tests/phpunit/includes/LinkFilterTest.php

   1 <?php
   2
   3 use Wikimedia\Rdbms\LikeMatch;
   4
   5 /**
   6  * @group Database
   7  */
   8 class LinkFilterTest extends MediaWikiLangTestCase {
   9
  10         protected function setUp() {
  11                 parent::setUp();
  12
  13                 $this->setMwGlobals( 'wgUrlProtocols', [
  14                         'http://',
  15                         'https://',
  16                         'ftp://',
  17                         'irc://',
  18                         'ircs://',
  19                         'gopher://',
  20                         'telnet://',
  21                         'nntp://',
  22                         'worldwind://',
  23                         'mailto:',
  24                         'news:',
  25                         'svn://',
  26                         'git://',
  27                         'mms://',
  28                         '//',
  29                 ] );
  30         }
  31
  32         /**
  33          * createRegexFromLike($like)
  34          *
  35          * Takes an array as created by LinkFilter::makeLikeArray() and creates a regex from it
  36          *
  37          * @param array $like Array as created by LinkFilter::makeLikeArray()
  38          * @return string Regex
  39          */
  40         function createRegexFromLIKE( $like ) {
  41                 $regex = '!^';
  42
  43                 foreach ( $like as $item ) {
  44                         if ( $item instanceof LikeMatch ) {
  45                                 if ( $item->toString() == '%' ) {
  46                                         $regex .= '.*';
  47                                 } elseif ( $item->toString() == '_' ) {
  48                                         $regex .= '.';
  49                                 }
  50                         } else {
  51                                 $regex .= preg_quote( $item, '!' );
  52                         }
  53
  54                 }
  55
  56                 $regex .= '$!';
  57
  58                 return $regex;
  59         }
  60
  61         /**
  62          * provideValidPatterns()
  63          *
  64          * @return array
  65          */
  66         public static function provideValidPatterns() {
  67                 return [
  68                         // Protocol, Search pattern, URL which matches the pattern
  69                         [ 'http://', '*.test.com', 'http://www.test.com' ],
  70                         [ 'http://', 'test.com:8080/dir/file', 'http://name:pass@test.com:8080/dir/file' ],
  71                         [ 'https://', '*.com', 'https://s.s.test..com:88/dir/file?a=1&b=2' ],
  72                         [ 'https://', '*.com', 'https://name:pass@secure.com/index.html' ],
  73                         [ 'http://', 'name:pass@test.com', 'http://test.com' ],
  74                         [ 'http://', 'test.com', 'http://name:pass@test.com' ],
  75                         [ 'http://', '*.test.com', 'http://a.b.c.test.com/dir/dir/file?a=6' ],
  76                         [ null, 'http://*.test.com', 'http://www.test.com' ],
  77                         [ 'mailto:', 'name@mail.test123.com', 'mailto:name@mail.test123.com' ],
  78                         [ '',
  79                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  80                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg'
  81                         ],
  82                         [ '', 'http://name:pass@*.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  83                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  84                         [ '', 'http://name:wrongpass@*.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]',
  85                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  86                         [ 'http://', 'name:pass@*.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  87                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  88                         [ '', 'http://name:pass@www.test.com:12345',
  89                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  90                         [ 'ftp://', 'user:pass@ftp.test.com:1233/home/user/file;type=efw',
  91                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ],
  92                         [ null, 'ftp://otheruser:otherpass@ftp.test.com:1233/home/user/file;type=',
  93                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ],
  94                         [ null, 'ftp://@ftp.test.com:1233/home/user/file;type=',
  95                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ],
  96                         [ null, 'ftp://ftp.test.com/',
