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 tearDown() {
  10                 MWTidy::destroySingleton();
  11                 parent::tearDown();
  12         }
  13
  14         /**
  15          * @covers Sanitizer::decodeCharReferences
  16          */
  17         public function testDecodeNamedEntities() {
  18                 $this->assertEquals(
  19                         "\xc3\xa9cole",
  20                         Sanitizer::decodeCharReferences( '&eacute;cole' ),
  21                         'decode named entities'
  22                 );
  23         }
  24
  25         /**
  26          * @covers Sanitizer::decodeCharReferences
  27          */
  28         public function testDecodeNumericEntities() {
  29                 $this->assertEquals(
  30                         "\xc4\x88io bonas dans l'\xc3\xa9cole!",
  31                         Sanitizer::decodeCharReferences( "&#x108;io bonas dans l'&#233;cole!" ),
  32                         'decode numeric entities'
  33                 );
  34         }
  35
  36         /**
  37          * @covers Sanitizer::decodeCharReferences
  38          */
  39         public function testDecodeMixedEntities() {
  40                 $this->assertEquals(
  41                         "\xc4\x88io bonas dans l'\xc3\xa9cole!",
  42                         Sanitizer::decodeCharReferences( "&#x108;io bonas dans l'&eacute;cole!" ),
  43                         'decode mixed numeric/named entities'
  44                 );
  45         }
  46
  47         /**
  48          * @covers Sanitizer::decodeCharReferences
  49          */
  50         public function testDecodeMixedComplexEntities() {
  51                 $this->assertEquals(
  52                         "\xc4\x88io bonas dans l'\xc3\xa9cole! (mais pas &#x108;io dans l'&eacute;cole)",
  53                         Sanitizer::decodeCharReferences(
  54                                 "&#x108;io bonas dans l'&eacute;cole! (mais pas &amp;#x108;io dans l'&#38;eacute;cole)"
  55                         ),
  56                         'decode mixed complex entities'
  57                 );
  58         }
  59
  60         /**
  61          * @covers Sanitizer::decodeCharReferences
  62          */
  63         public function testInvalidAmpersand() {
  64                 $this->assertEquals(
  65                         'a & b',
  66                         Sanitizer::decodeCharReferences( 'a & b' ),
  67                         'Invalid ampersand'
  68                 );
  69         }
  70
  71         /**
  72          * @covers Sanitizer::decodeCharReferences
  73          */
  74         public function testInvalidEntities() {
  75                 $this->assertEquals(
  76                         '&foo;',
  77                         Sanitizer::decodeCharReferences( '&foo;' ),
  78                         'Invalid named entity'
  79                 );
  80         }
  81
  82         /**
  83          * @covers Sanitizer::decodeCharReferences
  84          */
  85         public function testInvalidNumberedEntities() {
  86                 $this->assertEquals(
  87                         UtfNormal\Constants::UTF8_REPLACEMENT,
  88                         Sanitizer::decodeCharReferences( "&#88888888888888;" ),
  89                         'Invalid numbered entity'
  90                 );
  91         }
  92
  93         /**
  94          * @covers Sanitizer::removeHTMLtags
  95          * @dataProvider provideHtml5Tags
  96          *
  97          * @param string $tag Name of an HTML5 element (ie: 'video')
  98          * @param bool $escaped Whether sanitizer let the tag in or escape it (ie: '&lt;video&gt;')
  99          */
 100         public function testRemovehtmltagsOnHtml5Tags( $tag, $escaped ) {
 101                 MWTidy::setInstance( false );
 102
 103                 if ( $escaped ) {
 104                         $this->assertEquals( "&lt;$tag&gt;",
 105                                 Sanitizer::removeHTMLtags( "<$tag>" )
 106                         );
 107                 } else {
 108                         $this->assertEquals( "<$tag></$tag>\n",
 109                                 Sanitizer::removeHTMLtags( "<$tag>" )
 110                         );
 111                 }
 112         }
 113
 114         /**
 115          * Provide HTML5 tags
 116          */
 117         public static function provideHtml5Tags() {
 118                 $ESCAPED = true; # We want tag to be escaped
 119                 $VERBATIM = false; # We want to keep the tag
 120                 return array(
 121                         array( 'data', $VERBATIM ),
 122                         array( 'mark', $VERBATIM ),
 123                         array( 'time', $VERBATIM ),
 124                         array( 'video', $ESCAPED ),
 125                 );
 126         }
 127
 128         function dataRemoveHTMLtags() {
 129                 return array(
 130                         // former testSelfClosingTag
 131                         array(
 132                                 '<div>Hello world</div />',
 133                                 '<div>Hello world</div>',
 134                                 'Self-closing closing div'
 135                         ),
 136                         // Make sure special nested HTML5 semantics are not broken
 137                         // http://www.whatwg.org/html/text-level-semantics.html#the-kbd-element
 138                         array(
 139                                 '<kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd>',
 140                                 '<kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd>',
 141                                 'Nested <kbd>.'
