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