feature: add RemoveWord api

deps/cppjieba/DictTrie.hpp

@@ -61,6 +61,15 @@ class DictTrie {
     return true;
   }
 
+  bool DeleteUserWord(const string& word, const string& tag = UNKNOWN_TAG) {
+    DictUnit node_info;
+    if (!MakeNodeInfo(node_info, word, user_word_default_weight_, tag)) {
+      return false;
+    }
+    trie_->DeleteNode(node_info.word, &node_info);
+    return true;
+  }
+
   const DictUnit* Find(RuneStrArray::const_iterator begin, RuneStrArray::const_iterator end) const {
     return trie_->Find(begin, end);
   }

deps/cppjieba/Jieba.hpp

@@ -76,6 +76,10 @@ class Jieba {
     return dict_trie_.InsertUserWord(word,freq, tag);
   }
 
+  bool DeleteUserWord(const string& word, const string& tag = UNKNOWN_TAG) {
+    return dict_trie_.DeleteUserWord(word, tag);
+  }
+
   bool Find(const string& word)
   {
     return dict_trie_.Find(word);

deps/cppjieba/Trie.hpp

@@ -141,7 +141,33 @@ class Trie {
     assert(ptNode != NULL);
     ptNode->ptValue = ptValue;
   }
-
+  void DeleteNode(const Unicode& key, const DictUnit* ptValue) {
+      if (key.begin() == key.end()) {
+        return;
+      }
+      //定义一个NextMap迭代器
+      TrieNode::NextMap::const_iterator kmIter;
+      //定义一个指向root的TrieNode指针
+      TrieNode *ptNode = root_;
+      for (Unicode::const_iterator citer = key.begin(); citer != key.end(); ++citer) {
+        //链表不存在元素
+        if (NULL == ptNode->next) {
+          return;
+        }
+        kmIter = ptNode->next->find(*citer);
+        //如果map中不存在,跳出循环
+        if (ptNode->next->end() == kmIter) {
+              break;
+        }
+        //从unordered_map中擦除该项
+        ptNode->next->erase(*citer);
+        //删除该node
+        ptNode = kmIter->second;
+        delete ptNode;
+        break;
+      }
+      return;
+ }
  private:
   void CreateTrie(const vector<Unicode>& keys, const vector<const DictUnit*>& valuePointers) {
     if (valuePointers.empty() || keys.empty()) {

jieba.cpp

@@ -84,6 +84,10 @@ void AddWord(Jieba x, const char* word) {
   ((cppjieba::Jieba*)x)->InsertUserWord(word);
 }
 
+void RemoveWord(Jieba x, const char* word) {
+  ((cppjieba::Jieba*)x)->DeleteUserWord(word);
+}
+
 Word* Tokenize(Jieba x, const char* sentence, TokenizeMode mode, int is_hmm_used) {
   std::vector<cppjieba::Word> words;
   switch (mode) {

jieba.go

@@ -103,6 +103,12 @@ func (x *Jieba) AddWord(s string) {
 	C.AddWord(x.jieba, cstr)
 }
 
+func (x *Jieba) RemoveWord(s string) {
+	cstr := C.CString(s)
+	defer C.free(unsafe.Pointer(cstr))
+	C.RemoveWord(x.jieba, cstr)
+}
+
 func (x *Jieba) Tokenize(s string, mode TokenizeMode, hmm bool) []Word {
 	c_int_hmm := 0
 	if hmm {

jieba.h

@@ -28,6 +28,7 @@ char** CutAll(Jieba handle, const char* sentence);
 char** CutForSearch(Jieba handle, const char* sentence, int is_hmm_used);
 char** Tag(Jieba handle, const char* sentence);
 void AddWord(Jieba handle, const char* word);
+void RemoveWord(Jieba handle, const char* word);
 
 Word* Tokenize(Jieba x, const char* sentence, TokenizeMode mode, int is_hmm_used);