tests/phpunit/includes/LinkFilterTest.php

   1 <?php
   2
   3 /**
   4  * @group Database
   5  */
   6 class LinkFilterTest extends MediaWikiLangTestCase {
   7
   8         protected function setUp() {
   9
  10                 parent::setUp();
  11
  12                 $this->setMwGlobals( 'wgUrlProtocols', array(
  13                         'http://',
  14                         'https://',
  15                         'ftp://',
  16                         'irc://',
  17                         'ircs://',
  18                         'gopher://',
  19                         'telnet://',
  20                         'nntp://',
  21                         'worldwind://',
  22                         'mailto:',
  23                         'news:',
  24                         'svn://',
  25                         'git://',
  26                         'mms://',
  27                         '//',
  28                 ) );
  29
  30         }
  31
  32         /**
  33          * createRegexFromLike($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
  42                 $regex = '!^';
  43
  44                 foreach ( $like as $item ) {
  45
  46                         if ( $item instanceof LikeMatch ) {
  47                                 if ( $item->toString() == '%' ) {
  48                                         $regex .= '.*';
  49                                 } elseif ( $item->toString() == '_' ) {
  50                                         $regex .= '.';
  51                                 }
  52                         } else {
  53                                 $regex .= preg_quote( $item, '!' );
  54                         }
  55
  56                 }
  57
  58                 $regex .= '$!';
  59
  60                 return $regex;
  61
  62         }
  63
  64         /**
  65          * provideValidPatterns()
  66          *
  67          * @return array
  68          */
  69         public static function provideValidPatterns() {
  70
  71                 return array(
  72                         // Protocol, Search pattern, URL which matches the pattern
  73                         array( 'http://', '*.test.com', 'http://www.test.com' ),
  74                         array( 'http://', 'test.com:8080/dir/file', 'http://name:pass@test.com:8080/dir/file' ),
  75                         array( 'https://', '*.com', 'https://s.s.test..com:88/dir/file?a=1&b=2' ),
  76                         array( 'https://', '*.com', 'https://name:pass@secure.com/index.html' ),
  77                         array( 'http://', 'name:pass@test.com', 'http://test.com' ),
  78                         array( 'http://', 'test.com', 'http://name:pass@test.com' ),
  79                         array( 'http://', '*.test.com', 'http://a.b.c.test.com/dir/dir/file?a=6'),
  80                         array( null, 'http://*.test.com', 'http://www.test.com' ),
  81                         array( 'mailto:', 'name@mail.test123.com', 'mailto:name@mail.test123.com' ),
  82                         array( '',
  83                                 'http://name:pass@www.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                         ),
  86                         array( '', '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                         array( '', 'http://name:wrongpass@*.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]',
  89                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ),
  90                         array( 'http://', 'name:pass@*.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg',
  91                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ),
  92                         array( '', 'http://name:pass@www.test.com:12345',
  93                                 'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg' ),
  94                         array( 'ftp://', 'user:pass@ftp.test.com:1233/home/user/file;type=efw',
  95                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ),
  96                         array( null, 'ftp://otheruser:otherpass@ftp.test.com:1233/home/user/file;type=',
  97                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ),
  98                         array( null, 'ftp://@ftp.test.com:1233/home/user/file;type=',
  99                                 'ftp://user:pass@ftp.test.com:1233/home/user/file;type=efw' ),
 100                         array( null, 'ftp://ftp.test.com/',
 101                                 'ftp://user:pass@ftp.test.com/home/user/file;type=efw' ),
 102                         array( null, 'ftp://ftp.test.com/',
 103                                 'ftp://user:pass@ftp.test.com/home/user/file;type=efw' ),
 104                         array( null, 'ftp://*.test.com:222/',
 105                                 'ftp://user:pass@ftp.test.com:222/home' ),
 106                         array( 'irc://', '*.myserver:6667/', 'irc://test.myserver:6667/' ),
