== $info && !isset($info['missing']) && !isset($info['invalid']) && !isset($info['interwiki']); |
||
| 239 | } |
||
| 240 | |||
| 241 | /** |
||
| 242 | * Get the Project to which this page belongs. |
||
| 243 | * @return Project |
||
| 244 | */ |
||
| 245 | public function getProject(): Project |
||
| 246 | { |
||
| 247 | return $this->project; |
||
| 248 | } |
||
| 249 | |||
| 250 | /** |
||
| 251 | * Get the language code for this page. |
||
| 252 | * If not set, the language code for the project is returned. |
||
| 253 | * @return string |
||
| 254 | */ |
||
| 255 | public function getLang(): string |
||
| 256 | { |
||
| 257 | $info = $this->getPageInfo(); |
||
| 258 | if (isset($info['pagelanguage'])) { |
||
| 259 | return $info['pagelanguage']; |
||
| 260 | } else { |
||
| 261 | return $this->getProject()->getLang(); |
||
| 262 | } |
||
| 263 | } |
||
| 264 | |||
| 265 | /** |
||
| 266 | * Get the Wikidata ID of this page. |
||
| 267 | * @return string|null Null if none exists. |
||
| 268 | */ |
||
| 269 | public function getWikidataId(): ?string |
||
| 270 | { |
||
| 271 | $info = $this->getPageInfo(); |
||
| 272 | if (isset($info['pageprops']['wikibase_item'])) { |
||
| 273 | return $info['pageprops']['wikibase_item']; |
||
| 274 | } else { |
||
| 275 | return null; |
||
| 276 | } |
||
| 277 | } |
||
| 278 | |||
| 279 | /** |
||
| 280 | * Get the number of revisions the page has. |
||
| 281 | * @param ?User $user Optionally limit to those of this user. |
||
| 282 | * @param false|int $start |
||
| 283 | * @param false|int $end |
||
| 284 | * @return int |
||
| 285 | */ |
||
| 286 | public function getNumRevisions(?User $user = null, $start = false, $end = false): int |
||
| 287 | { |
||
| 288 | // If a user is given, we will not cache the result via instance variable. |
||
| 289 | if (null !== $user) { |
||
| 290 | return (int)$this->getRepository()->getNumRevisions($this, $user, $start, $end); |
||
| 291 | } |
||
| 292 | |||
| 293 | // Return cached value, if present. |
||
| 294 | if (null !== $this->numRevisions) { |
||
| 295 | return $this->numRevisions; |
||
| 296 | } |
||
| 297 | |||
| 298 | // Otherwise, return the count of all revisions if already present. |
||
| 299 | if (null !== $this->revisions) { |
||
| 300 | $this->numRevisions = count($this->revisions); |
||
| 301 | } else { |
||
| 302 | // Otherwise do a COUNT in the event fetching all revisions is not desired. |
||
| 303 | $this->numRevisions = (int)$this->getRepository()->getNumRevisions($this, null, $start, $end); |
||
| 304 | } |
||
| 305 | |||
| 306 | return $this->numRevisions; |
||
| 307 | } |
||
| 308 | |||
| 309 | /** |
||
| 310 | * Get all edits made to this page. |
||
| 311 | * @param User|null $user Specify to get only revisions by the given user. |
||
| 312 | * @param false|int $start |
||
| 313 | * @param false|int $end |
||
| 314 | * @return array |
||
| 315 | */ |
||
| 316 | public function getRevisions(?User $user = null, $start = false, $end = false): array |
||
| 317 | { |
||
| 318 | if ($this->revisions) { |
||
| 319 | return $this->revisions; |
||
| 320 | } |
||
| 321 | |||
| 322 | $this->revisions = $this->getRepository()->getRevisions($this, $user, $start, $end); |
||
| 323 | |||
| 324 | return $this->revisions; |
||
| 325 | } |
||
| 326 | |||
| 327 | /** |
||
| 328 | * Get the full page wikitext. |
||
| 329 | * @return string|null Null if nothing was found. |
||
| 330 | */ |
||
| 331 | public function getWikitext(): ?string |
||
| 332 | { |
||
| 333 | $content = $this->getRepository()->getPagesWikitext( |
||
| 334 | $this->getProject(), |
||
| 335 | [ $this->getTitle() ] |
||
