Merge pull request #71 from jaiminpan/master

add tag capbility for each segments

include/cppjieba/Jieba.hpp

@@ -2,7 +2,6 @@
 #define CPPJIEAB_JIEBA_H
 
 #include "QuerySegment.hpp"
-#include "PosTagger.hpp"
 //#include "LevelSegment.hpp"
 
 namespace cppjieba {
@@ -16,9 +15,9 @@ class Jieba {
       hmm_seg_(&model_),
       mix_seg_(&dict_trie_, &model_),
       full_seg_(&dict_trie_),
-      query_seg_(&dict_trie_, &model_),
+      query_seg_(&dict_trie_, &model_)
       //level_seg_(&dict_trie_),
-      pos_tagger_(&dict_trie_, &model_) {
+      {
   }
   ~Jieba() {
   }
@@ -61,7 +60,7 @@ class Jieba {
   }
   
   void Tag(const string& sentence, vector<pair<string, string> >& words) const {
-    pos_tagger_.Tag(sentence, words);
+    mix_seg_.Tag(sentence, words);
   }
   bool InsertUserWord(const string& word, const string& tag = UNKNOWN_TAG) {
     return dict_trie_.InsertUserWord(word, tag);
@@ -95,8 +94,6 @@ class Jieba {
   QuerySegment query_seg_;
   //LevelSegment level_seg_;
 
-  PosTagger pos_tagger_;
-  
 }; // class Jieba
 
 } // namespace cppjieba

include/cppjieba/MPSegment.hpp

@@ -6,11 +6,12 @@
 #include <cassert>
 #include "limonp/Logging.hpp"
 #include "DictTrie.hpp"
-#include "SegmentBase.hpp"
+#include "SegmentTagged.hpp"
+#include "PosTagger.hpp"
 
 namespace cppjieba {
 
-class MPSegment: public SegmentBase {
+class MPSegment: public SegmentTagged {
  public:
   MPSegment(const string& dictPath, const string& userDictPath = "")
     : dictTrie_(new DictTrie(dictPath, userDictPath)), isNeedDestroy_(true) {
@@ -25,9 +26,13 @@ class MPSegment: public SegmentBase {
     }
   }
 
+  void Cut(const string& sentence, vector<string>& words) const {
+    Cut(sentence, words, MAX_WORD_LENGTH);
+  }
+
   void Cut(const string& sentence,
         vector<string>& words,
-        size_t max_word_len = MAX_WORD_LENGTH) const {
+        size_t max_word_len) const {
     vector<Word> tmp;
     Cut(sentence, tmp, max_word_len);
     GetStringsFromWords(tmp, words);
@@ -64,6 +69,10 @@ class MPSegment: public SegmentBase {
     return dictTrie_;
   }
 
+  bool Tag(const string& src, vector<pair<string, string> >& res) const {
+    return tagger_.Tag(src, res, *this);
+  }
+
   bool IsUserDictSingleChineseWord(const Rune& value) const {
     return dictTrie_->IsUserDictSingleChineseWord(value);
   }
@@ -119,6 +128,8 @@ class MPSegment: public SegmentBase {
 
   const DictTrie* dictTrie_;
   bool isNeedDestroy_;
+  PosTagger tagger_;
+
 }; // class MPSegment
 
 } // namespace cppjieba

include/cppjieba/MixSegment.hpp

@@ -5,9 +5,10 @@
 #include "MPSegment.hpp"
 #include "HMMSegment.hpp"
 #include "limonp/StringUtil.hpp"
+#include "PosTagger.hpp"
 
 namespace cppjieba {
-class MixSegment: public SegmentBase {
+class MixSegment: public SegmentTagged {
  public:
   MixSegment(const string& mpSegDict, const string& hmmSegDict, 
         const string& userDict = "") 
@@ -20,7 +21,10 @@ class MixSegment: public SegmentBase {
   ~MixSegment() {
   }
 
-  void Cut(const string& sentence, vector<string>& words, bool hmm = true) const {
+  void Cut(const string& sentence, vector<string>& words) const {
+    Cut(sentence, words, true);
+  }
+  void Cut(const string& sentence, vector<string>& words, bool hmm) const {
     vector<Word> tmp;
     Cut(sentence, tmp, hmm);
     GetStringsFromWords(tmp, words);
@@ -84,9 +88,15 @@ class MixSegment: public SegmentBase {
   const DictTrie* GetDictTrie() const {
     return mpSeg_.GetDictTrie();
   }
+
+  bool Tag(const string& src, vector<pair<string, string> >& res) const {
+    return tagger_.Tag(src, res, *this);
+  }
+
  private:
   MPSegment mpSeg_;
   HMMSegment hmmSeg_;
+  PosTagger tagger_;
 
