[3.0.0] Fully implement CSS extraction and cleaning. See NEWS for more information, it is now a Filter.

- Some Lexer things were moved around

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1469 48356398-32a2-884e-a903-53898d9a118a

tests/HTMLPurifier/Filter/ExtractStyleBlocksTest.php

@@ -0,0 +1,112 @@
+<?php
+
+require_once 'HTMLPurifier/Filter/ExtractStyleBlocks.php';
+
+/**
+ * @todo Assimilate CSSTidy into our library
+ */
+class HTMLPurifier_Filter_ExtractStyleBlocksTest extends HTMLPurifier_Harness
+{
+    
+    // usual use case:
+    function test_tokenizeHTML_extractStyleBlocks() {
+        $purifier = new HTMLPurifier($this->config);
+        $purifier->addFilter(new HTMLPurifier_Filter_ExtractStyleBlocks());
+        $result = $purifier->purify('<style type="text/css">.foo {text-align:center;bogus:remove-me;}</style>Test<style>* {font-size:12pt;}</style>');
+        $this->assertIdentical($result, 'Test');
+        $this->assertIdentical($purifier->context->get('StyleBlocks'),
+            array(
+                ".foo {\ntext-align:center;\n}",
+                "* {\nfont-size:12pt;\n}"
+            )
+        );
+    }
+    
+    function assertExtractStyleBlocks($html, $expect = true, $styles = array()) {
+        $filter = new HTMLPurifier_Filter_ExtractStyleBlocks(false); // disable cleaning
+        if ($expect === true) $expect = $html;
+        $result = $filter->preFilter($html, $this->config, $this->context);
+        $this->assertIdentical($result, $expect);
+        $this->assertIdentical($this->context->get('StyleBlocks'), $styles);
+    }
+    
+    function test_extractStyleBlocks_preserve() {
+        $this->assertExtractStyleBlocks('Foobar');
+    }
+    
+    function test_extractStyleBlocks_allStyle() {
+        $this->assertExtractStyleBlocks('<style>foo</style>', '', array('foo'));
+    }
+    
+    function test_extractStyleBlocks_multipleBlocks() {
+        $this->assertExtractStyleBlocks(
+          "<style>1</style><style>2</style>NOP<style>4</style>",
+          "NOP",
+          array('1', '2', '4')
+        );
+    }
+    
+    function test_extractStyleBlocks_blockWithAttributes() {
+        $this->assertExtractStyleBlocks(
+          '<style type="text/css">css</style>',
+          '',
+          array('css')
+        );
+    }
+    
+    function test_extractStyleBlocks_styleWithPadding() {
+        $this->assertExtractStyleBlocks(
+          "Alas<styled>Awesome</styled>\n<style>foo</style> Trendy!",
+          "Alas<styled>Awesome</styled>\n Trendy!",
+          array('foo')
+        );
+    }
+    
+    function assertCleanCSS($input, $expect = true) {
+        $filter = new HTMLPurifier_Filter_ExtractStyleBlocks();
+        if ($expect === true) $expect = $input;
+        $result = $filter->cleanCSS($input, $this->config, $this->context);
+        $this->assertIdentical($result, $expect);
+    }
+    
+    function test_cleanCSS_malformed() {
+        $this->assertCleanCSS('</style>', '');
+    }
+    
+    function test_cleanCSS_selector() {
+        $this->assertCleanCSS("a .foo #id div.cl#foo {\nfont-weight:700;\n}");
+    }
+    
+    function test_cleanCSS_angledBrackets() {
+        $this->assertCleanCSS(
+            ".class {\nfont-family:'</style>';\n}",
+            ".class {\nfont-family:'\\3C /style\\3E ';\n}"
+        );
+    }
+    
+    function test_cleanCSS_angledBrackets2() {
+        // CSSTidy's behavior in this case is wrong, and should be fixed
+        //$this->assertCleanCSS(
+        //    "span[title=\"</style>\"] {\nfont-size:12pt;\n}",
+        //    "span[title=\"\\3C /style\\3E \"] {\nfont-size:12pt;\n}"
+        //);
+    }
+    
+    function test_cleanCSS_bogus() {
+        $this->assertCleanCSS("div {bogus:tree;}", "div {\n}");
+    }
+    
+    function test_cleanCSS_escapeCodes() {
+        $this->assertCleanCSS(
+            ".class {\nfont-family:'\\3C /style\\3E ';\n}"
+        );
+    }
+    
+    function test_cleanCSS_noEscapeCodes() {
+        $filter = new HTMLPurifier_Filter_ExtractStyleBlocks(null, true);
+        $input = ".class {\nfont-family:'</style>';\n}";
+        $result = $filter->cleanCSS($input, $this->config, $this->context);
+        $this->assertIdentical($result, $input);
+    }
+    
+}

tests/HTMLPurifier/LexerTest.php

@@ -31,48 +31,6 @@ class HTMLPurifier_LexerTest extends HTMLPurifier_Harness
         $this->assertIsA($lexer, 'HTMLPurifier_Lexer_DirectLex');
     }
     
