tests/phpunit/includes/LinkFilterTest.php

   1 <?php
   2
   3 use Wikimedia\Rdbms\LikeMatch;
   4
   5 /**
   6  * @covers LinkFilter
   7  * @group Database
   8  */
   9 class LinkFilterTest extends MediaWikiLangTestCase {
  10
  11         protected function setUp() {
  12                 parent::setUp();
  13
  14                 $this->setMwGlobals( 'wgUrlProtocols', [
  15                         'http://',
  16                         'https://',
  17                         'ftp://',
  18                         'irc://',
  19                         'ircs://',
  20                         'gopher://',
  21                         'telnet://',
  22                         'nntp://',
  23                         'worldwind://',
  24                         'mailto:',
  25                         'news:',
  26                         'svn://',
  27                         'git://',
  28                         'mms://',
  29                         '//',
  30                 ] );
  31         }
  32
  33         /**
  34          * createRegexFromLike($like)
  35          *
  36          * Takes an array as created by LinkFilter::makeLikeArray() and creates a regex from it
  37          *
  38          * @param array $like Array as created by LinkFilter::makeLikeArray()
  39          * @return string Regex
  40          */
  41         function createRegexFromLIKE( $like ) {
  42                 $regex = '!^';
  43
  44                 foreach ( $like as $item ) {
  45                         if ( $item instanceof LikeMatch ) {
  46                                 if ( $item->toString() == '%' ) {
  47                                         $regex .= '.*';
  48                                 } elseif ( $item->toString() == '_' ) {
  49                                         $regex .= '.';
  50                                 }
  51                         } else {
  52                                 $regex .= preg_quote( $item, '!' );
  53                         }
  54
  55                 }
  56
  57                 $regex .= '$!';
  58
  59                 return $regex;
  60         }
  61
  62         /**
  63          * provideValidPatterns()
  64          *
  65          * @return array
  66          */
  67         public static function provideValidPatterns() {
  68                 return [
  69                         // Protocol, Search pattern, URL which matches the pattern
  70                         [ 'http://', '*.test.com', 'http://www.test.com' ],
  71                         [ 'http://', 'test.com:8080/dir/file', 'http://name:pass@test.com:8080/dir/file' ],
  72                         [ 'https://', '*.com', 'https://s.s.test..com:88/dir/file?a=1&b=2' ],
  73                         [ 'https://', '*.com', 'https://name:pass@secure.com/index.html' ],
  74                         [ 'http://', 'name:pass@test.com', 'http://test.com' ],
  75                         [ 'http://', 'test.com', 'http://name:pass@test.com' ],
  76                         [ 'http://', '*.test.com', 'http://a.b.c.test.com/dir/dir/file?a=6' ],
  77                         [ null, 'http://*.test.com', 'http://www.test.com' ],
  78                         [ 'mailto:', 'name@mail.test123.com', 'mailto:name@mail.test123.com' ],
  79                         [ '',
  80                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  81                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg'
  82                         ],
  83                         [ '', 'http://name:pass@*.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  84                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  85                         [ '', 'http://name:wrongpass@*.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]',
  86                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  87                         [ 'http://', 'name:pass@*.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  88                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  89                         [ '', 'http://name:pass@www.test.com:12345',
  90                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ],
  91                         [ 'ftp://', 'user:pass@ftp.test.com:1233/home/user/file;type=efw',
  92                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ],
  93                         [ null, 'ftp://otheruser:otherpass@ftp.test.com:1233/home/user/file;type=',
  94                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ],
  95                         [ null, 'ftp://@ftp.test.com:1233/home/user/file;type=',
  96                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ],
  97                         [ null, 'ftp://ftp.test.com/',
