From 30072490990b4e9deaba64a15896260257fdaadd Mon Sep 17 00:00:00 2001
From: gwdwyy <wuyanyi09@gmail.com>
Date: Tue, 16 Jul 2013 22:40:57 +0800
Subject: [PATCH] add findPrefix into trie.cpp

---
 src/KeyWordExt.cpp |  2 +-
 src/Trie.cpp       | 45 ++++++++++++++++++++++++++++++++++++++++++---
 src/Trie.h         |  2 +-
 3 files changed, 44 insertions(+), 5 deletions(-)

diff --git a/src/KeyWordExt.cpp b/src/KeyWordExt.cpp
index 1c93b47..613d13c 100644
--- a/src/KeyWordExt.cpp
+++ b/src/KeyWordExt.cpp
@@ -286,7 +286,7 @@ namespace CppJieba
 		for(vector<WordInfo>::iterator it = wordInfos.begin(); it != wordInfos.end(); )
 		{
 			cout<<__FILE__<<__LINE__<<endl;
-			if(NULL != _priorPrefixTrie.findUtf8(it->word))
+			if(NULL != _priorPrefixTrie.findPrefix(it->word))
 			{
 				prior = *it;
 				it = wordInfos.erase(it);
diff --git a/src/Trie.cpp b/src/Trie.cpp
index d3fe34e..575fe41 100644
--- a/src/Trie.cpp
+++ b/src/Trie.cpp
@@ -112,8 +112,13 @@ namespace CppJieba
 		}
     }
 
-	const TrieNodeInfo* Trie::findUtf8(const string& utf8Str)
+	const TrieNodeInfo* Trie::findPrefix(const string& utf8Str)
 	{
+		if(NULL == _root)
+		{
+			LogFatal("trie not initted!");
+			return NULL;
+		}
 		if(utf8Str.empty())
 		{
 			LogError("utf8Str is empty");
@@ -122,10 +127,44 @@ namespace CppJieba
 		string uniStr = utf8ToUnicode(utf8Str);
 		if(uniStr.empty())
 		{
-			LogError("utf8ToUnicode return empty str");
+			LogError("utf8ToUnicode return empty star");
 			return NULL;
 		}
-		return find(uniStr);
+		if(uniStr.size() % 2)
+		{
+			LogError("utf8ToUnicode return uniStr illegal");
+			return NULL;
+		}
+		//find
+		TrieNode* p = _root;
+		TrieNodeInfo * res = NULL;
+		for(uint i = 0; i < uniStr.size(); i+=2)
+		{
+			ChUnicode chUni = twocharToUint16(uniStr[0], uniStr[i+1]);
+			if(p->isLeaf)
+			{
+				uint pos = p->nodeInfoVecPos;
+				if(pos < _nodeInfoVec.size())
+				{
+					res = &(_nodeInfoVec[pos]);
+				}
+				else
+				{
+					LogFatal("node's nodeInfoVecPos is out of _nodeInfoVec's range");
+					return NULL;
+				}
+				
+			}
+			if(p->hmap.find(chUni) == p->hmap.end())
+			{
+				break;
+			}
+			else
+			{
+				p = p->hmap[chUni];
+			}
+		}
+		return res;
 	}
 
 	const TrieNodeInfo* Trie::find(const string& uniStr)
diff --git a/src/Trie.h b/src/Trie.h
index f449e40..bead95b 100644
--- a/src/Trie.h
+++ b/src/Trie.h
@@ -70,9 +70,9 @@ namespace CppJieba
             void display();
 
 		public:
-			const TrieNodeInfo* findUtf8(const string& utf8Str);
 			const TrieNodeInfo* find(const string& uniStr);
 			const TrieNodeInfo* find(const ChUnicode* const chUniStr, size_t len);
+			const TrieNodeInfo* findPrefix(const string& utf8Str);
 
 		public:
 			double getWeight(const ChUnicode* uniStr, size_t len);