Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
87.25% |
130 / 149 |
|
75.00% |
6 / 8 |
CRAP | |
0.00% |
0 / 1 |
| AbuseLogger | |
87.25% |
130 / 149 |
|
75.00% |
6 / 8 |
44.49 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| addLogEntries | |
100.00% |
38 / 38 |
|
100.00% |
1 / 1 |
6 | |||
| buildLogTemplate | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
6 | |||
| newLocalLogEntryFromData | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
1 | |||
| insertLocalLogEntries | |
78.12% |
25 / 32 |
|
0.00% |
0 / 1 |
12.27 | |||
| insertCentralLogEntries | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
3 | |||
| storeVarDump | |
100.00% |
20 / 20 |
|
100.00% |
1 / 1 |
7 | |||
| publishEntry | |
0.00% |
0 / 12 |
|
0.00% |
0 / 1 |
30 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\AbuseFilter; |
| 4 | |
| 5 | use InvalidArgumentException; |
| 6 | use MediaWiki\CheckUser\Hooks; |
| 7 | use MediaWiki\Config\ServiceOptions; |
| 8 | use MediaWiki\Deferred\DeferredUpdates; |
| 9 | use MediaWiki\Extension\AbuseFilter\Parser\RuleCheckerFactory; |
| 10 | use MediaWiki\Extension\AbuseFilter\Variables\VariableHolder; |
| 11 | use MediaWiki\Extension\AbuseFilter\Variables\VariablesBlobStore; |
| 12 | use MediaWiki\Extension\AbuseFilter\Variables\VariablesManager; |
| 13 | use MediaWiki\Logging\ManualLogEntry; |
| 14 | use MediaWiki\Registration\ExtensionRegistry; |
| 15 | use MediaWiki\Title\Title; |
| 16 | use MediaWiki\User\User; |
| 17 | use MediaWiki\User\UserIdentityValue; |
| 18 | use Profiler; |
| 19 | use Wikimedia\IPUtils; |
| 20 | use Wikimedia\Rdbms\IDatabase; |
| 21 | use Wikimedia\Rdbms\LBFactory; |
| 22 | use Wikimedia\ScopedCallback; |
| 23 | |
| 24 | class AbuseLogger { |
| 25 | private string $action; |
| 26 | |
| 27 | /** @var string[][] A list of variable dumps generated by {@link self::storeVarDump} for de-duplication. */ |
| 28 | private array $varDumps = []; |
| 29 | |
| 30 | /** |
| 31 | * @internal Use {@link AbuseLoggerFactory::newLogger} instead |
| 32 | */ |
| 33 | public function __construct( |
| 34 | private readonly CentralDBManager $centralDBManager, |
| 35 | private readonly FilterLookup $filterLookup, |
| 36 | private readonly VariablesBlobStore $varBlobStore, |
| 37 | private readonly VariablesManager $varManager, |
| 38 | private readonly EditRevUpdater $editRevUpdater, |
| 39 | private readonly LBFactory $lbFactory, |
| 40 | private readonly RuleCheckerFactory $ruleCheckerFactory, |
| 41 | private readonly AbuseFilterPermissionManager $afPermissionManager, |
| 42 | private readonly ServiceOptions $options, |
| 43 | private readonly string $wikiID, |
| 44 | private readonly string $requestIP, |
| 45 | private readonly Title $title, |
| 46 | private readonly User $user, |
| 47 | private readonly VariableHolder $vars |
| 48 | ) { |
| 49 | if ( !$vars->varIsSet( 'action' ) ) { |
| 50 | throw new InvalidArgumentException( "The 'action' variable is not set." ); |
| 51 | } |
| 52 | $this->action = $vars->getComputedVariable( 'action' )->toString(); |
| 53 | } |
| 54 | |
| 55 | /** |
| 56 | * Create and publish log entries for taken actions |
| 57 | * |
| 58 | * @param array[] $actionsTaken |
| 59 | * @return array Shape is [ 'local' => int[], 'global' => int[] ], IDs of logged filters |
| 60 | * @phan-return array{local:int[],global:int[]} |
| 61 | */ |
| 62 | public function addLogEntries( array $actionsTaken ): array { |
| 63 | $dbw = $this->lbFactory->getPrimaryDatabase(); |
| 64 | $logTemplate = $this->buildLogTemplate(); |
| 65 | $centralLogTemplate = [ |
| 66 | 'afl_wiki' => $this->wikiID, |
| 67 | ]; |
| 68 | |
| 69 | $logRows = []; |
| 70 | $centralLogRows = []; |
| 71 | $loggedLocalFilters = []; |
| 72 | $loggedGlobalFilters = []; |
| 73 | |
| 74 | foreach ( $actionsTaken as $filter => $actions ) { |
