def run(self):
with self.threadingSum:
logging.debug("%s start" % self.url)
dbHandler = DbHandler()
if not dbHandler.hasQuestion(self.url):
# 插入新的问题
question = Question(self.url)
title = question.get_title()
detail = question.get_detail()
answerNum = question.get_answer_num()
followersNum = question.get_followers_num()
tags = ""
for tag in question.get_tags():
tags += tag + ";"
tags = tags[0: len(tags) - 1]
questionDict = {"url": self.url, "title": title,
"detail": detail, "followers": followersNum,
"answerNum": answerNum, "tags": tags}
dbHandler.insertNewQuestion(questionDict)
zh_qid = dbHandler.getQueIdByUrl(self.url)
# 插入新的答案
for answer_link in question.get_all_answer_link():
answer = Answer(answer_link)
author = answer.get_author()
votes = answer.get_votes()
answerDict = {"url": answer_link, "author": author, "zh_qid": zh_qid,
"votes": votes}
dbHandler.insertNewAnswer(answerDict)
# 插入图片地址
zh_aid = dbHandler.getAnsIdByUrl(answer_link)
for imgUrl in answer.get_all_pics():
dbHandler.insertNewImgUrl(zh_aid, imgUrl)
contents = answer.get_answer_content()
self.storeTheAnswer(zh_aid, contents)
dbHandler.close()
logging.debug("%s done" % self.url)
def question_test():
start = time.time()
question = Question("http://www.zhihu.com/question/33488763")
question.parser()
print "Title of the question is ", question.get_title()
print "Tags of the question is ", question.get_tags()
print "Details of the question is ", question.get_details()
print "Number of answer of the question is ", question.get_answer_num()
print "Number of collapsed answer of this question is ", question.get_collapsed_answer_num()
print "Number of follower of this question is ", question.get_follower_num()
print "Number of view of this question is ", question.get_view_num()
print "Number of comment of this question is ", question.get_comment_num()
print "Last activity time of this question is", question.get_last_activity_time()
print "Number of follower of related tags is", question.get_related_tags_follower_num()
print "First known follower of this question is ", question.get_first_known_follower()
print "Related questions are ", question.get_related_questions()
question.save_top_answers(3)
question.save_all_answers() # might be time-consuming
question.save_all_followers_profile() # might be time-consuming
end = time.time()
print "Time used is", end - start
from question import Question
from toHtml import ToHtml
question = Question("http://www.zhihu.com/question/27848661")
title = question.get_title()
print title
detail = question.get_detail()
print detail
answer_num = question.get_answer_num()
print answer_num
tags = question.get_tags()
for tag in tags:
print tag
# authors = question.get_all_authors()
# for author in authors:
# print author
# votes = question.get_all_votes()
# for vote in votes:
# print vote
# answers = question.get_all_answers()
# for answer in answers:
# print answer
# toHtml = ToHtml("http://www.zhihu.com/question/27848661")
# toHtml.answerToHtml()
# question = Question("http://www.zhihu.com/question/25029518")
# urls = question.get_all_pics()
# question.download_all_pics(urls)
