def grab(url, threadID):
print url
user_grab = User(url)
followers = user_grab.get_followers()
for i, user_grab in enumerate(followers):
user = Users()
while True:
try:
if Users.objects(data_id = user_grab.get_data_id()).count():
user = Users.objects(data_id = user_grab.get_data_id()).first()
break
except Exception, e:
logging.error("========error1")
logging.error(e)
time.sleep(300)
user.user_id = user_grab.get_user_id()
user.data_id = user_grab.get_data_id()
user.followees_num = user_grab.get_followees_num()
user.followers_num = user_grab.get_followers_num()
try:
user.asks_num = user_grab.get_asks_num()
user.answers_num = user_grab.get_answers_num()
user.collections_num = user_grab.get_collections_num()
except Exception, e:
logging.error("========error2")
logging.error(e)
logging.debug(user_grab.get_user_url())
def grab(url, threadID):
logging.info(url)
user_grab = User(url)
followees = user_grab.get_followees()
for i, user_grab in enumerate(followees):
user = Users()
flag = True
while True:
try:
if Users.objects(data_id = user_grab.get_data_id()).count():
user = Users.objects(data_id = user_grab.get_data_id()).first()
break
except Exception, e:
flag = False
logging.error("========error1")
logging.error(e)
time.sleep(300)
break
try:
user.user_id = user_grab.get_user_id()
user.data_id = user_grab.get_data_id()
user.followees_num = user_grab.get_followees_num()
user.followers_num = user_grab.get_followers_num()
user.asks_num = user_grab.get_asks_num()
user.answers_num = user_grab.get_answers_num()
user.collections_num = user_grab.get_collections_num()
user.agree_num = user_grab.get_agree_num()
user.thanks_num = user_grab.get_thanks_num()
user.url = user_grab.get_user_url()
user.modify_time = datetime.utcnow()
except Exception, e:
logging.error("========error2")
logging.error(e)
logging.debug(user_grab.get_user_url())
def main():
client = pymongo.MongoClient("localhost", 27017)
db = client.zhihu_user
urllist = db.urllist
userlist = db.userlist
origin_users = ["https://www.zhihu.com/people/jixin",
"https://www.zhihu.com/people/zhang-jia-wei",
"https://www.zhihu.com/people/zhu-xuan-86",
"https://www.zhihu.com/people/kaifulee",
"https://www.zhihu.com/people/e-miao-de-nai-ba"]
urls = urllist.distinct("user_url")
for u in origin_users:
if u in urls:
pass
else:
urllist.insert({"user_url": u,
"jlzt": "1"})
while 1:
item = urllist.find_one({'jlzt':'1'})
if item == None:
print u'已全部处理完成'
break
else:
user_url = item["user_url"]
starttime = datetime.datetime.now()
urllist.update({"user_url":user_url},{"$set":{"jlzt":"3"}})
try:
user = User(user_url)
zhihu_id = user.get_data_id()
# 用户唯一id
zhihu_name = user.get_user_id()
# 用户名
followees_num = user.get_followees_num()
# 用户关注人数
followers_num = user.get_followers_num()
# 用户关注者人数
gender = user.get_gender()
# 性别
# 提问数
asks_num = user.get_asks_num()
# 获取该用户回答的个数
answers_num = user.get_answers_num()
# 获取该用户收藏夹个数
collections_num = user.get_collections_num()
# 获取该用户获得的赞同数
agree_num = user.get_agree_num()
# 获取该用户获得的感谢数
thanks_num = user.get_thanks_num()
# 获取该用户的头像url
head_img_url = user.get_head_img_url()
# 关注的话题数
topics_num = user.get_topics_num()
# 获取该用户关注的人
followees = user.get_followees()
# 获取关注该用户的人
followers = user.get_followers()
# 获取提出的问题
questions = user.get_asks()
# 获取回答的问题
answers = user.get_answers()
# 获取话题
topics = user.get_topics()
print "start process " + zhihu_name + ";\n"
#print zhihu_id # 黄继新
#print zhihu_name # 614840
#print followees_num # 8408
#print followers_num # 1323
#print gender # 786
#print asks_num # 44
#print answers_num # 46387
#print collections_num # 11477
#print agree_num
#print thanks_num
#print head_img_url
#print topics_num
followee_l = []
follower_l = []
questions_l = []
answers_l = []
topics_l = []
print u'开始处理关注的人'
for followee in followees:
followee_l.append(followee.user_url.split('/')[4])
if len(followee_l) % 100 == 0:
print zhihu_name + "'s NO." + str(len(followee_l)) + " followee is being processed. please wait..."
time.sleep(0.05)
followee_list = ','.join(followee_l)
print u'添加完成'
print u'开始添加关注者至处理队列'
for follower in followers:
follower_l.append(follower.user_url.split('/')[4])
urls = urllist.distinct("user_url")
if follower.user_url in urls:
pass
# print "follower_url:'"+follower.user_url+"' passed"
else:
urllist.insert({"user_url": follower.user_url, "jlzt": "1"})
#print "follower_url:'" +follower.user_url + "' added"
if len(follower_l) % 100 == 0:
print zhihu_name + "'s NO." + str(len(follower_l)) + " follower is being processed. please wait..."
time.sleep(0.05)
print u'添加完成'
for q in questions:
questions_l.append("url=" + q.url + "|title=" + q.get_title())
time.sleep(0.01)
for a in answers:
answers_l.append(a.answer_url)
time.sleep(0.01)
for t in topics:
topics_l.append(t)
time.sleep(0.01)
user_data = {"zhihu_id":zhihu_id,
"zhihu_name":zhihu_name,
"followees_num":followees_num,
"followers_num":followers_num,
"followees":followee_l,
"followers":follower_l,
"questions":questions_l,
"gender":gender,
"asks_num":asks_num,
"answers_num":answers_num,
"ansers":answers_l,
"collections_num":collections_num,
"agree_num":agree_num,
"thanks_num":thanks_num,
"topics_num":topics_num,
"topics":topics_l,
"head_img_url":head_img_url
}
print "user_data prepared:"
urls = userlist.distinct("user_url")
if user_url in urls:
pass
else:
userlist.insert(user_data)
print "user_data inserted: \n"
urllist.update({"user_url":user_url},{"$set":{"jlzt":"2"}})
endtime = datetime.datetime.now()
interval=(endtime - starttime).seconds
print zhihu_name + "finnished. spent " + str(interval) + "seconds."
except:
traceback.print_exc()
time.sleep(10)
urllist.update({"user_url":user_url},{"$set":{"jlzt":"1"}})
continue
print "处理完毕"
