Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
74.34% |
84 / 113 |
|
62.50% |
5 / 8 |
CRAP | |
0.00% |
0 / 1 |
| SpamBlacklist | |
74.34% |
84 / 113 |
|
62.50% |
5 / 8 |
45.21 | |
0.00% |
0 / 1 |
| getBlacklistType | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| antiSpoof | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| filter | |
87.50% |
63 / 72 |
|
0.00% |
0 / 1 |
20.78 | |||
| getCurrentLinks | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
1 | |||
| warmCachesForFilter | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
2 | |||
| getRegexStart | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getRegexEnd | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| logFilterHit | |
7.14% |
1 / 14 |
|
0.00% |
0 / 1 |
16.81 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\SpamBlacklist; |
| 4 | |
| 5 | use ExtensionRegistry; |
| 6 | use LogPage; |
| 7 | use ManualLogEntry; |
| 8 | use MediaWiki\CheckUser\Hooks as CUHooks; |
| 9 | use MediaWiki\ExternalLinks\ExternalLinksLookup; |
| 10 | use MediaWiki\MediaWikiServices; |
| 11 | use MediaWiki\Title\Title; |
| 12 | use MediaWiki\User\User; |
| 13 | use ObjectCache; |
| 14 | use RequestContext; |
| 15 | use Wikimedia\AtEase\AtEase; |
| 16 | use Wikimedia\Rdbms\Database; |
| 17 | |
| 18 | class SpamBlacklist extends BaseBlacklist { |
| 19 | private const STASH_TTL = 180; |
| 20 | private const STASH_AGE_DYING = 150; |
| 21 | |
| 22 | /** |
| 23 | * Returns the code for the blacklist implementation |
| 24 | * |
| 25 | * @return string |
| 26 | */ |
| 27 | protected function getBlacklistType() { |
| 28 | return 'spam'; |
| 29 | } |
| 30 | |
| 31 | /** |
| 32 | * Apply some basic anti-spoofing to the links before they get filtered, |
| 33 | * see @bug 12896 |
| 34 | * |
| 35 | * @param string $text |
| 36 | * |
| 37 | * @return string |
| 38 | */ |
| 39 | protected function antiSpoof( $text ) { |
| 40 | $text = str_replace( '．', '.', $text ); |
| 41 | return $text; |
| 42 | } |
| 43 | |
| 44 | /** |
| 45 | * @param string[] $links An array of links to check against the blacklist |
| 46 | * @param ?Title $title The title of the page to which the filter shall be applied. |
| 47 | * This is used to load the old links already on the page, so |
| 48 | * the filter is only applied to links that got added. If not given, |
| 49 | * the filter is applied to all $links. |
| 50 | * @param User $user Relevant user |
| 51 | * @param bool $preventLog Whether to prevent logging of hits. Set to true when |
| 52 | * the action is testing the links rather than attempting to save them |
| 53 | * (e.g. the API spamblacklist action) |
| 54 | * @param string $mode Either 'check' or 'stash' |
| 55 | * |
| 56 | * @return string[]|bool Matched text(s) if the edit should not be allowed; false otherwise |
| 57 | */ |
| 58 | public function filter( |
| 59 | array $links, |
| 60 | ?Title $title, |
| 61 | User $user, |
| 62 | $preventLog = false, |
| 63 | $mode = 'check' |
| 64 | ) { |
| 65 | $statsd = MediaWikiServices::getInstance()->getStatsdDataFactory(); |
| 66 | $cache = ObjectCache::getLocalClusterInstance(); |
| 67 | |
| 68 | if ( !$links ) { |
| 69 | return false; |
| 70 | } |
| 71 | |
| 72 | sort( $links ); |
| 73 | $key = $cache->makeKey( |
| 74 | 'blacklist', |
| 75 | $this->getBlacklistType(), |
| 76 | 'pass', |
| 77 | sha1( implode( "\n", $links ) ), |
| 78 | md5( (string)$title ) |
| 79 | ); |
| 80 | // Skip blacklist checks if nothing matched during edit stashing... |
