Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
89.68% |
113 / 126 |
|
70.00% |
7 / 10 |
CRAP | |
0.00% |
0 / 1 |
| RunVariableGenerator | |
89.68% |
113 / 126 |
|
70.00% |
7 / 10 |
29.92 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| getStashEditVars | |
85.71% |
6 / 7 |
|
0.00% |
0 / 1 |
2.01 | |||
| getEditTextForFiltering | |
92.31% |
12 / 13 |
|
0.00% |
0 / 1 |
6.02 | |||
| newVariableHolderForEdit | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
3 | |||
| getEditVars | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
3 | |||
| setLastEditAge | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
3 | |||
| getMoveVars | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
1 | |||
| getDeleteVars | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| getUploadVars | |
66.67% |
22 / 33 |
|
0.00% |
0 / 1 |
7.33 | |||
| getAccountCreationVars | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
3 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\AbuseFilter\VariableGenerator; |
| 4 | |
| 5 | use LogicException; |
| 6 | use MediaWiki\Content\Content; |
| 7 | use MediaWiki\Extension\AbuseFilter\Hooks\AbuseFilterHookRunner; |
| 8 | use MediaWiki\Extension\AbuseFilter\TextExtractor; |
| 9 | use MediaWiki\Extension\AbuseFilter\Variables\VariableHolder; |
| 10 | use MediaWiki\Page\WikiPageFactory; |
| 11 | use MediaWiki\Revision\MutableRevisionRecord; |
| 12 | use MediaWiki\Revision\RevisionRecord; |
| 13 | use MediaWiki\Revision\SlotRecord; |
| 14 | use MediaWiki\Title\Title; |
| 15 | use MediaWiki\User\User; |
| 16 | use MediaWiki\User\UserFactory; |
| 17 | use MWFileProps; |
| 18 | use UploadBase; |
| 19 | use Wikimedia\Assert\PreconditionException; |
| 20 | use Wikimedia\Mime\MimeAnalyzer; |
| 21 | use WikiPage; |
| 22 | |
| 23 | /** |
| 24 | * This class contains the logic used to create variable holders before filtering |
| 25 | * an action. |
| 26 | */ |
| 27 | class RunVariableGenerator extends VariableGenerator { |
| 28 | /** |
| 29 | * @var User |
| 30 | */ |
| 31 | private $user; |
| 32 | |
| 33 | /** |
| 34 | * @var Title |
| 35 | */ |
| 36 | private $title; |
| 37 | |
| 38 | /** @var TextExtractor */ |
| 39 | private $textExtractor; |
| 40 | /** @var MimeAnalyzer */ |
| 41 | private $mimeAnalyzer; |
| 42 | /** @var WikiPageFactory */ |
| 43 | private $wikiPageFactory; |
| 44 | |
| 45 | /** |
| 46 | * @param AbuseFilterHookRunner $hookRunner |
| 47 | * @param UserFactory $userFactory |
| 48 | * @param TextExtractor $textExtractor |
| 49 | * @param MimeAnalyzer $mimeAnalyzer |
| 50 | * @param WikiPageFactory $wikiPageFactory |
| 51 | * @param User $user |
| 52 | * @param Title $title |
| 53 | * @param VariableHolder|null $vars |
| 54 | */ |
| 55 | public function __construct( |
| 56 | AbuseFilterHookRunner $hookRunner, |
| 57 | UserFactory $userFactory, |
| 58 | TextExtractor $textExtractor, |
| 59 | MimeAnalyzer $mimeAnalyzer, |
| 60 | WikiPageFactory $wikiPageFactory, |
| 61 | User $user, |
| 62 | Title $title, |
| 63 | ?VariableHolder $vars = null |
| 64 | ) { |
| 65 | parent::__construct( $hookRunner, $userFactory, $vars ); |
| 66 | $this->textExtractor = $textExtractor; |
| 67 | $this->mimeAnalyzer = $mimeAnalyzer; |
| 68 | $this->wikiPageFactory = $wikiPageFactory; |
| 69 | $this->user = $user; |
| 70 | $this->title = $title; |
| 71 | } |
| 72 | |
| 73 | /** |
| 74 | * Get variables for pre-filtering an edit during stash |
| 75 | * |
| 76 | * @param Content $content |
| 77 | * @param string $summary |
| 78 | * @param string $slot |
| 79 | * @param WikiPage $page |
| 80 | * @return VariableHolder|null |
| 81 | */ |
| 82 | public function getStashEditVars( |
| 83 | Content $content, |
| 84 | string $summary, |
| 85 | $slot, |
| 86 | WikiPage $page |
| 87 | ): ?VariableHolder { |
| 88 | $filterText = $this->getEditTextForFiltering( $page, $content, $slot ); |