| 75 | [ $filterID, $global ] = GlobalNameUtils::splitGlobalName( $filter ); |
| 76 | $thisLog = $logTemplate; |
| 77 | $thisLog['afl_filter_id'] = $filterID; |
| 78 | $thisLog['afl_global'] = (int)$global; |
| 79 | $thisLog['afl_actions'] = implode( ',', $actions ); |
| 80 | |
| 81 | // Don't log if we were only throttling. |
| 82 | // TODO This check should be removed or rewritten using Consequence objects |
| 83 | if ( $thisLog['afl_actions'] !== 'throttle' ) { |
| 84 | $logRows[] = $thisLog; |
| 85 | // Global logging |
| 86 | if ( $global ) { |
| 87 | $centralLog = $thisLog + $centralLogTemplate; |
| 88 | $centralLog['afl_filter_id'] = $filterID; |
| 89 | $centralLog['afl_global'] = 0; |
| 90 | $centralLog['afl_title'] = $this->title->getPrefixedText(); |
| 91 | $centralLog['afl_namespace'] = 0; |
| 92 | |
| 93 | $centralLogRows[] = $centralLog; |
| 94 | $loggedGlobalFilters[] = $filterID; |
| 95 | } else { |
| 96 | $loggedLocalFilters[] = $filterID; |
| 97 | } |
| 98 | } |
| 99 | } |
| 100 | |
| 101 | if ( !count( $logRows ) ) { |
| 102 | return [ 'local' => [], 'global' => [] ]; |
| 103 | } |
| 104 | |
| 105 | $localLogIDs = $this->insertLocalLogEntries( $logRows, $dbw ); |
| 106 | |
| 107 | $globalLogIDs = []; |
| 108 | if ( count( $loggedGlobalFilters ) ) { |
| 109 | $fdb = $this->centralDBManager->getConnection( DB_PRIMARY ); |
| 110 | $globalLogIDs = $this->insertCentralLogEntries( $centralLogRows, $fdb ); |
| 111 | } |
| 112 | |
| 113 | $this->editRevUpdater->setLogIdsForTarget( |
| 114 | $this->title, |
| 115 | [ 'local' => $localLogIDs, 'global' => $globalLogIDs ] |
| 116 | ); |
| 117 | |
| 118 | return [ 'local' => $loggedLocalFilters, 'global' => $loggedGlobalFilters ]; |
| 119 | } |
| 120 | |
| 121 | /** |
| 122 | * Creates a template to use for logging taken actions |
| 123 | */ |
| 124 | private function buildLogTemplate(): array { |
| 125 | // If $this->user isn't safe to load (e.g. a failure during |
| 126 | // AbortAutoAccount), create a dummy anonymous user instead. |
| 127 | $user = $this->user->isSafeToLoad() ? $this->user : new User; |
| 128 | // Create a template |
| 129 | $logTemplate = [ |
| 130 | 'afl_user' => $user->getId(), |
| 131 | 'afl_user_text' => $user->getName(), |
| 132 | 'afl_timestamp' => $this->lbFactory->getReplicaDatabase()->timestamp(), |
| 133 | 'afl_namespace' => $this->title->getNamespace(), |
| 134 | 'afl_title' => $this->title->getDBkey(), |
| 135 | 'afl_action' => $this->action, |
| 136 | 'afl_ip_hex' => $this->options->get( 'AbuseFilterLogIP' ) ? IPUtils::toHex( $this->requestIP ) : '', |
| 137 | ]; |
| 138 | // Hack to avoid revealing IPs of people creating accounts |
| 139 | if ( ( $this->action === 'createaccount' || $this->action === 'autocreateaccount' ) && !$user->getId() ) { |
| 140 | $logTemplate['afl_user_text'] = $this->vars->getComputedVariable( 'accountname' )->toString(); |
| 141 | } |
| 142 | return $logTemplate; |
| 143 | } |
| 144 | |
| 145 | private function newLocalLogEntryFromData( array $data ): ManualLogEntry { |
| 146 | // Give grep a chance to find the usages: |
| 147 | // logentry-abusefilter-hit |
| 148 | $entry = new ManualLogEntry( 'abusefilter', 'hit' ); |
| 149 | $user = new UserIdentityValue( $data['afl_user'], $data['afl_user_text'] ); |
| 150 | $entry->setPerformer( $user ); |
| 151 | $entry->setTarget( $this->title ); |
| 152 | $filterName = GlobalNameUtils::buildGlobalName( |
| 153 | $data['afl_filter_id'], |
| 154 | $data['afl_global'] === 1 |
| 155 | ); |
| 156 | // Additional info |
| 157 | $entry->setParameters( [ |
| 158 | 'action' => $data['afl_action'], |
| 159 | 'filter' => $filterName, |
| 160 | 'actions' => $data['afl_actions'], |
| 161 | 'log' => $data['afl_id'], |
| 162 | ] ); |
| 163 | return $entry; |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * @param array[] $logRows |
