typo3/sysext/core/Tests/Unit/Html/HtmlParserTest.php

   1 <?php
   2 namespace TYPO3\CMS\Core\Tests\Unit\Html;
   3
   4 /*
   5  * This file is part of the TYPO3 CMS project.
   6  *
   7  * It is free software; you can redistribute it and/or modify it under
   8  * the terms of the GNU General Public License, either version 2
   9  * of the License, or any later version.
  10  *
  11  * For the full copyright and license information, please read the
  12  * LICENSE.txt file that was distributed with this source code.
  13  *
  14  * The TYPO3 project - inspiring people to share!
  15  */
  16
  17 use TYPO3\CMS\Core\Html\HtmlParser;
  18
  19 /**
  20  * Testcase for \TYPO3\CMS\Core\Html\HtmlParser
  21  */
  22 class HtmlParserTest extends \TYPO3\CMS\Core\Tests\UnitTestCase {
  23
  24         /**
  25          * @var \TYPO3\CMS\Core\Html\HtmlParser
  26          */
  27         protected $subject = NULL;
  28
  29         protected function setUp() {
  30                 $this->subject = new HtmlParser();
  31         }
  32
  33         /**
  34          * Data provider for getSubpart
  35          *
  36          * @return array
  37          */
  38         public function getSubpartDataProvider() {
  39                 return array(
  40                         'No start marker' => array(
  41                                 '<body>text</body>',
  42                                 '###SUBPART###',
  43                                 ''
  44                         ),
  45                         'No stop marker' => array(
  46                                 '<body>
  47 <!-- ###SUBPART### Start -->
  48 text
  49 </body>',
  50                                 '###SUBPART###',
  51                                 ''
  52                         ),
  53                         'Start and stop marker in HTML comment' => array(
  54                                 '<body>
  55 <!-- ###SUBPART### Start -->
  56 text
  57 <!-- ###SUBPART### End -->
  58 </body>',
  59                                 '###SUBPART###',
  60                                 '
  61 text
  62 '
  63                         ),
  64                         'Stop marker in HTML comment' => array(
  65                                 '<body>
  66 ###SUBPART###
  67 text
  68 <!-- ###SUBPART### End -->
  69 </body>',
  70                                 '###SUBPART###',
  71                                 '
  72 text
  73 '
  74                         ),
  75                         'Start marker in HTML comment' => array(
  76                                 '<body>
  77 <!-- ###SUBPART### Start -->
  78 text
  79 ###SUBPART###
  80 </body>',
  81                                 '###SUBPART###',
  82                                 '
  83 text
  84 '
  85                         ),
  86                         'Start and stop marker direct' => array(
  87                                 '<body>
  88 ###SUBPART###
  89 text
  90 ###SUBPART###
  91 </body>',
  92                                 '###SUBPART###',
  93                                 '
  94 text
  95 '
  96                         ),
  97                 );
  98         }
  99
 100         /**
 101          * @test
 102          * @dataProvider getSubpartDataProvider
 103          */
 104         public function getSubpart($content, $marker, $expected) {
 105                 $this->assertSame($expected, HtmlParser::getSubpart($content, $marker));
 106         }
 107
 108         /**
 109          * Data provider for substituteMarkerAndSubpartArrayRecursiveResolvesMarkersAndSubpartsArray
 110          */
 111         public function substituteMarkerAndSubpartArrayRecursiveResolvesMarkersAndSubpartsArrayDataProvider() {
 112                 $template = '###SINGLEMARKER1###
 113 <!-- ###FOO### begin -->
 114 <!-- ###BAR### begin -->
 115 ###SINGLEMARKER2###
 116 <!-- ###BAR### end -->
 117 <!-- ###FOOTER### begin -->
 118 ###SINGLEMARKER3###
 119 <!-- ###FOOTER### end -->
 120 <!-- ###FOO### end -->';
 121
 122                 $expected ='Value 1
 123
 124
 125 Value 2.1
 126
 127 Value 2.2
 128
 129
 130 Value 3.1
 131
 132 Value 3.2
 133
 134 ';
 135
 136                 return array(
 137                         'Single marker' => array(
 138                                 '###SINGLEMARKER###',
 139                                 array(
 140                                         '###SINGLEMARKER###' => 'Value 1'
 141                                 ),
 142                                 '',
 143                                 FALSE,
 144                                 FALSE,
 145                                 'Value 1'
 146                         ),
 147                         'Subpart marker' => array(
 148                                 $template,
 149                                 array(
 150                                         '###SINGLEMARKER1###' => 'Value 1',
 151                                         '###FOO###' => array(
 152                                                 array(
 153                                                         '###BAR###' => array(
 154                                                                 array(
 155                                                                         '###SINGLEMARKER2###' => 'Value 2.1'
