def run(query):
#if __name__ == '__main__':
#初始化jb分词器
T.jieba_initialize()
#切换到语料库所在工作目录
mybot_path = './'
os.chdir(mybot_path)
mybot = aiml.Kernel()
mybot.learn(os.path.split(os.path.realpath(__file__))[0]+"/resources/std-startup.xml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/bye.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/tools.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/bad.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/funny.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/OrdinaryQuestion.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/Common conversation.aiml")
if len(query) > 60:
answer = '句子长度过长'
elif query.strip() == '':
answer = mybot.respond('无')
else:
message = T.wordSegment(query)
words = T.postag(query)
response = mybot.respond(message)
if response == '':
answer = mybot.respond('找不到答案')
elif response[0] == '#':
if response.__contains__("searchbaike"):
res = response.split(':')
entity = str(res[1]).replace(" ","")
attr = str(res[2]).replace(" ","")
ans = baike.query(entity, attr)
if '找不到' not in ans:
answer = ans
elif ans.__contains__('找不到'):
answer = search_summary.kwquery(query)
if len(answer) == 0:
answer = mybot.respond('找不到答案')
elif len(answer) == 1:
print(answer)
answer = answer[0].strip().replace(' ','').replace("\n","")
else:
answer = '找不到答案'
else:
answer = '找不到答案'
else:
answer = search_summary.kwquery(query)
if len(answer) == 0:
answer = mybot.respond('找不到答案')
elif len(answer) == 1:
answer = answer[0].strip().replace(' ','').replace("\n","")
else:
answer = '找不到答案'
return answer
def qa():
#初始化jb分词器
T.jieba_initialize()
#切换到语料库所在工作目录
mybot_path = './'
# os.chdir(mybot_path)
mybot = aiml.Kernel()
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/std-startup.xml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/Common conversation.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] + "/resources/bye.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] + "/resources/tools.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] + "/resources/bad.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] + "/resources/funny.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/OrdinaryQuestion.aiml")
# mybot.respond('Load Doc Snake')
#载入百科属性列表
print '''
.----------------. .-----------------. .----------------. .----------------. .----------------.
| .--------------. || .--------------. || .--------------. || .--------------. || .--------------. |
| | _______ | || | ____ _____ | || | __ | || | ___ ____ | || | _________ | |
| | / ___ | | || ||_ \|_ _| | || | / \ | || | |_ ||_ _| | || | |_ ___ | | |
| | | (__ \_| | || | | \ | | | || | / /\ \ | || | | |_/ / | || | | |_ \_| | |
| | '.___`-. | || | | |\ \| | | || | / /__\ \ | || | | __'. | || | | _| _ | |
| | |`\____) | | || | _| |_\ |_ | || | _/ / \ \_ | || | _| | \ \_ | || | _| |___/ | | |
| | |_______.' | || ||_____|\____| | || ||____| |____|| || | |____||____| | || | |_________| | |
| | | || | | || | | || | | || | | |
| '--------------' || '--------------' || '--------------' || '--------------' || '--------------' |
'----------------' '----------------' '----------------' '----------------' '----------------'
Eric：你好，我是Eric。╭(╯^╰)╮
'''
while True:
input_message = raw_input("Enter your message >> ")
if len(input_message) > 60:
print mybot.respond("句子长度过长")
continue
elif input_message.strip() == '':
print mybot.respond("无")
continue
print input_message
message = T.wordSegment(input_message)
# 去标点
print 'word Seg:' + message
print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message)
print "======="
print response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
# 百科搜索
elif response[0] == '#':
# 匹配百科
if response.__contains__("searchbaike"):
print "searchbaike"
print response
res = response.split(':')
#实体
entity = str(res[1]).replace(" ", "")
#属性
attr = str(res[2]).replace(" ", "")
print entity + '<---->' + attr
ans = baike.query(entity, attr)
# 如果命中答案
if type(ans) == list:
print 'Eric：' + QAT.ptranswer(ans, False)
continue
elif ans.decode('utf-8').__contains__(u'::找不到'):
#百度摘要+Bing摘要
print "通用搜索"
ans = search_summary.kwquery(input_message)
# 匹配不到模版，通用查询
elif response.__contains__("NoMatchingTemplate"):
print "NoMatchingTemplate"
ans = search_summary.kwquery(input_message)
if len(ans) == 0:
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
elif len(ans) > 1:
print "不确定候选答案"
print 'Eric: '
for a in ans:
print a.encode("utf8")
else:
print 'Eric：' + ans[0].encode("utf8")
# 匹配模版
else:
print 'Eric：' + response
def QA(input_message, mybot):
