def predict_nlu(model_path, text):
interpreter = Interpreter.load(model_path)
regex_interpreter = RegexInterpreter()
if text.startswith(INTENT_MESSAGE_PREFIX):
loop = asyncio.get_event_loop()
result = loop.run_until_complete(regex_interpreter.parse(text))
else:
result = interpreter.parse(text)
return result
def run_cmdline(model_path, component_builder=None):
interpreter = Interpreter.load(model_path, component_builder)
regex_interpreter = RegexInterpreter()
print_success(
"NLU model loaded. Type a message and press enter to parse it.")
while True:
print_success("Next message:")
message = input().strip()
if message.startswith(INTENT_MESSAGE_PREFIX):
loop = asyncio.get_event_loop()
result = loop.run_until_complete(regex_interpreter.parse(message))
else:
result = interpreter.parse(message)
print(json.dumps(result, indent=2))
class TrackerParse(object):
def __init__(self, agent=None):
self.interpreter = None
self.regexInterpreter = None
self.agent = agent
self.loop = asyncio.get_event_loop()
def nluModelLoad(self, nluModlePath):
self.interpreter = Interpreter.load(nluModlePath)
self.regexInterpreter = RegexInterpreter()
def nluPredict(self, message):
nluInfoDict = None
if message.startswith(INTENT_MESSAGE_PREFIX):
nluInfoDict = self.loop.run_until_complete(
self.regexInterpreter.parse(message))
else:
nluInfoDict = self.interpreter.parse(message)
return nluInfoDict
def corePredict(self, nluInfoDict, senderId):
responses, _ = self.loop.run_until_complete(
self.agent.handle_text(nlu_info=nluInfoDict,
sender_id=senderId,
output_channel=None))
# print("AI response message: {}".format(response))
responsesData = str()
for response in responses:
if 'text' in response.keys():
responsesData += response['text']
return responsesData
def coreModelLoad(self, coreModlePath):
action_endpoint = EndpointConfig(url="http://localhost:5055/webhook")
self.agent = HciAgent.load(coreModlePath,
action_endpoint=action_endpoint)
def trackerParse(self, senderId, fw):
"""提取对话状态内容
Args:
fw: 保存对话的文件
"""
tracker = self.agent.tracker_store.get_or_create_tracker(senderId)
# 解析对话管理状态
index = 0
for event in tracker.events:
index += 1
if index == 1:
continue
#name = event.type_name
if event.type_name == 'action':
actionName = event.action_name
actionContent = " - {}".format(actionName)
fw.write("{}\n".format(actionContent))
elif event.type_name == 'user':
intent = event.intent['name']
entities = {}
for entity in event.entities:
entities[entity['entity']] = entity['value']
if entities:
userContent = "* {}{}".format(intent, entities)
else:
userContent = "* {}".format(intent)
fw.write("{}\n".format(userContent))
elif event.type_name == 'slot':
slot = {}
slot[event.key] = event.value
slotContent = " - slot{}".format(slot)
fw.write("{}\n".format(slotContent))
elif event.type_name == 'restart':
# 标志一个对话story的结束
break
# 一个对话故事结束添加空行
fw.write("\n")
def getLatestTracker(self, senderId):
# 获取最新的AI和用户信息（AI的表达，用户的表达，用户的意图）
tracker = self.agent.tracker_store.get_or_create_tracker(senderId)
# 每个类都应有get和set方法
intent = tracker.latest_message.intent
entities = {}
for entity in tracker.latest_message.entities:
entities[entity['entity']] = entity['value']
return intent, entities
def getTestData(self, testTextPath):
"""将文本数据转为格式化数据"""
peopleTalkData = []
# 使用python正则来解析文本
storyPattern = '^ *## *'
# botPattern = '^ *[bB][oO][tT] *[:：] *'
userPattern = '^ *[yY][oO][uU] *[:：] *'
element = {}
storyId = None
elementUserTalk = []
index = 0
with open(testTextPath, 'r', encoding='utf-8') as fp:
for line in fp.readlines():
line = line.rstrip()
storyMatch = re.match(storyPattern, line)
userMsgMatch = re.match(userPattern, line)
if storyMatch:
element[storyId] = elementUserTalk
peopleTalkData.append(element)
element = {}
elementUserTalk = []
storyId = line[storyMatch.end():]
index += 1
elif userMsgMatch:
elementUserTalk.append(line[userMsgMatch.end():])
element[storyId] = elementUserTalk
peopleTalkData.append(element)
peopleTalkData = peopleTalkData[1:len(peopleTalkData)]
print("story number is: {}".format(str(index)))
return peopleTalkData
def storiesActionCompare(self, testStories, predictStories):
"""对比测试集与预测的差异，求取精准度"""
pass
