tests/phpunit/includes/HtmlFormatterTest.php

   1 <?php
   2
   3 /**
   4  * @group HtmlFormatter
   5  */
   6 class HtmlFormatterTest extends MediaWikiTestCase {
   7
   8         /**
   9          * Use TidySupport to check whether we should use $wgTidyInternal.
  10          *
  11          * The Tidy extension in HHVM does not support error text return, so it is
  12          * nominally usable, but does not pass tests which require error text from
  13          * Tidy.
  14          */
  15         protected function setUp() {
  16                 parent::setUp();
  17                 $tidySupport = new TidySupport();
  18                 $this->setMwGlobals( 'wgTidyInternal', $tidySupport->isInternal() );
  19         }
  20
  21         /**
  22          * @dataProvider getHtmlData
  23          *
  24          * @param string $input
  25          * @param string $expectedText
  26          * @param array $expectedRemoved
  27          * @param callable|bool $callback
  28          */
  29         public function testTransform( $input, $expectedText,
  30                 $expectedRemoved = array(), $callback = false
  31         ) {
  32                 $input = self::normalize( $input );
  33                 $formatter = new HtmlFormatter( HtmlFormatter::wrapHTML( $input ) );
  34                 if ( $callback ) {
  35                         $callback( $formatter );
  36                 }
  37                 $removedElements = $formatter->filterContent();
  38                 $html = $formatter->getText();
  39                 $removed = array();
  40                 foreach ( $removedElements as $removedElement ) {
  41                         $removed[] = self::normalize( $formatter->getText( $removedElement ) );
  42                 }
  43                 $expectedRemoved = array_map( 'self::normalize', $expectedRemoved );
  44
  45                 $this->assertValidHtmlSnippet( $html );
  46                 $this->assertEquals( self::normalize( $expectedText ), self::normalize( $html ) );
  47                 $this->assertEquals( asort( $expectedRemoved ), asort( $removed ) );
  48         }
  49
  50         private static function normalize( $s ) {
  51                 return str_replace( "\n", '',
  52                         str_replace( "\r", '', $s ) // "yay" to Windows!
  53                 );
  54         }
  55
  56         public function getHtmlData() {
  57                 $removeImages = function ( HtmlFormatter $f ) {
  58                         $f->setRemoveMedia();
  59                 };
  60                 $removeTags = function ( HtmlFormatter $f ) {
  61                         $f->remove( array( 'table', '.foo', '#bar', 'div.baz' ) );
  62                 };
  63                 $flattenSomeStuff = function ( HtmlFormatter $f ) {
  64                         $f->flatten( array( 's', 'div' ) );
  65                 };
  66                 $flattenEverything = function ( HtmlFormatter $f ) {
  67                         $f->flattenAllTags();
  68                 };
  69                 return array(
  70                         // remove images if asked
  71                         array(
  72                                 '<img src="/foo/bar.jpg" alt="Blah"/>',
  73                                 '',
  74                                 array( '<img src="/foo/bar.jpg" alt="Blah">' ),
  75                                 $removeImages,
  76                         ),
  77                         // basic tag removal
  78                         array(
  79                                 // @codingStandardsIgnoreStart Ignore long line warnings.
  80                                 '<table><tr><td>foo</td></tr></table><div class="foo">foo</div><div class="foo quux">foo</div><span id="bar">bar</span>
  81 <strong class="foo" id="bar">foobar</strong><div class="notfoo">test</div><div class="baz"/>
  82 <span class="baz">baz</span>',
  83                                 // @codingStandardsIgnoreEnd
  84                                 '<div class="notfoo">test</div>
  85 <span class="baz">baz</span>',
  86                                 array(
  87                                         '<table><tr><td>foo</td></tr></table>',
  88                                         '<div class="foo">foo</div>',
  89                                         '<div class="foo quux">foo</div>',
  90                                         '<span id="bar">bar</span>',
  91                                         '<strong class="foo" id="bar">foobar</strong>',
  92                                         '<div class="baz"/>',
  93                                 ),
  94                                 $removeTags,
  95                         ),
  96                         // don't flatten tags that start like chosen ones
  97                         array(
  98                                 '<div><s>foo</s> <span>bar</span></div>',
  99                                 'foo <span>bar</span>',
 100                                 array(),
 101                                 $flattenSomeStuff,
 102                         ),
 103                         // total flattening
 104                         array(
 105                                 '<div style="foo">bar<sup>2</sup></div>',
 106                                 'bar2',
 107                                 array(),
 108                                 $flattenEverything,
 109                         ),
 110                         // UTF-8 preservation and security
 111                         array(
 112                                 '<span title="&quot; \' &amp;">&lt;Тест!&gt;</span> &amp;&lt;&#38;&#0038;&#x26;&#x026;',
 113                                 '<span title="&quot; \' &amp;">&lt;Тест!&gt;</span> &amp;&lt;&amp;&amp;&amp;&amp;',
 114                                 array(),
 115                                 $removeTags, // Have some rules to trigger a DOM parse
 116                         ),
 117                         // https://phabricator.wikimedia.org/T55086
 118                         array(
 119                                 'Foo<sup id="cite_ref-1" class="reference"><a href="#cite_note-1">[1]</a></sup>'
 120                                         . ' <a href="/wiki/Bar" title="Bar" class="mw-redirect">Bar</a>',
 121                                 'Foo<sup id="cite_ref-1" class="reference"><a href="#cite_note-1">[1]</a></sup>'
 122                                         . ' <a href="/wiki/Bar" title="Bar" class="mw-redirect">Bar</a>',
 123                         ),
 124                 );
 125         }
 126
 127         public function testQuickProcessing() {
 128                 $f = new MockHtmlFormatter( 'foo' );
 129                 $f->filterContent();
 130                 $this->assertFalse( $f->hasDoc, 'HtmlFormatter should not needlessly parse HTML' );
 131         }
 132 }
 133
 134 class MockHtmlFormatter extends HtmlFormatter {
 135         public $hasDoc = false;
 136
 137         public function getDoc() {
 138                 $this->hasDoc = true;
 139                 return parent::getDoc();
 140         }
 141 }