tests/HTMLPurifier/Lexer/DirectLexTest.php

   1 <?php
   2
   3 class HTMLPurifier_Lexer_DirectLexTest extends HTMLPurifier_Harness
   4 {
   5
   6     protected $DirectLex;
   7
   8     function setUp() {
   9         $this->DirectLex = new HTMLPurifier_Lexer_DirectLex();
  10     }
  11
  12     // internals testing
  13     function test_parseAttributeString() {
  14
  15         $input[0] = 'href="about:blank" rel="nofollow"';
  16         $expect[0] = array('href'=>'about:blank', 'rel'=>'nofollow');
  17
  18         $input[1] = "href='about:blank'";
  19         $expect[1] = array('href'=>'about:blank');
  20
  21         // note that the single quotes aren't /really/ escaped
  22         $input[2] = 'onclick="javascript:alert(\'asdf\');"';
  23         $expect[2] = array('onclick' => "javascript:alert('asdf');");
  24
  25         $input[3] = 'selected';
  26         $expect[3] = array('selected'=>'selected');
  27
  28         // [INVALID]
  29         $input[4] = '="nokey"';
  30         $expect[4] = array();
  31
  32         // [SIMPLE]
  33         $input[5] = 'color=blue';
  34         $expect[5] = array('color' => 'blue');
  35
  36         // [INVALID]
  37         $input[6] = 'href="about:blank';
  38         $expect[6] = array('href' => 'about:blank');
  39
  40         // [INVALID]
  41         $input[7] = '"=';
  42         $expect[7] = array('"' => '');
  43         // we ought to get array()
  44
  45         $input[8] = 'href ="about:blank"rel ="nofollow"';
  46         $expect[8] = array('href' => 'about:blank', 'rel' => 'nofollow');
  47
  48         $input[9] = 'two bool';
  49         $expect[9] = array('two' => 'two', 'bool' => 'bool');
  50
  51         $input[10] = 'name="input" selected';
  52         $expect[10] = array('name' => 'input', 'selected' => 'selected');
  53
  54         $input[11] = '=""';
  55         $expect[11] = array();
  56
  57         $input[12] = '="" =""';
  58         $expect[12] = array('"' => ''); // tough to say, just don't throw a loop
  59
  60         $input[13] = 'href="';
  61         $expect[13] = array('href' => '');
  62
  63         $input[14] = 'href=" <';
  64         $expect[14] = array('href' => ' <');
  65
  66         $config = HTMLPurifier_Config::createDefault();
  67         $context = new HTMLPurifier_Context();
  68         $size = count($input);
  69         for($i = 0; $i < $size; $i++) {
  70             $result = $this->DirectLex->parseAttributeString($input[$i], $config, $context);
  71             $this->assertIdentical($expect[$i], $result, 'Test ' . $i . ': %s');
  72         }
  73
  74     }
  75
  76     function testLineNumbers() {
  77
  78         //       .  .     .     .  .     .     .           .      .             .
  79         //       01234567890123 01234567890123 0123456789012345 0123456789012   012345
  80         $html = "<b>Line 1</b>\n<i>Line 2</i>\nStill Line 2<br\n/>Now Line 4\n\n<br />";
  81
  82         $expect = array(
  83             // line 1
  84             0 => new HTMLPurifier_Token_Start('b')
  85            ,1 => new HTMLPurifier_Token_Text('Line 1')
  86            ,2 => new HTMLPurifier_Token_End('b')
  87            ,3 => new HTMLPurifier_Token_Text("\n")
  88             // line 2
  89            ,4 => new HTMLPurifier_Token_Start('i')
  90            ,5 => new HTMLPurifier_Token_Text('Line 2')
  91            ,6 => new HTMLPurifier_Token_End('i')
  92            ,7 => new HTMLPurifier_Token_Text("\nStill Line 2")
  93             // line 3
  94            ,8 => new HTMLPurifier_Token_Empty('br')
  95             // line 4
  96            ,9 => new HTMLPurifier_Token_Text("Now Line 4\n\n")
  97             // line SIX
  98            ,10 => new HTMLPurifier_Token_Empty('br')
  99         );
 100
 101         $context = new HTMLPurifier_Context();
 102         $config  = HTMLPurifier_Config::createDefault();
 103         $output = $this->DirectLex->tokenizeHTML($html, $config, $context);
 104
 105         $this->assertIdentical($output, $expect);
 106
 107         $context = new HTMLPurifier_Context();
 108         $config  = HTMLPurifier_Config::create(array(
 109             'Core.MaintainLineNumbers' => true
 110         ));
 111         $expect[0]->position(1, 0);
 112         $expect[1]->position(1, 3);
 113         $expect[2]->position(1, 9);
 114         $expect[3]->position(2, -1);
 115         $expect[4]->position(2, 0);
 116         $expect[5]->position(2, 3);
 117         $expect[6]->position(2, 9);
 118         $expect[7]->position(3, -1);
 119         $expect[8]->position(3, 12);
 120         $expect[9]->position(4, 2);
 121         $expect[10]->position(6, 0);
 122
 123         $output = $this->DirectLex->tokenizeHTML($html, $config, $context);
 124         $this->assertIdentical($output, $expect);
 125
 126     }
 127
 128 }
 129