x-tools /
xtools
| Total Complexity | 64 |
| Total Lines | 504 |
| Duplicated Lines | 0 % |
| Changes | 1 | ||
| Bugs | 0 | Features | 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 |
||
| 15 | class Page extends Model |
||
| 16 | { |
||
| 17 | /** @var string The page name as provided at instantiation. */ |
||
| 18 | protected string $unnormalizedPageName; |
||
| 19 | |||
| 20 | /** @var string[]|null Metadata about this page. */ |
||
| 21 | protected ?array $pageInfo; |
||
| 22 | |||
| 23 | /** @var string[] Revision history of this page. */ |
||
| 24 | protected array $revisions; |
||
| 25 | |||
| 26 | /** @var int Number of revisions for this page. */ |
||
| 27 | protected int $numRevisions; |
||
| 28 | |||
| 29 | /** @var string[] List of Wikidata sitelinks for this page. */ |
||
| 30 | protected array $wikidataItems; |
||
| 31 | |||
| 32 | /** @var int Number of Wikidata sitelinks for this page. */ |
||
| 33 | protected int $numWikidataItems; |
||
| 34 | |||
| 35 | /** @var int Length of the page in bytes. */ |
||
| 36 | protected int $length; |
||
| 37 | |||
| 38 | /** |
||
| 39 | * Page constructor. |
||
| 40 | * @param PageRepository $repository |
||
| 41 | * @param Project $project |
||
| 42 | * @param string $pageName |
||
| 43 | */ |
||
| 44 | public function __construct(PageRepository $repository, Project $project, string $pageName) |
||
| 45 | { |
||
| 46 | $this->repository = $repository; |
||
| 47 | $this->project = $project; |
||
| 48 | $this->unnormalizedPageName = $pageName; |
||
| 49 | } |
||
| 50 | |||
| 51 | /** |
||
| 52 | * Get a Page instance given a database row (either from or JOINed on the page table). |
||
| 53 | * @param PageRepository $repository |
||
| 54 | * @param Project $project |
||
| 55 | * @param array $row Must contain 'page_title' and 'page_namespace'. May contain 'page_len'. |
||
| 56 | * @return static |
||
| 57 | */ |
||
| 58 | public static function newFromRow(PageRepository $repository, Project $project, array $row): self |
||
| 59 | { |
||
| 60 | $pageTitle = $row['page_title']; |
||
| 61 | |||
| 62 | if (0 === (int)$row['page_namespace']) { |
||
| 63 | $fullPageTitle = $pageTitle; |
||
| 64 | } else { |
||
| 65 | $namespaces = $project->getNamespaces(); |
||
| 66 | $fullPageTitle = $namespaces[$row['page_namespace']].":$pageTitle"; |
||
| 67 | } |
||
| 68 | |||
| 69 | $page = new self($repository, $project, $fullPageTitle); |
||
| 70 | $page->pageInfo['ns'] = $row['page_namespace']; |
||
| 71 | if (isset($row['page_len'])) { |
||
| 72 | $page->length = (int)$row['page_len']; |
||
| 73 | } |
||
| 74 | |||
| 75 | return $page; |
||
| 76 | } |
||
| 77 | |||
| 78 | /** |
||
| 79 | * Unique identifier for this Page, to be used in cache keys. |
||
| 80 | * Use of md5 ensures the cache key does not contain reserved characters. |
||
| 81 | * @see Repository::getCacheKey() |
||
| 82 | * @return string |
||
| 83 | * @codeCoverageIgnore |
||
| 84 | */ |
||
| 85 | public function getCacheKey(): string |
||
| 86 | { |
||
| 87 | return md5((string)$this->getId()); |
||
| 88 | } |
||
| 89 | |||
| 90 | /** |
||
| 91 | * Get basic information about this page from the repository. |
||
| 92 | * @return array|null |
||
| 93 | */ |
||
| 94 | protected function getPageInfo(): ?array |
||
| 95 | { |
||
| 96 | if (!isset($this->pageInfo)) { |
||
| 97 | $this->pageInfo = $this->repository->getPageInfo($this->project, $this->unnormalizedPageName); |
||
|
Bug
introduced