  98                                 'ftp://user:pass@ftp.test.com/home/user/file;type=efw' ],
  99                         [ null, 'ftp://ftp.test.com/',
 100                                 'ftp://user:pass@ftp.test.com/home/user/file;type=efw' ],
 101                         [ null, 'ftp://*.test.com:222/',
 102                                 'ftp://user:pass@ftp.test.com:222/home' ],
 103                         [ 'irc://', '*.myserver:6667/', 'irc://test.myserver:6667/' ],
 104                         [ 'irc://', 'name:pass@*.myserver/', 'irc://test.myserver:6667/' ],
 105                         [ 'irc://', 'name:pass@*.myserver/', 'irc://other:@test.myserver:6667/' ],
 106                         [ '', 'irc://test/name,string,abc?msg=t', 'irc://test/name,string,abc?msg=test' ],
 107                         [ '', 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z',
 108                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z' ],
 109                         [ '', 'https://gerrit.wikimedia.org',
 110                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z' ],
 111                         [ 'mailto:', '*.test.com', 'mailto:name@pop3.test.com' ],
 112                         [ 'mailto:', 'test.com', 'mailto:name@test.com' ],
 113                         [ 'news:', 'test.1234afc@news.test.com', 'news:test.1234afc@news.test.com' ],
 114                         [ 'news:', '*.test.com', 'news:test.1234afc@news.test.com' ],
 115                         [ '', 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com',
 116                                 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com' ],
 117                         [ '', 'news:*.aol.com',
 118                                 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com' ],
 119                         [ '', 'git://github.com/prwef/abc-def.git', 'git://github.com/prwef/abc-def.git' ],
 120                         [ 'git://', 'github.com/', 'git://github.com/prwef/abc-def.git' ],
 121                         [ 'git://', '*.github.com/', 'git://a.b.c.d.e.f.github.com/prwef/abc-def.git' ],
 122                         [ '', 'gopher://*.test.com/', 'gopher://gopher.test.com/0/v2/vstat' ],
 123                         [ 'telnet://', '*.test.com', 'telnet://shell.test.com/~home/' ],
 124                         [ '', 'http://test.com', 'http://test.com/index?arg=1' ],
 125                         [ 'http://', '*.test.com', 'http://www.test.com/index?arg=1' ],
 126                         [ '' ,
 127                                 'http://xx23124:__ffdfdef__@www.test.com:12345/dir' ,
 128                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg'
 129                         ],
 130
 131                         // Tests for false positives
 132                         [ 'http://', 'test.com', 'http://www.test.com', false ],
 133                         [ 'http://', 'www1.test.com', 'http://www.test.com', false ],
 134                         [ 'http://', '*.test.com', 'http://www.test.t.com', false ],
 135                         [ '', 'http://test.com:8080', 'http://www.test.com:8080', false ],
 136                         [ '', 'https://test.com', 'http://test.com', false ],
 137                         [ '', 'http://test.com', 'https://test.com', false ],
 138                         [ 'http://', 'http://test.com', 'http://test.com', false ],
 139                         [ null, 'http://www.test.com', 'http://www.test.com:80', false ],
 140                         [ null, 'http://www.test.com:80', 'http://www.test.com', false ],
 141                         [ null, 'http://*.test.com:80', 'http://www.test.com', false ],
 142                         [ '', 'https://gerrit.wikimedia.org/r/#/XXX/status:open,n,z',
 143                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z', false ],
 144                         [ '', 'https://*.wikimedia.org/r/#/q/status:open,n,z',
 145                                 'https://gerrit.wikimedia.org/r/#/XXX/status:open,n,z', false ],
 146                         [ 'mailto:', '@test.com', '@abc.test.com', false ],
 147                         [ 'mailto:', 'mail@test.com', 'mail2@test.com', false ],
 148                         [ '', 'mailto:mail@test.com', 'mail2@test.com', false ],
 149                         [ '', 'mailto:@test.com', '@abc.test.com', false ],
 150                         [ 'ftp://', '*.co', 'ftp://www.co.uk', false ],
 151                         [ 'ftp://', '*.co', 'ftp://www.co.m', false ],
 152                         [ 'ftp://', '*.co/dir/', 'ftp://www.co/dir2/', false ],
 153                         [ 'ftp://', 'www.co/dir/', 'ftp://www.co/dir2/', false ],
 154                         [ 'ftp://', 'test.com/dir/', 'ftp://test.com/', false ],