 }; // class MixSegment
 

include/cppjieba/PosTagger.hpp

@@ -1,8 +1,8 @@
 #ifndef CPPJIEBA_POS_TAGGING_H
 #define CPPJIEBA_POS_TAGGING_H
 
-#include "MixSegment.hpp"
 #include "limonp/StringUtil.hpp"
+#include "SegmentTagged.hpp"
 #include "DictTrie.hpp"
 
 namespace cppjieba {
@@ -14,24 +14,18 @@ static const char* const POS_X = "x";
 
 class PosTagger {
  public:
-  PosTagger(const string& dictPath,
+  PosTagger() {
-    const string& hmmFilePath,
-    const string& userDictPath = "")
-    : segment_(dictPath, hmmFilePath, userDictPath) {
-  }
-  PosTagger(const DictTrie* dictTrie, const HMMModel* model) 
-    : segment_(dictTrie, model) {
   }
   ~PosTagger() {
   }
 
-  bool Tag(const string& src, vector<pair<string, string> >& res) const {
+  bool Tag(const string& src, vector<pair<string, string> >& res, const SegmentTagged& segment) const {
     vector<string> CutRes;
-    segment_.Cut(src, CutRes);
+    segment.Cut(src, CutRes);
 
     const DictUnit *tmp = NULL;
     RuneStrArray runes;
-    const DictTrie * dict = segment_.GetDictTrie();
+    const DictTrie * dict = segment.GetDictTrie();
     assert(dict != NULL);
     for (vector<string>::iterator itr = CutRes.begin(); itr != CutRes.end(); ++itr) {
       if (!DecodeRunesInString(*itr, runes)) {
@@ -71,7 +65,6 @@ class PosTagger {
     return POS_ENG;
   }
 
-  MixSegment segment_;
 }; // class PosTagger
 
 } // namespace cppjieba

include/cppjieba/QuerySegment.hpp

@@ -24,7 +24,11 @@ class QuerySegment: public SegmentBase {
   }
   ~QuerySegment() {
   }
-  void Cut(const string& sentence, vector<string>& words, bool hmm = true) const {
+
+  void Cut(const string& sentence, vector<string>& words) const {
+    Cut(sentence, words, true);
+  }
+  void Cut(const string& sentence, vector<string>& words, bool hmm) const {
     vector<Word> tmp;
     Cut(sentence, tmp, hmm);
     GetStringsFromWords(tmp, words);

include/cppjieba/SegmentBase.hpp

@@ -17,9 +17,11 @@ class SegmentBase {
   SegmentBase() {
     XCHECK(ResetSeparators(SPECIAL_SEPARATORS));
   }
-  ~SegmentBase() {
+  virtual ~SegmentBase() {
   }
 
+  virtual void Cut(const string& sentence, vector<string>& words) const = 0;
+
   bool ResetSeparators(const string& s) {
     symbols_.clear();
     RuneStrArray runes;

include/cppjieba/SegmentTagged.hpp

@@ -0,0 +1,25 @@
+#ifndef CPPJIEBA_SEGMENTTAGGED_H
+#define CPPJIEBA_SEGMENTTAGGED_H
+
+#include "SegmentBase.hpp"
+
+namespace cppjieba {
+
+struct DictTrie;
+
+class SegmentTagged : public SegmentBase{
+ public:
+  SegmentTagged() {
+  }
+  virtual ~SegmentTagged() {
+  }
+
+  virtual bool Tag(const string& src, vector<pair<string, string> >& res) const = 0;
+
+  virtual const struct DictTrie* GetDictTrie() const = 0;
+
+}; // class SegmentTagged
+
+} // cppjieba
+
+#endif

test/unittest/pos_tagger_test.cpp

@@ -1,4 +1,4 @@
-#include "cppjieba/PosTagger.hpp"
+#include "cppjieba/MixSegment.hpp"
 #include "gtest/gtest.h"
 
 using namespace cppjieba;
@@ -13,7 +13,7 @@ static const char * const ANS_TEST3 = "[iPhone6:eng, 手机:n, 的:uj, 最大:a,
 //static const char * const ANS_TEST3 = "";
 
 TEST(PosTaggerTest, Test) {
-  PosTagger tagger("../dict/jieba.dict.utf8", "../dict/hmm_model.utf8");
+  MixSegment tagger("../dict/jieba.dict.utf8", "../dict/hmm_model.utf8");
   {
     vector<pair<string, string> > res;
     tagger.Tag(QUERY_TEST1, res);
@@ -23,7 +23,7 @@ TEST(PosTaggerTest, Test) {
   }
 }
 TEST(PosTagger, TestUserDict) {
-  PosTagger tagger("../dict/jieba.dict.utf8", "../dict/hmm_model.utf8", "../test/testdata/userdict.utf8");
+  MixSegment tagger("../dict/jieba.dict.utf8", "../dict/hmm_model.utf8", "../test/testdata/userdict.utf8");
   {
     vector<pair<string, string> > res;
     tagger.Tag(QUERY_TEST2, res);