mirror/php-htmlpurifier
1
0
Fork 0
mirror of https://github.com/ezyang/htmlpurifier.git synced 2025-08-06 14:16:32 +02:00
Code Issues Releases Wiki Activity

Build another lexer from PHP5's DOM library. Extremely fast!

Browse Source 
git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@80 48356398-32a2-884e-a903-53898d9a118a
This commit is contained in:
Edward Z. Yang
2006-07-22 13:50:05 +00:00
parent 3e982c7f2c
commit 4bf3305dff
3 changed files with 130 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

59
tests/Lexer.php
View File

@@ -2,16 +2,20 @@
require_once 'HTMLPurifier/Lexer/DirectLex.php';
require_once 'HTMLPurifier/Lexer/PEARSax3.php';
require_once 'HTMLPurifier/Lexer/DOMLex.php';
class Test_HTMLPurifier_Lexer extends UnitTestCase
{
var $DirectLex;
var $PEARSax3;
var $DirectLex, $PEARSax3, $DOMLex;
var $_has_dom;
function setUp() {
$this->DirectLex =& new HTMLPurifier_Lexer_DirectLex();
$this->PEARSax3 =& new HTMLPurifier_Lexer_PEARSax3();
$this->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() {
@@ -67,6 +71,7 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
,new HTMLPurifier_Token_End('div')
);
// [XML-INVALID]
$input[4] = '<asdf></asdf><d></d><poOloka><poolasdf><ds></asdf></ASDF>';
$expect[4] = array(
new HTMLPurifier_Token_Start('asdf')
@@ -79,6 +84,17 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
,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] = '<a'."\t".'href="foobar.php"'."\n".'title="foo!">Link to <b id="asdf">foobar</b></a>';
$expect[5] = array(
@@ -95,7 +111,7 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
new HTMLPurifier_Token_Empty('br')
);
// [INVALID] [RECOVERABLE]
// [SGML-INVALID] [RECOVERABLE]
$input[7] = '<!-- Comment --> <!-- not so well formed --->';
$expect[7] = array(
new HTMLPurifier_Token_Comment(' Comment ')
@@ -104,7 +120,7 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
);
$sax_expect[7] = false; // we need to figure out proper comment output
// [INVALID]
// [SGML-INVALID]
$input[8] = '<a href=""';
$expect[8] = array(
new HTMLPurifier_Token_Text('<a href=""')
@@ -113,6 +129,10 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
$sax_expect[8] = array(
new HTMLPurifier_Token_Start('a', array('href'=>''))
);
// DOM parses it into an empty tag
$dom_expect[8] = array(
new HTMLPurifier_Token_Empty('a', array('href'=>''))
);
$input[9] = '&lt;b&gt;';
$expect[9] = array(
@@ -126,11 +146,15 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
// note that SAX can clump text nodes together. We won't be
// too picky though
// [INVALID]
// [SGML-INVALID]
$input[10] = '<a "=>';
$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]
$input[11] = '"';
@@ -144,27 +168,42 @@ class Test_HTMLPurifier_Lexer extends UnitTestCase
foreach($input as $i => $discard) {
$result = $this->DirectLex->tokenizeHTML($input[$i]);
$this->assertEqual($expect[$i], $result);
$this->assertEqual($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);
$this->assertEqual($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);
$this->assertEqual($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"';