| 89 | if ( $filterText === null ) { |
| 90 | return null; |
| 91 | } |
| 92 | [ $oldContent, $oldAfText, $text ] = $filterText; |
| 93 | return $this->newVariableHolderForEdit( |
| 94 | $page, $summary, $content, $text, $oldAfText, $oldContent |
| 95 | ); |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * Get the text of an edit to be used for filtering |
| 100 | * @todo Full support for multi-slots |
| 101 | * |
| 102 | * @param WikiPage $page |
| 103 | * @param Content $content |
| 104 | * @param string $slot |
| 105 | * @return array|null |
| 106 | */ |
| 107 | private function getEditTextForFiltering( WikiPage $page, Content $content, $slot ): ?array { |
| 108 | $oldRevRecord = $page->getRevisionRecord(); |
| 109 | if ( !$oldRevRecord ) { |
| 110 | return null; |
| 111 | } |
| 112 | |
| 113 | $oldContent = $oldRevRecord->getContent( SlotRecord::MAIN, RevisionRecord::RAW ); |
| 114 | if ( !$oldContent ) { |
| 115 | // @codeCoverageIgnoreStart |
| 116 | throw new LogicException( 'Content cannot be null' ); |
| 117 | // @codeCoverageIgnoreEnd |
| 118 | } |
| 119 | $oldAfText = $this->textExtractor->revisionToString( $oldRevRecord, $this->user ); |
| 120 | |
| 121 | // XXX: Recreate what the new revision will probably be so we can get the full AF |
| 122 | // text for all slots |
| 123 | $newRevision = MutableRevisionRecord::newFromParentRevision( $oldRevRecord ); |
| 124 | $newRevision->setContent( $slot, $content ); |
| 125 | $text = $this->textExtractor->revisionToString( $newRevision, $this->user ); |
| 126 | |
| 127 | // Don't trigger for null edits. Compare Content objects if available, but check the |
| 128 | // stringified contents as well, e.g. for line endings normalization (T240115). |
| 129 | // Don't treat content model change as null edit though. |
| 130 | if ( |
| 131 | $content->equals( $oldContent ) || |
| 132 | ( $oldContent->getModel() === $content->getModel() && strcmp( $oldAfText, $text ) === 0 ) |
| 133 | ) { |
| 134 | return null; |
| 135 | } |
| 136 | |
| 137 | return [ $oldContent, $oldAfText, $text ]; |
| 138 | } |
| 139 | |
| 140 | /** |
| 141 | * @param WikiPage $page |
| 142 | * @param string $summary |
| 143 | * @param Content $newcontent |
| 144 | * @param string $text |
| 145 | * @param string $oldtext |
| 146 | * @param Content|null $oldcontent |
| 147 | * @return VariableHolder |
| 148 | */ |
| 149 | private function newVariableHolderForEdit( |
| 150 | WikiPage $page, |
| 151 | string $summary, |
| 152 | Content $newcontent, |
| 153 | string $text, |
| 154 | string $oldtext, |
| 155 | ?Content $oldcontent = null |
| 156 | ): VariableHolder { |
| 157 | $this->addUserVars( $this->user ) |
| 158 | ->addTitleVars( $this->title, 'page' ); |
| 159 | $this->vars->setVar( 'action', 'edit' ); |
| 160 | $this->vars->setVar( 'summary', $summary ); |
| 161 | $this->setLastEditAge( $page->getRevisionRecord(), 'page' ); |
| 162 | |
| 163 | if ( $oldcontent instanceof Content ) { |
| 164 | $oldmodel = $oldcontent->getModel(); |
| 165 | } else { |
| 166 | $oldmodel = ''; |
| 167 | $oldtext = ''; |
| 168 | } |
| 169 | $this->vars->setVar( 'old_content_model', $oldmodel ); |
| 170 | $this->vars->setVar( 'new_content_model', $newcontent->getModel() ); |
| 171 | $this->vars->setVar( 'old_wikitext', $oldtext ); |
| 172 | $this->vars->setVar( 'new_wikitext', $text ); |
| 173 | |
| 174 | try { |
| 175 | $update = $page->getCurrentUpdate(); |
| 176 | $update->getParserOutputForMetaData(); |
| 177 | } catch ( PreconditionException | LogicException $exception ) { |
| 178 | // Temporary workaround until this becomes |
| 179 | // a hook parameter |
| 180 | $update = null; |
| 181 | } |
| 182 | $this->addEditVars( $page, $this->user, true, $update ); |
| 183 | |
| 184 | return $this->vars; |
| 185 | } |
| 186 | |
| 187 | /** |
