get favicons for custom substack publications.

This commit is contained in:
Jason Schwarzenberger
2020-11-24 10:36:31 +13:00
parent 3daae5fa1b
commit fe01ea52e5
5 changed files with 58 additions and 48 deletions
+15
View File
@@ -1,4 +1,19 @@
from bs4 import BeautifulSoup
def get_icons(markup, url):
soup = BeautifulSoup(markup, features='html.parser')
icon32 = soup.find_all('link', rel="icon", href=True, sizes="32x32")
icon16 = soup.find_all('link', rel="icon", href=True, sizes="16x16")
favicon = soup.find_all('link', rel="shortcut icon", href=True)
others = soup.find_all('link', rel="icon", href=True)
icons = icon32 + icon16 + favicon + others
base_url = '/'.join(url.split('/')[:3])
icons = list(set([i.get('href') for i in icons]))
icons = [i if i.startswith('http') else base_url + i for i in icons]
return icons
def parse_extruct(s, data):
rdfa_keys = {
'title': [