  97                                 'ftp://user:pass@ftp.test.com/home/user/file;type=efw' ],
  98                         [ null, 'ftp://ftp.test.com/',
  99                                 'ftp://user:pass@ftp.test.com/home/user/file;type=efw' ],
 100                         [ null, 'ftp://*.test.com:222/',
 101                                 'ftp://user:pass@ftp.test.com:222/home' ],
 102                         [ 'irc://', '*.myserver:6667/', 'irc://test.myserver:6667/' ],
 103                         [ 'irc://', 'name:pass@*.myserver/', 'irc://test.myserver:6667/' ],
 104                         [ 'irc://', 'name:pass@*.myserver/', 'irc://other:@test.myserver:6667/' ],
 105                         [ '', 'irc://test/name,string,abc?msg=t', 'irc://test/name,string,abc?msg=test' ],
 106                         [ '', 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z',
 107                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z' ],
 108                         [ '', 'https://gerrit.wikimedia.org',
 109                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z' ],
 110                         [ 'mailto:', '*.test.com', 'mailto:name@pop3.test.com' ],
 111                         [ 'mailto:', 'test.com', 'mailto:name@test.com' ],
 112                         [ 'news:', 'test.1234afc@news.test.com', 'news:test.1234afc@news.test.com' ],
 113                         [ 'news:', '*.test.com', 'news:test.1234afc@news.test.com' ],
 114                         [ '', 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com',
 115                                 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com' ],
 116                         [ '', 'news:*.aol.com',
 117                                 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com' ],
 118                         [ '', 'git://github.com/prwef/abc-def.git', 'git://github.com/prwef/abc-def.git' ],
 119                         [ 'git://', 'github.com/', 'git://github.com/prwef/abc-def.git' ],
 120                         [ 'git://', '*.github.com/', 'git://a.b.c.d.e.f.github.com/prwef/abc-def.git' ],
 121                         [ '', 'gopher://*.test.com/', 'gopher://gopher.test.com/0/v2/vstat' ],
 122                         [ 'telnet://', '*.test.com', 'telnet://shell.test.com/~home/' ],
 123                         [ '', 'http://test.com', 'http://test.com/index?arg=1' ],
 124                         [ 'http://', '*.test.com', 'http://www.test.com/index?arg=1' ],
 125                         [ '' ,
 126                                 'http://xx23124:__ffdfdef__@www.test.com:12345/dir' ,
 127                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg'
 128                         ],
 129
 130                         // Tests for false positives
 131                         [ 'http://', 'test.com', 'http://www.test.com', false ],
 132                         [ 'http://', 'www1.test.com', 'http://www.test.com', false ],
 133                         [ 'http://', '*.test.com', 'http://www.test.t.com', false ],
 134                         [ '', 'http://test.com:8080', 'http://www.test.com:8080', false ],
 135                         [ '', 'https://test.com', 'http://test.com', false ],
 136                         [ '', 'http://test.com', 'https://test.com', false ],
 137                         [ 'http://', 'http://test.com', 'http://test.com', false ],
 138                         [ null, 'http://www.test.com', 'http://www.test.com:80', false ],
 139                         [ null, 'http://www.test.com:80', 'http://www.test.com', false ],
 140                         [ null, 'http://*.test.com:80', 'http://www.test.com', false ],
 141                         [ '', 'https://gerrit.wikimedia.org/r/#/XXX/status:open,n,z',
 142                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z', false ],
 143                         [ '', 'https://*.wikimedia.org/r/#/q/status:open,n,z',
 144                                 'https://gerrit.wikimedia.org/r/#/XXX/status:open,n,z', false ],
 145                         [ 'mailto:', '@test.com', '@abc.test.com', false ],
 146                         [ 'mailto:', 'mail@test.com', 'mail2@test.com', false ],
 147                         [ '', 'mailto:mail@test.com', 'mail2@test.com', false ],
 148                         [ '', 'mailto:@test.com', '@abc.test.com', false ],
 149                         [ 'ftp://', '*.co', 'ftp://www.co.uk', false ],
 150                         [ 'ftp://', '*.co', 'ftp://www.co.m', false ],
 151                         [ 'ftp://', '*.co/dir/', 'ftp://www.co/dir2/', false ],
 152                         [ 'ftp://', 'www.co/dir/', 'ftp://www.co/dir2/', false ],
 153                         [ 'ftp://', 'test.com/dir/', 'ftp://test.com/', false ],