| 81 | $knownNonMatchAsOf = $cache->get( $key ); |
| 82 | if ( $mode === 'check' ) { |
| 83 | if ( $knownNonMatchAsOf ) { |
| 84 | $statsd->increment( 'spamblacklist.check-stash.hit' ); |
| 85 | |
| 86 | return false; |
| 87 | } else { |
| 88 | $statsd->increment( 'spamblacklist.check-stash.miss' ); |
| 89 | } |
| 90 | } elseif ( $mode === 'stash' ) { |
| 91 | if ( $knownNonMatchAsOf && ( time() - $knownNonMatchAsOf ) < self::STASH_AGE_DYING ) { |
| 92 | // OK; not about to expire soon |
| 93 | return false; |
| 94 | } |
| 95 | } |
| 96 | |
| 97 | $blacklists = $this->getBlacklists(); |
| 98 | $whitelists = $this->getWhitelists(); |
| 99 | |
| 100 | if ( count( $blacklists ) ) { |
| 101 | // poor man's anti-spoof, see bug 12896 |
| 102 | $newLinks = array_map( [ $this, 'antiSpoof' ], $links ); |
| 103 | |
| 104 | $oldLinks = []; |
| 105 | if ( $title !== null ) { |
| 106 | $oldLinks = $this->getCurrentLinks( $title ); |
| 107 | $addedLinks = array_diff( $newLinks, $oldLinks ); |
| 108 | } else { |
| 109 | // can't load old links, so treat all links as added. |
| 110 | $addedLinks = $newLinks; |
| 111 | } |
| 112 | |
| 113 | wfDebugLog( 'SpamBlacklist', "Old URLs: " . implode( ', ', $oldLinks ) ); |
| 114 | wfDebugLog( 'SpamBlacklist', "New URLs: " . implode( ', ', $newLinks ) ); |
| 115 | wfDebugLog( 'SpamBlacklist', "Added URLs: " . implode( ', ', $addedLinks ) ); |
| 116 | |
| 117 | $links = implode( "\n", $addedLinks ); |
| 118 | |
| 119 | # Strip whitelisted URLs from the match |
| 120 | if ( is_array( $whitelists ) ) { |
| 121 | wfDebugLog( 'SpamBlacklist', "Excluding whitelisted URLs from " . count( $whitelists ) . |
| 122 | " regexes: " . implode( ', ', $whitelists ) . "\n" ); |
| 123 | foreach ( $whitelists as $regex ) { |
| 124 | AtEase::suppressWarnings(); |
| 125 | $newLinks = preg_replace( $regex, '', $links ); |
| 126 | AtEase::restoreWarnings(); |
| 127 | if ( is_string( $newLinks ) ) { |
| 128 | // If there wasn't a regex error, strip the matching URLs |
| 129 | $links = $newLinks; |
| 130 | } |
| 131 | } |
| 132 | } |
| 133 | |
| 134 | # Do the match |
| 135 | wfDebugLog( 'SpamBlacklist', "Checking text against " . count( $blacklists ) . |
| 136 | " regexes: " . implode( ', ', $blacklists ) . "\n" ); |
| 137 | $retVal = false; |
| 138 | foreach ( $blacklists as $regex ) { |
| 139 | AtEase::suppressWarnings(); |
| 140 | $matches = []; |
| 141 | $check = ( preg_match_all( $regex, $links, $matches ) > 0 ); |
| 142 | AtEase::restoreWarnings(); |
| 143 | if ( $check ) { |
| 144 | wfDebugLog( 'SpamBlacklist', "Match!\n" ); |
| 145 | $ip = RequestContext::getMain()->getRequest()->getIP(); |
| 146 | $fullUrls = []; |
| 147 | $fullLineRegex = substr( $regex, 0, strrpos( $regex, '/' ) ) . '.*/Sim'; |
| 148 | preg_match_all( $fullLineRegex, $links, $fullUrls ); |
| 149 | $imploded = implode( ' ', $fullUrls[0] ); |
| 150 | wfDebugLog( 'SpamBlacklistHit', "$ip caught submitting spam: $imploded\n" ); |
| 151 | if ( !$preventLog && $title ) { |
| 152 | $this->logFilterHit( $user, $title, $imploded ); |
| 153 | } |
| 154 | if ( $retVal === false ) { |
| 155 | $retVal = []; |
| 156 | } |
| 157 | $retVal = array_merge( $retVal, $fullUrls[1] ); |
| 158 | } |
| 159 | } |
| 160 | if ( is_array( $retVal ) ) { |
| 161 | $retVal = array_unique( $retVal ); |
| 162 | } |
| 163 | } else { |
| 164 | $retVal = false; |
| 165 | } |
| 166 | |
| 167 | if ( $retVal === false ) { |
| 168 | // Cache the typical negative results |