 156                                                                 ),
 157                                                                 array(
 158                                                                         '###SINGLEMARKER2###' => 'Value 2.2'
 159                                                                 )
 160                                                         ),
 161                                                         '###FOOTER###' => array(
 162                                                                 array(
 163                                                                         '###SINGLEMARKER3###' => 'Value 3.1'
 164                                                                 ),
 165                                                                 array(
 166                                                                         '###SINGLEMARKER3###' => 'Value 3.2'
 167                                                                 )
 168                                                         )
 169                                                 )
 170                                         )
 171                                 ),
 172                                 '',
 173                                 FALSE,
 174                                 FALSE,
 175                                 $expected
 176                         ),
 177                         'Subpart marker with wrap' => array(
 178                                 $template,
 179                                 array(
 180                                         'SINGLEMARKER1' => 'Value 1',
 181                                         'FOO' => array(
 182                                                 array(
 183                                                         'BAR' => array(
 184                                                                 array(
 185                                                                         'SINGLEMARKER2' => 'Value 2.1'
 186                                                                 ),
 187                                                                 array(
 188                                                                         'SINGLEMARKER2' => 'Value 2.2'
 189                                                                 )
 190                                                         ),
 191                                                         'FOOTER' => array(
 192                                                                 array(
 193                                                                         'SINGLEMARKER3' => 'Value 3.1'
 194                                                                 ),
 195                                                                 array(
 196                                                                         'SINGLEMARKER3' => 'Value 3.2'
 197                                                                 )
 198                                                         )
 199                                                 )
 200                                         )
 201                                 ),
 202                                 '###|###',
 203                                 FALSE,
 204                                 FALSE,
 205                                 $expected
 206                         ),
 207                         'Subpart marker with lower marker array keys' => array(
 208                                 $template,
 209                                 array(
 210                                         '###singlemarker1###' => 'Value 1',
 211                                         '###foo###' => array(
 212                                                 array(
 213                                                         '###bar###' => array(
 214                                                                 array(
 215                                                                         '###singlemarker2###' => 'Value 2.1'
 216                                                                 ),
 217                                                                 array(
 218                                                                         '###singlemarker2###' => 'Value 2.2'
 219                                                                 )
 220                                                         ),
 221                                                         '###footer###' => array(
 222                                                                 array(
 223                                                                         '###singlemarker3###' => 'Value 3.1'
 224                                                                 ),
 225                                                                 array(
 226                                                                         '###singlemarker3###' => 'Value 3.2'
 227                                                                 )
 228                                                         )
 229                                                 )
 230                                         )
 231                                 ),
 232                                 '',
 233                                 TRUE,
