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 [
 121                         [ 'data', $VERBATIM ],
 122                         [ 'mark', $VERBATIM ],
 123                         [ 'time', $VERBATIM ],
 124                         [ 'video', $ESCAPED ],
 125                 ];
 126         }
 127
 128         function dataRemoveHTMLtags() {
 129                 return [
 130                         // former testSelfClosingTag
 131                         [
 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                         // https://html.spec.whatwg.org/multipage/semantics.html#the-kbd-element
 138                         [
 139                                 '<kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd>',
 140                                 '<kbd><kbd>Shift</kbd>+<kbd>F3</kbd></kbd>',
 141                                 'Nested <kbd>.'
 142                         ],
 143                         // https://html.spec.whatwg.org/multipage/semantics.html#the-sub-and-sup-elements
 144                         [
 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                         // https://html.spec.whatwg.org/multipage/semantics.html#the-dfn-element
 150                         [
 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 [
 180                         [ [ 'foo' => 'bar' ], 'foo=bar', 'Unquoted attribute' ],
 181                         [ [ 'foo' => 'bar' ], '    foo   =   bar    ', 'Spaced attribute' ],
 182                         [ [ 'foo' => 'bar' ], 'foo="bar"', 'Double-quoted attribute' ],
 183                         [ [ 'foo' => 'bar' ], 'foo=\'bar\'', 'Single-quoted attribute' ],
 184                         [
 185                                 [ 'foo' => 'bar', 'baz' => 'foo' ],
 186                                 'foo=\'bar\'   baz="foo"',
 187                                 'Several attributes'
 188                         ],
 189                         [
 190                                 [ 'foo' => 'bar', 'baz' => 'foo' ],
 191                                 'foo=\'bar\'   baz="foo"',
 192                                 'Several attributes'
 193                         ],
 194                         [
 195                                 [ 'foo' => 'bar', 'baz' => 'foo' ],
 196                                 'foo=\'bar\'   baz="foo"',
 197                                 'Several attributes'
 198                         ],
 199                         [ [ ':foo' => 'bar' ], ':foo=\'bar\'', 'Leading :' ],
 200                         [ [ '_foo' => 'bar' ], '_foo=\'bar\'', 'Leading _' ],
 201                         [ [ 'foo' => 'bar' ], 'Foo=\'bar\'', 'Leading capital' ],
 202                         [ [ 'foo' => 'BAR' ], 'FOO=BAR', 'Attribute keys are normalized to lowercase' ],
 203
 204                         # Invalid beginning
 205                         [ [], '-foo=bar', 'Leading - is forbidden' ],
 206                         [ [], '.foo=bar', 'Leading . is forbidden' ],
 207                         [ [ 'foo-bar' => 'bar' ], 'foo-bar=bar', 'A - is allowed inside the attribute' ],
 208                         [ [ 'foo-' => 'bar' ], 'foo-=bar', 'A - is allowed inside the attribute' ],
 209                         [ [ 'foo.bar' => 'baz' ], 'foo.bar=baz', 'A . is allowed inside the attribute' ],
 210                         [ [ 'foo.' => 'baz' ], 'foo.=baz', 'A . is allowed as last character' ],
 211                         [ [ '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 T29539)
 215                         [ [ '1foo' => 'baz' ], '1foo=baz', 'Leading numbers are allowed' ],
 216                         [ [], 'foo$=baz', 'Symbols are not allowed' ],
 217                         [ [], 'foo@=baz', 'Symbols are not allowed' ],
 218                         [ [], 'foo~=baz', 'Symbols are not allowed' ],
 219                         [
 220                                 [ 'foo' => '1[#^`*%w/(' ],
 221                                 'foo=1[#^`*%w/(',
 222                                 'All kind of characters are allowed as values'
 223                         ],
 224                         [
 225                                 [ 'foo' => '1[#^`*%\'w/(' ],
 226                                 'foo="1[#^`*%\'w/("',
 227                                 'Double quotes are allowed if quoted by single quotes'
 228                         ],
 229                         [
 230                                 [ 'foo' => '1[#^`*%"w/(' ],
 231                                 'foo=\'1[#^`*%"w/(\'',
 232                                 'Single quotes are allowed if quoted by double quotes'
 233                         ],
 234                         [ [ 'foo' => '&"' ], 'foo=&amp;&quot;', 'Special chars can be provided as entities' ],
 235                         [ [ '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                 /** [ <attribute>, <element>, [message] ] */
 252                 return [
 253                         [ 'clear="left"', 'br' ],
 254                         [ 'clear="all"', 'br' ],
 255                         [ 'width="100"', 'td' ],
 256                         [ 'nowrap="true"', 'td' ],
 257                         [ 'nowrap=""', 'td' ],
 258                         [ 'align="right"', 'td' ],
 259                         [ 'align="center"', 'table' ],
 260                         [ 'align="left"', 'tr' ],
 261                         [ 'align="center"', 'div' ],
 262                         [ 'align="left"', 'h1' ],
 263                         [ '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                 /** [ <expected>, <css>, [message] ] */