 142                         ),
 143                         // http://www.whatwg.org/html/text-level-semantics.html#the-sub-and-sup-elements
 144                         array(
 145                                 '<var>x<sub><var>i</var></sub></var>, <var>y<sub><var>i</var></sub></var>',
 146                                 '<var>x<sub><var>i</var></sub></var>, <var>y<sub><var>i</var></sub></var>',
 147                                 'Nested <var>.'
 148                         ),
 149                         // http://www.whatwg.org/html/text-level-semantics.html#the-dfn-element
 150                         array(
 151                                 '<dfn><abbr title="Garage Door Opener">GDO</abbr></dfn>',
 152                                 '<dfn><abbr title="Garage Door Opener">GDO</abbr></dfn>',
 153                                 '<abbr> inside <dfn>',
 154                         ),
 155                 );
 156         }
 157
 158         /**
 159          * @dataProvider dataRemoveHTMLtags
 160          * @covers Sanitizer::removeHTMLtags
 161          */
 162         public function testRemoveHTMLtags( $input, $output, $msg = null ) {
 163                 MWTidy::setInstance( false );
 164                 $this->assertEquals( $output, Sanitizer::removeHTMLtags( $input ), $msg );
 165         }
 166
 167         /**
 168          * @dataProvider provideTagAttributesToDecode
 169          * @covers Sanitizer::decodeTagAttributes
 170          */
 171         public function testDecodeTagAttributes( $expected, $attributes, $message = '' ) {
 172                 $this->assertEquals( $expected,
 173                         Sanitizer::decodeTagAttributes( $attributes ),
 174                         $message
 175                 );
 176         }
 177
 178         public static function provideTagAttributesToDecode() {
 179                 return array(
 180                         array( array( 'foo' => 'bar' ), 'foo=bar', 'Unquoted attribute' ),
 181                         array( array( 'foo' => 'bar' ), '    foo   =   bar    ', 'Spaced attribute' ),
 182                         array( array( 'foo' => 'bar' ), 'foo="bar"', 'Double-quoted attribute' ),
 183                         array( array( 'foo' => 'bar' ), 'foo=\'bar\'', 'Single-quoted attribute' ),
 184                         array(
 185                                 array( 'foo' => 'bar', 'baz' => 'foo' ),
 186                                 'foo=\'bar\'   baz="foo"',
 187                                 'Several attributes'
 188                         ),
 189                         array(
 190                                 array( 'foo' => 'bar', 'baz' => 'foo' ),
 191                                 'foo=\'bar\'   baz="foo"',
 192                                 'Several attributes'
 193                         ),
 194                         array(
 195                                 array( 'foo' => 'bar', 'baz' => 'foo' ),
 196                                 'foo=\'bar\'   baz="foo"',
 197                                 'Several attributes'
 198                         ),
 199                         array( array( ':foo' => 'bar' ), ':foo=\'bar\'', 'Leading :' ),
 200                         array( array( '_foo' => 'bar' ), '_foo=\'bar\'', 'Leading _' ),
 201                         array( array( 'foo' => 'bar' ), 'Foo=\'bar\'', 'Leading capital' ),
 202                         array( array( 'foo' => 'BAR' ), 'FOO=BAR', 'Attribute keys are normalized to lowercase' ),
 203
 204                         # Invalid beginning
 205                         array( array(), '-foo=bar', 'Leading - is forbidden' ),
 206                         array( array(), '.foo=bar', 'Leading . is forbidden' ),
 207                         array( array( 'foo-bar' => 'bar' ), 'foo-bar=bar', 'A - is allowed inside the attribute' ),
 208                         array( array( 'foo-' => 'bar' ), 'foo-=bar', 'A - is allowed inside the attribute' ),
 209                         array( array( 'foo.bar' => 'baz' ), 'foo.bar=baz', 'A . is allowed inside the attribute' ),
 210                         array( array( 'foo.' => 'baz' ), 'foo.=baz', 'A . is allowed as last character' ),
