tests/phpunit/includes/tidy/BalancerTest.php

   1 <?php
   2
   3 class BalancerTest extends MediaWikiTestCase {
   4         private $balancer;
   5
   6         /**
   7          * Anything that needs to happen before your tests should go here.
   8          */
   9         protected function setUp() {
  10                 // Be sure to do call the parent setup and teardown functions.
  11                 // This makes sure that all the various cleanup and restorations
  12                 // happen as they should (including the restoration for setMwGlobals).
  13                 parent::setUp();
  14                 $this->balancer = new MediaWiki\Tidy\Balancer( [
  15                         'strict' => false, /* not strict */
  16                         'allowedHtmlElements' => null, /* no sanitization */
  17                         'tidyCompat' => false, /* standard parser */
  18                 ] );
  19         }
  20
  21         /**
  22          * Anything cleanup you need to do should go here.
  23          */
  24         protected function tearDown() {
  25                 parent::tearDown();
  26         }
  27
  28         /**
  29          * @covers Balancer::balance
  30          * @dataProvider provideBalancerTests
  31          */
  32         public function testBalancer( $description, $input, $expected ) {
  33                 $output = $this->balancer->balance( $input );
  34
  35                 // Ignore self-closing tags
  36                 $output = preg_replace( '/\s*\/>/', '>', $output );
  37
  38                 $this->assertEquals( $expected, $output, $description );
  39         }
  40
  41         public static function provideBalancerTests() {
  42                 // Get the tests from html5lib-tests.json
  43                 $json = json_decode( file_get_contents(
  44                         __DIR__ . '/html5lib-tests.json'
  45                 ), true );
  46                 // Munge this slightly into the format phpunit expects
  47                 // for providers, and filter out HTML constructs which
  48                 // the balancer doesn't support.
  49                 $tests = [];
  50                 $start = '<html><head></head><body>';
  51                 $end = '</body></html>';
  52                 foreach ( $json as $filename => $cases ) {
  53                         foreach ( $cases as $case ) {
  54                                 $html = $case['document']['html'];
  55                                 if (
  56                                         substr( $html, 0, strlen( $start ) ) !== $start ||
  57                                         substr( $html, -strlen( $end ) ) !== $end
  58                                 ) {
  59                                         // Skip tests which involve stuff in the <head> or
  60                                         // weird doctypes.
  61                                         continue;
  62                                 }
  63                                 // We used to do this:
  64                                 //   $html = substr( $html, strlen( $start ), -strlen( $end ) );
  65                                 // But now we use a different field in the test case,
  66                                 // which reports how domino would parse this case in a
  67                                 // no-quirks <body> context.  (The original test case may
  68                                 // have had a different context, or relied on quirks mode.)
  69                                 $html = $case['document']['noQuirksBodyHtml'];
  70                                 // Normalize case of SVG attributes.
  71                                 $html = str_replace( 'foreignObject', 'foreignobject', $html );
  72
  73                                 if ( isset( $case['document']['props']['comment'] ) ) {
  74                                         // Skip tests which include HTML comments, which
  75                                         // the balancer requires to have been stripped.
  76                                         continue;
  77                                 }
  78                                 if ( strpos( $case['data'], '<![CDATA[' ) !== false ) {
  79                                         // Skip tests involving <![CDATA[ ]]> quoting.
  80                                         continue;
  81                                 }
  82                                 if ( stripos( $case['data'], '<!DOCTYPE' ) !== false ) {
  83                                         // Skip tests involving doctypes.
  84                                         continue;
  85                                 }
  86                                 if ( preg_match( ',</?(html|head|body|frame|plaintext)>|<rdar:,i', $case['data'] ) ) {
  87                                         // Skip tests involving some literal tags, which are
  88                                         // unsupported but don't show up in the expected output.
  89                                         continue;
  90                                 }
  91                                 if (
  92                                         isset( $case['document']['props']['tags']['form'] ) ||
  93                                         isset( $case['document']['props']['tags']['iframe'] ) ||
  94                                         isset( $case['document']['props']['tags']['noembed'] ) ||
  95                                         isset( $case['document']['props']['tags']['noscript'] ) ||
  96                                         isset( $case['document']['props']['tags']['script'] ) ||
  97                                         isset( $case['document']['props']['tags']['svg script'] ) ||
  98                                         isset( $case['document']['props']['tags']['svg title'] ) ||
  99                                         isset( $case['document']['props']['tags']['textarea'] ) ||
 100                                         isset( $case['document']['props']['tags']['title'] ) ||
 101                                         isset( $case['document']['props']['tags']['xmp'] )
 102                                 ) {
 103                                         // Skip tests with unsupported tags which *do* show
 104                                         // up in the expected output.
 105                                         continue;
 106                                 }
 107                                 if (
 108                                         $filename === 'entities01.dat' ||
 109                                         $filename === 'entities02.dat' ||
 110                                         preg_match( '/&([a-z]+|#x[0-9A-F]+);/i', $case['data'] ) ||
 111                                         preg_match( '/^(&|&#|&#X|&#x|&#45|&x-test|&AMP)$/', $case['data'] )
 112                                 ) {
 113                                         // Skip tests involving entity encoding.
 114                                         continue;
 115                                 }
 116                                 if (
 117                                         isset( $case['document']['props']['tagWithLt'] ) ||
 118                                         isset( $case['document']['props']['attrWithFunnyChar'] ) ||
 119                                         preg_match( ':^(</b test|<di|<foo bar=qux/>)$:', $case['data'] ) ||
 120                                         preg_match( ':</p<p>:', $case['data'] )
 121                                 ) {
 122                                         // Skip tests with funny tag or attribute names,
 123                                         // which are really tests of the HTML tokenizer, not
 124                                         // the tree builder.
 125                                         continue;
 126                                 }
 127                                 if (
 128                                         stripos( $case['data'], 'encoding=" text/html "' ) !== false
 129                                 ) {
 130                                         // The Sanitizer normalizes whitespace in attribute
 131                                         // values, which makes this test case invalid.
 132                                         continue;
 133                                 }
 134                                 if ( $filename === 'plain-text-unsafe.dat' ) {
 135                                         // Skip tests with ASCII null, etc.
 136                                         continue;
 137                                 }
 138                                 $tests[] = [
 139                                         $filename, # use better description?
 140                                         $case['data'],
 141                                         $html
 142                                 ];
 143                         }
 144                 }
 145                 return $tests;
 146         }
 147 }