tests/phpunit/includes/SanitizerTest.php

   1 <?php
   2
   3 /**
   4  * @todo Tests covering decodeCharReferences can be refactored into a single
   5  * method and dataprovider.
   6  */
   7 class SanitizerTest extends MediaWikiTestCase {
   8
   9         protected function setUp() {
  10                 parent::setUp();
  11
  12                 AutoLoader::loadClass( 'Sanitizer' );
  13         }
  14
  15         /**
  16          * @covers Sanitizer::decodeCharReferences
  17          */
  18         public function testDecodeNamedEntities() {
  19                 $this->assertEquals(
  20                         "\xc3\xa9cole",
  21                         Sanitizer::decodeCharReferences( '&eacute;cole' ),
  22                         'decode named entities'
  23                 );
  24         }
  25
  26         /**
  27          * @covers Sanitizer::decodeCharReferences
  28          */
  29         public function testDecodeNumericEntities() {
  30                 $this->assertEquals(
  31                         "\xc4\x88io bonas dans l'\xc3\xa9cole!",
  32                         Sanitizer::decodeCharReferences( "&#x108;io bonas dans l'&#233;cole!" ),
  33                         'decode numeric entities'
  34                 );
  35         }
  36
  37         /**
  38          * @covers Sanitizer::decodeCharReferences
  39          */
  40         public function testDecodeMixedEntities() {
  41                 $this->assertEquals(
  42                         "\xc4\x88io bonas dans l'\xc3\xa9cole!",
  43                         Sanitizer::decodeCharReferences( "&#x108;io bonas dans l'&eacute;cole!" ),
  44                         'decode mixed numeric/named entities'
  45                 );
  46         }
  47
  48         /**
  49          * @covers Sanitizer::decodeCharReferences
  50          */
  51         public function testDecodeMixedComplexEntities() {
  52                 $this->assertEquals(
  53                         "\xc4\x88io bonas dans l'\xc3\xa9cole! (mais pas &#x108;io dans l'&eacute;cole)",
  54                         Sanitizer::decodeCharReferences(
  55                                 "&#x108;io bonas dans l'&eacute;cole! (mais pas &amp;#x108;io dans l'&#38;eacute;cole)"
  56                         ),
  57                         'decode mixed complex entities'
  58                 );
  59         }
  60
  61         /**
  62          * @covers Sanitizer::decodeCharReferences
  63          */
  64         public function testInvalidAmpersand() {
  65                 $this->assertEquals(
  66                         'a & b',
  67                         Sanitizer::decodeCharReferences( 'a & b' ),
  68                         'Invalid ampersand'
  69                 );
  70         }
  71
  72         /**
  73          * @covers Sanitizer::decodeCharReferences
  74          */
  75         public function testInvalidEntities() {
  76                 $this->assertEquals(
  77                         '&foo;',
  78                         Sanitizer::decodeCharReferences( '&foo;' ),
  79                         'Invalid named entity'
  80                 );
  81         }
  82
  83         /**
  84          * @covers Sanitizer::decodeCharReferences
  85          */
  86         public function testInvalidNumberedEntities() {
  87                 $this->assertEquals(
  88                         UTF8_REPLACEMENT,
  89                         Sanitizer::decodeCharReferences( "&#88888888888888;" ),
  90                         'Invalid numbered entity'
  91                 );
  92         }
  93
  94         /**
  95          * @covers Sanitizer::removeHTMLtags
  96          * @dataProvider provideHtml5Tags
  97          *
  98          * @param string $tag Name of an HTML5 element (ie: 'video')
  99          * @param bool $escaped Whether sanitizer let the tag in or escape it (ie: '&lt;video&gt;')
 100          */
 101         public function testRemovehtmltagsOnHtml5Tags( $tag, $escaped ) {
 102                 $this->setMwGlobals( array(
 103                         'wgUseTidy' => false
 104                 ) );
 105
 106                 if ( $escaped ) {
 107                         $this->assertEquals( "&lt;$tag&gt;",
 108                                 Sanitizer::removeHTMLtags( "<$tag>" )
 109                         );
 110                 } else {
 111                         $this->assertEquals( "<$tag></$tag>\n",
