From 8e8a68352b3b8376313610e9fc06450fe4897655 Mon Sep 17 00:00:00 2001
From: wyy <wuyanyi09@gmail.com>
Date: Fri, 6 Dec 2013 22:52:59 -0800
Subject: [PATCH] modify some gbk enc for more robust

---
 src/ChineseFilter.hpp | 10 +++++++++-
 src/SegmentBase.hpp   |  7 ++++++-
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/src/ChineseFilter.hpp b/src/ChineseFilter.hpp
index c969607..6a5dd66 100644
--- a/src/ChineseFilter.hpp
+++ b/src/ChineseFilter.hpp
@@ -112,7 +112,7 @@ namespace CppJieba
         }
         char x = 0x80;
         int resFlag = (str[0] & x ? 1 : 0);
-        resLen = 1;
+        resLen = 0;
         if(!resFlag)
         {
             while(resLen < len && !(str[resLen] & x))
@@ -124,9 +124,17 @@ namespace CppJieba
         {
             while(resLen < len && (str[resLen] & x))
             {
+#ifdef CPPJIEBA_GBK
+                resLen += 2;
+#else
                 resLen ++;
+#endif
             }
         }
+        if(resLen > len)
+        {
+            return -1;
+        }
         return resFlag;
     }
 }
diff --git a/src/SegmentBase.hpp b/src/SegmentBase.hpp
index 5536047..114f2f3 100644
--- a/src/SegmentBase.hpp
+++ b/src/SegmentBase.hpp
@@ -31,7 +31,7 @@ namespace CppJieba
                     LogError("not inited.");
                     return false;
                 }
-                const char * cstr = str.c_str();
+                const char * const cstr = str.c_str();
                 uint size = str.size();
                 uint offset = 0;
                 string subs;
@@ -60,6 +60,11 @@ namespace CppJieba
                         }
                         cut(unico.begin(), unico.end(), res);
                     }
+                    if(len == 0)
+                    {
+                        LogFatal("str[%s] illegal.", cstr);
+                        return false;
+                    }
                     offset += len;
                 }
                 return true;