 211                         array( array( 'foo6' => 'baz' ), 'foo6=baz', 'Numbers are allowed' ),
 212
 213                         # This bit is more relaxed than XML rules, but some extensions use
 214                         # it, like ProofreadPage (see bug 27539)
 215                         array( array( '1foo' => 'baz' ), '1foo=baz', 'Leading numbers are allowed' ),
 216                         array( array(), 'foo$=baz', 'Symbols are not allowed' ),
 217                         array( array(), 'foo@=baz', 'Symbols are not allowed' ),
 218                         array( array(), 'foo~=baz', 'Symbols are not allowed' ),
 219                         array(
 220                                 array( 'foo' => '1[#^`*%w/(' ),
 221                                 'foo=1[#^`*%w/(',
 222                                 'All kind of characters are allowed as values'
 223                         ),
 224                         array(
 225                                 array( 'foo' => '1[#^`*%\'w/(' ),
 226                                 'foo="1[#^`*%\'w/("',
 227                                 'Double quotes are allowed if quoted by single quotes'
 228                         ),
 229                         array(
 230                                 array( 'foo' => '1[#^`*%"w/(' ),
 231                                 'foo=\'1[#^`*%"w/(\'',
 232                                 'Single quotes are allowed if quoted by double quotes'
 233                         ),
 234                         array( array( 'foo' => '&"' ), 'foo=&amp;&quot;', 'Special chars can be provided as entities' ),
 235                         array( array( 'foo' => '&foobar;' ), 'foo=&foobar;', 'Entity-like items are accepted' ),
 236                 );
 237         }
 238
 239         /**
 240          * @dataProvider provideDeprecatedAttributes
 241          * @covers Sanitizer::fixTagAttributes
 242          */
 243         public function testDeprecatedAttributesUnaltered( $inputAttr, $inputEl, $message = '' ) {
 244                 $this->assertEquals( " $inputAttr",
 245                         Sanitizer::fixTagAttributes( $inputAttr, $inputEl ),
 246                         $message
 247                 );
 248         }
 249
 250         public static function provideDeprecatedAttributes() {
 251                 /** array( <attribute>, <element>, [message] ) */
 252                 return array(
 253                         array( 'clear="left"', 'br' ),
 254                         array( 'clear="all"', 'br' ),
 255                         array( 'width="100"', 'td' ),
 256                         array( 'nowrap="true"', 'td' ),
 257                         array( 'nowrap=""', 'td' ),
 258                         array( 'align="right"', 'td' ),
 259                         array( 'align="center"', 'table' ),
 260                         array( 'align="left"', 'tr' ),
 261                         array( 'align="center"', 'div' ),
 262                         array( 'align="left"', 'h1' ),
 263                         array( 'align="left"', 'p' ),
 264                 );
 265         }
 266
 267         /**
 268          * @dataProvider provideCssCommentsFixtures
 269          * @covers Sanitizer::checkCss
 270          */
 271         public function testCssCommentsChecking( $expected, $css, $message = '' ) {
 272                 $this->assertEquals( $expected,
 273                         Sanitizer::checkCss( $css ),
 274                         $message
 275                 );
 276         }
 277
 278         public static function provideCssCommentsFixtures() {
 279                 /** array( <expected>, <css>, [message] ) */
 280                 return array(
 281                         // Valid comments spanning entire input
 282                         array( '/**/', '/**/' ),
 283                         array( '/* comment */', '/* comment */' ),
 284                         // Weird stuff
 285                         array( ' ', '/****/' ),
 286                         array( ' ', '/* /* */' ),
 287                         array( 'display: block;', "display:/* foo */block;" ),
 288                         array( 'display: block;', "display:\\2f\\2a foo \\2a\\2f block;",
