def generate_user_seeds(self, request_times=1, user_accessed_set=None):
if self._url is None:
print "I'm anonymous user."
return 0
else:
if self.soup is None:
self.parser()
soup = self.soup
seed_list = []
for i in range(request_times):
post_url = "https://www.zhihu.com/lookup/suggest_member"
_xsrf = soup.find("input", attrs={'name': '_xsrf'})["value"]
data = {
'ids': ",,",
'_xsrf': _xsrf
}
post_data = urlencode(data)
r_post = zhihu_util.post(post_url, post_data)
suggent_member_list = json.loads(r_post)["msg"]
for suggent_member in suggent_member_list:
suggent_member_soup = zhihu_util.get_soup(suggent_member)
suggent_member_str = suggent_member_soup.find("a", class_="image-link")\
.get("href").split("/")[-1]
seed_list.append(suggent_member_str)
seed_set = set(seed_list)
if user_accessed_set:
seed_set.difference_update(user_accessed_set)
return seed_set
def test_post_repeatedly(self):
count = 0
post_url = 'https://www.zhihu.com/node/TopicsPlazzaListV2'
post_data = 'method=next¶ms=%7B%22topic_id%22%3A686%2C%22offset%22%3A80%2C%22hash_id%22%3A%22dced108689287057f5cc3b5e85cb8289%22%7D&_xsrf=c6946d5914172133e875956a711be3ad'
while count < 3:
response = zhihu_util.post(post_url, post_data)
print "...post count:%s" % count
self.assertTrue(response != "FAIL", "post count is {0}".format(count))
count += 1
def get_followees(self):
if self._url is None:
print "I'm anonymous user."
return
yield
else:
followees_num = self.get_followees_num()
if followees_num == 0:
return
yield
else:
followee_url = self._url + "/followees"
r = zhihu_util.get_content(followee_url)
# print "r:%s" % r
soup = zhihu_util.get_soup(r)
for i in xrange((followees_num - 1) / 20 + 1):
if i == 0:
user_url_list = soup.find_all("h2", class_="zm-list-content-title")
for j in xrange(min(followees_num, 20)):
try:
yield User(user_url_list[j].a["href"],
user_url_list[j].a.string.encode("utf-8"))
except:
print("...get followee error ,just skip...")
return
yield
else:
post_url = "http://www.zhihu.com/node/ProfileFolloweesListV2"
_xsrf = soup.find("input", attrs={'name': '_xsrf'})["value"]
offset = i * 20
hash_id = re.findall("hash_id": "(.*)"},", r)[0]
params = json.dumps(
{"offset": offset, "order_by": "created", "hash_id": hash_id})
data = {
'_xsrf': _xsrf,
'method': "next",
'params': params
}
post_data = urlencode(data)
r_post = zhihu_util.post(post_url, post_data)
followee_list = json.loads(r_post)["msg"]
for j in xrange(min(followees_num - i * 20, 20)):
try:
followee_soup = zhihu_util.get_soup(followee_list[j])
user_link = followee_soup.find("h2", class_="zm-list-content-title").a
yield User(user_link["href"], user_link.string.encode("utf-8"))
except:
print("...get followee error ,just skip...")
return
yield
def test_post(self):
post_url = 'https://www.zhihu.com/node/TopicsPlazzaListV2'
post_data = 'method=next¶ms=%7B%22topic_id%22%3A253%2C%22offset%22%3A60%2C%22hash_id%22%3A%22dced108689287057f5cc3b5e85cb8289%22%7D&_xsrf=c6946d5914172133e875956a711be3ad'
zhihu_util.post(post_url, post_data)