findAns = False
reply = ''
ansdict = {}
dbname = 'zwgx' #数据库名
dbip = 'localhost' #数据库IPlocalhost
dbport = 3306 #数据库端口
dbusername = '******' #数据库用户名
dbpassword = '******' #数据库密码root
schoolname = ''
intention = ''
if len(input_message) > 60:
reply = mybot.respond("句子长度过长")
findAns = True
elif input_message.strip() == '无':
reply = mybot.respond("无")
findAns = True
if (findAns == False):
#print input_message
#传入一个b''未加工的对象
message = T.wordSegment(input_message)
# 分词去标点
if message == 'q':
exit()
else:
print 'word Seg:' + message
print utf_to_bgk('词性：')
words = T.postag(input_message) #词性标注 格式 词/词性
for w in words:
print w.word, w.flag
if w.flag == 'school':
try:
#先用自定义的分词处理得到对应的词性，然后根据词性到数据库查询
db = pymysql.connect(host=dbip,
user=dbusername,
passwd=dbpassword,
db=dbname,
charset="utf8")
cursor = db.cursor()
sql = u"SELECT `学校名` FROM 学校简称 WHERE `简称`='" + w.word + "'"
# 执行SQL语句
cursor.execute(sql)
# 获取所有记录列表,已验证数据库查询跑通
results = cursor.fetchall()
#print results
#替换简称
if len(results) > 0:
input_message = input_message.replace(
w.word, results[0][0]).__str__()
w.flag = 'nt'
w.word = results[0][0]
print utf_to_bgk(input_message), utf_to_bgk(
w.word), utf_to_bgk(w.flag)
# 关闭数据库连接
db.close()
except Exception as e:
print(e)
# 识别学校简称并配对数据库中已存内容
if w.flag == 'x' or w.flag == 'nt':
try:
db = pymysql.connect(host=dbip,
user=dbusername,
passwd=dbpassword,
db=dbname,
charset="utf8")
cursor = db.cursor()
sql = u"SELECT `属性`,`内容` FROM school WHERE `学校`='" + w.word + "'"
# 执行SQL语句
cursor.execute(sql)
# 获取所有记录列表
results = cursor.fetchall()
#print u'flag转化之后的查询',results
if len(results) > 0:
for row in results:
ansdict[row[0]] = row[1]
#print row[0],row[1]
# reply +=row[0].encode("utf8")
# reply+=" ".encode("utf8")
# shuxing=raw_input('Frank：你想了解什么属性 ' + reply+">>")
# sql = u"SELECT `内容` FROM school WHERE `学校`='" + w.word + u"'AND `属性`='"+shuxing+"'"
# cursor.execute(sql)
# results = cursor.fetchall()
# if len(results)>0:
# print "Frank： "+results[0][0].encode("utf8")
# reply=results[0][0].encode("utf8")
# return reply
# 关闭数据库连接
#print u'查询之后的结果储存',ansdict 将数据库中的所有信息写入到ansdict中
db.close()
except Exception as e:
print(e)
#todo： 每个词去找查数据库可以优化一下 加一下词性判断
#获得学校的名称
if FindSchool(dbip, dbusername, dbpassword, dbname,
w.word) != "":
schoolname = FindSchool(dbip, dbusername, dbpassword,
dbname, w.word)
uni = input_message.strip().decode('utf-8')
print u'查看返回值', uni, utf_to_bgk(input_message.strip())
response = mybot.respond(
input_message.strip()) #如果未给传入参数转化为utf8则报错
print "======="
#print response
print "=======+"
if response == "":
reply = mybot.respond('找不到答案')
findAns = True
print 'Frank1：' + utf_to_bgk(reply)
# *********************************************************************************
# 百科搜索 aiml机器人没有没有
elif response[0] == '#':
# 匹配百科
# if response.__contains__("searchbaike"):
# print "searchbaike"
# print response
# res = response.split(':')
# # 实体
# entity = str(res[1]).replace(" ", "")
# # 属性
# attr = str(res[2]).replace(" ", "")
# print entity + '<---->' + attr
#
# ans = baike.query(entity, attr)
# # 如果命中答案
# if type(ans) == list:
# print 'Frank：' + QAT.ptranswer(ans, False)
# reply = QAT.ptranswer(ans, False)
# findAns = True
# elif ans.decode('utf-8').__contains__(u'::找不到'):
# # 百度摘要+Bing摘要
# print "通用搜索"
# ans = search_summary.kwquery(input_message)
#
# # 匹配不到模版，通用查询
# elif response.__contains__("NoMatchingTemplate"):
# print "NoMatchingTemplate"
#
#当复杂问题时，通过分类器模型进行分类再查询
if (schoolname != ""):
sock = socket(AF_INET, SOCK_STREAM)
sock.connect(('127.0.0.1', 50009))
sock.sendall(input_message.encode("utf-8"))
intention = sock.recv(1024)
sock.close()
print utf_to_bgk(intention), u'经过分类器处理后的结果'
#经过dl识别后分类问题，如果问题在数据库中，即把问题分类为数据库的一个属性，再调用属性值，可以增加数据库的属性分类和值
if unicode(intention) in ansdict:
reply = ansdict[unicode(intention)]
#print 'Frank：' + reply.encode("utf8")
#print 'Frank2：' + utf_to_bgk(reply)
#如果问题没有在数据库预存储
else:
TempDict = search_summary.kwquery(input_message, intention,
schoolname)
ansdict['schoolname'] = TempDict['schoolname']
ansdict['intention'] = TempDict['intention']
ansdict['index'] = TempDict['index']
ans = TempDict['answer']
if (findAns == False):
if len(ans) == 0:
ans = mybot.respond('找不到答案')
#print 'Frank3：' + utf_to_bgk(ans)
reply = ans
findAns = True
elif len(ans) > 1:
print u"不确定候选答案"
print 'Frank4: '
for a in ans:
print a.encode("utf8")
reply += a.encode("utf8") + '\n'
findAns = True
else:
#print 'Frank5：' + ans[0].encode("utf8")
reply = ans[0].encode("utf8")
findAns = True
# 匹配模版
else:
print 'Frank6：' + utf_to_bgk(response)
reply = response
findAns = True
ansdict['baidu'] = reply