 234                                 FALSE,
 235                                 $expected
 236                         ),
 237                         'Subpart marker with unused markers' => array(
 238                                 $template,
 239                                 array(
 240                                         '###FOO###' => array(
 241                                                 array(
 242                                                         '###BAR###' => array(
 243                                                                 array(
 244                                                                         '###SINGLEMARKER2###' => 'Value 2.1'
 245                                                                 )
 246                                                         ),
 247                                                         '###FOOTER###' => array(
 248                                                                 array(
 249                                                                         '###SINGLEMARKER3###' => 'Value 3.1'
 250                                                                 )
 251                                                         )
 252                                                 )
 253                                         )
 254                                 ),
 255                                 '',
 256                                 FALSE,
 257                                 TRUE,
 258                                 '
 259
 260
 261 Value 2.1
 262
 263
 264 Value 3.1
 265
 266 '
 267                         ),
 268                         'Subpart marker with empty subpart' => array(
 269                                 $template,
 270                                 array(
 271                                         '###SINGLEMARKER1###' => 'Value 1',
 272                                         '###FOO###' => array(
 273                                                 array(
 274                                                         '###BAR###' => array(
 275                                                                 array(
 276                                                                         '###SINGLEMARKER2###' => 'Value 2.1'
 277                                                                 ),
 278                                                                 array(
 279                                                                         '###SINGLEMARKER2###' => 'Value 2.2'
 280                                                                 )
 281                                                         ),
 282                                                         '###FOOTER###' => array()
 283                                                 )
 284                                         )
 285                                 ),
 286                                 '',
 287                                 FALSE,
 288                                 FALSE,
 289                                 'Value 1
 290
 291
 292 Value 2.1
 293
 294 Value 2.2
 295
 296
 297 '
 298                         )
 299                 );
 300         }
 301
 302         /**
 303          * @test
 304          * @dataProvider substituteMarkerAndSubpartArrayRecursiveResolvesMarkersAndSubpartsArrayDataProvider
 305          */
 306         public function substituteMarkerAndSubpartArrayRecursiveResolvesMarkersAndSubpartsArray($template, $markersAndSubparts, $wrap, $uppercase, $deleteUnused, $expected) {
 307                 $this->assertSame($expected, HtmlParser::substituteMarkerAndSubpartArrayRecursive($template, $markersAndSubparts, $wrap, $uppercase, $deleteUnused));
 308         }
 309
 310         /**
 311          * @return array
 312          */
 313         public function cDataWillRemainUnmodifiedDataProvider() {
 314                 return array(
 315                         'single-line CDATA' => array(
 316                                 '/*<![CDATA[*/ <hello world> /*]]>*/',
 317                                 '/*<![CDATA[*/ <hello world> /*]]>*/',
 318                         ),
 319                         'multi-line CDATA #1' => array(
 320                                 '/*<![CDATA[*/' . LF . '<hello world> /*]]>*/',
 321                                 '/*<![CDATA[*/' . LF . '<hello world> /*]]>*/',
 322                         ),
 323                         'multi-line CDATA #2' => array(
 324                                 '/*<![CDATA[*/ <hello world>' . LF . '/*]]>*/',
 325                                 '/*<![CDATA[*/ <hello world>' . LF . '/*]]>*/',
 326                         ),
 327                         'multi-line CDATA #3' => array(
 328                                 '/*<![CDATA[*/' . LF . '<hello world>' . LF . '/*]]>*/',
 329                                 '/*<![CDATA[*/' . LF . '<hello world>' . LF . '/*]]>*/',
 330                         ),
 331                 );
 332         }
 333
 334         /**
 335          * @test
 336          * @param string $source
 337          * @param string $expected
 338          * @dataProvider cDataWillRemainUnmodifiedDataProvider
 339          */
 340         public function xHtmlCleaningDoesNotModifyCDATA($source, $expected) {
