[Xmlstorage-commits] SF.net SVN: xmlstorage: [75] trunk/c++

[<mar...@us...>]

Revision: 75
          http://xmlstorage.svn.sourceforge.net/xmlstorage/?rev=75&view=rev
Author:   martinfuchs
Date:     2008-01-30 15:05:56 -0800 (Wed, 30 Jan 2008)

Log Message:
-----------
allow UTF-8 encoded umlaut characters in tag names

Modified Paths:
--------------
    trunk/c++/xmlstorage.cpp
    trunk/c++/xs-native.cpp

Modified: trunk/c++/xmlstorage.cpp
===================================================================
--- trunk/c++/xmlstorage.cpp	2008-01-19 18:00:57 UTC (rev 74)
+++ trunk/c++/xmlstorage.cpp	2008-01-30 23:05:56 UTC (rev 75)
@@ -2,7 +2,7 @@
  //
  // XML storage C++ classes version 1.2
  //
- // Copyright (c) 2004, 2005, 2006, 2007 Martin Fuchs <mar...@gm...>
+ // Copyright (c) 2004, 2005, 2006, 2007, 2008 Martin Fuchs <mar...@gm...>
  //
 
  /// \file xmlstorage.cpp
@@ -534,18 +534,33 @@
 }
 
 
+const char* get_xmlsym_end_utf8(const char* p)
+{
+	for(; *p; ++p) {
+		char c = *p;
+
+		if (c == '\xC3')	// UTF-8 escape character
+			++p;	//TODO only continue on umlaut characters
+		else if (!isalnum(c) && c!='_' && c!='-')
+			break;
+	}
+
+	return p;
+}
+
+
 void DocType::parse(const char* p)
 {
 	while(isspace((unsigned char)*p)) ++p;
 
 	const char* start = p;
-	while(isxmlsym(*p)) ++p;
+	p = get_xmlsym_end_utf8(p);
 	_name.assign(start, p-start);
 
 	while(isspace((unsigned char)*p)) ++p;
 
 	start = p;
-	while(isxmlsym(*p)) ++p;
+	p = get_xmlsym_end_utf8(p);
 	std::string keyword(p, p-start);	// "PUBLIC" or "SYSTEM"
 
 	while(isspace((unsigned char)*p)) ++p;

Modified: trunk/c++/xs-native.cpp
===================================================================
--- trunk/c++/xs-native.cpp	2008-01-19 18:00:57 UTC (rev 74)
+++ trunk/c++/xs-native.cpp	2008-01-30 23:05:56 UTC (rev 75)
@@ -93,7 +93,7 @@
 		_buffer_str.erase();
 	}
 
-	void append(char c)
+	void append(int c)
 	{
 		size_t wpos = _wptr-_buffer;
 
@@ -103,7 +103,7 @@
 			_wptr = _buffer + wpos;
 		}
 
-		*_wptr++ = c;
+		*_wptr++ = static_cast<char>(c);
 	}
 
 	const std::string& str(bool utf8)	// returns UTF-8 encoded buffer content
@@ -148,8 +148,7 @@
 		if (*q == '?')
 			++q;
 
-		while(isxmlsym(*q))
-			++q;
+		q = get_xmlsym_end_utf8(q);
 
 #ifdef XS_STRING_UTF8
 		return XS_String(p, q-p);
@@ -174,8 +173,7 @@
 		else if (*p == '?')
 			++p;
 
-		while(isxmlsym(*p))
-			++p;
+		p = get_xmlsym_end_utf8(p);
 
 		 // read attributes from buffer
 		while(*p && *p!='>' && *p!='/') {
@@ -184,8 +182,7 @@
 
 			const char* attr_name = p;
 
-			while(isxmlsym(*p))
-				++p;
+			p = get_xmlsym_end_utf8(p);
 
 			if (*p != '=')
 				break;	//@TODO error handling


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.