 107                         array( 'irc://', 'name:pass@*.myserver/', 'irc://test.myserver:6667/' ),
 108                         array( 'irc://', 'name:pass@*.myserver/', 'irc://other:@test.myserver:6667/' ),
 109                         array( '', 'irc://test/name,string,abc?msg=t', 'irc://test/name,string,abc?msg=test' ),
 110                         array( '', 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z',
 111                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z' ),
 112                         array( '', 'https://gerrit.wikimedia.org',
 113                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z' ),
 114                         array( 'mailto:', '*.test.com', 'mailto:name@pop3.test.com' ),
 115                         array( 'mailto:', 'test.com', 'mailto:name@test.com' ),
 116                         array( 'news:', 'test.1234afc@news.test.com', 'news:test.1234afc@news.test.com' ),
 117                         array( 'news:', '*.test.com', 'news:test.1234afc@news.test.com' ),
 118                         array( '', 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com',
 119                                 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com' ),
 120                         array( '', 'news:*.aol.com',
 121                                 'news:4df8kh$iagfewewf(at)newsbf02aaa.news.aol.com' ),
 122                         array( '', 'git://github.com/prwef/abc-def.git', 'git://github.com/prwef/abc-def.git' ),
 123                         array( 'git://', 'github.com/', 'git://github.com/prwef/abc-def.git' ),
 124                         array( 'git://', '*.github.com/', 'git://a.b.c.d.e.f.github.com/prwef/abc-def.git' ),
 125                         array( '', 'gopher://*.test.com/', 'gopher://gopher.test.com/0/v2/vstat'),
 126                         array( 'telnet://', '*.test.com', 'telnet://shell.test.com/~home/'),
 127
 128                         //
 129                         // The following only work in PHP >= 5.3.7, due to a bug in parse_url which eats
 130                         // the path from the url (https://bugs.php.net/bug.php?id=54180)
 131                         //
 132                         // array( '', 'http://test.com', 'http://test.com/index?arg=1' ),
 133                         // array( 'http://', '*.test.com', 'http://www.test.com/index?arg=1' ),
 134                         // array( '' ,
 135                         //    'http://xx23124:__ffdfdef__@www.test.com:12345/dir' ,
 136                         //    'http://name:pass@www.test.com:12345/dir/dir/file.xyz.php#__se__?arg1=_&arg2[]=4rtg'
 137                         // ),
 138                         //
 139
 140                         //
 141                         // Tests for false positives
 142                         //
 143                         array( 'http://', 'test.com', 'http://www.test.com', false ),
 144                         array( 'http://', 'www1.test.com', 'http://www.test.com', false ),
 145                         array( 'http://', '*.test.com', 'http://www.test.t.com', false ),
 146                         array( '', 'http://test.com:8080', 'http://www.test.com:8080', false ),
 147                         array( '', 'https://test.com', 'http://test.com', false ),
 148                         array( '', 'http://test.com', 'https://test.com', false ),
 149                         array( 'http://', 'http://test.com', 'http://test.com', false ),
 150                         array( null, 'http://www.test.com', 'http://www.test.com:80', false ),
 151                         array( null, 'http://www.test.com:80', 'http://www.test.com', false ),
 152                         array( null, 'http://*.test.com:80', 'http://www.test.com', false ),
 153                         array( '', 'https://gerrit.wikimedia.org/r/#/XXX/status:open,n,z',
 154                                 'https://gerrit.wikimedia.org/r/#/q/status:open,n,z', false ),
 155                         array( '', 'https://*.wikimedia.org/r/#/q/status:open,n,z',
 156                                 'https://gerrit.wikimedia.org/r/#/XXX/status:open,n,z', false ),
 157                         array( 'mailto:', '@test.com', '@abc.test.com', false ),
 158                         array( 'mailto:', 'mail@test.com', 'mail2@test.com', false ),
 159                         array( '', 'mailto:mail@test.com', 'mail2@test.com', false ),
 160                         array( '', 'mailto:@test.com', '@abc.test.com', false ),
 161                         array( 'ftp://', '*.co', 'ftp://www.co.uk', false ),
 162                         array( 'ftp://', '*.co', 'ftp://www.co.m', false ),
 163                         array( 'ftp://', '*.co/dir/', 'ftp://www.co/dir2/', false ),
 164                         array( 'ftp://', 'www.co/dir/', 'ftp://www.co/dir2/', false ),
 165                         array( 'ftp://', 'test.com/dir/', 'ftp://test.com/', false ),
