DirectLex = new HTMLPurifier_Lexer_DirectLex(); $this->PEARSax3 = new HTMLPurifier_Lexer_PEARSax3(); $this->DOMLex = new HTMLPurifier_Lexer_DOMLex(); $this->_has_dom = version_compare(PHP_VERSION, '5', '>='); } function test_nextWhiteSpace() { $HP =& $this->DirectLex; $this->assertIdentical(false, $HP->nextWhiteSpace('asdf')); $this->assertIdentical(0, $HP->nextWhiteSpace(' asdf')); $this->assertIdentical(0, $HP->nextWhiteSpace("\nasdf")); $this->assertIdentical(1, $HP->nextWhiteSpace("a\tsdf")); $this->assertIdentical(4, $HP->nextWhiteSpace("asdf\r")); $this->assertIdentical(2, $HP->nextWhiteSpace("as\t\r\nasdf as")); } function test_parseData() { $HP =& $this->DirectLex; $this->assertIdentical('asdf', $HP->parseData('asdf')); $this->assertIdentical('&', $HP->parseData('&')); $this->assertIdentical('"', $HP->parseData('"')); $this->assertIdentical("'", $HP->parseData(''')); $this->assertIdentical('-', $HP->parseData('-')); // UTF-8 needed!!! } function test_tokenizeHTML() { $input = array(); $expect = array(); $sax_expect = array(); $input[0] = ''; $expect[0] = array(); $input[1] = 'This is regular text.'; $expect[1] = array( new HTMLPurifier_Token_Text('This is regular text.') ); $input[2] = 'This is bold text'; $expect[2] = array( new HTMLPurifier_Token_Text('This is ') ,new HTMLPurifier_Token_Start('b', array()) ,new HTMLPurifier_Token_Text('bold') ,new HTMLPurifier_Token_End('b') ,new HTMLPurifier_Token_Text(' text') ); $input[3] = '
Totally rad dude. asdf
'; $expect[3] = array( new HTMLPurifier_Token_Start('DIV', array()) ,new HTMLPurifier_Token_Text('Totally rad dude. ') ,new HTMLPurifier_Token_Start('b', array()) ,new HTMLPurifier_Token_Text('asdf') ,new HTMLPurifier_Token_End('b') ,new HTMLPurifier_Token_End('div') ); // [XML-INVALID] $input[4] = ''; $expect[4] = array( new HTMLPurifier_Token_Start('asdf') ,new HTMLPurifier_Token_End('asdf') ,new HTMLPurifier_Token_Start('d') ,new HTMLPurifier_Token_End('d') ,new HTMLPurifier_Token_Start('poOloka') ,new HTMLPurifier_Token_Start('poolasdf') ,new HTMLPurifier_Token_Start('ds') ,new HTMLPurifier_Token_End('asdf') ,new HTMLPurifier_Token_End('ASDF') ); // DOM is different because it condenses empty tags into REAL empty ones // as well as makes it well-formed $dom_expect[4] = array( new HTMLPurifier_Token_Empty('asdf') ,new HTMLPurifier_Token_Empty('d') ,new HTMLPurifier_Token_Start('pooloka') ,new HTMLPurifier_Token_Start('poolasdf') ,new HTMLPurifier_Token_Empty('ds') ,new HTMLPurifier_Token_End('poolasdf') ,new HTMLPurifier_Token_End('pooloka') ); $input[5] = 'Link to foobar'; $expect[5] = array( new HTMLPurifier_Token_Start('a',array('href'=>'foobar.php','title'=>'foo!')) ,new HTMLPurifier_Token_Text('Link to ') ,new HTMLPurifier_Token_Start('b',array('id'=>'asdf')) ,new HTMLPurifier_Token_Text('foobar') ,new HTMLPurifier_Token_End('b') ,new HTMLPurifier_Token_End('a') ); $input[6] = '