 154                         [ '', 'http://test.com:8080/dir/', 'http://test.com:808/dir/', false ],
 155                         [ '', 'http://test.com/dir/index.html', 'http://test.com/dir/index.php', false ],
 156
 157                         // These are false positives too and ideally shouldn't match, but that
 158                         // would require using regexes and RLIKE instead of LIKE
 159                         // [ null, 'http://*.test.com', 'http://www.test.com:80', false ],
 160                         // [ '', 'https://*.wikimedia.org/r/#/q/status:open,n,z',
 161                         //      'https://gerrit.wikimedia.org/XXX/r/#/q/status:open,n,z', false ],
 162                 ];
 163         }
 164
 165         /**
 166          * testMakeLikeArrayWithValidPatterns()
 167          *
 168          * Tests whether the LIKE clause produced by LinkFilter::makeLikeArray($pattern, $protocol)
 169          * will find one of the URL indexes produced by wfMakeUrlIndexes($url)
 170          *
 171          * @dataProvider provideValidPatterns
 172          *
 173          * @param string $protocol Protocol, e.g. 'http://' or 'mailto:'
 174          * @param string $pattern Search pattern to feed to LinkFilter::makeLikeArray
 175          * @param string $url URL to feed to wfMakeUrlIndexes
 176          * @param bool $shouldBeFound Should the URL be found? (defaults true)
 177          */
 178         function testMakeLikeArrayWithValidPatterns( $protocol, $pattern, $url, $shouldBeFound = true ) {
 179                 $indexes = wfMakeUrlIndexes( $url );
 180                 $likeArray = LinkFilter::makeLikeArray( $pattern, $protocol );
 181
 182                 $this->assertTrue( $likeArray !== false,
 183                         "LinkFilter::makeLikeArray('$pattern', '$protocol') returned false on a valid pattern"
 184                 );
 185
 186                 $regex = $this->createRegexFromLIKE( $likeArray );
 187                 $debugmsg = "Regex: '" . $regex . "'\n";
 188                 $debugmsg .= count( $indexes ) . " index(es) created by wfMakeUrlIndexes():\n";
 189
 190                 $matches = 0;
 191
 192                 foreach ( $indexes as $index ) {
 193                         $matches += preg_match( $regex, $index );
 194                         $debugmsg .= "\t'$index'\n";
 195                 }
 196
 197                 if ( $shouldBeFound ) {
 198                         $this->assertTrue(
 199                                 $matches > 0,
 200                                 "Search pattern '$protocol$pattern' does not find url '$url' \n$debugmsg"
 201                         );
 202                 } else {
 203                         $this->assertFalse(
 204                                 $matches > 0,
 205                                 "Search pattern '$protocol$pattern' should not find url '$url' \n$debugmsg"
 206                         );
 207                 }
 208         }
 209
 210         /**
 211          * provideInvalidPatterns()
 212          *
 213          * @return array
 214          */
 215         public static function provideInvalidPatterns() {
 216                 return [
 217                         [ '' ],
 218                         [ '*' ],
 219                         [ 'http://*' ],
 220                         [ 'http://*/' ],
 221                         [ 'http://*/dir/file' ],
 222                         [ 'test.*.com' ],
 223                         [ 'http://test.*.com' ],
 224                         [ 'test.*.com' ],
 225                         [ 'http://*.test.*' ],
 226                         [ 'http://*test.com' ],
 227                         [ 'https://*' ],
 228                         [ '*://test.com' ],
 229                         [ 'mailto:name:pass@t*est.com' ],
 230                         [ 'http://*:888/' ],
 231                         [ '*http://' ],
 232                         [ 'test.com/*/index' ],
 233                         [ 'test.com/dir/index?arg=*' ],
 234                 ];
 235         }
 236
 237         /**
 238          * testMakeLikeArrayWithInvalidPatterns()
 239          *
 240          * Tests whether LinkFilter::makeLikeArray($pattern) will reject invalid search patterns
 241          *
 242          * @dataProvider provideInvalidPatterns
 243          *
 244          * @param string $pattern Invalid search pattern
 245          */
 246         function testMakeLikeArrayWithInvalidPatterns( $pattern ) {
 247                 $this->assertFalse(
 248                         LinkFilter::makeLikeArray( $pattern ),
 249                         "'$pattern' is not a valid pattern and should be rejected"
 250                 );
 251         }
 252
 253 }