 112                                 Sanitizer::removeHTMLtags( "<$tag>" )
 113                         );
 114                 }
 115         }
 116
 117         /**
 118          * Provide HTML5 tags
 119          */
 120         public static function provideHtml5Tags() {
 121                 $ESCAPED = true; # We want tag to be escaped
 122                 $VERBATIM = false; # We want to keep the tag
 123                 return array(
 124                         array( 'data', $VERBATIM ),
 125                         array( 'mark', $VERBATIM ),
 126                         array( 'time', $VERBATIM ),
 127                         array( 'video', $ESCAPED ),
 128                 );
 129         }
 130
 131         function dataRemoveHTMLtags() {
 132                 return array(
 133                         // former testSelfClosingTag
 134                         array(
 135                                 '<div>Hello world</div />',
 136                                 '<div>Hello world</div>',
 137                                 'Self-closing closing div'
 138                         ),
 139                         // Make sure special nested HTML5 semantics are not broken
 140                         // http://www.whatwg.org/html/text-level-semantics.html#the-kbd-element
 141                         array(
 142                                 '<kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd>',
 143                                 '<kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd>',
 144                                 'Nested <kbd>.'
 145                         ),
 146                         // http://www.whatwg.org/html/text-level-semantics.html#the-sub-and-sup-elements
 147                         array(
 148                                 '<var>x<sub><var>i</var></sub></var>, <var>y<sub><var>i</var></sub></var>',
 149                                 '<var>x<sub><var>i</var></sub></var>, <var>y<sub><var>i</var></sub></var>',
 150                                 'Nested <var>.'
 151                         ),
 152                         // http://www.whatwg.org/html/text-level-semantics.html#the-dfn-element
 153                         array(
 154                                 '<dfn><abbr title="Garage Door Opener">GDO</abbr></dfn>',
 155                                 '<dfn><abbr title="Garage Door Opener">GDO</abbr></dfn>',
 156                                 '<abbr> inside <dfn>',
 157                         ),
 158                 );
 159         }
 160
 161         /**
 162          * @dataProvider dataRemoveHTMLtags
 163          * @covers Sanitizer::removeHTMLtags
 164          */
 165         public function testRemoveHTMLtags( $input, $output, $msg = null ) {
 166                 $GLOBALS['wgUseTidy'] = false;
 167                 $this->assertEquals( $output, Sanitizer::removeHTMLtags( $input ), $msg );
 168         }
 169
 170         /**
 171          * @dataProvider provideTagAttributesToDecode
 172          * @covers Sanitizer::decodeTagAttributes
 173          */
 174         public function testDecodeTagAttributes( $expected, $attributes, $message = '' ) {
 175                 $this->assertEquals( $expected,
 176                         Sanitizer::decodeTagAttributes( $attributes ),
 177                         $message
 178                 );
 179         }
 180
 181         public static function provideTagAttributesToDecode() {
 182                 return array(
 183                         array( array( 'foo' => 'bar' ), 'foo=bar', 'Unquoted attribute' ),
 184                         array( array( 'foo' => 'bar' ), '    foo   =   bar    ', 'Spaced attribute' ),
 185                         array( array( 'foo' => 'bar' ), 'foo="bar"', 'Double-quoted attribute' ),
 186                         array( array( 'foo' => 'bar' ), 'foo=\'bar\'', 'Single-quoted attribute' ),
 187                         array(
 188                                 array( 'foo' => 'bar', 'baz' => 'foo' ),
 189                                 'foo=\'bar\'   baz="foo"',
 190                                 'Several attributes'
 191                         ),
 192                         array(
 193                                 array( 'foo' => 'bar', 'baz' => 'foo' ),
 194                                 'foo=\'bar\'   baz="foo"',
 195                                 'Several attributes'
 196                         ),
 197                         array(
