[Pingus-CVS] r3975 - trunk/pingus/src

[grumbel at BerliOS]

Author: grumbel
Date: 2009-02-26 15:07:49 +0100 (Thu, 26 Feb 2009)
New Revision: 3975

Modified:
   trunk/pingus/src/utf8_iterator.cpp
   trunk/pingus/src/utf8_iterator.hpp
Log:
Some docu added, fixing added missing throw

Modified: trunk/pingus/src/utf8_iterator.cpp
===================================================================
--- trunk/pingus/src/utf8_iterator.cpp  2009-02-26 14:05:48 UTC (rev 3974)
+++ trunk/pingus/src/utf8_iterator.cpp  2009-02-26 14:07:49 UTC (rev 3975)
@@ -119,44 +119,53 @@
 {
   uint32_t c1 = (unsigned char) text[p+0];
 
-  if (has_multibyte_mark(c1)) std::runtime_error("Malformed utf-8 sequence");
-
-  if ((c1 & 0200) == 0000) {
-    // 0xxx.xxxx: 1 byte sequence
-    p+=1;
-    return c1;
-  }
-  else if ((c1 & 0340) == 0300) {
-    // 110x.xxxx: 2 byte sequence
-    if(p+1 >= text.size()) throw std::range_error("Malformed utf-8 sequence");
-    uint32_t c2 = (unsigned char) text[p+1];
-    if (!has_multibyte_mark(c2)) throw std::runtime_error("Malformed utf-8 
sequence");
-    p+=2;
-    return (c1 & 0037) << 6 | (c2 & 0077);
-  }
-  else if ((c1 & 0360) == 0340) {
-    // 1110.xxxx: 3 byte sequence
-    if(p+2 >= text.size()) throw std::range_error("Malformed utf-8 sequence");
-    uint32_t c2 = (unsigned char) text[p+1];
-    uint32_t c3 = (unsigned char) text[p+2];
-    if (!has_multibyte_mark(c2)) throw std::runtime_error("Malformed utf-8 
sequence");
-    if (!has_multibyte_mark(c3)) throw std::runtime_error("Malformed utf-8 
sequence");
-    p+=3;
-    return (c1 & 0017) << 12 | (c2 & 0077) << 6 | (c3 & 0077);
-  }
-  else if ((c1 & 0370) == 0360) {
-    // 1111.0xxx: 4 byte sequence
-    if(p+3 >= text.size()) throw std::range_error("Malformed utf-8 sequence");
-    uint32_t c2 = (unsigned char) text[p+1];
-    uint32_t c3 = (unsigned char) text[p+2];
-    uint32_t c4 = (unsigned char) text[p+4];
-    if (!has_multibyte_mark(c2)) throw std::runtime_error("Malformed utf-8 
sequence");
-    if (!has_multibyte_mark(c3)) throw std::runtime_error("Malformed utf-8 
sequence");
-    if (!has_multibyte_mark(c4)) throw std::runtime_error("Malformed utf-8 
sequence");
-    p+=4;
-    return (c1 & 0007) << 18 | (c2 & 0077) << 12 | (c3 & 0077) << 6 | (c4 & 
0077);
-  }
-  throw std::runtime_error("Malformed utf-8 sequence");
+  if (has_multibyte_mark(c1)) 
+    {
+      throw std::runtime_error("Malformed utf-8 sequence");
+    }
+  else if ((c1 & 0200) == 0000) 
+    {
+      // 0xxx.xxxx: 1 byte sequence
+      p+=1;
+      return c1;
+    }
+  else if ((c1 & 0340) == 0300) 
+    {
+      // 110x.xxxx: 2 byte sequence
+      if(p+1 >= text.size()) throw std::range_error("Malformed utf-8 
sequence");
+      uint32_t c2 = (unsigned char) text[p+1];
+      if (!has_multibyte_mark(c2)) throw std::runtime_error("Malformed utf-8 
sequence");
+      p+=2;
+      return (c1 & 0037) << 6 | (c2 & 0077);
+    }
+  else if ((c1 & 0360) == 0340) 
+    {
+      // 1110.xxxx: 3 byte sequence
+      if(p+2 >= text.size()) throw std::range_error("Malformed utf-8 
sequence");
+      uint32_t c2 = (unsigned char) text[p+1];
+      uint32_t c3 = (unsigned char) text[p+2];
+      if (!has_multibyte_mark(c2)) throw std::runtime_error("Malformed utf-8 
sequence");
+      if (!has_multibyte_mark(c3)) throw std::runtime_error("Malformed utf-8 
sequence");
+      p+=3;
+      return (c1 & 0017) << 12 | (c2 & 0077) << 6 | (c3 & 0077);
+    }
+  else if ((c1 & 0370) == 0360) 
+    {
+      // 1111.0xxx: 4 byte sequence
+      if(p+3 >= text.size()) throw std::range_error("Malformed utf-8 
sequence");
+      uint32_t c2 = (unsigned char) text[p+1];
+      uint32_t c3 = (unsigned char) text[p+2];
+      uint32_t c4 = (unsigned char) text[p+4];
+      if (!has_multibyte_mark(c2)) throw std::runtime_error("Malformed utf-8 
sequence");
+      if (!has_multibyte_mark(c3)) throw std::runtime_error("Malformed utf-8 
sequence");
+      if (!has_multibyte_mark(c4)) throw std::runtime_error("Malformed utf-8 
sequence");
+      p+=4;
+      return (c1 & 0007) << 18 | (c2 & 0077) << 12 | (c3 & 0077) << 6 | (c4 & 
0077);
+    }
+  else
+    {
+      throw std::runtime_error("Malformed utf-8 sequence");
+    }
 }
 
 // FIXME: Get rid of exceptions in this code

Modified: trunk/pingus/src/utf8_iterator.hpp
===================================================================
--- trunk/pingus/src/utf8_iterator.hpp  2009-02-26 14:05:48 UTC (rev 3974)
+++ trunk/pingus/src/utf8_iterator.hpp  2009-02-26 14:07:49 UTC (rev 3975)
@@ -24,11 +24,17 @@
 class UTF8
 {
 public:
+  /** 
+   * Returns the number of characters in a UTF-8 string 
+   */
   static std::string::size_type length(const std::string& str);
+
   static std::string substr(const std::string& text, std::string::size_type 
pos, std::string::size_type n);
   static std::string::const_iterator advance(std::string::const_iterator it, 
std::string::size_type n = 1);
 
-  /** return true if a linebreak is allowed after this character */
+  /** 
+   * return true if a linebreak is allowed after this character 
+   */
   static bool is_linebreak_character(uint32_t unicode);
 
   /**