| 168 | * @param IDatabase $dbw |
| 169 | * @return int[] |
| 170 | */ |
| 171 | private function insertLocalLogEntries( array $logRows, IDatabase $dbw ): array { |
| 172 | $loggedIDs = []; |
| 173 | foreach ( $logRows as $data ) { |
| 174 | $data['afl_var_dump'] = $this->storeVarDump( $data['afl_filter_id'], (bool)$data['afl_global'], false ); |
| 175 | $dbw->newInsertQueryBuilder() |
| 176 | ->insertInto( 'abuse_filter_log' ) |
| 177 | ->row( $data ) |
| 178 | ->caller( __METHOD__ ) |
| 179 | ->execute(); |
| 180 | $loggedIDs[] = $data['afl_id'] = $dbw->insertId(); |
| 181 | |
| 182 | // Send data to CheckUser if installed and we |
| 183 | // aren't already sending a notification to recentchanges |
| 184 | if ( ExtensionRegistry::getInstance()->isLoaded( 'CheckUser' ) |
| 185 | && !str_contains( $this->options->get( 'AbuseFilterNotifications' ) ?: '', 'rc' ) |
| 186 | ) { |
| 187 | $entry = $this->newLocalLogEntryFromData( $data ); |
| 188 | $user = $entry->getPerformerIdentity(); |
| 189 | // Invert the hack from ::buildLogTemplate because CheckUser attempts |
| 190 | // to assign an actor id to the non-existing user |
| 191 | if ( |
| 192 | ( $this->action === 'createaccount' || $this->action === 'autocreateaccount' ) |
| 193 | && !$user->getId() |
| 194 | ) { |
| 195 | $entry->setPerformer( new UserIdentityValue( 0, $this->requestIP ) ); |
| 196 | } |
| 197 | $rc = $entry->getRecentChange(); |
| 198 | // We need to send the entries on POSTSEND to ensure that the user definitely exists, as a temporary |
| 199 | // account being created by this edit may not exist until after AbuseFilter processes the edit. |
| 200 | DeferredUpdates::addCallableUpdate( static function () use ( $rc ) { |
| 201 | // Silence the TransactionProfiler warnings for performing write queries (T359648). |
| 202 | $trxProfiler = Profiler::instance()->getTransactionProfiler(); |
| 203 | $scope = $trxProfiler->silenceForScope( $trxProfiler::EXPECTATION_REPLICAS_ONLY ); |
| 204 | Hooks::updateCheckUserData( $rc ); |
| 205 | ScopedCallback::consume( $scope ); |
| 206 | } ); |
| 207 | } |
| 208 | |
| 209 | if ( $this->options->get( 'AbuseFilterNotifications' ) !== false ) { |
| 210 | $filterID = $data['afl_filter_id']; |
| 211 | $global = $data['afl_global']; |
| 212 | if ( |
| 213 | !$this->options->get( 'AbuseFilterNotificationsPrivate' ) && |
| 214 | $this->filterLookup->getFilter( $filterID, $global )->isHidden() |
| 215 | ) { |
| 216 | continue; |
| 217 | } |
| 218 | $entry = $this->newLocalLogEntryFromData( $data ); |
| 219 | $this->publishEntry( $dbw, $entry ); |
| 220 | } |
| 221 | } |
| 222 | return $loggedIDs; |
| 223 | } |
| 224 | |
| 225 | /** |
| 226 | * @param array[] $centralLogRows |
| 227 | * @param IDatabase $fdb |
| 228 | * @return int[] |
| 229 | */ |
| 230 | private function insertCentralLogEntries( array $centralLogRows, IDatabase $fdb ): array { |
| 231 | $this->varManager->computeDBVars( $this->vars ); |
| 232 | foreach ( $centralLogRows as $index => $data ) { |
| 233 | $centralLogRows[$index]['afl_var_dump'] = $this->storeVarDump( |
| 234 | $data['afl_filter_id'], |
| 235 | // All the filters logged centrally are global. Note, this must not use `afl_global`, because that is |
| 236 | // in the perspective of the central wiki, hence false: what we consider global on the current wiki is |
| 237 | // local to the central wiki. |
| 238 | true, |
| 239 | true |
| 240 | ); |
| 241 | } |
| 242 | |
| 243 | $loggedIDs = []; |
| 244 | foreach ( $centralLogRows as $row ) { |
| 245 | $fdb->newInsertQueryBuilder() |
| 246 | ->insertInto( 'abuse_filter_log' ) |
| 247 | ->row( $row ) |
| 248 | ->caller( __METHOD__ ) |
| 249 | ->execute(); |
| 250 | $loggedIDs[] = $fdb->insertId(); |
| 251 | } |
| 252 | return $loggedIDs; |
| 253 | } |
