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

List:       kfm-devel
Subject:    Patch: allow KURL to parse #someReference?aQuery fault-tolerant
From:       Tobias Anton <anton () stud ! fbi ! fh-darmstadt ! de>
Date:       2001-07-28 11:29:59
[Download RAW message or body]

Hi!

Here's a patch that allows kurl to parse urls with reversed order of query 
and reference - forbidden according to the rfc, but there are sites out there 
using it this way. consistent to Mozilla's behaviour, and, i think, IE would 
tolerate it, too.

in combination with the second patch, this fixes konq's failure in the 
mozilla-testsuite:
http://mozilla.org/quality/browser/standards/javascript/tcmatrix/are001.html

Cheers,
Tobias



["kurl.patch" (text/x-diff)]

? kurl.patch
Index: kurl.cpp
===================================================================
RCS file: /home/kde/kdelibs/kdecore/kurl.cpp,v
retrieving revision 1.168
diff -u -r1.168 kurl.cpp
--- kurl.cpp	2001/07/19 07:42:41	1.168
+++ kurl.cpp	2001/07/28 08:56:53
@@ -521,6 +521,9 @@
   QChar* orig = buf;
   memcpy( buf, _url.unicode(), len * sizeof( QChar ) );
 
+  QChar delim;
+  QString tmp;
+
   uint pos = 0;
 
   // Node 1: Accept alpha or slash
@@ -554,7 +557,7 @@
     {
       m_strProtocol = QString( orig, pos ).lower();
       pos++;
-      goto Node11;
+      goto Node9;
     }
   else
     goto NodeErr;
@@ -687,36 +690,39 @@
     goto NodeOk;
   start = pos++;
 
-  // Node 9: Accept any character and # or terminate
- Node9:
-  while( buf[pos] != '#' && pos < len ) pos++;
-  if ( pos == len )
-    {
-      QString tmp( buf + start, len - start );
-      setEncodedPathAndQuery( tmp, encoding_hint );
-      // setEncodedPathAndQuery( QString( buf + start, pos - start ) );
+ Node9: // parse path until query or reference reached
+
+  while( buf[pos] != '#' && buf[pos]!='?' && pos < len ) pos++;
+
+  tmp = QString( buf + start, pos - start );
+  setEncodedPathAndQuery( tmp, encoding_hint );
+
+  if ( pos >= len-1 )
       goto NodeOk;
-    }
-  else if ( buf[pos] != '#' )
-    goto NodeErr;
-  setEncodedPathAndQuery( QString( buf + start, pos - start ), encoding_hint );
-  pos++;
 
-  // Node 10: Accept all the rest
-  m_strRef_encoded = QString( buf + pos, len - pos );
-  goto NodeOk;
+  start = pos + 1;
 
-  // Node 11 We need at least one character
- Node11:
-  start = pos;
-  if ( pos++ == len )
-    goto NodeOk; // Wrong, but since a fix was applied up top it is a non-issue here!!!!
-                 // Just for the record an opaque URL such as "mailto:" is always required
-                 // to have at least one more character other than a '/' following the colon.
-  // Node 12: Accept the res
-  setEncodedPathAndQuery( QString( buf + start, len - start ), encoding_hint );
-  goto NodeOk;
+ Node10: // parse query or reference depending on what comes first
+  delim = (buf[pos++]=='#'?'?':'#');
+
+  while(buf[pos]!=delim && pos < len) pos++;
 
+  if (pos == len)
+      goto NodeOk;
+
+  tmp = QString(buf + start, pos - start);
+  if (delim=='#')
+      m_strQuery_encoded = tmp;
+  else
+      m_strRef_encoded = tmp;
+
+ Node11: // feed the rest into the remaining variable
+  tmp = QString( buf + pos + 1, len - pos - 1);
+  if (delim == '#')
+      m_strRef_encoded = tmp;
+  else
+      m_strQuery_encoded = tmp;
+
  NodeOk:
   delete []orig;
   m_bIsMalformed = false; // Valid URL
@@ -985,6 +991,17 @@
   return tmp;
 }
 
+void KURL::setEncodedPath( const QString& _txt, int encoding_hint )
+{
+    m_strPath_encoded = _txt;
+
+  bool keepEncoded;
+  m_strPath = decode( m_strPath_encoded, &keepEncoded, encoding_hint );
+  if (!keepEncoded)
+     m_strPath_encoded = QString::null;
+}
+
+
 void KURL::setEncodedPathAndQuery( const QString& _txt, int encoding_hint )
 {
   int pos = _txt.find( '?' );
Index: kurl.h
===================================================================
RCS file: /home/kde/kdelibs/kdecore/kurl.h,v
retrieving revision 1.76
diff -u -r1.76 kurl.h
--- kurl.h	2001/07/19 07:42:41	1.76
+++ kurl.h	2001/07/28 08:56:54
@@ -229,6 +229,8 @@
    */
   void setEncodedPathAndQuery( const QString& _txt, int encoding_hint = 0 );
 
+  void setEncodedPath(const QString& _txt, int encoding_hint = 0 );
+
   /**
    * @return The concatenation if the encoded path , '?' and the encoded query.
    *

["kjs_html.diff" (text/x-diff)]

Index: ecma/kjs_html.cpp
===================================================================
RCS file: /home/kde/kdelibs/khtml/ecma/kjs_html.cpp,v
retrieving revision 1.104
diff -u -r1.104 kjs_html.cpp
--- ecma/kjs_html.cpp	2001/07/18 10:37:57	1.104
+++ ecma/kjs_html.cpp	2001/07/28 09:06:03
@@ -45,7 +45,9 @@
 #include <xml/dom_docimpl.h>
 #include <xml/dom2_eventsimpl.h>
 #include <khtmlview.h>
+#include <xml/dom_stringimpl.h>
 
+
 #include <kjs/operations.h>
 #include "kjs_dom.h"
 #include "kjs_html.h"
@@ -305,6 +307,7 @@
       if      (p == "disabled")        return Boolean(link.disabled());
       else if (p == "charset")         return getString(link.charset());
       else if (p == "href")            return getString(link.href());
+      else if (p == "hash")            return getString('#'+KURL(link.href().string()).ref());
       else if (p == "hreflang")        return getString(link.hreflang());
       else if (p == "media")           return getString(link.media());
       else if (p == "rel")             return getString(link.rel());
@@ -621,6 +624,7 @@
       else if (p == "coords")          return getString(anchor.coords());
       else if (p == "href")            return getString(anchor.href());
       else if (p == "hreflang")        return getString(anchor.hreflang());
+      else if (p == "hash")            return getString('#'+KURL(anchor.href().string()).ref());
       else if (p == "name")            return getString(anchor.name());
       else if (p == "rel")             return getString(anchor.rel());
       else if (p == "rev")             return getString(anchor.rev());
@@ -709,6 +713,7 @@
       else if (p == "alt")             return getString(area.alt());
       else if (p == "coords")          return getString(area.coords());
       else if (p == "href")            return getString(area.href());
+      else if (p == "hash")            return getString('#'+KURL(area.href().string()).ref());
       else if (p == "noHref")          return Boolean(area.noHref());
       else if (p == "shape")           return getString(area.shape());
       else if (p == "tabIndex")        return Number(area.tabIndex());


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