Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 106 |
|
0.00% |
0 / 10 |
CRAP | |
0.00% |
0 / 1 |
| ReferenceRecorder | |
0.00% |
0 / 106 |
|
0.00% |
0 / 10 |
1406 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| onAfterLoad | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| onAfterInsert | |
0.00% |
0 / 14 |
|
0.00% |
0 / 1 |
30 | |||
| calculateChangesFromExisting | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| calculateChangesFromTopic | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
30 | |||
| isHidden | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| collectTopicRevisions | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
12 | |||
| getReferencesFromRevisionContent | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
132 | |||
| getExistingReferences | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
2 | |||
| referencesDifference | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
56 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Flow\Data\Listener; |
| 4 | |
| 5 | use Flow\Data\ManagerGroup; |
| 6 | use Flow\Exception\FlowException; |
| 7 | use Flow\Exception\InvalidDataException; |
| 8 | use Flow\LinksTableUpdater; |
| 9 | use Flow\Model\AbstractRevision; |
| 10 | use Flow\Model\PostRevision; |
| 11 | use Flow\Model\PostSummary; |
| 12 | use Flow\Model\Reference; |
| 13 | use Flow\Model\UUID; |
| 14 | use Flow\Model\Workflow; |
| 15 | use Flow\Parsoid\ReferenceExtractor; |
| 16 | use Flow\Repository\TreeRepository; |
| 17 | use MediaWiki\WikiMap\WikiMap; |
| 18 | use SplQueue; |
| 19 | |
| 20 | /** |
| 21 | * Listens for new revisions to be inserted. Calculates the difference in |
| 22 | * references(URLs, images, etc) between this new version and the previous |
| 23 | * revision. Uses calculated difference to update links tables to match the new revision. |
| 24 | */ |
| 25 | class ReferenceRecorder extends AbstractListener { |
| 26 | /** |
| 27 | * @var ReferenceExtractor |
| 28 | */ |
| 29 | protected $referenceExtractor; |
| 30 | |
| 31 | /** |
| 32 | * @var ManagerGroup |
| 33 | */ |
| 34 | protected $storage; |
| 35 | |
| 36 | /** |
| 37 | * @var LinksTableUpdater |
| 38 | */ |
| 39 | protected $linksTableUpdater; |
| 40 | |
| 41 | /** |
| 42 | * @var TreeRepository Used to query for the posts within a topic when moderation |
| 43 | * changes the visibility of a topic. |
| 44 | */ |
| 45 | protected $treeRepository; |
| 46 | |
| 47 | /** |
| 48 | * @var SplQueue |
| 49 | */ |
| 50 | protected $deferredQueue; |
| 51 | |
| 52 | public function __construct( |
| 53 | ReferenceExtractor $referenceExtractor, |
| 54 | LinksTableUpdater $linksTableUpdater, |
| 55 | ManagerGroup $storage, |
| 56 | TreeRepository $treeRepository, |
| 57 | SplQueue $deferredQueue |
| 58 | ) { |
| 59 | $this->referenceExtractor = $referenceExtractor; |
| 60 | $this->linksTableUpdater = $linksTableUpdater; |
| 61 | $this->storage = $storage; |
| 62 | $this->treeRepository = $treeRepository; |
| 63 | $this->deferredQueue = $deferredQueue; |
| 64 | } |
| 65 | |
| 66 | public function onAfterLoad( $object, array $old ) { |
| 67 | // Nuthin |
| 68 | } |
| 69 | |
| 70 | public function onAfterInsert( $revision, array $new, array $metadata ) { |
| 71 | if ( !isset( $metadata['workflow'] ) ) { |
| 72 | return; |
| 73 | } |
| 74 | if ( !$revision instanceof AbstractRevision ) { |
| 75 | throw new InvalidDataException( 'ReferenceRecorder can only attach to AbstractRevision storage' ); |
| 76 | } |
| 77 | /** @var Workflow $workflow */ |
| 78 | $workflow = $metadata['workflow']; |
| 79 | |
| 80 | if ( $revision instanceof PostRevision && $revision->isTopicTitle() ) { |
| 81 | [ $added, $removed ] = $this->calculateChangesFromTopic( $workflow, $revision ); |
| 82 | } else { |
| 83 | [ $added, $removed ] = $this->calculateChangesFromExisting( $workflow, $revision ); |
| 84 | } |
| 85 | |
| 86 | $this->storage->multiPut( $added ); |
| 87 | $this->storage->multiRemove( $removed ); |
| 88 | |
| 89 | // Data has not yet been committed at this point, so let's delay |
| 90 | // updating `categorylinks`, `externallinks`, etc. |
| 91 | $linksTableUpdater = $this->linksTableUpdater; |
| 92 | $this->deferredQueue->push( static function () use ( $linksTableUpdater, $workflow ) { |
| 93 | $linksTableUpdater->doUpdate( $workflow ); |