| 254 | |
| 255 | /** |
| 256 | * Returns a string to be used as the value of afl_var_dump in an abuse_filter_log row. This may either |
| 257 | * be a BlobStore address or a JSON string (see {@link VariablesBlobStore::storeVarDump} for more detail). |
| 258 | * |
| 259 | * This method removes protected variables from the var dump that are not used in the filter |
| 260 | * associated with the AbuseFilter log to be created. It also de-duplicates var dumps where |
| 261 | * this is possible. |
| 262 | * |
| 263 | * @param int $filterId The filter associated with the AbuseFilter log entry |
| 264 | * @param bool $isGlobalFilter If the filter associated with the AbuseFilter log entry is global |
| 265 | * @param bool $useCentralDB Whether the dump should be stored in the central database |
| 266 | * @return string |
| 267 | */ |
| 268 | private function storeVarDump( int $filterId, bool $isGlobalFilter, bool $useCentralDB ): string { |
| 269 | // Generate a key for the varDumps instance cache used to de-duplicate var dumps where possible. |
| 270 | // The key for this cache is the protected variables used in the filter along with whether the |
| 271 | // var dump is global. |
| 272 | $filter = $this->filterLookup->getFilter( $filterId, $isGlobalFilter ); |
| 273 | $usedVariables = $this->ruleCheckerFactory->newRuleChecker()->getUsedVars( $filter->getRules() ); |
| 274 | $usedProtectedVariables = $this->afPermissionManager->getUsedProtectedVariables( $usedVariables ); |
| 275 | if ( count( $usedProtectedVariables ) ) { |
| 276 | sort( $usedProtectedVariables ); |
| 277 | $variablesKey = implode( ',', $usedProtectedVariables ); |
| 278 | } else { |
| 279 | $variablesKey = 0; |
| 280 | } |
| 281 | $centralDBKey = (int)$useCentralDB; |
| 282 | |
| 283 | // Create a new var dump if the instance cache does not have this key. |
| 284 | if ( |
| 285 | !array_key_exists( $centralDBKey, $this->varDumps ) || |
| 286 | !array_key_exists( $variablesKey, $this->varDumps[$centralDBKey] ) |
| 287 | ) { |
| 288 | // Filter out all protected variables that are not used in the current filter. Any other filter with |
| 289 | // the same list of protected filters will also use this var dump |
| 290 | $filteredVars = VariableHolder::newFromArray( $this->vars->getVars() ); |
| 291 | $protectedVariables = $this->afPermissionManager->getProtectedVariables(); |
| 292 | foreach ( array_keys( $filteredVars->getVars() ) as $varName ) { |
| 293 | if ( in_array( $varName, $protectedVariables ) && !in_array( $varName, $usedProtectedVariables ) ) { |
| 294 | $filteredVars->removeVar( $varName ); |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | $this->varDumps[$centralDBKey][$variablesKey] = $this->varBlobStore->storeVarDump( |
| 299 | $filteredVars, |
| 300 | $useCentralDB |
| 301 | ); |
| 302 | } |
| 303 | |
| 304 | return $this->varDumps[$centralDBKey][$variablesKey]; |
| 305 | } |
| 306 | |
| 307 | /** |
| 308 | * Like ManualLogEntry::publish, but doesn't require an ID (which we don't have) and skips the |
| 309 | * tagging part |
| 310 | * |
| 311 | * @param IDatabase $dbw To cancel the callback if the log insertion fails |
| 312 | * @param ManualLogEntry $entry |
| 313 | */ |
| 314 | private function publishEntry( IDatabase $dbw, ManualLogEntry $entry ): void { |
| 315 | DeferredUpdates::addCallableUpdate( |
| 316 | function () use ( $entry ) { |
| 317 | $rc = $entry->getRecentChange(); |
| 318 | $to = $this->options->get( 'AbuseFilterNotifications' ); |
| 319 | |
| 320 | if ( $to === 'rc' || $to === 'rcandudp' ) { |
| 321 | $rc->save( $rc::SEND_NONE ); |
| 322 | } |
| 323 | if ( $to === 'udp' || $to === 'rcandudp' ) { |
| 324 | $rc->notifyRCFeeds(); |
| 325 | } |
| 326 | }, |
| 327 | DeferredUpdates::POSTSEND, |
| 328 | $dbw |
| 329 | ); |
| 330 | } |
| 331 | |
| 332 | } |