| 188 | * Get variables for filtering an edit. |
| 189 | * |
| 190 | * @param Content $content |
| 191 | * @param string $summary |
| 192 | * @param string $slot |
| 193 | * @param WikiPage $page |
| 194 | * @return VariableHolder|null |
| 195 | */ |
| 196 | public function getEditVars( |
| 197 | Content $content, |
| 198 | string $summary, |
| 199 | $slot, |
| 200 | WikiPage $page |
| 201 | ): ?VariableHolder { |
| 202 | if ( $this->title->exists() ) { |
| 203 | $filterText = $this->getEditTextForFiltering( $page, $content, $slot ); |
| 204 | if ( $filterText === null ) { |
| 205 | return null; |
| 206 | } |
| 207 | [ $oldContent, $oldAfText, $text ] = $filterText; |
| 208 | } else { |
| 209 | // Optimization |
| 210 | $oldContent = null; |
| 211 | $oldAfText = ''; |
| 212 | $text = $this->textExtractor->contentToString( $content ); |
| 213 | } |
| 214 | |
| 215 | return $this->newVariableHolderForEdit( |
| 216 | $page, $summary, $content, $text, $oldAfText, $oldContent |
| 217 | ); |
| 218 | } |
| 219 | |
| 220 | /** |
| 221 | * @param RevisionRecord|Title|null $from |
| 222 | * @param string $prefix |
| 223 | */ |
| 224 | private function setLastEditAge( $from, string $prefix ): void { |
| 225 | $varName = "{$prefix}_last_edit_age"; |
| 226 | if ( $from instanceof RevisionRecord ) { |
| 227 | $this->vars->setVar( |
| 228 | $varName, |
| 229 | (int)wfTimestamp( TS_UNIX ) - (int)wfTimestamp( TS_UNIX, $from->getTimestamp() ) |
| 230 | ); |
| 231 | } elseif ( $from instanceof Title ) { |
| 232 | $this->vars->setLazyLoadVar( |
| 233 | $varName, |
| 234 | 'revision-age', |
| 235 | [ 'title' => $from, 'asof' => wfTimestampNow() ] |
| 236 | ); |
| 237 | } else { |
| 238 | $this->vars->setVar( $varName, null ); |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | /** |
| 243 | * Get variables used to filter a move. |
| 244 | * |
| 245 | * @param Title $newTitle |
| 246 | * @param string $reason |
| 247 | * @return VariableHolder |
| 248 | */ |
| 249 | public function getMoveVars( |
| 250 | Title $newTitle, |
| 251 | string $reason |
| 252 | ): VariableHolder { |
| 253 | $this->addUserVars( $this->user ) |
| 254 | ->addTitleVars( $this->title, 'moved_from' ) |
| 255 | ->addTitleVars( $newTitle, 'moved_to' ); |
| 256 | |
| 257 | $this->vars->setVar( 'summary', $reason ); |
| 258 | $this->vars->setVar( 'action', 'move' ); |
| 259 | $this->setLastEditAge( $this->title, 'moved_from' ); |
| 260 | $this->setLastEditAge( $newTitle, 'moved_to' ); |
| 261 | // TODO: add old_wikitext etc. (T320347) |
| 262 | return $this->vars; |
| 263 | } |
| 264 | |
| 265 | /** |
| 266 | * Get variables for filtering a deletion. |
| 267 | * |
| 268 | * @param string $reason |
| 269 | * @return VariableHolder |
| 270 | */ |
| 271 | public function getDeleteVars( |
| 272 | string $reason |
| 273 | ): VariableHolder { |
| 274 | $this->addUserVars( $this->user ) |
| 275 | ->addTitleVars( $this->title, 'page' ); |
| 276 | |
| 277 | $this->vars->setVar( 'summary', $reason ); |
| 278 | $this->vars->setVar( 'action', 'delete' ); |
| 279 | // FIXME: this is an unnecessary round-trip, we could obtain WikiPage from |
| 280 | // the hook and call WikiPage::getRevisionRecord, but then ProofreadPage tests fail |
| 281 | $this->setLastEditAge( $this->title, 'page' ); |
| 282 | // TODO: add old_wikitext etc. (T173663) |
| 283 | return $this->vars; |
| 284 | } |
| 285 | |
| 286 | /** |
| 287 | * Get variables for filtering an upload. |
| 288 | * |
| 289 | * @param string $action |
| 290 | * @param UploadBase $upload |
| 291 | * @param string|null $summary |
| 292 | * @param string|null $text |
| 293 | * @param array|null $props |
| 294 | * @return VariableHolder|null |
| 295 | */ |
| 296 | public function getUploadVars( |
| 297 | string $action, |
| 298 | UploadBase $upload, |
| 299 | ?string $summary, |
| 300 | ?string $text, |
