From 599c130bd9a5790740e575777bda05aecc98b483 Mon Sep 17 00:00:00 2001
From: aholic <ruochen.xu@gmail.com>
Date: Thu, 28 Nov 2013 10:49:40 +0800
Subject: [PATCH] make MixSegment looks better

---
 src/FullSegment.hpp  |  19 +++-----
 src/MixSegment.hpp   | 102 ++++++++++++++++++++++++++-----------------
 src/QuerySegment.hpp |  17 +++-----
 3 files changed, 73 insertions(+), 65 deletions(-)

diff --git a/src/FullSegment.hpp b/src/FullSegment.hpp
index 5fc3604..f1182fc 100644
--- a/src/FullSegment.hpp
+++ b/src/FullSegment.hpp
@@ -61,16 +61,12 @@ namespace CppJieba
         bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<Unicode>& res) const
         {
             assert(_getInitFlag());
-            //if (!_getInitFlag())
-            //{
-            //    LogError("not inited.");
-            //    return false;
-            //}
             if (begin >= end)
             {
                 LogError("begin >= end");
                 return false;
             }
+
             //resut of searching in trie tree
             vector<pair<uint, const TrieNodeInfo*> > tRes;
 
@@ -117,24 +113,21 @@ namespace CppJieba
 
         bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<string>& res) const
         {
-            if (!_getInitFlag())
+            assert(_getInitFlag());
+            if (begin >= end)
             {
-                LogError("not inited.");
-                return false;
-            }
-            if (begin > end)
-            {
-                LogError("begin > end");
+                LogError("begin >= end");
                 return false;
             }
+
             vector<Unicode> uRes;
             if (!cut(begin, end, uRes))
             {
                 LogError("get unicode cut result error.");
                 return false;
             }
-            string tmp;
 
+            string tmp;
             for (vector<Unicode>::const_iterator uItr = uRes.begin(); uItr != uRes.end(); uItr++)
             {
                 if (TransCode::encode(*uItr, tmp))
diff --git a/src/MixSegment.hpp b/src/MixSegment.hpp
index 272a57f..79c40d1 100644
--- a/src/MixSegment.hpp
+++ b/src/MixSegment.hpp
@@ -55,73 +55,95 @@ namespace CppJieba
         public:
             using SegmentBase::cut;
         public:
-
-            virtual bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<string>& res)const
+            virtual bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<Unicode>& res) const
             {
-                //if(!_getInitFlag())
-                //{
-                //    LogError("not inited.");
-                //    return false;
-                //}
                 assert(_getInitFlag());
-                if(begin == end)
+                if(begin >= end)
                 {
+                    LogError("begin >= end");
                     return false;
                 }
+
                 vector<TrieNodeInfo> infos;
                 if(!_mpSeg.cut(begin, end, infos))
                 {
                     LogError("mpSeg cutDAG failed.");
                     return false;
                 }
-                Unicode unico;
+
                 vector<Unicode> hmmRes;
-                string tmp;
-                for(uint i= 0; i < infos.size(); i++)
+                Unicode piece;
+                for (uint i = 0, j = 0; i < infos.size(); i++)
                 {
-                    TransCode::encode(infos[i].word,tmp);
-                    if(1 == infos[i].word.size())
+                    //if mp get a word, it's ok, put it into result
+                    if (1 != infos[i].word.size())
                     {
-                        unico.push_back(infos[i].word[0]);
+                        res.push_back(infos[i].word);
+                        continue;
                     }
-                    else
+
+                    // if mp get a single one, collect it in sequence
+                    j = i;
+                    while (j < infos.size() && infos[j].word.size() == 1)
                     {
-                        if(!unico.empty())
-                        {
-                            hmmRes.clear();
-                            if(!_hmmSeg.cut(unico.begin(), unico.end(), hmmRes))
-                            {
-                                LogError("_hmmSeg cut failed.");
-                                return false;
-                            }
-                            for(uint j = 0; j < hmmRes.size(); j++)
-                            {
-                                TransCode::encode(hmmRes[j], tmp);
-                                res.push_back(tmp);
-                            }
-                        }
-                        unico.clear();
-                        TransCode::encode(infos[i].word, tmp);
-                        res.push_back(tmp);
+                        piece.push_back(infos[j].word[0]);
+                        j++;
                     }
-                }
-                if(!unico.empty())
-                {
-                    hmmRes.clear();
-                    if(!_hmmSeg.cut(unico.begin(), unico.end(), hmmRes))
+
+                    // cut the sequence with hmm
+                    if (!_hmmSeg.cut(piece.begin(), piece.end(), hmmRes))
                     {
                         LogError("_hmmSeg cut failed.");
                         return false;
                     }
-                    for(uint j = 0; j < hmmRes.size(); j++)
+
+                    //put hmm result to return
+                    for (uint k = 0; k < hmmRes.size(); k++)
                     {
-                        TransCode::encode(hmmRes[j], tmp);
-                        res.push_back(tmp);
+                        res.push_back(hmmRes[k]);
                     }
+
+                    //clear tmp vars
+                    piece.clear();
+                    hmmRes.clear();
+
+                    //let i jump over this piece
+                    i = j - 1;
                 }
 
                 return true;
             }
+
+            virtual bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<string>& res)const
+            {
+                assert(_getInitFlag());
+                if(begin >= end)
+                {
+                    LogError("begin >= end");
+                    return false;
+                }
+
+                vector<Unicode> uRes;
+                if (!cut(begin, end, uRes))
+                {
+                    LogError("get unicode cut result error.");
+                    return false;
+                }
+
+                string tmp;
+                for (vector<Unicode>::const_iterator uItr = uRes.begin(); uItr != uRes.end(); uItr++)
+                {
+                    if (TransCode::encode(*uItr, tmp))
+                    {
+                        res.push_back(tmp);
+                    }
+                    else
+                    {
+                        LogError("encode failed.");
+                    }
+                }
+                return true;
+            }
     };
 }
 
diff --git a/src/QuerySegment.hpp b/src/QuerySegment.hpp
index de268c5..ae5987e 100644
--- a/src/QuerySegment.hpp
+++ b/src/QuerySegment.hpp
@@ -62,16 +62,12 @@ namespace CppJieba
         bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<Unicode>& res) const
         {
             assert(_getInitFlag());
-            //if (!_getInitFlag())
-            //{
-            //    LogError("not inited.");
-            //    return false;
-            //}
             if (begin >= end)
             {
                 LogError("begin >= end");
                 return false;
             }
+
             //use hmm cut first
             vector<Unicode> hmmRes;
             if (!_hmmSeg.cut(begin, end, hmmRes))
@@ -107,16 +103,13 @@ namespace CppJieba
 
         bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<string>& res) const
         {
-            if (!_getInitFlag())
+            assert(_getInitFlag());
+            if (begin >= end)
             {
-                LogError("not inited.");
-                return false;
-            }
-            if (begin > end)
-            {
-                LogError("begin > end");
+                LogError("begin >= end");
                 return false;
             }
+
             vector<Unicode> uRes;
             if (!cut(begin, end, uRes))
             {