 289                                 'Backslash-escaped comments must be stripped (bug 28450)' ),
 290                         array( '', '/* unfinished comment structure',
 291                                 'Remove anything after a comment-start token' ),
 292                         array( '', "\\2f\\2a unifinished comment'",
 293                                 'Remove anything after a backslash-escaped comment-start token' ),
 294                         array(
 295                                 '/* insecure input */',
 296                                 'filter: progid:DXImageTransform.Microsoft.AlphaImageLoader'
 297                                         . '(src=\'asdf.png\',sizingMethod=\'scale\');'
 298                         ),
 299                         array(
 300                                 '/* insecure input */',
 301                                 '-ms-filter: "progid:DXImageTransform.Microsoft.AlphaImageLoader'
 302                                         . '(src=\'asdf.png\',sizingMethod=\'scale\')";'
 303                         ),
 304                         array( '/* insecure input */', 'width: expression(1+1);' ),
 305                         array( '/* insecure input */', 'background-image: image(asdf.png);' ),
 306                         array( '/* insecure input */', 'background-image: -webkit-image(asdf.png);' ),
 307                         array( '/* insecure input */', 'background-image: -moz-image(asdf.png);' ),
 308                         array( '/* insecure input */', 'background-image: image-set("asdf.png" 1x, "asdf.png" 2x);' ),
 309                         array(
 310                                 '/* insecure input */',
 311                                 'background-image: -webkit-image-set("asdf.png" 1x, "asdf.png" 2x);'
 312                         ),
 313                         array(
 314                                 '/* insecure input */',
 315                                 'background-image: -moz-image-set("asdf.png" 1x, "asdf.png" 2x);'
 316                         ),
 317                 );
 318         }
 319
 320         /**
 321          * Test for support or lack of support for specific attributes in the attribute whitelist.
 322          */
 323         public static function provideAttributeSupport() {
 324                 /** array( <attributes>, <expected>, <message> ) */
 325                 return array(
 326                         array(
 327                                 'div',
 328                                 ' role="presentation"',
 329                                 ' role="presentation"',
 330                                 'Support for WAI-ARIA\'s role="presentation".'
 331                         ),
 332                         array( 'div', ' role="main"', '', "Other WAI-ARIA roles are currently not supported." ),
 333                 );
 334         }
 335
 336         /**
 337          * @dataProvider provideAttributeSupport
 338          * @covers Sanitizer::fixTagAttributes
 339          */
 340         public function testAttributeSupport( $tag, $attributes, $expected, $message ) {
 341                 $this->assertEquals( $expected,
 342                         Sanitizer::fixTagAttributes( $attributes, $tag ),
 343                         $message
 344                 );
 345         }
 346
 347         /**
 348          * @dataProvider provideEscapeHtmlAllowEntities
 349          * @covers Sanitizer::escapeHtmlAllowEntities
 350          */
 351         public function testEscapeHtmlAllowEntities( $expected, $html ) {
 352                 $this->assertEquals(
 353                         $expected,
 354                         Sanitizer::escapeHtmlAllowEntities( $html )
 355                 );
 356         }
 357
 358         public static function provideEscapeHtmlAllowEntities() {
 359                 return array(
 360                         array( 'foo', 'foo' ),
 361                         array( 'a¡b', 'a&#161;b' ),
 362                         array( 'foo&#039;bar', "foo'bar" ),
 363                         array( '&lt;script&gt;foo&lt;/script&gt;', '<script>foo</script>' ),
 364                 );
 365         }
 366 }