From: Mikko Rasa <tdb@tdb.fi>
Date: Mon, 23 Mar 2009 11:49:11 +0000 (+0000)
Subject: Fix the UTF-8 decoder
X-Git-Tag: strings-1.1~3
X-Git-Url: http://git.tdb.fi/?p=libs%2Fcore.git;a=commitdiff_plain;h=a0a5c796a6ec11a13c49912672a82bf1857bbc71

Fix the UTF-8 decoder
---

diff --git a/source/utf8.cpp b/source/utf8.cpp
index 08dd69a..dd01150 100644
--- a/source/utf8.cpp
+++ b/source/utf8.cpp
@@ -71,22 +71,14 @@ UnicodeChar Utf8::Decoder::decode_char(const string &str, string::const_iterator
 		UnicodeChar result=(*j++)&(mask-1);
 
 		unsigned k;
-		for(k=1; (k<bytes && j!=str.end()); ++k)
-		{
-			if((*j&0xC0)!=0x80)
-			{
-				result=error("Incomplete UTF-8 character");
-				i=j;
-				return result;
-			}
+		for(k=1; (k<bytes && j!=str.end() && (*j&0xC0)==0x80); ++k)
 			result=(result<<6) | ((*j++)&0x3F);
-		}
 
 		if(k<bytes)
-			result=error("Incomplete UTF-8 character at end of input");
-		else if(!result>>(bytes*6-6))
+			result=error("Incomplete UTF-8 character");
+		else if(!(result>>(bytes*5-4)) || !(result>>7))
 			result=error("Denormalized UTF-8 multibyte sequence");
-		else if(result>0x10FFFF)
+		else if(result>0x10FFFF || (result>=0xD800 && result<=0xDFFF))
 			result=error("Invalid Unicode code point");
 
 		i=j;