[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-commits
Subject:    KDE/kdelibs/khtml/css
From:       Maks Orlovich <maksim () kde ! org>
Date:       2010-04-27 23:56:43
Message-ID: 20100428000008.071F3AC8A5 () svn ! kde ! org
[Download RAW message or body]

SVN commit 1119920 by orlovich:

Reject invalid pseudos here.

This brings our test results on Selectors API test suite to:
CSS2.1: 98.4%: 992 passed, 16 failed
CSS3:   98.9%: 2163 passed, 25 failed

16 failures on both are due to the tests expecting foo(undefined) to behave 
differently from foo(), which is a pretty odd thing to do in ECMAScript.

The 9 CSS3-only ones have to do with nested :not, probably needs a special reject. 
(Will need to bug someone for advice on that)

We do lack NAMESPACE_ERR production, though. (Which surprisingly isn't covered, and is also annoying to 
implement)

But man, this is way easier than XPath (less than a day of work!) and likely a lot faster.
Good thing I decided to put it into the feature plan, I guess.



 M  +18 -0     cssparser.cpp  


--- trunk/KDE/kdelibs/khtml/css/cssparser.cpp #1119919:1119920
@@ -322,6 +322,24 @@
     selectors.clear();
     setupParser("@-khtml-selectors{", string, "} ");
     runParser();
+
+    // Make sure to detect problems with pseudos, too
+    bool ok = true;
+    for (int i = 0; i < selectors.size(); ++i) {
+        DOM::CSSSelector* sel = selectors[i];
+        if(sel->match == CSSSelector::PseudoClass || sel->match == CSSSelector::PseudoElement) {
+            if (sel->pseudoType() == CSSSelector::PseudoOther) {
+                ok = false;
+                break;
+            }
+        }
+    }
+
+    if (!ok) {
+        qDeleteAll(selectors);
+        selectors.clear();
+    }
+    
     return selectors;
 }
 
[prev in list] [next in list] [prev in thread] [next in thread]