|
|
@@ -23,6 +23,7 @@ from html import unescape
|
|
|
from collections import OrderedDict
|
|
|
|
|
|
from util import (
|
|
|
+ scheme,
|
|
|
merge_links,
|
|
|
check_link_structure,
|
|
|
check_links_structure,
|
|
|
@@ -52,7 +53,7 @@ def validate_links(links):
|
|
|
def archivable_links(links):
|
|
|
"""remove chrome://, about:// or other schemed links that cant be archived"""
|
|
|
for link in links:
|
|
|
- scheme_is_valid = scheme(url) in ('http', 'https', 'ftp)
|
|
|
+ scheme_is_valid = scheme(link['url']) in ('http', 'https', 'ftp')
|
|
|
not_blacklisted = (not URL_BLACKLIST.match(link['url'])) if URL_BLACKLIST else True
|
|
|
if scheme_is_valid and not_blacklisted:
|
|
|
yield link
|