 155                         [ '', 'http://test.com:8080/dir/', 'http://test.com:808/dir/', false ],
 156                         [ '', 'http://test.com/dir/index.html', 'http://test.com/dir/index.php', false ],
 157
 158                         // These are false positives too and ideally shouldn't match, but that
 159                         // would require using regexes and RLIKE instead of LIKE
 160                         // [ null, 'http://*.test.com', 'http://www.test.com:80', false ],
 161                         // [ '', 'https://*.wikimedia.org/r/#/q/status:open,n,z',
 162                         //      'https://gerrit.wikimedia.org/XXX/r/#/q/status:open,n,z', false ],
 163                 ];
 164         }
 165
 166         /**
 167          * testMakeLikeArrayWithValidPatterns()
 168          *
 169          * Tests whether the LIKE clause produced by LinkFilter::makeLikeArray($pattern, $protocol)
 170          * will find one of the URL indexes produced by wfMakeUrlIndexes($url)
 171          *
 172          * @dataProvider provideValidPatterns
 173          *
 174          * @param string $protocol Protocol, e.g. 'http://' or 'mailto:'
 175          * @param string $pattern Search pattern to feed to LinkFilter::makeLikeArray
 176          * @param string $url URL to feed to wfMakeUrlIndexes
 177          * @param bool $shouldBeFound Should the URL be found? (defaults true)
 178          */
 179         function testMakeLikeArrayWithValidPatterns( $protocol, $pattern, $url, $shouldBeFound = true ) {
 180                 $indexes = wfMakeUrlIndexes( $url );
 181                 $likeArray = LinkFilter::makeLikeArray( $pattern, $protocol );
 182
 183                 $this->assertTrue( $likeArray !== false,
 184                         "LinkFilter::makeLikeArray('$pattern', '$protocol') returned false on a valid pattern"
 185                 );
 186
 187                 $regex = $this->createRegexFromLIKE( $likeArray );
 188                 $debugmsg = "Regex: '" . $regex . "'\n";
 189                 $debugmsg .= count( $indexes ) . " index(es) created by wfMakeUrlIndexes():\n";
 190
 191                 $matches = 0;
 192
 193                 foreach ( $indexes as $index ) {
 194                         $matches += preg_match( $regex, $index );
 195                         $debugmsg .= "\t'$index'\n";
 196                 }
 197
 198                 if ( $shouldBeFound ) {
 199                         $this->assertTrue(
 200                                 $matches > 0,
 201                                 "Search pattern '$protocol$pattern' does not find url '$url' \n$debugmsg"
 202                         );
 203                 } else {
 204                         $this->assertFalse(
 205                                 $matches > 0,
 206                                 "Search pattern '$protocol$pattern' should not find url '$url' \n$debugmsg"
 207                         );
 208                 }
 209         }
 210
 211         /**
 212          * provideInvalidPatterns()
 213          *
 214          * @return array
 215          */
 216         public static function provideInvalidPatterns() {
 217                 return [
 218                         [ '' ],
 219                         [ '*' ],
 220                         [ 'http://*' ],
 221                         [ 'http://*/' ],
 222                         [ 'http://*/dir/file' ],
 223                         [ 'test.*.com' ],
 224                         [ 'http://test.*.com' ],
 225                         [ 'test.*.com' ],
 226                         [ 'http://*.test.*' ],
 227                         [ 'http://*test.com' ],
 228                         [ 'https://*' ],
 229                         [ '*://test.com' ],
 230                         [ 'mailto:name:pass@t*est.com' ],
 231                         [ 'http://*:888/' ],
 232                         [ '*http://' ],
 233                         [ 'test.com/*/index' ],
 234                         [ 'test.com/dir/index?arg=*' ],
 235                 ];
 236         }
 237
 238         /**
 239          * testMakeLikeArrayWithInvalidPatterns()
 240          *
 241          * Tests whether LinkFilter::makeLikeArray($pattern) will reject invalid search patterns
 242          *
 243          * @dataProvider provideInvalidPatterns
 244          *
 245          * @param string $pattern Invalid search pattern
 246          */
 247         function testMakeLikeArrayWithInvalidPatterns( $pattern ) {
 248                 $this->assertFalse(
 249                         LinkFilter::makeLikeArray( $pattern ),
 250                         "'$pattern' is not a valid pattern and should be rejected"
 251                 );
 252         }
 253
 254 }