Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 60 |
|
0.00% |
0 / 7 |
CRAP | |
0.00% |
0 / 1 |
| SpamRegexBatch | |
0.00% |
0 / 60 |
|
0.00% |
0 / 7 |
506 | |
0.00% |
0 / 1 |
| buildRegexes | |
0.00% |
0 / 21 |
|
0.00% |
0 / 1 |
42 | |||
| validateRegexes | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| stripLines | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
2 | |||
| buildSafeRegexes | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| getBadLines | |
0.00% |
0 / 13 |
|
0.00% |
0 / 1 |
42 | |||
| regexesFromText | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| regexesFromMessage | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\SpamBlacklist; |
| 4 | |
| 5 | use Wikimedia\AtEase\AtEase; |
| 6 | |
| 7 | /** |
| 8 | * Utility class for working with blacklists |
| 9 | */ |
| 10 | class SpamRegexBatch { |
| 11 | /** |
| 12 | * Build a set of regular expressions matching URLs with the list of regex fragments. |
| 13 | * Returns an empty list if the input list is empty. |
| 14 | * |
| 15 | * @param string[] $lines list of fragments which will match in URLs |
| 16 | * @param BaseBlacklist $blacklist |
| 17 | * @param int $batchSize largest allowed batch regex; |
| 18 | * if 0, will produce one regex per line |
| 19 | * @return string[] |
| 20 | */ |
| 21 | private static function buildRegexes( array $lines, BaseBlacklist $blacklist, $batchSize = 4096 ) { |
| 22 | # Make regex |
| 23 | # It's faster using the S modifier even though it will usually only be run once |
| 24 | // $regex = 'https?://+[a-z0-9_\-.]*(' . implode( '|', $lines ) . ')'; |
| 25 | // return '/' . str_replace( '/', '\/', preg_replace('|\\\*/|', '/', $regex) ) . '/Sim'; |
| 26 | $regexes = []; |
| 27 | $regexStart = $blacklist->getRegexStart(); |
| 28 | $regexEnd = $blacklist->getRegexEnd( $batchSize ); |
| 29 | $build = false; |
| 30 | foreach ( $lines as $line ) { |
| 31 | if ( substr( $line, -1, 1 ) == "\\" ) { |
| 32 | // Final \ will break silently on the batched regexes. |
| 33 | // Skip it here to avoid breaking the next line; |
| 34 | // warnings from getBadLines() will still trigger on |
| 35 | // edit to keep new ones from floating in. |
| 36 | continue; |
| 37 | } |
| 38 | // FIXME: not very robust size check, but should work. :) |
| 39 | if ( $build === false ) { |
| 40 | $build = $line; |
| 41 | } elseif ( strlen( $build ) + strlen( $line ) > $batchSize ) { |
| 42 | $regexes[] = $regexStart . |
| 43 | str_replace( '/', '\/', preg_replace( '|\\\*/|u', '/', $build ) ) . |
| 44 | $regexEnd; |
| 45 | $build = $line; |
| 46 | } else { |
| 47 | $build .= '|'; |
| 48 | $build .= $line; |
| 49 | } |
| 50 | } |
| 51 | if ( $build !== false ) { |
| 52 | $regexes[] = $regexStart . |
| 53 | str_replace( '/', '\/', preg_replace( '|\\\*/|u', '/', $build ) ) . |
| 54 | $regexEnd; |
| 55 | } |
| 56 | return $regexes; |
| 57 | } |
| 58 | |
| 59 | /** |
| 60 | * Confirm that a set of regexes is either empty or valid. |
| 61 | * |
| 62 | * @param string[] $regexes set of regexes |
| 63 | * @return bool true if ok, false if contains invalid lines |
| 64 | */ |
| 65 | private static function validateRegexes( $regexes ) { |
| 66 | foreach ( $regexes as $regex ) { |
| 67 | AtEase::suppressWarnings(); |
| 68 | // @phan-suppress-next-line PhanParamSuspiciousOrder False positive |
| 69 | $ok = preg_match( $regex, '' ); |
| 70 | AtEase::restoreWarnings(); |
| 71 | |
| 72 | if ( $ok === false ) { |
| 73 | return false; |
| 74 | } |
| 75 | } |
| 76 | return true; |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * Strip comments and whitespace, then remove blanks |
| 81 | * |
| 82 | * @param string[] $lines |
| 83 | * @return string[] |
| 84 | */ |
| 85 | private static function stripLines( array $lines ) { |
