tests/HTMLPurifier/Lexer/DirectLexTest.php

   1 <?php
   2
   3 class HTMLPurifier_Lexer_DirectLexTest extends HTMLPurifier_Harness
   4 {
   5
   6     protected $DirectLex;
   7
   8     public function setUp()
   9     {
  10         $this->DirectLex = new HTMLPurifier_Lexer_DirectLex();
  11     }
  12
  13     // internals testing
  14     public function test_parseAttributeString()
  15     {
  16         $input[0] = 'href="about:blank" rel="nofollow"';
  17         $expect[0] = array('href'=>'about:blank', 'rel'=>'nofollow');
  18
  19         $input[1] = "href='about:blank'";
  20         $expect[1] = array('href'=>'about:blank');
  21
  22         // note that the single quotes aren't /really/ escaped
  23         $input[2] = 'onclick="javascript:alert(\'asdf\');"';
  24         $expect[2] = array('onclick' => "javascript:alert('asdf');");
  25
  26         $input[3] = 'selected';
  27         $expect[3] = array('selected'=>'selected');
  28
  29         // [INVALID]
  30         $input[4] = '="nokey"';
  31         $expect[4] = array();
  32
  33         // [SIMPLE]
  34         $input[5] = 'color=blue';
  35         $expect[5] = array('color' => 'blue');
  36
  37         // [INVALID]
  38         $input[6] = 'href="about:blank';
  39         $expect[6] = array('href' => 'about:blank');
  40
  41         // [INVALID]
  42         $input[7] = '"=';
  43         $expect[7] = array('"' => '');
  44         // we ought to get array()
  45
  46         $input[8] = 'href ="about:blank"rel ="nofollow"';
  47         $expect[8] = array('href' => 'about:blank', 'rel' => 'nofollow');
  48
  49         $input[9] = 'two bool';
  50         $expect[9] = array('two' => 'two', 'bool' => 'bool');
  51
  52         $input[10] = 'name="input" selected';
  53         $expect[10] = array('name' => 'input', 'selected' => 'selected');
  54
  55         $input[11] = '=""';
  56         $expect[11] = array();
  57
  58         $input[12] = '="" =""';
  59         $expect[12] = array(); // tough to say, just don't throw a loop
  60
  61         $input[13] = 'href="';
  62         $expect[13] = array('href' => '');
  63
  64         $input[14] = 'href=" <';
  65         $expect[14] = array('href' => ' <');
  66
  67         $config = HTMLPurifier_Config::createDefault();
  68         $context = new HTMLPurifier_Context();
  69         $size = count($input);
  70         for($i = 0; $i < $size; $i++) {
  71             $result = $this->DirectLex->parseAttributeString($input[$i], $config, $context);
  72             $this->assertIdentical($expect[$i], $result, 'Test ' . $i . ': %s');
  73         }
  74
  75     }
  76
  77     public function testLineNumbers()
  78     {
  79         //       .  .     .     .  .     .     .           .      .             .
  80         //       01234567890123 01234567890123 0123456789012345 0123456789012   012345
  81         $html = "<b>Line 1</b>\n<i>Line 2</i>\nStill Line 2<br\n/>Now Line 4\n\n<br />";
  82
  83         $expect = array(
  84             // line 1
  85             0 => new HTMLPurifier_Token_Start('b')
  86            ,1 => new HTMLPurifier_Token_Text('Line 1')
  87            ,2 => new HTMLPurifier_Token_End('b')
  88            ,3 => new HTMLPurifier_Token_Text("\n")
  89             // line 2
  90            ,4 => new HTMLPurifier_Token_Start('i')
  91            ,5 => new HTMLPurifier_Token_Text('Line 2')
  92            ,6 => new HTMLPurifier_Token_End('i')
  93            ,7 => new HTMLPurifier_Token_Text("\nStill Line 2")
  94             // line 3
  95            ,8 => new HTMLPurifier_Token_Empty('br')
  96             // line 4
  97            ,9 => new HTMLPurifier_Token_Text("Now Line 4\n\n")
  98             // line SIX
  99            ,10 => new HTMLPurifier_Token_Empty('br')
 100         );
 101
 102         $context = new HTMLPurifier_Context();
 103         $config  = HTMLPurifier_Config::createDefault();
 104         $output = $this->DirectLex->tokenizeHTML($html, $config, $context);
 105
 106         $this->assertIdentical($output, $expect);
 107
 108         $context = new HTMLPurifier_Context();
 109         $config  = HTMLPurifier_Config::create(array(
 110             'Core.MaintainLineNumbers' => true
 111         ));
 112         $expect[0]->position(1, 0);
 113         $expect[1]->position(1, 3);
 114         $expect[2]->position(1, 9);
 115         $expect[3]->position(2, -1);
 116         $expect[4]->position(2, 0);
 117         $expect[5]->position(2, 3);
 118         $expect[6]->position(2, 9);
 119         $expect[7]->position(3, -1);
 120         $expect[8]->position(3, 12);
 121         $expect[9]->position(4, 2);
 122         $expect[10]->position(6, 0);
 123
 124         $output = $this->DirectLex->tokenizeHTML($html, $config, $context);
 125         $this->assertIdentical($output, $expect);
 126
 127     }
 128
 129 }
 130
 131 // vim: et sw=4 sts=4