by
|
|||
| 98 | } |
||
| 99 | return $this->pageInfo; |
||
| 100 | } |
||
| 101 | |||
| 102 | /** |
||
| 103 | * Get the page's title. |
||
| 104 | * @param bool $useUnnormalized Use the unnormalized page title to avoid an API call. This should be used only if |
||
| 105 | * you fetched the page title via other means (SQL query), and is not from user input alone. |
||
| 106 | * @return string |
||
| 107 | */ |
||
| 108 | public function getTitle(bool $useUnnormalized = false): string |
||
| 109 | { |
||
| 110 | if ($useUnnormalized) { |
||
| 111 | return $this->unnormalizedPageName; |
||
| 112 | } |
||
| 113 | $info = $this->getPageInfo(); |
||
| 114 | return $info['title'] ?? $this->unnormalizedPageName; |
||
| 115 | } |
||
| 116 | |||
| 117 | /** |
||
| 118 | * Get the page's title without the namespace. |
||
| 119 | * @return string |
||
| 120 | */ |
||
| 121 | public function getTitleWithoutNamespace(): string |
||
| 122 | { |
||
| 123 | $info = $this->getPageInfo(); |
||
| 124 | $title = $info['title'] ?? $this->unnormalizedPageName; |
||
| 125 | $nsName = $this->getNamespaceName(); |
||
| 126 | return $nsName |
||
| 127 | ? str_replace($nsName . ':', '', $title) |
||
| 128 | : $title; |
||
| 129 | } |
||
| 130 | |||
| 131 | /** |
||
| 132 | * Get this page's database ID. |
||
| 133 | * @return int|null Null if nonexistent. |
||
| 134 | */ |
||
| 135 | public function getId(): ?int |
||
| 136 | { |
||
| 137 | $info = $this->getPageInfo(); |
||
| 138 | return isset($info['pageid']) ? (int)$info['pageid'] : null; |
||
| 139 | } |
||
| 140 | |||
| 141 | /** |
||
| 142 | * Get this page's length in bytes. |
||
| 143 | * @return int|null Null if nonexistent. |
||
| 144 | */ |
||
| 145 | public function getLength(): ?int |
||
| 146 | { |
||
| 147 | if (isset($this->length)) { |
||
| 148 | return $this->length; |
||
| 149 | } |
||
| 150 | $info = $this->getPageInfo(); |
||
| 151 | $this->length = isset($info['length']) ? (int)$info['length'] : null; |
||
| 152 | return $this->length; |
||
| 153 | } |
||
| 154 | |||
| 155 | /** |
||
| 156 | * Get HTML for the stylized display of the title. |
||
| 157 | * The text will be the same as Page::getTitle(). |
||
| 158 | * @return string |
||
| 159 | */ |
||
| 160 | public function getDisplayTitle(): string |
||
| 161 | { |
||
| 162 | $info = $this->getPageInfo(); |
||
| 163 | if (isset($info['displaytitle'])) { |
||
| 164 | return $info['displaytitle']; |
||
| 165 | } |
||
| 166 | return $this->getTitle(); |
||
| 167 | } |
||
| 168 | |||
| 169 | /** |
||
| 170 | * Get the full URL of this page. |
||
| 171 | * @return string|null Null if nonexistent. |
||
| 172 | */ |
||
| 173 | public function getUrl(): ?string |
||
| 177 | } |
||
| 178 | |||
| 179 | /** |
||
| 180 | * Get the numerical ID of the namespace of this page. |
||
| 181 | * @return int|null Null if page doesn't exist. |
||
| 182 | */ |
||
| 183 | public function getNamespace(): ?int |
||
| 184 | { |
||
| 185 | if (isset($this->pageInfo['ns']) && is_numeric($this->pageInfo['ns'])) { |
||
| 186 | return (int)$this->pageInfo['ns']; |
||
| 187 | } |
||
| 188 | $info = $this->getPageInfo(); |
||
| 189 | return isset($info['ns']) ? (int)$info['ns'] : null; |
||
| 190 | } |
||
| 191 | |||
| 192 | /** |
||
| 193 | * Get the name of the namespace of this page. |
||
| 194 | * @return string|null Null if could not be determined. |
||
| 195 | */ |
||
| 196 | public function getNamespaceName(): ?string |
