def get_fans_or_follows(html, uid, type):
"""
Get fans or follows and store their relationships
:param html: current page source
:param uid: current user id
:param type: type of relations, 1 stands for fans，2 stands for follows
:return: list of fans or followers
"""
if html == '':
return list()
pattern = re.compile(r'FM.view\((.*)\)')
soup = BeautifulSoup(html, "html.parser")
scripts = soup.find_all('script')
user_ids = list()
relations = list()
for script in scripts:
m = re.search(pattern, script.string)
if m and 'pl.content.followTab.index' in script.string:
all_info = m.group(1)
cont = json.loads(all_info).get('html', '')
soup = BeautifulSoup(cont, 'html.parser')
follows = soup.find(attrs={
'class': 'follow_box'
}).find_all(attrs={'class': 'follow_item'})
pattern = 'uid=(.*?)&'
for follow in follows:
m = re.search(pattern, str(follow))
if m:
r = m.group(1)
# filter invalid ids
if r.isdigit():
user_ids.append(r)
relations.append(UserRelation(uid, r, type))
save_relations(relations)
return user_ids
def get_fans_or_follows(html, uid, type):
"""
Get fans or follows and store their relationships
:param html: current page source
:param uid: current user id
:param type: type of relations, 1 stands for fans，2 stands for follows
:return: list of fans or followers
"""
if html == '':
return list()
pattern = re.compile(r'FM.view\((.*)\)')
soup = BeautifulSoup(html, "html.parser")
scripts = soup.find_all('script')
user_ids = list()
relations = list()
for script in scripts:
m = re.search(pattern, script.string)
if m and 'pl.content.followTab.index' in script.string:
all_info = m.group(1)
cont = json.loads(all_info).get('html', '')
soup = BeautifulSoup(cont, 'html.parser')
follows = soup.find(attrs={'class': 'follow_box'}).find_all(attrs={'class': 'follow_item'})
pattern = 'uid=(.*?)&'
for follow in follows:
m = re.search(pattern, str(follow))
if m:
r = m.group(1)
# filter invalid ids
if r.isdigit():
user_ids.append(r)
relations.append(UserRelation(uid, r, type))
save_relations(relations)
return user_ids