json_s = json.dumps(ansdict)
return json_s
while True:
input_message = raw_input("Enter your message >> ")
if len(input_message) > 60:
print mybot.respond("句子长度过长")
continue
elif input_message.strip() == '':
print mybot.respond("无")
continue
print input_message
message = T.wordSegment(input_message)
# 去标点
print 'word Seg:' + message
print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message)
print "======="
print response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
# 百科搜索
elif response[0] == '#':
def ws():
user_socket = request.environ.get('wsgi.websocket') # type:WebSocket
while 1:
msg =user_socket.receive()
question = json.loads(msg)
q = question['data']['mine']['content']
msg =q
input_message = str(msg).encode('utf-8')
if len(input_message) > 60:
answer = mybot.respond("句子长度过长")
# continue
elif input_message.strip() == '':
answer = mybot.respond("无话可说")
# continue
# print input_message
message = T.wordSegment(input_message)
# 去标点
# print 'word Seg:'+ message
# print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message) # 在AIML数据集里寻找答案
print "======="
if response[0] == '#':
print response + 'mark'
else:
answer = response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
answer = ans
# 百科搜索
elif response[0] == '#' or len(response) < 1:
# 匹配百科
if response.__contains__("searchbaike"):
print "searchbaike"
print response
res = response.split(':')
# 实体
entity = str(res[1]).replace(" ", "")
# 属性
attr = str(res[2]).replace(" ", "")
print entity + '<---->' + attr
ans = baike.query(entity, attr)
# 如果命中答案
print type(ans)
if type(ans) == list:
answer = '回答：' + QAT.ptranswer(ans, False)
# continue
elif ans.decode('utf-8').__contains__(u'::找不到'):
# 百度摘要+Bing摘要
print "通用搜索"
answer = search_summary.kwquery(input_message)
# 匹配不到模版，通用查询
elif response.__contains__("NoMatchingTemplate"):
print "NoMatchingTemplate"
ans = search_summary.kwquery(input_message)
print type(ans)
if len(ans) == 0:
ans = mybot.respond('找不到答案')
answer = '回答：' + ans
elif len(ans) > 1:
print "不确定候选答案"
answer = ans[0]
print 'Eric: '
for a in ans:
print a.encode("utf-8")
else:
answer = '回答：' + ans[0].encode("utf-8")
# 匹配模版
else:
answer = '回答：' + response
s = '展开全部'
print type(answer).__name__
if (type(answer).__name__ == 'list' and len(answer)>0) or '唔... 怎么回答...'in answer or '天气' in msg:
answer = geta(msg)
answer = answer
print 'wocao'
else:
#print answer
if s in str(answer):
print answer
answer = str(answer).replace('\n', '').replace('展开全部', "").split('已赞过')[0]
print 'OS' +answer
print user_socket,msg
res =answer
a = {
"username": "******",
"avatar": "https://robot.rszhang.top/images/icon/nv/0.jpg",
"id": "-2", # //消息的来源ID（如果是私聊，则是用户id，如果是群聊，则是群组id）
"type": "friend", # //聊天窗口来源类型，从发送消息传递的to里面获取
"content": res, # //消息内容
"cid": 0, # //消息id，可不传。除非你要对消息进行一些操作（如撤回）
"mine": True, # //是否我发送的消息，如果为true，则会显示在右方
"fromid": "100000", # /消息的发送者id（比如群组中的某个消息发送者），可用于自动解决浏览器多窗口时的一些问题
"timestamp": 1467475443306, # //服务端时间戳毫秒数。注意：如果你返回的是标准的 unix 时间戳，记得要 *1000
}
user_socket.send(json.dumps(a))
def kwquery(query):
#分词 去停用词 抽取关键词
keywords = []
words = T.postag(query)
for k in words:
# 只保留名词
if k.flag.__contains__("n"):
# print k.flag
# print k.word
keywords.append(k.word)
answer = []
text = ''
# 找到百科的答案就置1
flag = 0
# 抓取百度前10条的摘要
soup_baidu = To.get_html_baidu('https://www.baidu.com/s?wd=' +
quote(query))
for i in range(1, 10):
if soup_baidu == None:
break
results = soup_baidu.find(id=i)
if results == None:
print "百度摘要找不到答案"
break
# print '============='
# print results.attrs
# print type(results.attrs)
# print results['class']
#判断是否有mu,如果第一个是百度知识图谱的 就直接命中答案
if results.attrs.has_key('mu') and i == 1:
# print results.attrs["mu"]
r = results.find(class_='op_exactqa_s_answer')
# print r
if r == None:
print "百度知识图谱找不到答案"
# continue
else:
# print r.get_text()
print "百度知识图谱找到答案"
answer.append(r.get_text().strip())
flag = 1
break
#古诗词判断
if results.attrs.has_key('mu') and i == 1:
r = results.find(class_="op_exactqa_detail_s_answer")
if r == None:
print "百度诗词找不到答案"
# continue
else:
# print r.get_text()
print "百度诗词找到答案"
answer.append(r.get_text().strip())
flag = 1
break
#计算器
if results.attrs.has_key(
'mu') and i == 1 and results.attrs['mu'].__contains__(
'http://open.baidu.com/static/calculator/calculator.html'):
r = results.find('div').find_all('td')[1].find_all('div')[1]
if r == None:
print "计算器找不到答案"
# continue
else:
# print r.get_text()
print "计算器找到答案"
answer.append(r.get_text().strip())
flag = 1
break
# 百度知道答案
if results.attrs.has_key('mu') and i == 1:
r = results.find(class_='op_best_answer_question_link')