 198                                 array( 'foo' => 'bar', 'baz' => 'foo' ),
 199                                 'foo=\'bar\'   baz="foo"',
 200                                 'Several attributes'
 201                         ),
 202                         array( array( ':foo' => 'bar' ), ':foo=\'bar\'', 'Leading :' ),
 203                         array( array( '_foo' => 'bar' ), '_foo=\'bar\'', 'Leading _' ),
 204                         array( array( 'foo' => 'bar' ), 'Foo=\'bar\'', 'Leading capital' ),
 205                         array( array( 'foo' => 'BAR' ), 'FOO=BAR', 'Attribute keys are normalized to lowercase' ),
 206
 207                         # Invalid beginning
 208                         array( array(), '-foo=bar', 'Leading - is forbidden' ),
 209                         array( array(), '.foo=bar', 'Leading . is forbidden' ),
 210                         array( array( 'foo-bar' => 'bar' ), 'foo-bar=bar', 'A - is allowed inside the attribute' ),
 211                         array( array( 'foo-' => 'bar' ), 'foo-=bar', 'A - is allowed inside the attribute' ),
 212                         array( array( 'foo.bar' => 'baz' ), 'foo.bar=baz', 'A . is allowed inside the attribute' ),
 213                         array( array( 'foo.' => 'baz' ), 'foo.=baz', 'A . is allowed as last character' ),
 214                         array( array( 'foo6' => 'baz' ), 'foo6=baz', 'Numbers are allowed' ),
 215
 216                         # This bit is more relaxed than XML rules, but some extensions use
 217                         # it, like ProofreadPage (see bug 27539)
 218                         array( array( '1foo' => 'baz' ), '1foo=baz', 'Leading numbers are allowed' ),
 219                         array( array(), 'foo$=baz', 'Symbols are not allowed' ),
 220                         array( array(), 'foo@=baz', 'Symbols are not allowed' ),
 221                         array( array(), 'foo~=baz', 'Symbols are not allowed' ),
 222                         array(
 223                                 array( 'foo' => '1[#^`*%w/(' ),
 224                                 'foo=1[#^`*%w/(',
 225                                 'All kind of characters are allowed as values'
 226                         ),
 227                         array(
 228                                 array( 'foo' => '1[#^`*%\'w/(' ),
 229                                 'foo="1[#^`*%\'w/("',
 230                                 'Double quotes are allowed if quoted by single quotes'
 231                         ),
 232                         array(
 233                                 array( 'foo' => '1[#^`*%"w/(' ),
 234                                 'foo=\'1[#^`*%"w/(\'',
 235                                 'Single quotes are allowed if quoted by double quotes'
 236                         ),
 237                         array( array( 'foo' => '&"' ), 'foo=&amp;&quot;', 'Special chars can be provided as entities' ),
 238                         array( array( 'foo' => '&foobar;' ), 'foo=&foobar;', 'Entity-like items are accepted' ),
 239                 );
 240         }
 241
 242         /**
 243          * @dataProvider provideDeprecatedAttributes
 244          * @covers Sanitizer::fixTagAttributes
 245          */
 246         public function testDeprecatedAttributesUnaltered( $inputAttr, $inputEl, $message = '' ) {
 247                 $this->assertEquals( " $inputAttr",
 248                         Sanitizer::fixTagAttributes( $inputAttr, $inputEl ),
 249                         $message
 250                 );
 251         }
 252
 253         public static function provideDeprecatedAttributes() {
 254                 /** array( <attribute>, <element>, [message] ) */
 255                 return array(
 256                         array( 'clear="left"', 'br' ),
 257                         array( 'clear="all"', 'br' ),
 258                         array( 'width="100"', 'td' ),
 259                         array( 'nowrap="true"', 'td' ),
 260                         array( 'nowrap=""', 'td' ),
 261                         array( 'align="right"', 'td' ),
 262                         array( 'align="center"', 'table' ),
 263                         array( 'align="left"', 'tr' ),
 264                         array( 'align="center"', 'div' ),
 265                         array( 'align="left"', 'h1' ),
 266                         array( 'align="left"', 'p' ),
 267                 );
 268         }
 269
 270         /**
