tests/phpunit/includes/HtmlFormatterTest.php

   1 <?php
   2
   3 /**
   4  * @group HtmlFormatter
   5  */
   6 class HtmlFormatterTest extends MediaWikiTestCase {
   7         /**
   8          * @dataProvider getHtmlData
   9          * @covers HtmlFormatter::getText
  10          */
  11         public function testTransform( $input, $expectedText,
  12                 $expectedRemoved = array(), $callback = false
  13         ) {
  14                 $input = self::normalize( $input );
  15                 $formatter = new HtmlFormatter( HtmlFormatter::wrapHTML( $input ) );
  16                 if ( $callback ) {
  17                         $callback( $formatter );
  18                 }
  19                 $removedElements = $formatter->filterContent();
  20                 $html = $formatter->getText();
  21                 $removed = array();
  22                 foreach ( $removedElements as $removedElement ) {
  23                         $removed[] = self::normalize( $formatter->getText( $removedElement ) );
  24                 }
  25                 $expectedRemoved = array_map( 'self::normalize', $expectedRemoved );
  26
  27                 $this->assertValidHtmlSnippet( $html );
  28                 $this->assertEquals( self::normalize( $expectedText ), self::normalize( $html ) );
  29                 $this->assertEquals( asort( $expectedRemoved ), asort( $removed ) );
  30         }
  31
  32         private static function normalize( $s ) {
  33                 return str_replace( "\n", '',
  34                         str_replace( "\r", '', $s ) // "yay" to Windows!
  35                 );
  36         }
  37
  38         public function getHtmlData() {
  39                 $removeImages = function( HtmlFormatter $f ) {
  40                         $f->setRemoveMedia();
  41                 };
  42                 $removeTags = function( HtmlFormatter $f ) {
  43                         $f->remove( array( 'table', '.foo', '#bar', 'div.baz' ) );
  44                 };
  45                 $flattenSomeStuff = function( HtmlFormatter $f ) {
  46                         $f->flatten( array( 's', 'div' ) );
  47                 };
  48                 $flattenEverything = function( HtmlFormatter $f ) {
  49                         $f->flattenAllTags();
  50                 };
  51                 return array(
  52                         // remove images if asked
  53                         array(
  54                                 '<img src="/foo/bar.jpg" alt="Blah"/>',
  55                                 '',
  56                                 array( '<img src="/foo/bar.jpg" alt="Blah">' ),
  57                                 $removeImages,
  58                         ),
  59                         // basic tag removal
  60                         array(
  61                                 // @codingStandardsIgnoreStart Ignore long line warnings.
  62                                 '<table><tr><td>foo</td></tr></table><div class="foo">foo</div><div class="foo quux">foo</div><span id="bar">bar</span>
  63 <strong class="foo" id="bar">foobar</strong><div class="notfoo">test</div><div class="baz"/>
  64 <span class="baz">baz</span>',
  65                                 // @codingStandardsIgnoreEnd
  66                                 '<div class="notfoo">test</div>
  67 <span class="baz">baz</span>',
  68                                 array(
  69                                         '<table><tr><td>foo</td></tr></table>',
  70                                         '<div class="foo">foo</div>',
  71                                         '<div class="foo quux">foo</div>',
  72                                         '<span id="bar">bar</span>',
  73                                         '<strong class="foo" id="bar">foobar</strong>',
  74                                         '<div class="baz"/>',
  75                                 ),
  76                                 $removeTags,
  77                         ),
  78                         // don't flatten tags that start like chosen ones
  79                         array(
  80                                 '<div><s>foo</s> <span>bar</span></div>',
  81                                 'foo <span>bar</span>',
  82                                 array(),
  83                                 $flattenSomeStuff,
  84                         ),
  85                         // total flattening
  86                         array(
  87                                 '<div style="foo">bar<sup>2</sup></div>',
  88                                 'bar2',
  89                                 array(),
  90                                 $flattenEverything,
  91                         ),
  92                         // UTF-8 preservation and security
  93                         array(
  94                                 '<span title="&quot; \' &amp;">&lt;Тест!&gt;</span> &amp;&lt;&#38;&#0038;&#x26;&#x026;',
  95                                 '<span title="&quot; \' &amp;">&lt;Тест!&gt;</span> &amp;&lt;&amp;&amp;&amp;&amp;',
  96                         ),
  97                         // https://bugzilla.wikimedia.org/show_bug.cgi?id=53086
  98                         array(
  99                                 'Foo<sup id="cite_ref-1" class="reference"><a href="#cite_note-1">[1]</a></sup>'
 100                                         . ' <a href="/wiki/Bar" title="Bar" class="mw-redirect">Bar</a>',
 101                                 'Foo<sup id="cite_ref-1" class="reference"><a href="#cite_note-1">[1]</a></sup>'
 102                                         . ' <a href="/wiki/Bar" title="Bar" class="mw-redirect">Bar</a>',
 103                         ),
 104                 );
 105         }
 106 }