if r == None:
print "百度知道图谱找不到答案"
else:
print "百度知道图谱找到答案"
url = r['href']
zhidao_soup = To.get_html_zhidao(url)
r = zhidao_soup.find(class_='bd answer').find('pre')
answer.append(r.get_text())
flag = 1
break
if results.find("h3") != None:
# 百度知道
if results.find("h3").find("a").get_text().__contains__(
u"百度知道") and i == 1:
url = results.find("h3").find("a")['href']
if url == None:
print "百度知道图谱找不到答案"
continue
else:
print "百度知道图谱找到答案"
zhidao_soup = To.get_html_zhidao(url)
r = zhidao_soup.find(class_='bd answer')
if r == None:
continue
else:
r = r.find('pre')
answer.append(r.get_text().strip())
flag = 1
break
# 百度百科
if results.find("h3").find("a").get_text().__contains__(
u"百度百科") and i == 1:
url = results.find("h3").find("a")['href']
if url == None:
print "百度百科找不到答案"
continue
else:
print "百度百科找到答案"
baike_soup = To.get_html_baike(url)
r = baike_soup.find(class_='lemma-summary')
if r == None:
continue
else:
r = r.get_text().replace("\n", "").strip()
answer.append(r)
flag = 1
break
text += results.get_text()
if flag == 1:
return answer
#获取bing的摘要
soup_bing = To.get_html_bing('https://www.bing.com/search?q=' +
quote(query))
# 判断是否在Bing的知识图谱中
# bingbaike = soup_bing.find(class_="b_xlText b_emphText")
bingbaike = soup_bing.find(class_="bm_box")
if bingbaike != None:
if bingbaike.find_all(class_="b_vList")[1] != None:
if bingbaike.find_all(class_="b_vList")[1].find("li") != None:
print "Bing知识图谱找到答案"
flag = 1
answer.append(bingbaike.get_text())
# print "====="
# print answer
# print "====="
return answer
else:
print "Bing知识图谱找不到答案"
results = soup_bing.find(id="b_results")
bing_list = results.find_all('li')
for bl in bing_list:
temp = bl.get_text()
if temp.__contains__(u" - 必应网典"):
print "查找Bing网典"
url = bl.find("h2").find("a")['href']
if url == None:
print "Bing网典找不到答案"
continue
else:
print "Bing网典找到答案"
bingwd_soup = To.get_html_bingwd(url)
r = bingwd_soup.find(class_='bk_card_desc').find("p")
if r == None:
continue
else:
r = r.get_text().replace("\n", "").strip()
answer.append(r)
flag = 1
break
if flag == 1:
return answer
text += results.get_text()
# print text
# 如果再两家搜索引擎的知识图谱中都没找到答案，那么就分析摘要
if flag == 0:
#分句
cutlist = [u"。", u"?", u".", u"_", u"-", u":", u"！", u"？"]
temp = ''
sentences = []
for i in range(0, len(text)):
if text[i] in cutlist:
if temp == '':
continue
else:
# print temp
sentences.append(temp)
temp = ''
else:
temp += text[i]
# 找到含有关键词的句子,去除无关的句子
key_sentences = {}
for s in sentences:
for k in keywords:
if k in s:
key_sentences[s] = 1
# 根据问题制定规则
# 识别人名
target_list = {}
for ks in key_sentences:
# print ks
words = T.postag(ks)
for w in words:
# print "====="
# print w.word
if w.flag == ("nr"):
if target_list.has_key(w.word):
target_list[w.word] += 1
else:
target_list[w.word] = 1
# 找出最大词频
sorted_lists = sorted(target_list.items(),
lambda x, y: cmp(x[1], y[1]),
reverse=True)
# print len(target_list)
#去除问句中的关键词
sorted_lists2 = []
# 候选队列
for i, st in enumerate(sorted_lists):
# print st[0]
if st[0] in keywords:
continue
else:
sorted_lists2.append(st)
print "返回前n个词频"
answer = []
for i, st in enumerate(sorted_lists2):
# print st[0]
# print st[1]
if i < 3:
# print st[0]
# print st[1]
answer.append(st[0])
# print answer
return answer
def kwquery(query):
#分词 去停用词 抽取关键词
keywords = []
words = T.postag(query)
for k in words:
# 只保留名词
if k.flag.__contains__("n"):
# print k.flag
# print k.word
keywords.append(k.word)
answer = []
text = ''
# 找到答案就置1
flag = 0
# 抓取百度前10条的摘要
soup_baidu = To.get_html_baidu('https://www.baidu.com/s?wd='+quote(query))
for i in range(1,10):
if soup_baidu == None:
break
results = soup_baidu.find(id=i)
if results == None:
print "百度摘要找不到答案"
break
# print '============='
# print results.attrs
# print type(results.attrs)
# print results['class']
#判断是否有mu,如果第一个是百度知识图谱的 就直接命中答案
if results.attrs.has_key('mu') and i == 1:
# print results.attrs["mu"]
r = results.find(class_='op_exactqa_s_answer')
if r == None:
print "百度知识图谱找不到答案"
else:
# print r.get_text()
print "百度知识图谱找到答案"
answer.append(r.get_text().strip())
flag = 1
break
#古诗词判断
if results.attrs.has_key('mu') and i == 1:
r = results.find(class_="op_exactqa_detail_s_answer")
if r == None:
print "百度诗词找不到答案"
else:
# print r.get_text()
print "百度诗词找到答案"
answer.append(r.get_text().strip())
flag = 1
break
#万年历 & 日期
if results.attrs.has_key('mu') and i == 1 and results.attrs['mu'].__contains__('http://open.baidu.com/calendar'):
r = results.find(class_="op-calendar-content")
if r == None:
print "百度万年历找不到答案"
else:
# print r.get_text()
print "百度万年历找到答案"
answer.append(r.get_text().strip().replace("\n","").replace(" ",""))
flag = 1