| 336 | ); |
||
| 337 | |||
| 338 | return $content[$this->getTitle()] ?? null; |
||
| 339 | } |
||
| 340 | |||
| 341 | /** |
||
| 342 | * Get the statement for a single revision, so that you can iterate row by row. |
||
| 343 | * @see PageRepository::getRevisionsStmt() |
||
| 344 | * @param User|null $user Specify to get only revisions by the given user. |
||
| 345 | * @param ?int $limit Max number of revisions to process. |
||
| 346 | * @param ?int $numRevisions Number of revisions, if known. This is used solely to determine the |
||
| 347 | * OFFSET if we are given a $limit. If $limit is set and $numRevisions is not set, a |
||
| 348 | * separate query is ran to get the nuber of revisions. |
||
| 349 | * @param false|int $start |
||
| 350 | * @param false|int $end |
||
| 351 | * @return ResultStatement |
||
| 352 | */ |
||
| 353 | public function getRevisionsStmt( |
||
| 354 | ?User $user = null, |
||
| 355 | ?int $limit = null, |
||
| 356 | ?int $numRevisions = null, |
||
| 357 | $start = false, |
||
| 358 | $end = false |
||
| 359 | ): ResultStatement { |
||
| 360 | // If we have a limit, we need to know the total number of revisions so that PageRepo |
||
| 361 | // will properly set the OFFSET. See PageRepository::getRevisionsStmt() for more info. |
||
| 362 | if (isset($limit) && null === $numRevisions) { |
||
| 363 | $numRevisions = $this->getNumRevisions($user, $start, $end); |
||
| 364 | } |
||
| 365 | return $this->getRepository()->getRevisionsStmt($this, $user, $limit, $numRevisions, $start, $end); |
||
| 366 | } |
||
| 367 | |||
| 368 | /** |
||
| 369 | * Get the revision ID that immediately precedes the given date. |
||
| 370 | * @param DateTime $date |
||
| 371 | * @return int|null Null if none found. |
||
| 372 | */ |
||
| 373 | public function getRevisionIdAtDate(DateTime $date): ?int |
||
| 374 | { |
||
| 375 | return $this->getRepository()->getRevisionIdAtDate($this, $date); |
||
| 376 | } |
||
| 377 | |||
| 378 | /** |
||
| 379 | * Get CheckWiki errors for this page |
||
| 380 | * @return string[] See getErrors() for format |
||
| 381 | */ |
||
| 382 | public function getCheckWikiErrors(): array |
||
| 383 | { |
||
| 384 | return $this->getRepository()->getCheckWikiErrors($this); |
||
| 385 | } |
||
| 386 | |||
| 387 | /** |
||
| 388 | * Get Wikidata errors for this page |
||
| 389 | * @return string[] See getErrors() for format |
||
| 390 | */ |
||
| 391 | public function getWikidataErrors(): array |
||
| 392 | { |
||
| 393 | $errors = []; |
||
| 394 | |||
| 395 | if (empty($this->getWikidataId())) { |
||
| 396 | return []; |
||
| 397 | } |
||
| 398 | |||
| 399 | $wikidataInfo = $this->getRepository()->getWikidataInfo($this); |
||
| 400 | |||
| 401 | $terms = array_map(function ($entry) { |
||
| 402 | return $entry['term']; |
||
| 403 | }, $wikidataInfo); |
||
| 404 | |||
| 405 | $lang = $this->getLang(); |
||
| 406 | |||
| 407 | if (!in_array('label', $terms)) { |
||
| 408 | $errors[] = [ |
||
| 409 | 'prio' => 2, |
||
| 410 | 'name' => 'Wikidata', |
||
| 411 | 'notice' => "Label for language <em>$lang</em> is missing", // FIXME: i18n |
||
| 412 | 'explanation' => "See: <a target='_blank' " . |
||
| 413 | "href='//www.wikidata.org/wiki/Help:Label'>Help:Label</a>", |
||
| 414 | ]; |
||
| 415 | } |
||
| 416 | |||
| 417 | if (!in_array('description', $terms)) { |
||
| 418 | $errors[] = [ |
||
| 419 | 'prio' => 3, |
||
| 420 | 'name' => 'Wikidata', |
||
| 421 | 'notice' => "Description for language <em>$lang</em> is missing", // FIXME: i18n |
||
| 422 | 'explanation' => "See: <a target='_blank' " . |
||
| 423 | "href='//www.wikidata.org/wiki/Help:Description'>Help:Description</a>", |
