tests/phpunit/includes/HtmlFormatterTest.php

   1 <?php
   2
   3 /**
   4  * @group HtmlFormatter
   5  */
   6 class HtmlFormatterTest extends MediaWikiTestCase {
   7         /**
   8          * @dataProvider getHtmlData
   9          *
  10          * @param string $input
  11          * @param string $expectedText
  12          * @param array $expectedRemoved
  13          * @param callable|bool $callback
  14          */
  15         public function testTransform( $input, $expectedText,
  16                 $expectedRemoved = array(), $callback = false
  17         ) {
  18                 $input = self::normalize( $input );
  19                 $formatter = new HtmlFormatter( HtmlFormatter::wrapHTML( $input ) );
  20                 if ( $callback ) {
  21                         $callback( $formatter );
  22                 }
  23                 $removedElements = $formatter->filterContent();
  24                 $html = $formatter->getText();
  25                 $removed = array();
  26                 foreach ( $removedElements as $removedElement ) {
  27                         $removed[] = self::normalize( $formatter->getText( $removedElement ) );
  28                 }
  29                 $expectedRemoved = array_map( 'self::normalize', $expectedRemoved );
  30
  31                 $this->assertValidHtmlSnippet( $html );
  32                 $this->assertEquals( self::normalize( $expectedText ), self::normalize( $html ) );
  33                 $this->assertEquals( asort( $expectedRemoved ), asort( $removed ) );
  34         }
  35
  36         private static function normalize( $s ) {
  37                 return str_replace( "\n", '',
  38                         str_replace( "\r", '', $s ) // "yay" to Windows!
  39                 );
  40         }
  41
  42         public function getHtmlData() {
  43                 $removeImages = function ( HtmlFormatter $f ) {
  44                         $f->setRemoveMedia();
  45                 };
  46                 $removeTags = function ( HtmlFormatter $f ) {
  47                         $f->remove( array( 'table', '.foo', '#bar', 'div.baz' ) );
  48                 };
  49                 $flattenSomeStuff = function ( HtmlFormatter $f ) {
  50                         $f->flatten( array( 's', 'div' ) );
  51                 };
  52                 $flattenEverything = function ( HtmlFormatter $f ) {
  53                         $f->flattenAllTags();
  54                 };
  55                 return array(
  56                         // remove images if asked
  57                         array(
  58                                 '<img src="/foo/bar.jpg" alt="Blah"/>',
  59                                 '',
  60                                 array( '<img src="/foo/bar.jpg" alt="Blah">' ),
  61                                 $removeImages,
  62                         ),
  63                         // basic tag removal
  64                         array(
  65                                 // @codingStandardsIgnoreStart Ignore long line warnings.
  66                                 '<table><tr><td>foo</td></tr></table><div class="foo">foo</div><div class="foo quux">foo</div><span id="bar">bar</span>
  67 <strong class="foo" id="bar">foobar</strong><div class="notfoo">test</div><div class="baz"/>
  68 <span class="baz">baz</span>',
  69                                 // @codingStandardsIgnoreEnd
  70                                 '<div class="notfoo">test</div>
  71 <span class="baz">baz</span>',
  72                                 array(
  73                                         '<table><tr><td>foo</td></tr></table>',
  74                                         '<div class="foo">foo</div>',
  75                                         '<div class="foo quux">foo</div>',
  76                                         '<span id="bar">bar</span>',
  77                                         '<strong class="foo" id="bar">foobar</strong>',
  78                                         '<div class="baz"/>',
  79                                 ),
  80                                 $removeTags,
  81                         ),
  82                         // don't flatten tags that start like chosen ones
  83                         array(
  84                                 '<div><s>foo</s> <span>bar</span></div>',
  85                                 'foo <span>bar</span>',
  86                                 array(),
  87                                 $flattenSomeStuff,
  88                         ),
  89                         // total flattening
  90                         array(
  91                                 '<div style="foo">bar<sup>2</sup></div>',
  92                                 'bar2',
  93                                 array(),
  94                                 $flattenEverything,
  95                         ),
  96                         // UTF-8 preservation and security
  97                         array(
  98                                 '<span title="&quot; \' &amp;">&lt;Тест!&gt;</span> &amp;&lt;&#38;&#0038;&#x26;&#x026;',
  99                                 '<span title="&quot; \' &amp;">&lt;Тест!&gt;</span> &amp;&lt;&amp;&amp;&amp;&amp;',
 100                                 array(),
 101                                 $removeTags, // Have some rules to trigger a DOM parse
 102                         ),
 103                         // https://bugzilla.wikimedia.org/show_bug.cgi?id=53086
 104                         array(
 105                                 'Foo<sup id="cite_ref-1" class="reference"><a href="#cite_note-1">[1]</a></sup>'
 106                                         . ' <a href="/wiki/Bar" title="Bar" class="mw-redirect">Bar</a>',
 107                                 'Foo<sup id="cite_ref-1" class="reference"><a href="#cite_note-1">[1]</a></sup>'
 108                                         . ' <a href="/wiki/Bar" title="Bar" class="mw-redirect">Bar</a>',
 109                         ),
 110                 );
 111         }
 112
 113         public function testQuickProcessing() {
 114                 $f = new MockHtmlFormatter( 'foo' );
 115                 $f->filterContent();
 116                 $this->assertFalse( $f->hasDoc, 'HtmlFormatter should not needlessly parse HTML' );
 117         }
 118 }
 119
 120 class MockHtmlFormatter extends HtmlFormatter {
 121         public $hasDoc = false;
 122
 123         public function getDoc() {
 124                 $this->hasDoc = true;
 125                 return parent::getDoc();
 126         }
 127 }