break
if results.attrs.has_key('tpl') and i == 1 and results.attrs['tpl'].__contains__('calendar_new'):
r = results.attrs['fk'].replace("6018_","")
print r
if r == None:
print "百度万年历新版找不到答案"
# continue
else:
# print r.get_text()
print "百度万年历新版找到答案"
answer.append(r)
flag = 1
break
#计算器
if results.attrs.has_key('mu') and i == 1 and results.attrs['mu'].__contains__('http://open.baidu.com/static/calculator/calculator.html'):
r = results.find('div').find_all('td')[1].find_all('div')[1]
if r == None:
print "计算器找不到答案"
# continue
else:
# print r.get_text()
print "计算器找到答案"
answer.append(r.get_text().strip())
flag = 1
break
# 百度知道答案
if results.attrs.has_key('mu') and i == 1:
r = results.find(class_='op_best_answer_question_link')
if r == None:
print "百度知道图谱找不到答案"
else:
print "百度知道图谱找到答案"
url = r['href']
zhidao_soup = To.get_html_zhidao(url)
r = zhidao_soup.find(class_='bd answer').find('pre')
if r == None:
r = zhidao_soup.find(class_='bd answer').find(class_='line content')
answer.append(r.get_text())
flag = 1
break
if results.find("h3") != None:
# 百度知道
if results.find("h3").find("a").get_text().__contains__(u"百度知道") and (i == 1 or i ==2):
url = results.find("h3").find("a")['href']
if url == None:
print "百度知道图谱找不到答案"
continue
else:
print "百度知道图谱找到答案"
zhidao_soup = To.get_html_zhidao(url)
r = zhidao_soup.find(class_='bd answer')
if r == None:
continue
else:
r = r.find('pre')
if r == None :
r = zhidao_soup.find(class_='bd answer').find(class_='line content')
answer.append(r.get_text().strip())
flag = 1
break
# 百度百科
if results.find("h3").find("a").get_text().__contains__(u"百度百科") and (i == 1 or i ==2):
url = results.find("h3").find("a")['href']
if url == None:
print "百度百科找不到答案"
continue
else:
print "百度百科找到答案"
baike_soup = To.get_html_baike(url)
r = baike_soup.find(class_='lemma-summary')
if r == None:
continue
else:
r = r.get_text().replace("\n","").strip()
answer.append(r)
flag = 1
break
text += results.get_text()
if flag == 1:
return answer
#获取bing的摘要
soup_bing = To.get_html_bing('https://www.bing.com/search?q='+quote(query))
# 判断是否在Bing的知识图谱中
# bingbaike = soup_bing.find(class_="b_xlText b_emphText")
bingbaike = soup_bing.find(class_="bm_box")
if bingbaike != None:
if bingbaike.find_all(class_="b_vList")[1] != None:
if bingbaike.find_all(class_="b_vList")[1].find("li") != None:
print "Bing知识图谱找到答案"
flag = 1
answer.append(bingbaike.get_text())
# print "====="
# print answer
# print "====="
return answer
else:
print "Bing知识图谱找不到答案"
results = soup_bing.find(id="b_results")
bing_list = results.find_all('li')
for bl in bing_list:
temp = bl.get_text()
if temp.__contains__(u" - 必应网典"):
print "查找Bing网典"
url = bl.find("h2").find("a")['href']
if url == None:
print "Bing网典找不到答案"
continue
else:
print "Bing网典找到答案"
bingwd_soup = To.get_html_bingwd(url)
r = bingwd_soup.find(class_='bk_card_desc').find("p")
if r == None:
continue
else:
r = r.get_text().replace("\n","").strip()
answer.append(r)
flag = 1
break
if flag == 1:
return answer
text += results.get_text()
# print text
# 如果再两家搜索引擎的知识图谱中都没找到答案，那么就分析摘要
if flag == 0:
#分句
cutlist = [u"。",u"?",u".", u"_", u"-",u":",u"！",u"？"]
temp = ''
sentences = []
for i in range(0,len(text)):
if text[i] in cutlist:
if temp == '':
continue
else:
# print temp
sentences.append(temp)
temp = ''
else:
temp += text[i]
# 找到含有关键词的句子,去除无关的句子
key_sentences = {}
for s in sentences:
for k in keywords:
if k in s:
key_sentences[s]=1
# 根据问题制定规则
# 识别人名
target_list = {}
for ks in key_sentences:
# print ks
words = T.postag(ks)
for w in words:
# print "====="
# print w.word
if w.flag == ("nr"):
if target_list.has_key(w.word):
target_list[w.word] += 1
else:
target_list[w.word] = 1
# 找出最大词频
sorted_lists = sorted(target_list.items(), lambda x, y: cmp(x[1], y[1]), reverse=True)
# print len(target_list)
#去除问句中的关键词
sorted_lists2 = []
# 候选队列
for i, st in enumerate(sorted_lists):
# print st[0]
if st[0] in keywords:
continue
else:
sorted_lists2.append(st)
print "返回前n个词频"
answer = []
for i,st in enumerate(sorted_lists2):
# print st[0]
# print st[1]
if i< 3:
# print st[0]
# print st[1]
answer.append(st[0])
# print answer
return answer
def QA(input_message, mybot):
findAns = False
reply = ''
ansdict = {}
dbname = 'zwgx' #数据库名
dbip = '106.14.124.221' #数据库IPlocalhost
dbport = 3306 #数据库端口
dbusername = '******' #数据库用户名
dbpassword = '******' #数据库密码root
schoolname = ''
intention = ''
if len(input_message) > 60:
reply = mybot.respond("句子长度过长")
findAns = True
elif input_message.strip() == '无':
reply = mybot.respond("无")
findAns = True
if (findAns == False):
# print input_message
message = T.wordSegment(input_message)