'; $expect[6] = array( new HTMLPurifier_Token_Empty('br') ); // [SGML-INVALID] [RECOVERABLE] $input[7] = ' '; $expect[7] = array( new HTMLPurifier_Token_Comment(' Comment ') ,new HTMLPurifier_Token_Text(' ') ,new HTMLPurifier_Token_Comment(' not so well formed -') ); $sax_expect[7] = false; // we need to figure out proper comment output // [SGML-INVALID] $input[8] = ''')) ); // DOM parses it into an empty tag $dom_expect[8] = array( new HTMLPurifier_Token_Empty('a', array('href'=>'')) ); $input[9] = '<b>'; $expect[9] = array( new HTMLPurifier_Token_Text('') ); $sax_expect[9] = array( new HTMLPurifier_Token_Text('<') ,new HTMLPurifier_Token_Text('b') ,new HTMLPurifier_Token_Text('>') ); // note that SAX can clump text nodes together. We won't be // too picky though // [SGML-INVALID] $input[10] = ''; $expect[10] = array( new HTMLPurifier_Token_Start('a', array('"' => '')) ); // DOM doesn't register an invalid attribute $dom_expect[10] = array( new HTMLPurifier_Token_Empty('a') ); // [INVALID] [RECOVERABLE] $input[11] = '"'; $expect[11] = array( new HTMLPurifier_Token_Text('"') ); // compare with this valid one: $input[12] = '"'; $expect[12] = array( new HTMLPurifier_Token_Text('"') ); $sax_expect[12] = false; // SAX chokes on this? We do have entity parsing on, so it should work! foreach($input as $i => $discard) { $result = $this->DirectLex->tokenizeHTML($input[$i]); $this->assertEqual($expect[$i], $result, 'Test '.$i.': %s'); paintIf($result, $expect[$i] != $result); // assert unless I say otherwise $sax_result = $this->PEARSax3->tokenizeHTML($input[$i]); if (!isset($sax_expect[$i])) { // by default, assert with normal result $this->assertEqual($expect[$i], $sax_result, 'Test '.$i.': %s'); paintIf($sax_result, $expect[$i] != $sax_result); } elseif ($sax_expect[$i] === false) { // assertions were turned off, optionally dump // paintIf($sax_expect, $i == NUMBER); } else { // match with a custom SAX result array $this->assertEqual($sax_expect[$i], $sax_result, 'Test '.$i.': %s'); paintIf($sax_result, $sax_expect[$i] != $sax_result); } if ($this->_has_dom) { $dom_result = $this->DOMLex->tokenizeHTML($input[$i]); // same structure as SAX if (!isset($dom_expect[$i])) { $this->assertEqual($expect[$i], $dom_result, 'Test '.$i.': %s'); paintIf($dom_result, $expect[$i] != $dom_result); } elseif ($dom_expect[$i] === false) { // paintIf($dom_result, $i == NUMBER); } else { $this->assertEqual($dom_expect[$i], $dom_result, 'Test '.$i.': %s'); paintIf($dom_result, $dom_expect[$i] != $dom_result); } } } } // internals testing function test_tokenizeAttributeString() { $input[] = 'href="asdf" boom="assdf"'; $expect[] = array('href'=>'asdf', 'boom'=>'assdf'); $input[] = "href='r'"; $expect[] = array('href'=>'r'); $input[] = 'onclick="javascript:alert(\'asdf\');"'; $expect[] = array('onclick' => "javascript:alert('asdf');"); $input[] = 'selected'; $expect[] = array('selected'=>'selected'); $input[] = '="asdf"'; $expect[] = array(); $input[] = 'missile=launch'; $expect[] = array('missile' => 'launch'); $input[] = 'href="foo'; $expect[] = array('href' => 'foo'); $size = count($input); for($i = 0; $i < $size; $i++) { $result = $this->DirectLex->tokenizeAttributeString($input[$i]); $this->assertEqual($expect[$i], $result); paintIf($result, $expect[$i] != $result); } } } ?>