 271          * @dataProvider provideCssCommentsFixtures
 272          * @covers Sanitizer::checkCss
 273          */
 274         public function testCssCommentsChecking( $expected, $css, $message = '' ) {
 275                 $this->assertEquals( $expected,
 276                         Sanitizer::checkCss( $css ),
 277                         $message
 278                 );
 279         }
 280
 281         public static function provideCssCommentsFixtures() {
 282                 /** array( <expected>, <css>, [message] ) */
 283                 return array(
 284                         // Valid comments spanning entire input
 285                         array( '/**/', '/**/' ),
 286                         array( '/* comment */', '/* comment */' ),
 287                         // Weird stuff
 288                         array( ' ', '/****/' ),
 289                         array( ' ', '/* /* */' ),
 290                         array( 'display: block;', "display:/* foo */block;" ),
 291                         array( 'display: block;', "display:\\2f\\2a foo \\2a\\2f block;",
 292                                 'Backslash-escaped comments must be stripped (bug 28450)' ),
 293                         array( '', '/* unfinished comment structure',
 294                                 'Remove anything after a comment-start token' ),
 295                         array( '', "\\2f\\2a unifinished comment'",
 296                                 'Remove anything after a backslash-escaped comment-start token' ),
 297                         array(
 298                                 '/* insecure input */',
 299                                 'filter: progid:DXImageTransform.Microsoft.AlphaImageLoader'
 300                                         . '(src=\'asdf.png\',sizingMethod=\'scale\');'
 301                         ),
 302                         array(
 303                                 '/* insecure input */',
 304                                 '-ms-filter: "progid:DXImageTransform.Microsoft.AlphaImageLoader'
 305                                         . '(src=\'asdf.png\',sizingMethod=\'scale\')";'
 306                         ),
 307                         array( '/* insecure input */', 'width: expression(1+1);' ),
 308                         array( '/* insecure input */', 'background-image: image(asdf.png);' ),
 309                         array( '/* insecure input */', 'background-image: -webkit-image(asdf.png);' ),
 310                         array( '/* insecure input */', 'background-image: -moz-image(asdf.png);' ),
 311                         array( '/* insecure input */', 'background-image: image-set("asdf.png" 1x, "asdf.png" 2x);' ),
 312                         array(
 313                                 '/* insecure input */',
 314                                 'background-image: -webkit-image-set("asdf.png" 1x, "asdf.png" 2x);'
 315                         ),
 316                         array(
 317                                 '/* insecure input */',
 318                                 'background-image: -moz-image-set("asdf.png" 1x, "asdf.png" 2x);'
 319                         ),
 320                 );
 321         }
 322
 323         /**
 324          * Test for support or lack of support for specific attributes in the attribute whitelist.
 325          */
 326         public static function provideAttributeSupport() {
 327                 /** array( <attributes>, <expected>, <message> ) */
 328                 return array(
 329                         array(
 330                                 'div',
 331                                 ' role="presentation"',
 332                                 ' role="presentation"',
 333                                 'Support for WAI-ARIA\'s role="presentation".'
 334                         ),
 335                         array( 'div', ' role="main"', '', "Other WAI-ARIA roles are currently not supported." ),
 336                 );
 337         }
 338
 339         /**
 340          * @dataProvider provideAttributeSupport
 341          * @covers Sanitizer::fixTagAttributes
 342          */
 343         public function testAttributeSupport( $tag, $attributes, $expected, $message ) {
 344                 $this->assertEquals( $expected,
 345                         Sanitizer::fixTagAttributes( $attributes, $tag ),
 346                         $message
 347                 );
 348         }
 349 }