-    // HTMLPurifier_Lexer->extractStyleBlocks() --------------------------------
-    
-    function assertExtractStyleBlocks($html, $expect = true, $styles = array()) {
-        $lexer = HTMLPurifier_Lexer::create($this->config);
-        if ($expect === true) $expect = $html;
-        $result = $lexer->extractStyleBlocks($html, $this->config, $this->context);
-        $this->assertIdentical($result, $expect);
-        $this->assertIdentical($this->context->get('StyleBlocks'), $styles);
-    }
-    
-    function test_extractStyleBlocks_preserve() {
-        $this->assertExtractStyleBlocks('Foobar');
-    }
-    
-    function test_extractStyleBlocks_allStyle() {
-        $this->assertExtractStyleBlocks('<style>foo</style>', '', array('foo'));
-    }
-    
-    function test_extractStyleBlocks_multipleBlocks() {
-        $this->assertExtractStyleBlocks(
-          "<style>1</style><style>2</style>NOP<style>4</style>",
-          "NOP",
-          array('1', '2', '4')
-        );
-    }
-    
-    function test_extractStyleBlocks_blockWithAttributes() {
-        $this->assertExtractStyleBlocks(
-          '<style type="text/css">css</style>',
-          '',
-          array('css')
-        );
-    }
-    
-    function test_extractStyleBlocks_styleWithPadding() {
-        $this->assertExtractStyleBlocks(
-          "Alas<styled>Awesome</styled>\n<style>foo</style> Trendy!",
-          "Alas<styled>Awesome</styled>\n Trendy!",
-          array('foo')
-        );
-    }
-    
     // HTMLPurifier_Lexer->parseData() -----------------------------------------
     
     function assertParseData($input, $expect = true) {
@@ -553,17 +511,6 @@ class HTMLPurifier_LexerTest extends HTMLPurifier_Harness
         );
     }
     
-    function test_tokenizeHTML_extractStyleBlocks() {
-        $this->config->set('HTML', 'ExtractStyleBlocks', true);
-        $this->assertTokenization(
-            '<style type="text/css">.foo {text-align:center;}</style>Test',
-            array(
-                new HTMLPurifier_Token_Text('Test')
-            )
-        );
-        $this->assertIdentical($this->context->get('StyleBlocks'), array('.foo {text-align:center;}'));
-    }
-    
     /*
     
     function test_tokenizeHTML_() {

tests/index.php

@@ -16,6 +16,7 @@ $GLOBALS['HTMLPurifierTest'] = array();
 $GLOBALS['HTMLPurifierTest']['PEAR'] = false; // do PEAR tests
 $GLOBALS['HTMLPurifierTest']['PH5P'] = version_compare(PHP_VERSION, "5", ">=") && class_exists('DOMDocument');
 $simpletest_location = 'simpletest/'; // reasonable guess
+$csstidy_location = false;
 
 // load SimpleTest
 if (file_exists('../conf/test-settings.php')) include '../conf/test-settings.php';
@@ -24,6 +25,11 @@ require_once $simpletest_location . 'unit_tester.php';
 require_once $simpletest_location . 'reporter.php';
 require_once $simpletest_location . 'mock_objects.php';
 
+if ($csstidy_location !== false) {
+    require_once $csstidy_location . 'class.csstidy.php';
+    require_once $csstidy_location . 'class.csstidy_print.php';
+}
+
 error_reporting(E_ALL | E_STRICT); // after SimpleTest is loaded, turn on compile time errors
 
 // load Debugger

tests/test_files.php

@@ -75,6 +75,7 @@ $test_files[] = 'HTMLPurifier/ErrorCollectorTest.php';
 $test_files[] = 'HTMLPurifier/EncoderTest.php';
 $test_files[] = 'HTMLPurifier/EntityLookupTest.php';
 $test_files[] = 'HTMLPurifier/EntityParserTest.php';
+$test_files[] = 'HTMLPurifier/Filter/ExtractStyleBlocksTest.php';
 $test_files[] = 'HTMLPurifier/GeneratorTest.php';
 $test_files[] = 'HTMLPurifier/HTMLDefinitionTest.php';
 $test_files[] = 'HTMLPurifier/HTMLModuleManagerTest.php';