 280                 return [
 281                         // Valid comments spanning entire input
 282                         [ '/**/', '/**/' ],
 283                         [ '/* comment */', '/* comment */' ],
 284                         // Weird stuff
 285                         [ ' ', '/****/' ],
 286                         [ ' ', '/* /* */' ],
 287                         [ 'display: block;', "display:/* foo */block;" ],
 288                         [ 'display: block;', "display:\\2f\\2a foo \\2a\\2f block;",
 289                                 'Backslash-escaped comments must be stripped (T30450)' ],
 290                         [ '', '/* unfinished comment structure',
 291                                 'Remove anything after a comment-start token' ],
 292                         [ '', "\\2f\\2a unifinished comment'",
 293                                 'Remove anything after a backslash-escaped comment-start token' ],
 294                         [
 295                                 '/* insecure input */',
 296                                 'filter: progid:DXImageTransform.Microsoft.AlphaImageLoader'
 297                                         . '(src=\'asdf.png\',sizingMethod=\'scale\');'
 298                         ],
 299                         [
 300                                 '/* insecure input */',
 301                                 '-ms-filter: "progid:DXImageTransform.Microsoft.AlphaImageLoader'
 302                                         . '(src=\'asdf.png\',sizingMethod=\'scale\')";'
 303                         ],
 304                         [ '/* insecure input */', 'width: expression(1+1);' ],
 305                         [ '/* insecure input */', 'background-image: image(asdf.png);' ],
 306                         [ '/* insecure input */', 'background-image: -webkit-image(asdf.png);' ],
 307                         [ '/* insecure input */', 'background-image: -moz-image(asdf.png);' ],
 308                         [ '/* insecure input */', 'background-image: image-set("asdf.png" 1x, "asdf.png" 2x);' ],
 309                         [
 310                                 '/* insecure input */',
 311                                 'background-image: -webkit-image-set("asdf.png" 1x, "asdf.png" 2x);'
 312                         ],
 313                         [
 314                                 '/* insecure input */',
 315                                 'background-image: -moz-image-set("asdf.png" 1x, "asdf.png" 2x);'
 316                         ],
 317                         [ '/* insecure input */', 'foo: attr( title, url );' ],
 318                         [ '/* insecure input */', 'foo: attr( title url );' ],
 319                 ];
 320         }
 321
 322         /**
 323          * @dataProvider provideEscapeHtmlAllowEntities
 324          * @covers Sanitizer::escapeHtmlAllowEntities
 325          */
 326         public function testEscapeHtmlAllowEntities( $expected, $html ) {
 327                 $this->assertEquals(
 328                         $expected,
 329                         Sanitizer::escapeHtmlAllowEntities( $html )
 330                 );
 331         }
 332
 333         public static function provideEscapeHtmlAllowEntities() {
 334                 return [
 335                         [ 'foo', 'foo' ],
 336                         [ 'a¡b', 'a&#161;b' ],
 337                         [ 'foo&#039;bar', "foo'bar" ],
 338                         [ '&lt;script&gt;foo&lt;/script&gt;', '<script>foo</script>' ],
 339                 ];
 340         }
 341
 342         /**
 343          * Test escapeIdReferenceList for consistency with escapeId
 344          *
 345          * @dataProvider provideEscapeIdReferenceList
 346          * @covers Sanitizer::escapeIdReferenceList
 347          */
 348         public function testEscapeIdReferenceList( $referenceList, $id1, $id2 ) {
 349                 $this->assertEquals(
 350                         Sanitizer::escapeIdReferenceList( $referenceList, 'noninitial' ),
 351                         Sanitizer::escapeId( $id1, 'noninitial' )
 352                                 . ' '
 353                                 . Sanitizer::escapeId( $id2, 'noninitial' )
 354                 );
 355         }
 356
 357         public static function provideEscapeIdReferenceList() {
 358                 /** [ <reference list>, <individual id 1>, <individual id 2> ] */
 359                 return [
 360                         [ 'foo bar', 'foo', 'bar' ],
 361                         [ '#1 #2', '#1', '#2' ],
 362                         [ '+1 +2', '+1', '+2' ],
 363                 ];
 364         }
 365
 366         /**
 367          * @dataProvider provideIsReservedDataAttribute
 368          */
 369         public function testIsReservedDataAttribute( $attr, $expected ) {
 370                 $this->assertSame( $expected, Sanitizer::isReservedDataAttribute( $attr ) );
 371         }
 372
 373         public static function provideIsReservedDataAttribute() {
 374                 return [
 375                         [ 'foo', false ],
 376                         [ 'data', false ],
 377                         [ 'data-foo', false ],
 378                         [ 'data-mw', true ],
 379                         [ 'data-ooui', true ],
 380                         [ 'data-parsoid', true ],
 381                         [ 'data-mw-foo', true ],
 382                         [ 'data-ooui-foo', true ],
 383                         [ 'data-mwfoo', true ], // could be false but this is how it's implemented currently
 384                 ];
 385         }
 386 }