| 86 | return array_filter( |
| 87 | array_map( 'trim', |
| 88 | preg_replace( '/#.*$/', '', |
| 89 | $lines ) |
| 90 | ) |
| 91 | ); |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * Do a sanity check on the batch regex. |
| 96 | * |
| 97 | * @param string[] $lines unsanitized input lines |
| 98 | * @param BaseBlacklist $blacklist |
| 99 | * @param bool|string $fileName optional for debug reporting |
| 100 | * @return string[] of regexes |
| 101 | */ |
| 102 | private static function buildSafeRegexes( array $lines, BaseBlacklist $blacklist, $fileName = false ) { |
| 103 | $lines = self::stripLines( $lines ); |
| 104 | $regexes = self::buildRegexes( $lines, $blacklist ); |
| 105 | if ( self::validateRegexes( $regexes ) ) { |
| 106 | return $regexes; |
| 107 | } else { |
| 108 | // _Something_ broke... rebuild line-by-line; it'll be |
| 109 | // slower if there's a lot of blacklist lines, but one |
| 110 | // broken line won't take out hundreds of its brothers. |
| 111 | if ( $fileName ) { |
| 112 | wfDebugLog( 'SpamBlacklist', "Spam blacklist warning: bogus line in $fileName\n" ); |
| 113 | } |
| 114 | return self::buildRegexes( $lines, $blacklist, 0 ); |
| 115 | } |
| 116 | } |
| 117 | |
| 118 | /** |
| 119 | * Returns an array of invalid lines |
| 120 | * |
| 121 | * @param string[] $lines |
| 122 | * @param BaseBlacklist $blacklist |
| 123 | * @return string[] of input lines which produce invalid input, or empty array if no problems |
| 124 | */ |
| 125 | public static function getBadLines( $lines, BaseBlacklist $blacklist ) { |
| 126 | $lines = self::stripLines( $lines ); |
| 127 | |
| 128 | $badLines = []; |
| 129 | foreach ( $lines as $line ) { |
| 130 | if ( substr( $line, -1, 1 ) == "\\" ) { |
| 131 | // Final \ will break silently on the batched regexes. |
| 132 | $badLines[] = $line; |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | $regexes = self::buildRegexes( $lines, $blacklist ); |
| 137 | if ( self::validateRegexes( $regexes ) ) { |
| 138 | // No other problems! |
| 139 | return $badLines; |
| 140 | } |
| 141 | |
| 142 | // Something failed in the batch, so check them one by one. |
| 143 | foreach ( $lines as $line ) { |
| 144 | $regexes = self::buildRegexes( [ $line ], $blacklist ); |
| 145 | if ( !self::validateRegexes( $regexes ) ) { |
| 146 | $badLines[] = $line; |
| 147 | } |
| 148 | } |
| 149 | return $badLines; |
| 150 | } |
| 151 | |
| 152 | /** |
| 153 | * Build a set of regular expressions from the given multiline input text, |
| 154 | * with empty lines and comments stripped. |
| 155 | * |
| 156 | * @param string $source |
| 157 | * @param BaseBlacklist $blacklist |
| 158 | * @param bool|string $fileName optional, for reporting of bad files |
| 159 | * @return string[] of regular expressions, potentially empty |
| 160 | */ |
| 161 | public static function regexesFromText( $source, BaseBlacklist $blacklist, $fileName = false ) { |
| 162 | $lines = explode( "\n", $source ); |
| 163 | return self::buildSafeRegexes( $lines, $blacklist, $fileName ); |
| 164 | } |
| 165 | |
| 166 | /** |
| 167 | * Build a set of regular expressions from a MediaWiki message. |
| 168 | * Will be correctly empty if the message isn't present. |
| 169 | * |
| 170 | * @param string $message |
| 171 | * @param BaseBlacklist $blacklist |
| 172 | * @return string[] of regular expressions, potentially empty |
| 173 | */ |
| 174 | public static function regexesFromMessage( $message, BaseBlacklist $blacklist ) { |
| 175 | $source = wfMessage( $message )->inContentLanguage(); |
| 176 | if ( !$source->isDisabled() ) { |
| 177 | return self::regexesFromText( $source->plain(), $blacklist ); |
| 178 | } else { |
| 179 | return []; |
| 180 | } |
| 181 | } |
| 182 | } |