 341                 $result = $this->subject->XHTML_clean($source);
 342                 $this->assertSame($expected, $result);
 343         }
 344
 345         /**
 346          * Data provider for spanTagCorrectlyRemovedWhenRmTagIfNoAttribIsConfigured
 347          */
 348         public static function spanTagCorrectlyRemovedWhenRmTagIfNoAttribIsConfiguredDataProvider() {
 349                 return array(
 350                         'Span tag with no attrib' => array(
 351                                 '<span>text</span>',
 352                                 'text'
 353                         ),
 354                         'Span tag with allowed id attrib' => array(
 355                                 '<span id="id">text</span>',
 356                                 '<span id="id">text</span>'
 357                         ),
 358                         'Span tag with disallowed style attrib' => array(
 359                                 '<span style="line-height: 12px;">text</span>',
 360                                 'text'
 361                         )
 362                 );
 363         }
 364
 365         /**
 366          * @test
 367          * @dataProvider spanTagCorrectlyRemovedWhenRmTagIfNoAttribIsConfiguredDataProvider
 368          */
 369         public function tagCorrectlyRemovedWhenRmTagIfNoAttribIsConfigured($content, $expectedResult) {
 370                 $tsConfig = array(
 371                         'allowTags' => 'span',
 372                         'tags.' => array(
 373                                 'span.' => array(
 374                                         'allowedAttribs' => 'id',
 375                                         'rmTagIfNoAttrib' => 1
 376                                 )
 377                         )
 378                 );
 379                 $this->assertEquals($expectedResult, $this->parseConfigAndCleanHtml($tsConfig, $content));
 380         }
 381
 382         /**
 383          * @test
 384          */
 385         public function rmTagIfNoAttribIsConfiguredDoesNotChangeNestingType() {
 386                 $tsConfig = array(
 387                         'allowTags' => 'div,span',
 388                         'rmTagIfNoAttrib' => 'span',
 389                         'globalNesting' => 'div,span'
 390                 );
 391                 $content = '<span></span><span id="test"><div></span></div>';
 392                 $expectedResult = '<span id="test"></span>';
 393                 $this->assertEquals($expectedResult, $this->parseConfigAndCleanHtml($tsConfig, $content));
 394         }
 395
 396         /**
 397          * Data provider for localNestingCorrectlyRemovesInvalidTags
 398          */
 399         public static function localNestingCorrectlyRemovesInvalidTagsDataProvider() {
 400                 return array(
 401                         'Valid nesting is untouched' => array(
 402                                 '<B><I></B></I>',
 403                                 '<B><I></B></I>'
 404                         ),
 405                         'Valid nesting with content is untouched' => array(
 406                                 'testa<B>test1<I>test2</B>test3</I>testb',
 407                                 'testa<B>test1<I>test2</B>test3</I>testb'
 408                         ),
 409                         'Superflous tags are removed' => array(
 410                                 '</B><B><I></B></I></B>',
 411                                 '<B><I></B></I>'
 412                         ),
 413                         'Superflous tags with content are removed' => array(
 414                                 'test1</B>test2<B>test3<I>test4</B>test5</I>test6</B>test7',
 415                                 'test1test2<B>test3<I>test4</B>test5</I>test6test7'
 416                         ),
 417                         'Another valid nesting test' => array(
 418                                 '<span><div></span></div>',
 419                                 '<span><div></span></div>',
 420                         ),
 421                 );
 422         }
 423
 424         /**
 425          * @test
 426          * @dataProvider localNestingCorrectlyRemovesInvalidTagsDataProvider
 427          * @param string $content
 428          * @param string $expectedResult
 429          */
 430         public function localNestingCorrectlyRemovesInvalidTags($content, $expectedResult) {
 431                 $tsConfig = array(
 432                         'allowTags' => 'div,span,b,i',
 433                         'localNesting' => 'div,span,b,i',
 434                 );
 435                 $this->assertEquals($expectedResult, $this->parseConfigAndCleanHtml($tsConfig, $content));
 436         }
 437
 438         /**
 439          * Data provider for globalNestingCorrectlyRemovesInvalidTags
 440          */
 441         public static function globalNestingCorrectlyRemovesInvalidTagsDataProvider() {
 442                 return array(
 443                         'Valid nesting is untouched' => array(
 444                                 '<B><I></I></B>',
 445                                 '<B><I></I></B>'
 446                         ),
 447                         'Valid nesting with content is untouched' => array(
 448                                 'testa<B>test1<I>test2</I>test3</B>testb',