| 94 | } ); |
| 95 | } |
| 96 | |
| 97 | /** |
| 98 | * Compares the references contained within $revision against those stored for |
| 99 | * that revision. Returns the differences. |
| 100 | * |
| 101 | * @param Workflow $workflow |
| 102 | * @param AbstractRevision $revision |
| 103 | * @param PostRevision|null $root |
| 104 | * @return array Two nested arrays, first the references that were added and |
| 105 | * second the references that were removed. |
| 106 | */ |
| 107 | protected function calculateChangesFromExisting( |
| 108 | Workflow $workflow, |
| 109 | AbstractRevision $revision, |
| 110 | ?PostRevision $root = null |
| 111 | ) { |
| 112 | $prevReferences = $this->getExistingReferences( |
| 113 | $revision->getRevisionType(), |
| 114 | $revision->getCollectionId() |
| 115 | ); |
| 116 | $references = $this->getReferencesFromRevisionContent( $workflow, $revision, $root ); |
| 117 | |
| 118 | return $this->referencesDifference( $prevReferences, $references ); |
| 119 | } |
| 120 | |
| 121 | /** |
| 122 | * Topic titles themselves only support minimal wikitext, and references in the |
| 123 | * title itself are not tracked. |
| 124 | * |
| 125 | * However, moderation actions change what references are visible. When |
| 126 | * transitioning from or to a generically visible state (unmoderated or locked) the |
| 127 | * entire topic + summary needs to be re-evaluated. |
| 128 | * |
| 129 | * @param Workflow $workflow |
| 130 | * @param PostRevision $current Topic revision object that was inserted |
| 131 | * @return array Contains two arrays, first the references to add a second |
| 132 | * the references to remove |
| 133 | * @throws FlowException |
| 134 | */ |
| 135 | protected function calculateChangesFromTopic( Workflow $workflow, PostRevision $current ) { |
| 136 | if ( $current->isFirstRevision() ) { |
| 137 | return [ [], [] ]; |
| 138 | } |
| 139 | $previous = $this->storage->get( 'PostRevision', $current->getPrevRevisionId() ); |
| 140 | if ( !$previous ) { |
| 141 | throw new FlowException( 'Expected previous revision of ' . $current->getPrevRevisionId()->getAlphadecimal() ); |
| 142 | } |
| 143 | |
| 144 | $isHidden = self::isHidden( $current ); |
| 145 | $wasHidden = self::isHidden( $previous ); |
| 146 | |
| 147 | if ( $isHidden === $wasHidden ) { |
| 148 | return [ [], [] ]; |
| 149 | } |
| 150 | |
| 151 | // re-run |
| 152 | $revisions = $this->collectTopicRevisions( $workflow ); |
| 153 | $added = []; |
| 154 | $removed = []; |
| 155 | foreach ( $revisions as $revision ) { |
| 156 | [ $add, $remove ] = $this->calculateChangesFromExisting( $workflow, $revision, $current ); |
| 157 | $added = array_merge( $added, $add ); |
| 158 | $removed = array_merge( $removed, $remove ); |
| 159 | } |
| 160 | |
| 161 | return [ $added, $removed ]; |
| 162 | } |
| 163 | |
| 164 | protected static function isHidden( AbstractRevision $revision ) { |
| 165 | return $revision->isModerated() && $revision->getModerationState() !== $revision::MODERATED_LOCKED; |
| 166 | } |
| 167 | |
| 168 | /** |
| 169 | * Gets all the 'top' revisions within the topic, namely the posts and the |
| 170 | * summary. These are used when a topic changes is visibility via moderation |
| 171 | * to add or remove the relevant references. |
| 172 | * |
| 173 | * @param Workflow $workflow |
| 174 | * @return AbstractRevision[] |
| 175 | */ |
| 176 | protected function collectTopicRevisions( Workflow $workflow ) { |
| 177 | $found = $this->treeRepository->fetchSubtreeNodeList( [ $workflow->getId() ] ); |
| 178 | $queries = []; |
| 179 | foreach ( reset( $found ) as $uuid ) { |
| 180 | $queries[] = [ 'rev_type_id' => $uuid ]; |
| 181 | } |
| 182 | |
| 183 | $posts = $this->storage->findMulti( |
| 184 | 'PostRevision', |
| 185 | $queries, |
| 186 | [ 'sort' => 'rev_id', 'order' => 'DESC', 'limit' => 1 ] |
| 187 | ); |
| 188 | |
| 189 | // we also need the most recent topic summary if it exists |
| 190 | $summaries = $this->storage->find( |
| 191 | 'PostSummary', |
| 192 | [ 'rev_type_id' => $workflow->getId() ], |
| 193 | [ 'sort' => 'rev_id', 'order' => 'DESC', 'limit' => 1 ] |
| 194 | ); |
| 195 | |
| 196 | $result = $summaries; |
| 197 | // we have to unwrap the posts since we used findMulti, it returns |
| 198 | // a separate result set for each query |