| 169 | $cache->set( $key, time(), self::STASH_TTL ); |
| 170 | if ( $mode === 'stash' ) { |
| 171 | $statsd->increment( 'spamblacklist.check-stash.store' ); |
| 172 | } |
| 173 | } |
| 174 | |
| 175 | return $retVal; |
| 176 | } |
| 177 | |
| 178 | /** |
| 179 | * Look up the links currently in the article, so we can |
| 180 | * ignore them on a second run. |
| 181 | * |
| 182 | * WARNING: I can add more *of the same link* with no problem here. |
| 183 | * @param Title $title |
| 184 | * @return array |
| 185 | */ |
| 186 | public function getCurrentLinks( Title $title ) { |
| 187 | $cache = MediaWikiServices::getInstance()->getMainWANObjectCache(); |
| 188 | $fname = __METHOD__; |
| 189 | return $cache->getWithSetCallback( |
| 190 | // Key is warmed via warmCachesForFilter() from ApiStashEdit |
| 191 | $cache->makeKey( 'external-link-list', $title->getLatestRevID() ), |
| 192 | $cache::TTL_MINUTE, |
| 193 | static function ( $oldValue, &$ttl, array &$setOpts ) use ( $title, $fname ) { |
| 194 | $dbr = MediaWikiServices::getInstance()->getConnectionProvider()->getReplicaDatabase(); |
| 195 | $setOpts += Database::getCacheSetOptions( $dbr ); |
| 196 | return ExternalLinksLookup::getExternalLinksForPage( |
| 197 | $title->getArticleID(), |
| 198 | $dbr, |
| 199 | $fname |
| 200 | ); |
| 201 | } |
| 202 | ); |
| 203 | } |
| 204 | |
| 205 | public function warmCachesForFilter( Title $title, array $entries, User $user ) { |
| 206 | $this->filter( |
| 207 | $entries, |
| 208 | $title, |
| 209 | $user, |
| 210 | // no logging |
| 211 | true, |
| 212 | 'stash' |
| 213 | ); |
| 214 | } |
| 215 | |
| 216 | /** |
| 217 | * Returns the start of the regex for matches |
| 218 | * |
| 219 | * @return string |
| 220 | */ |
| 221 | public function getRegexStart() { |
| 222 | return '/(?:https?:)?\/\/+[a-z0-9_\-.]*('; |
| 223 | } |
| 224 | |
| 225 | /** |
| 226 | * Returns the end of the regex for matches |
| 227 | * |
| 228 | * @param int $batchSize |
| 229 | * @return string |
| 230 | */ |
| 231 | public function getRegexEnd( $batchSize ) { |
| 232 | return ')' . parent::getRegexEnd( $batchSize ); |
| 233 | } |
| 234 | |
| 235 | /** |
| 236 | * Logs the filter hit to Special:Log if |
| 237 | * $wgLogSpamBlacklistHits is enabled. |
| 238 | * |
| 239 | * @param User $user |
| 240 | * @param Title $title |
| 241 | * @param string $url URL that the user attempted to add |
| 242 | */ |
| 243 | public function logFilterHit( User $user, $title, $url ) { |
| 244 | global $wgLogSpamBlacklistHits; |
| 245 | if ( $wgLogSpamBlacklistHits ) { |
| 246 | $logEntry = new ManualLogEntry( 'spamblacklist', 'hit' ); |
| 247 | $logEntry->setPerformer( $user ); |
| 248 | $logEntry->setTarget( $title ); |
| 249 | $logEntry->setParameters( [ |
| 250 | '4::url' => $url, |
| 251 | ] ); |
| 252 | $logid = $logEntry->insert(); |
| 253 | $log = new LogPage( 'spamblacklist' ); |
| 254 | if ( $log->isRestricted() ) { |
| 255 | // Make sure checkusers can see this action if the log is restricted |
| 256 | // (which is the default) |
| 257 | if ( ExtensionRegistry::getInstance()->isLoaded( 'CheckUser' ) ) { |
| 258 | $rc = $logEntry->getRecentChange( $logid ); |
| 259 | CUHooks::updateCheckUserData( $rc ); |
| 260 | } |
| 261 | } else { |
| 262 | // If the log is unrestricted, publish normally to RC, |
| 263 | // which will also update checkuser |
| 264 | $logEntry->publish( $logid, "rc" ); |
| 265 | } |
| 266 | } |
| 267 | } |
| 268 | } |