||
| 197 | { |
||
| 198 | $info = $this->getPageInfo(); |
||
| 199 | return isset($info['ns']) |
||
| 200 | ? ($this->getProject()->getNamespaces()[$info['ns']] ?? null) |
||
| 201 | : null; |
||
| 202 | } |
||
| 203 | |||
| 204 | /** |
||
| 205 | * Get the number of page watchers. |
||
| 206 | * @return int|null Null if unknown. |
||
| 207 | */ |
||
| 208 | public function getWatchers(): ?int |
||
| 209 | { |
||
| 210 | $info = $this->getPageInfo(); |
||
| 211 | return isset($info['watchers']) ? (int)$info['watchers'] : null; |
||
| 212 | } |
||
| 213 | |||
| 214 | /** |
||
| 215 | * Get the HTML content of the body of the page. |
||
| 216 | * @param DateTime|int $target If a DateTime object, the |
||
| 217 | * revision at that time will be returned. If an integer, it is |
||
| 218 | * assumed to be the actual revision ID. |
||
| 219 | * @return string |
||
| 220 | */ |
||
| 221 | public function getHTMLContent($target = null): string |
||
| 222 | { |
||
| 223 | if (is_a($target, 'DateTime')) { |
||
| 224 | $target = $this->repository->getRevisionIdAtDate($this, $target); |
||
| 225 | } |
||
| 226 | return $this->repository->getHTMLContent($this, $target); |
||
| 227 | } |
||
| 228 | |||
| 229 | /** |
||
| 230 | * Whether or not this page exists. |
||
| 231 | * @return bool |
||
| 232 | */ |
||
| 233 | public function exists(): bool |
||
| 234 | { |
||
| 235 | $info = $this->getPageInfo(); |
||
| 236 | return null !== $info && !isset($info['missing']) && !isset($info['invalid']) && !isset($info['interwiki']); |
||
| 237 | } |
||
| 238 | |||
| 239 | /** |
||
| 240 | * Get the Project to which this page belongs. |
||
| 241 | * @return Project |
||
| 242 | */ |
||
| 243 | public function getProject(): Project |
||
| 244 | { |
||
| 245 | return $this->project; |
||
| 246 | } |
||
| 247 | |||
| 248 | /** |
||
| 249 | * Get the language code for this page. |
||
| 250 | * If not set, the language code for the project is returned. |
||
| 251 | * @return string |
||
| 252 | */ |
||
| 253 | public function getLang(): string |
||
| 254 | { |
||
| 255 | $info = $this->getPageInfo(); |
||
| 256 | return $info['pagelanguage'] ?? $this->getProject()->getLang(); |
||
| 257 | } |
||
| 258 | |||
| 259 | /** |
||
| 260 | * Get the Wikidata ID of this page. |
||
| 261 | * @return string|null Null if none exists. |
||
| 262 | */ |
||
| 263 | public function getWikidataId(): ?string |
||
| 264 | { |
||
| 265 | $info = $this->getPageInfo(); |
||
| 266 | return $info['pageprops']['wikibase_item'] ?? null; |
||
| 267 | } |
||
| 268 | |||
| 269 | /** |
||
| 270 | * Get the number of revisions the page has. |
||
| 271 | * @param ?User $user Optionally limit to those of this user. |
||
| 272 | * @param false|int $start |
||
| 273 | * @param false|int $end |
||
| 274 | * @return int |
||
| 275 | */ |
||
| 276 | public function getNumRevisions(?User $user = null, $start = false, $end = false): int |
||
| 277 | { |
||
| 278 | // If a user is given, we will not cache the result via instance variable. |
||
| 279 | if (null !== $user) { |
||
| 280 | return $this->repository->getNumRevisions($this, $user, $start, $end); |
||
| 281 | } |
||
| 282 | |||
| 283 | // Return cached value, if present. |
||
| 284 | if (isset($this->numRevisions)) { |
||
| 285 | return $this->numRevisions; |
||
| 286 | } |
||
| 287 | |||
| 288 | // Otherwise, return the count of all revisions if already present. |
||
| 289 | if (isset($this->revisions)) { |