# 分词去标点
if message == 'q':
exit()
else:
print 'word Seg:' + message
print '词性：'
words = T.postag(input_message)
for w in words:
print w.word, w.flag
if w.flag == 'school':
try:
db = pymysql.connect(host=dbip,
user=dbusername,
passwd=dbpassword,
db=dbname,
charset="utf8")
cursor = db.cursor()
sql = u"SELECT `学校名` FROM 学校简称 WHERE `简称`='" + w.word + "'"
# 执行SQL语句
cursor.execute(sql)
# 获取所有记录列表
results = cursor.fetchall()
#替换简称
if len(results) > 0:
input_message = input_message.replace(
w.word, results[0][0]).__str__()
w.flag = 'nt'
w.word = results[0][0]
# 关闭数据库连接
db.close()
except Exception as e:
print(e)
# 识别学校简称并配对数据库中已存内容
if w.flag == 'x' or w.flag == 'nt':
try:
db = pymysql.connect(host=dbip,
user=dbusername,
passwd=dbpassword,
db=dbname,
charset="utf8")
cursor = db.cursor()
sql = u"SELECT `属性`,`内容` FROM school WHERE `学校`='" + w.word + "'"
# 执行SQL语句
cursor.execute(sql)
# 获取所有记录列表
results = cursor.fetchall()
if len(results) > 0:
for row in results:
ansdict[row[0]] = row[1]
#print row[0],row[1]
# reply +=row[0].encode("utf8")
# reply+=" ".encode("utf8")
# shuxing=raw_input('Frank：你想了解什么属性 ' + reply+">>")
# sql = u"SELECT `内容` FROM school WHERE `学校`='" + w.word + u"'AND `属性`='"+shuxing+"'"
# cursor.execute(sql)
# results = cursor.fetchall()
# if len(results)>0:
# print "Frank： "+results[0][0].encode("utf8")
# reply=results[0][0].encode("utf8")
# return reply
# 关闭数据库连接
db.close()
except Exception as e:
print(e)
#todo： 每个词去找查数据库可以优化一下 加一下词性判断
if FindSchool(dbip, dbusername, dbpassword, dbname,
w.word) != "":
schoolname = FindSchool(dbip, dbusername, dbpassword,
dbname, w.word)
response = mybot.respond(input_message.strip())
print "======="
print response
print "======="
if response == "":
reply = mybot.respond('找不到答案')
findAns = True
print 'Frank：' + reply
# *********************************************************************************
# 百科搜索
elif response[0] == '#':
# 匹配百科
# if response.__contains__("searchbaike"):
# print "searchbaike"
# print response
# res = response.split(':')
# # 实体
# entity = str(res[1]).replace(" ", "")
# # 属性
# attr = str(res[2]).replace(" ", "")
# print entity + '<---->' + attr
#
# ans = baike.query(entity, attr)
# # 如果命中答案
# if type(ans) == list:
# print 'Frank：' + QAT.ptranswer(ans, False)
# reply = QAT.ptranswer(ans, False)
# findAns = True
# elif ans.decode('utf-8').__contains__(u'::找不到'):
# # 百度摘要+Bing摘要
# print "通用搜索"
# ans = search_summary.kwquery(input_message)
#
# # 匹配不到模版，通用查询
# elif response.__contains__("NoMatchingTemplate"):
# print "NoMatchingTemplate"
if (schoolname != ""):
sock = socket(AF_INET, SOCK_STREAM)
sock.connect(('127.0.0.1', 50009))
sock.sendall(input_message.encode("utf-8"))
intention = sock.recv(1024)
sock.close()
print intention
if unicode(intention) in ansdict:
reply = ansdict[unicode(intention)]
print 'Frank：' + reply.encode("utf8")
else:
TempDict = search_summary.kwquery(input_message, intention,
schoolname)
ansdict['schoolname'] = TempDict['schoolname']
ansdict['intention'] = TempDict['intention']
ansdict['index'] = TempDict['index']
ans = TempDict['answer']
#*********************************************************************************
if (findAns == False):
if len(ans) == 0:
ans = mybot.respond('找不到答案')
print 'Frank：' + ans
reply = ans
findAns = True
elif len(ans) > 1:
print "不确定候选答案"
print 'Frank: '
for a in ans:
print a.encode("utf8")
reply += a.encode("utf8") + '\n'
findAns = True
else:
print 'Frank：' + ans[0].encode("utf8")
reply = ans[0].encode("utf8")
findAns = True
# 匹配模版
else:
print 'Frank：' + response
reply = response
findAns = True
ansdict['baidu'] = reply
json_s = json.dumps(ansdict)
return json_s
while True:
input_message = raw_input("Enter your message >> ")
if len(input_message) > 60:
print mybot.respond("句子长度过长")
continue
elif input_message.strip() == '':
print mybot.respond("无")
continue
print input_message
message = T.wordSegment(input_message)
# 去标点
print 'word Seg:'+ message
print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message)
print "======="
print response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
# 百科搜索
def run(question):
# 初始化jb分词器
T.jieba_initialize()
# 切换到语料库所在工作目录
mybot_path = './'
os.chdir(mybot_path)
mybot = aiml.Kernel()
if os.path.isfile("bot_brain.brn"):
mybot.bootstrap(brainFile="bot_brain.brn")
else:
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/std-startup.xml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/tuling.xml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/bye.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/tools.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/bad.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/funny.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/OrdinaryQuestion.aiml")