||
| 424 | ]; |
||
| 425 | } |
||
| 426 | |||
| 427 | return $errors; |
||
| 428 | } |
||
| 429 | |||
| 430 | /** |
||
| 431 | * Get Wikidata and CheckWiki errors, if present |
||
| 432 | * @return string[] List of errors in the format: |
||
| 433 | * [[ |
||
| 434 | * 'prio' => int, |
||
| 435 | * 'name' => string, |
||
| 436 | * 'notice' => string (HTML), |
||
| 437 | * 'explanation' => string (HTML) |
||
| 438 | * ], ... ] |
||
| 439 | */ |
||
| 440 | public function getErrors(): array |
||
| 441 | { |
||
| 442 | // Includes label and description |
||
| 443 | $wikidataErrors = $this->getWikidataErrors(); |
||
| 444 | |||
| 445 | $checkWikiErrors = $this->getCheckWikiErrors(); |
||
| 446 | |||
| 447 | return array_merge($wikidataErrors, $checkWikiErrors); |
||
| 448 | } |
||
| 449 | |||
| 450 | /** |
||
| 451 | * Get all wikidata items for the page, not just languages of sister projects |
||
| 452 | * @return string[] |
||
| 453 | */ |
||
| 454 | public function getWikidataItems(): array |
||
| 455 | { |
||
| 456 | if (!is_array($this->wikidataItems)) { |
||
| 457 | $this->wikidataItems = $this->getRepository()->getWikidataItems($this); |
||
| 458 | } |
||
| 459 | return $this->wikidataItems; |
||
| 460 | } |
||
| 461 | |||
| 462 | /** |
||
| 463 | * Count wikidata items for the page, not just languages of sister projects |
||
| 464 | * @return int Number of records. |
||
| 465 | */ |
||
| 466 | public function countWikidataItems(): int |
||
| 467 | { |
||
| 468 | if (is_array($this->wikidataItems)) { |
||
| 469 | $this->numWikidataItems = count($this->wikidataItems); |
||
| 470 | } elseif (null === $this->numWikidataItems) { |
||
| 471 | $this->numWikidataItems = (int)$this->getRepository()->countWikidataItems($this); |
||
| 472 | } |
||
| 473 | return $this->numWikidataItems; |
||
| 474 | } |
||
| 475 | |||
| 476 | /** |
||
| 477 | * Get number of in and outgoing links and redirects to this page. |
||
| 478 | * @return string[] Counts with keys 'links_ext_count', 'links_out_count', 'links_in_count' and 'redirects_count'. |
||
| 479 | */ |
||
| 480 | public function countLinksAndRedirects(): array |
||
| 481 | { |
||
| 482 | return $this->getRepository()->countLinksAndRedirects($this); |
||
| 483 | } |
||
| 484 | |||
| 485 | /** |
||
| 486 | * Get the sum of pageviews for the given page and timeframe. |
||
| 487 | * @param string|DateTime $start In the format YYYYMMDD |
||
| 488 | * @param string|DateTime $end In the format YYYYMMDD |
||
| 489 | * @return int |
||
| 490 | */ |
||
| 491 | public function getPageviews($start, $end): int |
||
| 492 | { |
||
| 493 | try { |
||
| 494 | $pageviews = $this->getRepository()->getPageviews($this, $start, $end); |
||
| 495 | } catch (\GuzzleHttp\Exception\ClientException $e) { |
||
| 496 | // 404 means zero pageviews |
||
| 497 | return 0; |
||
| 498 | } |
||
| 499 | |||
| 500 | return array_sum(array_map(function ($item) { |
||
| 501 | return (int)$item['views']; |
||
| 502 | }, $pageviews['items'])); |
||
| 503 | } |
||
| 504 | |||
| 505 | /** |
||
| 506 | * Get the sum of pageviews over the last N days |
||
| 507 | * @param int $days Default 30 |
||
| 508 | * @return int Number of pageviews |
||
| 509 | */ |
||
| 510 | public function getLastPageviews(int $days = 30): int |
||
| 515 | } |
||
| 516 | |||
| 517 | /** |
||
| 518 | * Is the page the project's Main Page? |
||
| 519 | * @return bool |
||
| 520 | */ |
||
| 521 | public function isMainPage(): bool |
||
| 522 | { |
||
| 523 | return $this->getProject()->getMainPage() === $this->getTitle(); |
||
| 524 | } |
||
| 525 | } |
||
| 526 |
Loading history...