 166                         array( '', 'http://test.com:8080/dir/', 'http://test.com:808/dir/', false ),
 167                         array( '', 'http://test.com/dir/index.html', 'http://test.com/dir/index.php', false ),
 168
 169                         //
 170                         // These are false positives too and ideally shouldn't match, but that
 171                         // would require using regexes and RLIKE instead of LIKE
 172                         //
 173                         // array( null, 'http://*.test.com', 'http://www.test.com:80', false ),
 174                         // array( '', 'https://*.wikimedia.org/r/#/q/status:open,n,z',
 175                         //      'https://gerrit.wikimedia.org/XXX/r/#/q/status:open,n,z', false ),
 176                 );
 177
 178         }
 179
 180         /**
 181          * testMakeLikeArrayWithValidPatterns()
 182          *
 183          * Tests whether the LIKE clause produced by LinkFilter::makeLikeArray($pattern, $protocol)
 184          * will find one of the URL indexes produced by wfMakeUrlIndexes($url)
 185          *
 186          * @dataProvider provideValidPatterns
 187          *
 188          * @param string $protocol Protocol, e.g. 'http://' or 'mailto:'
 189          * @param string $pattern Search pattern to feed to LinkFilter::makeLikeArray
 190          * @param string $url URL to feed to wfMakeUrlIndexes
 191          * @param bool $shouldBeFound Should the URL be found? (defaults true)
 192          */
 193         function testMakeLikeArrayWithValidPatterns( $protocol, $pattern, $url, $shouldBeFound = true ) {
 194
 195                 $indexes = wfMakeUrlIndexes( $url );
 196                 $likeArray = LinkFilter::makeLikeArray( $pattern, $protocol );
 197
 198                 $this->assertTrue( $likeArray !== false,
 199                         "LinkFilter::makeLikeArray('$pattern', '$protocol') returned false on a valid pattern"
 200                 );
 201
 202                 $regex = $this->createRegexFromLIKE( $likeArray );
 203                 $debugmsg = "Regex: '" . $regex . "'\n";
 204                 $debugmsg .= count( $indexes ) . " index(es) created by wfMakeUrlIndexes():\n";
 205
 206                 $matches = 0;
 207
 208                 foreach ( $indexes as $index ) {
 209                         $matches += preg_match( $regex, $index );
 210                         $debugmsg .= "\t'$index'\n";
 211                 }
 212
 213                 if ( $shouldBeFound ) {
 214                         $this->assertTrue(
 215                                 $matches > 0,
 216                                 "Search pattern '$protocol$pattern' does not find url '$url' \n$debugmsg"
 217                         );
 218                 } else {
 219                         $this->assertFalse(
 220                                 $matches > 0,
 221                                 "Search pattern '$protocol$pattern' should not find url '$url' \n$debugmsg"
 222                         );
 223                 }
 224
 225         }
 226
 227         /**
 228          * provideInvalidPatterns()
 229          *
 230          * @return array
 231          */
 232         public static function provideInvalidPatterns() {
 233
 234                 return array(
 235                         array( '' ),
 236                         array( '*' ),
 237                         array( 'http://*' ),
 238                         array( 'http://*/' ),
 239                         array( 'http://*/dir/file' ),
 240                         array( 'test.*.com' ),
 241                         array( 'http://test.*.com' ),
 242                         array( 'test.*.com' ),
 243                         array( 'http://*.test.*' ),
 244                         array( 'http://*test.com' ),
 245                         array( 'https://*' ),
 246                         array( '*://test.com'),
 247                         array( 'mailto:name:pass@t*est.com' ),
 248                         array( 'http://*:888/'),
 249                         array( '*http://'),
 250                         array( 'test.com/*/index' ),
 251                         array( 'test.com/dir/index?arg=*' ),
 252                 );
 253
 254         }
 255
 256         /**
 257          * testMakeLikeArrayWithInvalidPatterns()
 258          *
 259          * Tests whether LinkFilter::makeLikeArray($pattern) will reject invalid search patterns
 260          *
 261          * @dataProvider provideInvalidPatterns
 262          *
 263          * @param string $pattern Invalid search pattern
 264          */
 265         function testMakeLikeArrayWithInvalidPatterns( $pattern ) {
 266
 267                 $this->assertFalse(
 268                         LinkFilter::makeLikeArray( $pattern ),
 269                         "'$pattern' is not a valid pattern and should be rejected"
 270                 );
 271
 272         }
 273
 274 }