| 199 | foreach ( $posts as $found ) { |
| 200 | $result[] = reset( $found ); |
| 201 | } |
| 202 | return $result; |
| 203 | } |
| 204 | |
| 205 | /** |
| 206 | * Pulls references from a revision's content |
| 207 | * |
| 208 | * @param Workflow $workflow The Workflow that the revision is attached to. |
| 209 | * @param AbstractRevision $revision The Revision to pull references from. |
| 210 | * @param PostRevision|null $root |
| 211 | * @return Reference[] Array of References. |
| 212 | */ |
| 213 | public function getReferencesFromRevisionContent( |
| 214 | Workflow $workflow, |
| 215 | AbstractRevision $revision, |
| 216 | ?PostRevision $root = null |
| 217 | ) { |
| 218 | // Locked is the only moderated state we still collect references for. |
| 219 | if ( self::isHidden( $revision ) ) { |
| 220 | return []; |
| 221 | } |
| 222 | |
| 223 | // We also do not track references in topic titles. |
| 224 | if ( $revision instanceof PostRevision && $revision->isTopicTitle() ) { |
| 225 | return []; |
| 226 | } |
| 227 | |
| 228 | // If this is attached to a topic we also need to check its permissions |
| 229 | if ( $root === null ) { |
| 230 | try { |
| 231 | if ( $revision instanceof PostRevision && !$revision->isTopicTitle() ) { |
| 232 | $root = $revision->getCollection()->getRoot()->getLastRevision(); |
| 233 | } elseif ( $revision instanceof PostSummary ) { |
| 234 | $root = $revision->getCollection()->getPost()->getRoot()->getLastRevision(); |
| 235 | } |
| 236 | } catch ( FlowException $e ) { |
| 237 | // Do nothing - we're likely in a unit test where no root can |
| 238 | // be resolved because the revision is created on the fly |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | if ( $root && ( self::isHidden( $root ) ) ) { |
| 243 | return []; |
| 244 | } |
| 245 | |
| 246 | return $this->referenceExtractor->getReferences( |
| 247 | $workflow, |
| 248 | $revision->getRevisionType(), |
| 249 | $revision->getCollectionId(), |
| 250 | $revision->getContent( 'html' ) |
| 251 | ); |
| 252 | } |
| 253 | |
| 254 | /** |
| 255 | * Retrieves references that are already stored in the database for a given revision |
| 256 | * |
| 257 | * @param string $revType The value returned from Revision::getRevisionType() for the revision. |
| 258 | * @param UUID $objectId The revision's Object ID. |
| 259 | * @return Reference[] Array of References. |
| 260 | */ |
| 261 | public function getExistingReferences( $revType, UUID $objectId ) { |
| 262 | $prevWikiReferences = $this->storage->find( 'WikiReference', [ |
| 263 | 'ref_src_wiki' => WikiMap::getCurrentWikiId(), |
| 264 | 'ref_src_object_type' => $revType, |
| 265 | 'ref_src_object_id' => $objectId, |
| 266 | ] ); |
| 267 | |
| 268 | $prevUrlReferences = $this->storage->find( 'URLReference', [ |
| 269 | 'ref_src_wiki' => WikiMap::getCurrentWikiId(), |
| 270 | 'ref_src_object_type' => $revType, |
| 271 | 'ref_src_object_id' => $objectId, |
| 272 | ] ); |
| 273 | |
| 274 | return array_merge( (array)$prevWikiReferences, (array)$prevUrlReferences ); |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * Compares two arrays of references |
| 279 | * |
| 280 | * Would be protected if not for testing. |
| 281 | * |
| 282 | * @param Reference[] $old The old references. |
| 283 | * @param Reference[] $new The new references. |
| 284 | * @return array Array with two elements: added and removed references. |
| 285 | */ |
| 286 | public function referencesDifference( array $old, array $new ) { |
| 287 | $newReferences = []; |
| 288 | |
| 289 | foreach ( $new as $ref ) { |
| 290 | $newReferences[$ref->getIdentifier()] = $ref; |
| 291 | } |
| 292 | |
| 293 | $oldReferences = []; |
| 294 | |
| 295 | foreach ( $old as $ref ) { |
| 296 | $oldReferences[$ref->getIdentifier()] = $ref; |
| 297 | } |
| 298 | |
| 299 | $addReferences = []; |
| 300 | |
| 301 | foreach ( $newReferences as $identifier => $ref ) { |
| 302 | if ( !isset( $oldReferences[$identifier] ) ) { |
| 303 | $addReferences[] = $ref; |
| 304 | } |
| 305 | } |
| 306 | |
| 307 | $removeReferences = []; |
| 308 | |
| 309 | foreach ( $oldReferences as $identifier => $ref ) { |
| 310 | if ( !isset( $newReferences[$identifier] ) ) { |
| 311 | $removeReferences[] = $ref; |
| 312 | } |
| 313 | } |
| 314 | |
| 315 | return [ $addReferences, $removeReferences ]; |
| 316 | } |
| 317 | } |