||
| 290 | $this->numRevisions = count($this->revisions); |
||
| 291 | } else { |
||
| 292 | // Otherwise do a COUNT in the event fetching all revisions is not desired. |
||
| 293 | $this->numRevisions = $this->repository->getNumRevisions($this, null, $start, $end); |
||
| 294 | } |
||
| 295 | |||
| 296 | return $this->numRevisions; |
||
| 297 | } |
||
| 298 | |||
| 299 | /** |
||
| 300 | * Get all edits made to this page. |
||
| 301 | * @param User|null $user Specify to get only revisions by the given user. |
||
| 302 | * @param false|int $start |
||
| 303 | * @param false|int $end |
||
| 304 | * @return array |
||
| 305 | */ |
||
| 306 | public function getRevisions(?User $user = null, $start = false, $end = false): array |
||
| 307 | { |
||
| 308 | if (isset($this->revisions)) { |
||
| 309 | return $this->revisions; |
||
| 310 | } |
||
| 311 | |||
| 312 | $this->revisions = $this->repository->getRevisions($this, $user, $start, $end); |
||
| 313 | |||
| 314 | return $this->revisions; |
||
| 315 | } |
||
| 316 | |||
| 317 | /** |
||
| 318 | * Get the full page wikitext. |
||
| 319 | * @return string|null Null if nothing was found. |
||
| 320 | */ |
||
| 321 | public function getWikitext(): ?string |
||
| 322 | { |
||
| 323 | $content = $this->repository->getPagesWikitext( |
||
| 324 | $this->getProject(), |
||
| 325 | [ $this->getTitle() ] |
||
| 326 | ); |
||
| 327 | |||
| 328 | return $content[$this->getTitle()] ?? null; |
||
| 329 | } |
||
| 330 | |||
| 331 | /** |
||
| 332 | * Get the statement for a single revision, so that you can iterate row by row. |
||
| 333 | * @see PageRepository::getRevisionsStmt() |
||
| 334 | * @param User|null $user Specify to get only revisions by the given user. |
||
| 335 | * @param ?int $limit Max number of revisions to process. |
||
| 336 | * @param ?int $numRevisions Number of revisions, if known. This is used solely to determine the |
||
| 337 | * OFFSET if we are given a $limit. If $limit is set and $numRevisions is not set, a |
||
| 338 | * separate query is ran to get the nuber of revisions. |
||
| 339 | * @param false|int $start |
||
| 340 | * @param false|int $end |
||
| 341 | * @return ResultStatement |
||
| 342 | */ |
||
| 343 | public function getRevisionsStmt( |
||
| 344 | ?User $user = null, |
||
| 345 | ?int $limit = null, |
||
| 346 | ?int $numRevisions = null, |
||
| 347 | $start = false, |
||
| 348 | $end = false |
||
| 349 | ): ResultStatement { |
||
| 350 | // If we have a limit, we need to know the total number of revisions so that PageRepo |
||
| 351 | // will properly set the OFFSET. See PageRepository::getRevisionsStmt() for more info. |
||
| 352 | if (isset($limit) && null === $numRevisions) { |
||
| 353 | $numRevisions = $this->getNumRevisions($user, $start, $end); |
||
| 354 | } |
||
| 355 | return $this->repository->getRevisionsStmt($this, $user, $limit, $numRevisions, $start, $end); |
||
| 356 | } |
||
| 357 | |||
| 358 | /** |
||
| 359 | * Get the revision ID that immediately precedes the given date. |
||
| 360 | * @param DateTime $date |
||
| 361 | * @return int|null Null if none found. |
||
| 362 | */ |
||
| 363 | public function getRevisionIdAtDate(DateTime $date): ?int |
||
| 364 | { |
||
| 365 | return $this->repository->getRevisionIdAtDate($this, $date); |
||
| 366 | } |
||
| 367 | |||
| 368 | /** |
||
| 369 | * Get CheckWiki errors for this page |
||
| 370 | * @return string[] See getErrors() for format |
||
| 371 | */ |
