From 81c2d3caf1ca7c069dfdef8af2eb6b92cdd85611 Mon Sep 17 00:00:00 2001
From: wyy <wuyanyi09@gmail.com>
Date: Sat, 7 Dec 2013 07:45:06 -0800
Subject: [PATCH] modify calcDAG try to speed up

---
 src/MPSegment.hpp | 49 +++++++++++++++++++++++++++--------------------
 src/Trie.hpp      | 16 ++++++++--------
 2 files changed, 36 insertions(+), 29 deletions(-)

diff --git a/src/MPSegment.hpp b/src/MPSegment.hpp
index 42d8e5c..e195b9a 100644
--- a/src/MPSegment.hpp
+++ b/src/MPSegment.hpp
@@ -126,13 +126,9 @@ namespace CppJieba
                     return false;
                 }
                 SegmentContext segContext;
-                for(Unicode::const_iterator it = begin; it != end; it++)
-                {
-                    segContext.push_back(SegmentChar(*it));
-                }
 
                 //calc DAG
-                if(!_calcDAG(segContext))
+                if(!_calcDAG(begin, end, segContext))
                 {
                     LogError("_calcDAG failed.");
                     return false;
@@ -154,36 +150,47 @@ namespace CppJieba
             }
 
         private:
-            bool _calcDAG(SegmentContext& segContext)const
+            bool _calcDAG(Unicode::const_iterator begin, Unicode::const_iterator end, SegmentContext& segContext) const
             {
-                if(segContext.empty())
+                if(begin >= end)
                 {
-                    LogError("segContext empty.");
+                    LogError("begin >= end.");
                     return false;
                 }
-
-                Unicode unicode;
-                vector<pair<uint, const TrieNodeInfo*> > vp;
-                for(uint i = 0; i < segContext.size(); i++)
+                for(Unicode::const_iterator it = begin; it != end; it++)
                 {
-                    unicode.clear();
-                    for(uint j = i ; j < segContext.size(); j++)
-                    {
-                        unicode.push_back(segContext[j].uniCh);
-                    }
+                    segContext.push_back(SegmentChar(*it));
+                }
 
+                vector<pair<uint, const TrieNodeInfo*> > vp;
+                //for(Unicode::const_iterator it = begin; it != end; it++)
+                //{
+                //    segContext.push_back(SegmentChar(*it));
+                //}
+                for(Unicode::const_iterator it = begin; it != end; it++)
+                //for(uint i = 0; i < segContext.size(); i++)
+                {
+                    //unicode.clear();
+                    //for(uint j = i; j < segContext.size(); j++)
+                    //{
+                    //    unicode.push_back(segContext[j].uniCh);
+                    //}
+                    
+                    segContext.push_back(SegmentChar(*it));
+                    SegmentChar& back = segContext.back();
+                    int i = it - begin;
                     vp.clear();
-                    if(_trie.find(unicode, vp))
+                    if(_trie.find(it, end, vp))
                     {
                         for(uint j = 0; j < vp.size(); j++)
                         {
                             uint nextp = vp[j].first + i;
-                            segContext[i].dag[nextp] = vp[j].second; 
+                            back.dag[nextp] = vp[j].second; 
                         }
                     }
-                    if(segContext[i].dag.end() == segContext[i].dag.find(i))
+                    if(back.dag.end() == back.dag.find(i))
                     {
-                        segContext[i].dag[i] = NULL;
+                        back.dag[i] = NULL;
                     }
                 }
                 return true;
diff --git a/src/Trie.hpp b/src/Trie.hpp
index 8282587..c39277a 100644
--- a/src/Trie.hpp
+++ b/src/Trie.hpp
@@ -244,14 +244,14 @@ namespace CppJieba
                 return !res.empty();
             }
 
-            bool find(const Unicode& unico, vector<pair<uint, const TrieNodeInfo*> >& res)const
-            {
-                if (!unico.empty())
-                {
-                    return find(unico.begin(), unico.end(), res);
-                }
-                return false;
-            }
+            //bool find(const Unicode& unico, vector<pair<uint, const TrieNodeInfo*> >& res)const
+            //{
+            //    if (!unico.empty())
+            //    {
+            //        return find(unico.begin(), unico.end(), res);
+            //    }
+            //    return false;
+            //}
 
         public:
             double getMinLogFreq()const{return _minLogFreq;};