 449                                 'testa<B>test1<I>test2</I>test3</B>testb'
 450                         ),
 451                         'Invalid nesting is cleaned' => array(
 452                                 '</B><B><I></B></I></B>',
 453                                 '<B></B>'
 454                         ),
 455                         'Invalid nesting with content is cleaned' => array(
 456                                 'test1</B>test2<B>test3<I>test4</B>test5</I>test6</B>test7',
 457                                 'test1test2<B>test3test4</B>test5test6test7'
 458                         ),
 459                         'Another invalid nesting test' => array(
 460                                 '<span><div></span></div>',
 461                                 '<span></span>',
 462                         ),
 463                 );
 464         }
 465
 466         /**
 467          * @test
 468          * @dataProvider globalNestingCorrectlyRemovesInvalidTagsDataProvider
 469          * @param string $content
 470          * @param string $expectedResult
 471          */
 472         public function globalNestingCorrectlyRemovesInvalidTags($content, $expectedResult) {
 473                 $tsConfig = array(
 474                         'allowTags' => 'span,div,b,i',
 475                         'globalNesting' => 'span,div,b,i',
 476                 );
 477                 $this->assertEquals($expectedResult, $this->parseConfigAndCleanHtml($tsConfig, $content));
 478         }
 479
 480         /**
 481          * @return array
 482          */
 483         public function emptyTagsDataProvider() {
 484                 return array(
 485                         array(0 , NULL, FALSE, '<h1></h1>', '<h1></h1>'),
 486                         array(1 , NULL, FALSE, '<h1></h1>', ''),
 487                         array(1 , NULL, FALSE, '<h1>hallo</h1>', '<h1>hallo</h1>'),
 488                         array(1 , NULL, FALSE, '<h1 class="something"></h1>', ''),
 489                         array(1 , NULL, FALSE, '<h1 class="something"></h1><h2></h2>', ''),
 490                         array(1 , 'h2', FALSE, '<h1 class="something"></h1><h2></h2>', '<h1 class="something"></h1>'),
 491                         array(1 , 'h2, h1', FALSE, '<h1 class="something"></h1><h2></h2>', ''),
 492                         array(1 , NULL, FALSE, '<div><p></p></div>', ''),
 493                         array(1 , NULL, FALSE, '<div><p>&nbsp;</p></div>', '<div><p>&nbsp;</p></div>'),
 494                         array(1 , NULL, TRUE, '<div><p>&nbsp;&nbsp;</p></div>', ''),
 495                         array(1 , NULL, TRUE, '<div>&nbsp;&nbsp;<p></p></div>', ''),
 496                         array(1 , NULL, FALSE, '<div>Some content<p></p></div>', '<div>Some content</div>'),
 497                         array(1 , NULL, TRUE, '<div>Some content<p></p></div>', '<div>Some content</div>'),
 498                         array(1 , NULL, FALSE, '<div>Some content</div>', '<div>Some content</div>'),
 499                         array(1 , NULL, TRUE, '<div>Some content</div>', '<div>Some content</div>'),
 500                         array(1 , NULL, FALSE, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
 501                         array(1 , NULL, TRUE, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
 502                 );
 503         }
 504
 505         /**
 506          * @test
 507          * @dataProvider emptyTagsDataProvider
 508          * @param bool $stripOn TRUE if stripping should be activated.
 509          * @param string $tagList Comma seperated list of tags that should be stripped.
 510          * @param bool $treatNonBreakingSpaceAsEmpty If TRUE &nbsp; will be considered empty.
 511          * @param string $content The HTML code that should be modified.
 512          * @param string $expectedResult The expected HTML code result.
 513          */
 514         public function stripEmptyTags($stripOn, $tagList, $treatNonBreakingSpaceAsEmpty, $content, $expectedResult) {
 515                 $tsConfig = array(
 516                         'keepNonMatchedTags' => 1,
 517                         'stripEmptyTags' => $stripOn,
 518                         'stripEmptyTags.' => array(
 519                                 'tags' => $tagList,
 520                                 'treatNonBreakingSpaceAsEmpty' => $treatNonBreakingSpaceAsEmpty
 521                         ),
 522                 );
 523
 524                 $result = $this->parseConfigAndCleanHtml($tsConfig, $content);
 525                 $this->assertEquals($expectedResult, $result);
 526         }
 527
 528         /**
 529          * Calls HTMLparserConfig() and passes the generated config to the HTMLcleaner() method on the current subject.
 530          *
 531          * @param array $tsConfig The TypoScript that should be used to generate the HTML parser config.
 532          * @param string $content The content that should be parsed by the HTMLcleaner.
 533          * @return string The parsed content.
 534          */
 535         protected function parseConfigAndCleanHtml(array $tsConfig, $content) {
 536                 $config = $this->subject->HTMLparserConfig($tsConfig);
 537                 return $this->subject->HTMLcleaner($content, $config[0], $config[1], $config[2], $config[3]);
 538         }
 539 }