Use User class instead of dictionary
This commit is contained in:
parent
7792e5cbda
commit
0eedc67144
37
scrape.py
37
scrape.py
@ -37,6 +37,16 @@ print('WebDriver initialised')
|
||||
|
||||
##### Do stuff ##############################################
|
||||
|
||||
# User class, holds all the data we will be collecting from user pages
|
||||
# It's being used here as a more structured alternative to a dictionary
|
||||
class User:
|
||||
url_username = ''
|
||||
username = ''
|
||||
track_count = ''
|
||||
like_count = ''
|
||||
following_count = ''
|
||||
follower_count = ''
|
||||
|
||||
user_dict = {}
|
||||
users_to_process = ['thomotron', 'lacheque', 'slynk']
|
||||
|
||||
@ -45,26 +55,29 @@ for user_str in users_to_process:
|
||||
if user_str in user_dict.keys():
|
||||
continue
|
||||
|
||||
# Initialise our user object to store our values in
|
||||
user = User()
|
||||
user.url_username = user_str
|
||||
|
||||
# Get the user's profile page and soup it
|
||||
driver.get('https://soundcloud.com/' + user_str)
|
||||
soup = BeautifulSoup(driver.page_source, 'lxml')
|
||||
|
||||
# Initialise our user object to store our values in
|
||||
user = {}
|
||||
|
||||
# Grab their username
|
||||
# Use stripped_strings generator as workaround for users with premium badge
|
||||
username = next(soup.find(class_='profileHeaderInfo__userName').stripped_strings)
|
||||
user['username'] = username
|
||||
print(username)
|
||||
user.username = username
|
||||
|
||||
track_count = int(soup.find('a', href='/' + username.lower() + '/tracks', class_='infoStats__statLink').div.string)
|
||||
user['track_count'] = track_count
|
||||
print(track_count)
|
||||
# Grab their track count
|
||||
track_count = soup.find('a', href='/' + user.url_username + '/tracks', class_='infoStats__statLink').div.string
|
||||
user.track_count = track_count
|
||||
|
||||
# Grab their like count
|
||||
like_count = soup.find('a', href='/' + user.url_username + '/likes').find(class_='sidebarHeader__actualTitle').string.split(' ')[0]
|
||||
user.like_count = like_count
|
||||
|
||||
like_count = int(soup.find('a', href='/' + username.lower() + '/likes').find(class_='sidebarHeader__actualTitle').string.split(' ')[0])
|
||||
user['like_count'] = like_count
|
||||
print(like_count)
|
||||
|
||||
# Finally add the user to the dictionary
|
||||
user_dict['user_str'] = user
|
||||
user_dict[user.url_username] = user
|
||||
|
||||
driver.quit()
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user