def wordeee(self, text, topic, question):
lol = mongDB.Logger()
dic_1 = {}
dic_11 = {}
from collections import Counter
numall = 0
words = []
#words = [word for word in jieba.cut(text, cut_all=True) if len(word) >= 2]
for word in jieba.cut(text, cut_all=True):
if len(word) >= 2:
words.append(word)
numall += 1
c = Counter(words)
p = 0.0
listword = []
for word_freq in c.most_common(20):
word, freq = word_freq
if str(word).find('\n') != -1:
pass
else:
listword.append(word)
p += float(freq) / float(numall)
dic_1['answer_words'] = listword
dic_1['p'] = p * 100 #百分比
fileq = open(question)
qq = fileq.read()
dic_1['title'] = qq
self.questionlist.append(dic_1)
dic_11['questionlist'] = self.questionlist
dic_11['itstopic'] = topic
#print word, freq
#lol.db_table7.insert(dic_11)###########
return dic_11
def __init__(self):
#"""Do nothing, by default."""
self.qqcount = 0
self.dic_name = {}
self.ltable1 = []
self.list_p = []
self.topic = getopic()
self.lo = mongDB.Logger()
def getpaint(authors, topic):
lol = mongDB.Logger()
for author in authors:
dic_paint = {}
dic_paint['name'] = author.name
dic_paint['num'] = author.follower_count
dic_paint['topic'] = topic
#######
lol.db_table1.insert(dic_paint)
def xiangsidu(self):
lol = mongDB.Logger()
count = 1
dic_11={}
for i in range(10):
files1 = self.xs.analyse_nr('F:/zhihu/answer1' +str(count)+ '/')
#print files1
text = self.xs.change(files1)
dic_11=self.xs.wordeee(text, self.aa.topic,'F:/zhihu/answer1' +str(count)+ '/'+"q.txt")
count += 1
#print dic_11
lol.db_table7.insert(dic_11) ###########
def getpaint(authors, topic):
list1 = []
count = 0
lol = mongDB.Logger()
for author in authors:
dic_paint = {}
dic_follow = {}
dic_following = {}
if author.over:
continue
else:
if author.followings:
for _, ii in zip(range(20), author.followings):
if ii.over:
name2 = str(ii.over_reason)
continue
name2 = str(ii.name)
if name2.find('.') != -1:
name2 = name2.replace('.', u"(点)")
#print name2,'----------------------------'
dic_following[name2] = ii.follower_count
if author.followers:
for _, follower in zip(range(20), author.followers):
if follower.over:
name1 = str(follower.over_reason)
continue
name1 = str(follower.name)
if name1.find('.') != -1:
name1 = name1.replace('.', u"(点)")
#print name1
dic_follow[name1] = follower.follower_count
dic_paint['name'] = author.name
dic_paint['num'] = author.follower_count
dic_paint['follower'] = dic_follow
dic_paint['following'] = dic_following
dic_paint['topic'] = topic
####################
#js1=json.dumps(dic_paint)
#js2=json.loads(js1)
lol.db_table1.insert(dic_paint)
def __init__(self):
self.aa=pzhn.get("")#输入话题
self.xs=SimilarityDegree.XiangSi()
self.lo1=mongDB.Logger()