mybot.learn(
os.path.split(os.path.realpath(__file__))[0] +
"/resources/Common conversation.aiml")
#mybot.bootstrap(learnFiles="std-startup.xml", commands="load aiml b")
mybot.saveBrain("bot_brain.brn")
# 载入百科属性列表
print '''
Eric：你好，我是问答机器人。╭(╯^╰)╮
'''
input_message = question
if len(input_message) > 60:
return mybot.respond("句子长度过长")
#continue
elif input_message.strip() == '':
return mybot.respond("无话可说")
#continue
# print input_message
message = T.wordSegment(input_message)
# 去标点
# print 'word Seg:'+ message
# print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message) # 在AIML数据集里寻找答案
print "======="
if response[0] == '#':
print response + 'mark'
pass
else:
return response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
# 百科搜索
elif response[0] == '#' or len(response) < 1:
# 匹配百科
if response.__contains__("searchbaike"):
print "searchbaike"
print response
res = response.split(':')
# 实体
entity = str(res[1]).replace(" ", "")
# 属性
attr = str(res[2]).replace(" ", "")
print entity + '<---->' + attr
ans = baike.query(entity, attr)
# 如果命中答案
if type(ans) == list:
return '回答：' + QAT.ptranswer(ans, False)
#continue
elif ans.decode('utf-8').__contains__(u'::找不到'):
# 百度摘要+Bing摘要
print "通用搜索"
ans = search_summary.kwquery(input_message)
# 匹配不到模版，通用查询
elif response.__contains__("NoMatchingTemplate"):
print "NoMatchingTemplate"
ans = search_summary.kwquery(input_message)
if len(ans) == 0:
ans = mybot.respond('找不到答案')
return '回答：' + ans
elif len(ans) > 1:
print "不确定候选答案"
return ans
print 'Eric: '
for a in ans:
print a.encode("utf-8")
else:
return '回答：' + ans[0].encode("utf-8")
# 匹配模版
else:
return '回答：' + response
def handle(self):
conn = self.request
conn.sendall('欢迎访问智能百科问答系统')
Flag =True
data = conn.recv(4096)
while Flag:
input_message = data
print "input_message====="
print input_message
print "=========="
reply = ''
if len(input_message) > 60:
print mybot.respond("句子长度过长")
reply = mybot.respond("句子长度过长")
conn.sendall(reply)
Flag =False
continue
elif input_message.strip() == '无':
print mybot.respond("无")
reply = mybot.respond("无")
conn.sendall(reply)
Flag = False
continue
print input_message
message = T.wordSegment(input_message)
# 去标点
print 'word Seg:' + message
print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message)
print "======="
print response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
reply = mybot.respond('找不到答案')
conn.sendall(reply)
Flag = False
# 百科搜索
elif response[0] == '#':
# 匹配百科
if response.__contains__("searchbaike"):
print "searchbaike"
print response
res = response.split(':')
# 实体
entity = str(res[1]).replace(" ", "")
# 属性
attr = str(res[2]).replace(" ", "")
print entity + '<---->' + attr
ans = baike.query(entity, attr)
# 如果命中答案
if type(ans) == list:
print 'Eric：' + QAT.ptranswer(ans, False)
reply = QAT.ptranswer(ans, False)
conn.sendall(reply)
Flag = False
continue
elif ans.decode('utf-8').__contains__(u'::找不到'):
# 百度摘要+Bing摘要
print "通用搜索"
ans = search_summary.kwquery(input_message)
# 匹配不到模版，通用查询
elif response.__contains__("NoMatchingTemplate"):
print "NoMatchingTemplate"
ans = search_summary.kwquery(input_message)
if len(ans) == 0:
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
reply = ans
conn.sendall(reply)
Flag = False
elif len(ans) > 1:
print "不确定候选答案"
print 'Eric: '
for a in ans:
print a.encode("utf8")
reply += a.encode("utf8") + '\n'
conn.sendall(reply)
Flag = False
else:
print 'Eric：' + ans[0].encode("utf8")
reply = ans[0].encode("utf8")
conn.sendall(reply)
Flag = False
# 匹配模版
else:
print 'Eric：' + response
reply = response
conn.sendall(reply)
Flag = False
def qa(question):
#初始化jb分词器
T.jieba_initialize()
#切换到语料库所在工作目录
mybot_path = './'
# os.chdir(mybot_path)
mybot = aiml.Kernel()
mybot.learn(os.path.split(os.path.realpath(__file__))[0]+"/resources/std-startup.xml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/bye.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/tools.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/bad.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/funny.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/OrdinaryQuestion.aiml")
mybot.learn(os.path.split(os.path.realpath(__file__))[0] + "/resources/Common conversation.aiml")
# mybot.respond('Load Doc Snake')
#载入百科属性列表
print '''
.----------------. .-----------------. .----------------. .----------------. .----------------.