| 301 | ?array $props |
| 302 | ): ?VariableHolder { |
| 303 | if ( !$props ) { |
| 304 | $props = ( new MWFileProps( $this->mimeAnalyzer ) )->getPropsFromPath( |
| 305 | $upload->getTempPath(), |
| 306 | true |
| 307 | ); |
| 308 | } |
| 309 | |
| 310 | $this->addUserVars( $this->user ) |
| 311 | ->addTitleVars( $this->title, 'page' ); |
| 312 | $this->vars->setVar( 'action', $action ); |
| 313 | |
| 314 | // We use the hexadecimal version of the file sha1. |
| 315 | // Use UploadBase::getTempFileSha1Base36 so that we don't have to calculate the sha1 sum again |
| 316 | $sha1 = \Wikimedia\base_convert( $upload->getTempFileSha1Base36(), 36, 16, 40 ); |
| 317 | |
| 318 | // This is the same as AbuseFilterRowVariableGenerator::addUploadVars, but from a different source |
| 319 | $this->vars->setVar( 'file_sha1', $sha1 ); |
| 320 | $this->vars->setVar( 'file_size', $upload->getFileSize() ); |
| 321 | |
| 322 | $this->vars->setVar( 'file_mime', $props['mime'] ); |
| 323 | $this->vars->setVar( 'file_mediatype', $this->mimeAnalyzer->getMediaType( null, $props['mime'] ) ); |
| 324 | $this->vars->setVar( 'file_width', $props['width'] ); |
| 325 | $this->vars->setVar( 'file_height', $props['height'] ); |
| 326 | $this->vars->setVar( 'file_bits_per_channel', $props['bits'] ); |
| 327 | |
| 328 | // We only have the upload comment and page text when using the UploadVerifyUpload hook |
| 329 | if ( $summary !== null && $text !== null ) { |
| 330 | // This block is adapted from self::getEditTextForFiltering() |
| 331 | $page = $this->wikiPageFactory->newFromTitle( $this->title ); |
| 332 | if ( $this->title->exists() ) { |
| 333 | $revRec = $page->getRevisionRecord(); |
| 334 | if ( !$revRec ) { |
| 335 | return null; |
| 336 | } |
| 337 | |
| 338 | $this->setLastEditAge( $revRec, 'page' ); |
| 339 | $oldcontent = $revRec->getContent( SlotRecord::MAIN, RevisionRecord::RAW ); |
| 340 | '@phan-var Content $oldcontent'; |
| 341 | $oldtext = $this->textExtractor->contentToString( $oldcontent ); |
| 342 | |
| 343 | // Page text is ignored for uploads when the page already exists |
| 344 | $text = $oldtext; |
| 345 | } else { |
| 346 | $oldtext = ''; |
| 347 | $this->setLastEditAge( null, 'page' ); |
| 348 | } |
| 349 | |
| 350 | // Load vars for filters to check |
| 351 | $this->vars->setVar( 'summary', $summary ); |
| 352 | $this->vars->setVar( 'old_wikitext', $oldtext ); |
| 353 | $this->vars->setVar( 'new_wikitext', $text ); |
| 354 | // TODO: set old_content_model and new_content_model vars, use them |
| 355 | $this->addEditVars( $page, $this->user, true ); |
| 356 | } |
| 357 | return $this->vars; |
| 358 | } |
| 359 | |
| 360 | /** |
| 361 | * Get variables for filtering an account creation |
| 362 | * |
| 363 | * @param User $createdUser This is the user being created, not the creator (which is $this->user) |
| 364 | * @param bool $autocreate |
| 365 | * @return VariableHolder |
| 366 | */ |
| 367 | public function getAccountCreationVars( |
| 368 | User $createdUser, |
| 369 | bool $autocreate |
| 370 | ): VariableHolder { |
| 371 | // generateUserVars records $this->user->getName() which would be the IP for unregistered users |
| 372 | if ( $this->user->isRegistered() ) { |
| 373 | $this->addUserVars( $this->user ); |
| 374 | } else { |
| 375 | // Set the user_type for IP users, so that filters can distinguish between account |
| 376 | // creations from temporary accounts and those from IP addresses. |
| 377 | $this->vars->setLazyLoadVar( |
| 378 | 'user_type', |
| 379 | 'user-type', |
| 380 | [ 'user-identity' => $this->user ] |
| 381 | ); |
| 382 | } |
| 383 | |
| 384 | $this->vars->setVar( 'action', $autocreate ? 'autocreateaccount' : 'createaccount' ); |
| 385 | $this->vars->setVar( 'accountname', $createdUser->getName() ); |
| 386 | return $this->vars; |
| 387 | } |
| 388 | } |