Update module for custom matching.

.gitignore

@@ -5,4 +5,4 @@ __pycache__
 *.model
 *.log
 .DS_Store
-*log.txt
+*log.txt

Chatbot/RuleMatcher/customRuleBase.py

@@ -0,0 +1,63 @@
+import json
+
+from .rulebase import RuleBase
+
+class CustomRuleBase(RuleBase):
+
+    """
+    用於客製化比對的規則庫，每次比對完即清空規則
+    """
+
+    #TODO 客製化的「階段式對話」
+
+    def customMatch(self, sentence, apiKey):
+
+        """
+        比對 sentence 與用戶自定義的規則
+
+        Args:
+            - sentence : 用戶輸入
+            - apiKey   : 該名會員的聊天機器人金鑰
+        """
+        # 清空之前讀入的規則
+        self.rules.clear()
+
+        # 重新建構規則表
+        customRules = self.getCustomDomainRules(apiKey)
+        customRules = json.loads(customRules)
+        self.buildCustomRules(customRules)
+
+        # 進行比對
+        return self.match(sentence, threshold=customThreshold, root=apiKey)
+
+    def buildCustomRules(self, rules):
+
+        """
+        將讀入的規則從字典轉換為 Rule Class 型式
+
+        Args:
+            - rules: 由 json.loads 導出的字典型式的規則
+        """
+        assert self.model is not None, "Please load the model before loading rules."
+
+        for rule in rules:
+
+            domain = rule["domain"]
+            concepts_list = rule["concepts"]
+            children_list = rule["children"]
+            response = rule["response"]
+
+            if domain not in self.rules:
+                rule = Rule(domain, concepts_list, children_list, response, self.model)
+                self.rules[domain] = rule
+                if is_root:
+                    self.forest_base_roots.append(rule)
+            else:
+                print("[Rules]: Detect a duplicate domain name '%s'." % domain)
+
+    def getCustomDomainRules(self, key):
+        """
+        依照 apiKey 取得該用戶的規則集
+        """
+        #TODO
+        return None

Chatbot/RuleMatcher/rule/rule.json

@@ -44,7 +44,7 @@
 
   		],
   		"concepts": [
-  			"吃","餓","好吃","喝","渴","玩","好玩","逛","看","好看","去","走"
+  			"吃","餓","喝","好玩","逛"
   		],
   		"children": [
   			"吃",
@@ -172,7 +172,7 @@
         "domain": "病症",
         "response": [],
         "concepts": [
-            "病症","症狀","病徵"
+            "病症","症狀","病徵","嘔吐"
         ],
         "children": [
             "過敏",

Chatbot/RuleMatcher/rulebase.py

@@ -204,6 +204,9 @@ def match(self, sentence, topk=1, threshold=0, root=None):
         Return:
             a list holds the top k-th rules and the classification tree travel path.
         """
+
+        log = open("matching_log.txt",'w',encoding='utf-8')
+
         assert self.model is not None, "Please load the model before any match."
 
         result_list  = []
@@ -225,6 +228,14 @@ def match(self, sentence, topk=1, threshold=0, root=None):
             result_list = sorted(result_list, reverse=True , key=lambda k: k[0])
             top_domain  = result_list[0][1] # get the best matcher's term.
 
+            #Output matching_log.
+            log.write("---")
+            for result in result_list:
+                s,d,m = result
+                log.write("Sim: %f, Domain: %s, Matchee: %s\n" % (s,d,m))
+            log.write("---")
+
+
             if self.rules[top_domain].has_child():
                 result_list = []
                 term_trans += top_domain+'>'

Chatbot/chatbot.py

@@ -158,11 +158,6 @@ def get_task_handler(self, domain=None):
 
         return handler
 
-    def getCustomDomainRules(self, key):
-        """
-        """
-        #TODO
-        return None
 
     def getCustomQARules(self, key):
         """

Chatbot/console.py

@@ -5,21 +5,26 @@
 
 import jieba
 import jieba.analyse
+
 import RuleMatcher.rulebase as rulebase
+import RuleMatcher.customRuleBase as crb
 
 def main():
     console = Console()
     console.listen() # goto interactive mode.
 
 class Console(object):
 
+    """
+    Build some nlp function as an package.
+    """
+
     def __init__(self,model_path="model/ch-corpus-3sg.bin",
                  rule_path="RuleMatcher/rule/",
                  stopword="jieba_dict/stopword.txt",
                  jieba_dic="jieba_dict/dict.txt.big",
                  jieba_user_dic="jieba_dict/userdict.txt"):
         print("[Console] Building a console...")
-        print("*********************************")
         try:
 
             cur_dir = os.getcwd()
@@ -34,11 +39,8 @@ def __init__(self,model_path="model/ch-corpus-3sg.bin",
             self.rb = rulebase.RuleBase()
             print("[Console] Loading vector model...")
             self.rb.load_model(model_path)
-            print("[Console] Vector model has loaded.")
             print("[Console] Loading pre-defined rules.")
             self.rb.load_rules_from_dic(rule_path)
-            print("[Console] Rules have loaded.")
-            print("*********************************")
             print("[Console] Initialized successfully :>")
 
             os.chdir(cur_dir)
@@ -48,6 +50,9 @@ def __init__(self,model_path="model/ch-corpus-3sg.bin",
             print(repr(e))
             exit()
 
+        self.cusRuleBase = crb.CustomRuleBase()
+        self.cusRuleBase.model = self.rb.model
+
     def listen(self):
         #into interactive console
         while True:
@@ -141,7 +146,7 @@ def word_segment(self, sentence):
                 keyword.append(word)
         return keyword
 
-    def rule_match(self, sentence, best_only=False, search_from=None, segmented=False):
+    def rule_match(self, sentence, best_only=False, search_from=None, segmented=False, api_key=None):
 
         """
         Match the sentence with rules.
@@ -152,6 +157,7 @@ def rule_match(self, sentence, best_only=False, search_from=None, segmented=Fals
             - root      : a domain name, then the rule match will start
                           at searching from that domain, not from forest roots.
             - segmented : the sentence is segmented or not.
+            - api_key   : a key to fetch custom rules in the database.
         Return:
             - a list of candiate rule
             - the travel path of classification tree.
@@ -162,16 +168,20 @@ def rule_match(self, sentence, best_only=False, search_from=None, segmented=Fals
         else:
             keyword = self.word_segment(sentence)
 
-        if search_from is None: # use for rule matching.
-            result_list,path = self.rb.match(keyword,threshold=0.1)
-        else:  # use for reasoning.
-            result_list,path = self.rb.match(keyword,threshold=0.1,root=search_from)
+        if api_key is None:
+            if search_from is None: # use for classification (rule matching).
+                result_list,path = self.rb.match(keyword,threshold=0.1)
+            else:  # use for reasoning.
+                result_list,path = self.rb.match(keyword,threshold=0.1,root=search_from)
+        else:
+            result_list,path = self.cusRuleBase.customMatch()
 
         if best_only:
             return [result_list[0], path]
         else:
             return [result_list, path]
 
+
     def get_response(self, rule_id):
 
         """

Chatbot/jieba_dict/stopword.txt

@@ -1217,5 +1217,6 @@ $
 好像
 幾個
 想要
+想
 覺得
 直到