| .--------------. || .--------------. || .--------------. || .--------------. || .--------------. |
| | _______ | || | ____ _____ | || | __ | || | ___ ____ | || | _________ | |
| | / ___ | | || ||_ \|_ _| | || | / \ | || | |_ ||_ _| | || | |_ ___ | | |
| | | (__ \_| | || | | \ | | | || | / /\ \ | || | | |_/ / | || | | |_ \_| | |
| | '.___`-. | || | | |\ \| | | || | / /__\ \ | || | | __'. | || | | _| _ | |
| | |`\____) | | || | _| |_\ |_ | || | _/ / \ \_ | || | _| | \ \_ | || | _| |___/ | | |
| | |_______.' | || ||_____|\____| | || ||____| |____|| || | |____||____| | || | |_________| | |
| | | || | | || | | || | | || | | |
| '--------------' || '--------------' || '--------------' || '--------------' || '--------------' |
'----------------' '----------------' '----------------' '----------------' '----------------'
Eric：你好，我是Eric。╭(╯^╰)╮
'''
input_message = question
if len(input_message) > 60:
print mybot.respond("句子长度过长")
elif input_message.strip() == '':
print mybot.respond("无")
print input_message
message = T.wordSegment(input_message)
# 去标点
print 'word Seg:'+ message
print '词性：'
words = T.postag(input_message)
if message == 'q':
exit()
else:
response = mybot.respond(message)
print "======="
print response
print "======="
if response == "":
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
# 百科搜索
elif response[0] == '#':
# 匹配百科
if response.__contains__("searchbaike"):
print "searchbaike"
print response
res = response.split(':')
#实体
entity = str(res[1]).replace(" ","")
#属性
attr = str(res[2]).replace(" ","")
print entity+'<---->'+attr
ans = baike.query(entity, attr)
# 如果命中答案
if type(ans) == list:
print 'Eric：' + QAT.ptranswer(ans,False)
elif ans.decode('utf-8').__contains__(u'::找不到'):
#百度摘要+Bing摘要
print "通用搜索"
ans = search_summary.kwquery(input_message)
# 匹配不到模版，通用查询
elif response.__contains__("NoMatchingTemplate"):
print "NoMatchingTemplate"
ans = search_summary.kwquery(input_message)
if len(ans) == 0:
ans = mybot.respond('找不到答案')
print 'Eric：' + ans
elif len(ans) >1:
print "不确定候选答案"
print 'Eric: '
for a in ans:
print a.encode("utf8")
else:
print 'Eric：' + ans[0].encode("utf8")
# 匹配模版
else:
print 'Eric：' + response
def answer(question):
if len(question) > 600:
print mybot.respond("句子长度过长")
raise Exception("Too Long")
elif question.strip() == '':
print mybot.respond("无")
raise Exception("No Input")
print question
message = T.wordSegment(question)
# 去标点
print 'word Seg:'+ message
print '词性：'
words = T.postag(question)
if message == 'q':
exit()
else:
response = mybot.respond(message)
print response
if response == "":
raise Exception("No Answer")
# 百科搜索
elif response[0] == '#':
# 匹配百科
if response.__contains__("searchbaike"):
print "searchbaike"
print response
res = response.split(':')
#实体
entity = str(res[1]).replace(" ","")
#属性
attr = str(res[2]).replace(" ","")
print entity+'<---->'+attr
ans = baike.query(entity, attr)
# 如果命中答案
if type(ans) == list:
print 'Eric：' + QAT.ptranswer(ans,False)
return [QAT.ptranswer(ans,False)]
elif ans.decode('utf-8').__contains__(u'::找不到'):
#百度摘要+Bing摘要
print "通用搜索"
ans = search_summary.kwquery(question)
# 匹配不到模版，通用查询
elif response.__contains__("NoMatchingTemplate"):
print "NoMatchingTemplate"
ans = search_summary.kwquery(question)
if len(ans) == 0:
raise Exception("No Answer")
elif len(ans) >1:
print "不确定候选答案"
print 'Eric: '
for a in ans:
print a.encode("utf8")
return [a.encode("utf8") for a in ans]
else:
print 'Eric：' + ans[0].encode("utf8")
return [ans[0].encode("utf8")]
# 匹配模版
else:
print 'Eric：' + response
return [response]
def kwquery(query):
#分词 去停用词 抽取关键词
keywords = []
words = T.postag(query)
for k in words:
#print(k)
# 只保留名词
if k.flag.__contains__("n"):
# #print k.flag
# #print k.word
keywords.append(k.word)
answer = []
text = ''
# 找到答案就置1
flag = 0
# 抓取百度前10条的摘要
soup_baidu = To.get_html_baidu('https://www.baidu.com/s?wd=' +
quote(query))
for i in range(1, 10):
if soup_baidu == None:
break
results = soup_baidu.find(id=i)
if results == None:
##print("百度百科找不到答案")
break
if results.attrs.get('mu') and i == 1:
# #print results.attrs["mu"]
r = results.find(class_='op_exactqa_s_answer')
if r == None:
pass
##print("百度知识图谱找不到答案")
else:
##print("百度知识图谱找到答案")
answer.append(r.get_text().strip())
flag = 1
break
if flag == 0 and len(keywords) > 1:
##print(keywords)
ans_lst = list(kgquery_entity(keywords[0], keywords[1]))
if len(ans_lst) != 0:
answer = ans_lst
flag = 1
if results.find("h3") != None and flag == 0:
if results.find("h3").find("a").get_text().__contains__(
u"百度百科") and (i == 1 or i == 2):
url = results.find("h3").find("a")['href']
if url == None:
##print("百度百科找不到答案")
continue
else:
#print("百度百科找到答案")
baike_soup = To.get_html_baike(url)
r = baike_soup.find(class_='lemma-summary')
if r == None:
continue
else:
r = r.get_text().replace("\n", "").strip()
answer.append(r)
flag = 1
break
text += results.get_text().strip()
if flag == 0:
#分句
cutlist = [u"。", u"?", u".", u"_", u"-", u":", u"！", u"？"]
temp = ''
sentences = []
for i in range(0, len(text)):
if text[i] in cutlist:
if temp == '':
continue
else:
# #print temp
sentences.append(temp)
temp = ''
else:
temp += text[i]
# 找到含有关键词的句子,去除无关的句子
key_sentences = {}
for s in sentences:
for k in keywords:
if k in s:
key_sentences[s] = 1
# 识别人名
target_list = {}
for ks in key_sentences:
# #print ks
words = T.postag(ks)
for w in words:
# #print "====="
# #print w.word
if w.flag == ("nr"):
if target_list.get(w.word):
target_list[w.word] += 1
else:
target_list[w.word] = 1
sorted_lists = sorted(target_list.items(),
key=operator.itemgetter(1),
reverse=True)
sorted_lists2 = []
# 候选队列
for i, st in enumerate(sorted_lists):
# #print st[0]
if st[0] in keywords:
continue
else:
sorted_lists2.append(st)
#print("返回前3个词频")
answer = []
for i, st in enumerate(sorted_lists2):
# #print st[0]
# #print st[1]
if i < 3:
# #print st[0]
# #print st[1]
answer.append(st[0])
##print(answer)
return answer