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