||
| 372 | public function getCheckWikiErrors(): array |
||
| 373 | { |
||
| 374 | return $this->repository->getCheckWikiErrors($this); |
||
| 375 | } |
||
| 376 | |||
| 377 | /** |
||
| 378 | * Get Wikidata errors for this page |
||
| 379 | * @return string[][] See getErrors() for format |
||
| 380 | */ |
||
| 381 | public function getWikidataErrors(): array |
||
| 418 | } |
||
| 419 | |||
| 420 | /** |
||
| 421 | * Get Wikidata and CheckWiki errors, if present |
||
| 422 | * @return string[][] List of errors in the format: |
||
| 423 | * [[ |
||
| 424 | * 'prio' => int, |
||
| 425 | * 'name' => string, |
||
| 426 | * 'notice' => string (HTML), |
||
| 427 | * 'explanation' => string (HTML) |
||
| 428 | * ], ... ] |
||
| 429 | */ |
||
| 430 | public function getErrors(): array |
||
| 431 | { |
||
| 432 | // Includes label and description |
||
| 433 | $wikidataErrors = $this->getWikidataErrors(); |
||
| 434 | |||
| 435 | $checkWikiErrors = $this->getCheckWikiErrors(); |
||
| 436 | |||
| 437 | return array_merge($wikidataErrors, $checkWikiErrors); |
||
| 438 | } |
||
| 439 | |||
| 440 | /** |
||
| 441 | * Get all wikidata items for the page, not just languages of sister projects |
||
| 442 | * @return string[] |
||
| 443 | */ |
||
| 444 | public function getWikidataItems(): array |
||
| 445 | { |
||
| 446 | if (!isset($this->wikidataItems)) { |
||
| 447 | $this->wikidataItems = $this->repository->getWikidataItems($this); |
||
| 448 | } |
||
| 449 | return $this->wikidataItems; |
||
| 450 | } |
||
| 451 | |||
| 452 | /** |
||
| 453 | * Count wikidata items for the page, not just languages of sister projects |
||
| 454 | * @return int Number of records. |
||
| 455 | */ |
||
| 456 | public function countWikidataItems(): int |
||
| 464 | } |
||
| 465 | |||
| 466 | /** |
||
| 467 | * Get number of in and outgoing links and redirects to this page. |
||
| 468 | * @return string[] Counts with keys 'links_ext_count', 'links_out_count', 'links_in_count' and 'redirects_count'. |
||
| 469 | */ |
||
| 470 | public function countLinksAndRedirects(): array |
||
| 473 | } |
||
| 474 | |||
| 475 | /** |
||
| 476 | * Get the sum of pageviews for the given page and timeframe. |
||
| 477 | * @param string|DateTime $start In the format YYYYMMDD |
||
| 478 | * @param string|DateTime $end In the format YYYYMMDD |
||
| 479 | * @return int|null Total pageviews or null if data is unavailable. |
||
| 480 | */ |
||
| 481 | public function getPageviews($start, $end): ?int |
||
| 482 | { |
||
| 483 | try { |
||
| 484 | $pageviews = $this->repository->getPageviews($this, $start, $end); |
||
| 485 | } catch (ClientException $e) { |
||
| 486 | // 404 means zero pageviews |
||
| 487 | return 0; |
||
| 488 | } |
||
| 489 | |||
| 490 | if (null === $pageviews) { |
||
| 491 | return null; |
||
| 497 | } |
||
| 498 | |||
| 499 | /** |
||
| 500 | * Get the sum of pageviews over the last N days |
||
| 501 | * @param int $days Default ArticleInfoApi::PAGEVIEWS_OFFSET |
||
| 502 | * @see ArticleInfoApi::PAGEVIEWS_OFFSET |
||
| 503 | * @return int|null Number of pageviews or null if data is unavailable. |
||
| 504 | */ |
||
| 505 | public function getLatestPageviews(int $days = ArticleInfoApi::PAGEVIEWS_OFFSET): ?int |
||
| 510 | } |
||
| 511 | |||
| 512 | /** |
||
| 513 | * Is the page the project's Main Page? |
||
| 514 | * @return bool |
||
| 515 | */ |
||
| 516 | public function isMainPage(): bool |
||
| 521 |
Loading history...
