mirror of
https://github.com/netzbegruenung/green-spider.git
synced 2024-05-01 16:44:51 +02:00
Add rating for contact and social media links
This commit is contained in:
parent
f555781735
commit
3cb7bbbec6
|
@ -6,6 +6,7 @@ criteria based on information gather by checks before.
|
|||
import logging
|
||||
|
||||
from rating import canonical_url
|
||||
from rating import contact_link
|
||||
from rating import favicon
|
||||
from rating import feeds
|
||||
from rating import https
|
||||
|
@ -15,6 +16,7 @@ from rating import reachable
|
|||
from rating import resolvable
|
||||
from rating import response_duration
|
||||
from rating import responsive_layout
|
||||
from rating import social_media_links
|
||||
from rating import use_specific_fonts
|
||||
from rating import www_optional
|
||||
|
||||
|
@ -30,6 +32,7 @@ def calculate_rating(results):
|
|||
# The raters to execute.
|
||||
rating_modules = {
|
||||
'CANONICAL_URL': canonical_url,
|
||||
'CONTACT_LINK': contact_link,
|
||||
'DNS_RESOLVABLE_IPV4': resolvable,
|
||||
'FAVICON': favicon,
|
||||
'FEEDS': feeds,
|
||||
|
@ -39,6 +42,7 @@ def calculate_rating(results):
|
|||
'NO_SCRIPT_ERRORS': no_script_errors,
|
||||
'RESPONSIVE': responsive_layout,
|
||||
'SITE_REACHABLE': reachable,
|
||||
'SOCIAL_MEDIA_LINKS': social_media_links,
|
||||
'USE_SPECIFIC_FONTS': use_specific_fonts,
|
||||
'WWW_OPTIONAL': www_optional,
|
||||
}
|
||||
|
|
41
rating/contact_link.py
Normal file
41
rating/contact_link.py
Normal file
|
@ -0,0 +1,41 @@
|
|||
"""
|
||||
Checks whether the pages has a link "Kontakt"
|
||||
"""
|
||||
|
||||
from rating.abstract_rater import AbstractRater
|
||||
|
||||
class Rater(AbstractRater):
|
||||
|
||||
rating_type = 'boolean'
|
||||
default_value = False
|
||||
depends_on_checks = ['hyperlinks']
|
||||
max_score = 1
|
||||
|
||||
def __init__(self, check_results):
|
||||
super().__init__(check_results)
|
||||
|
||||
def rate(self):
|
||||
value = self.default_value
|
||||
score = 0
|
||||
|
||||
urls = 0
|
||||
urls_with_contact_link = 0
|
||||
|
||||
for url in self.check_results['hyperlinks']:
|
||||
|
||||
urls += 1
|
||||
|
||||
for link in self.check_results['hyperlinks'][url]['links']:
|
||||
if link['text'].lower() == 'kontakt':
|
||||
urls_with_contact_link += 1
|
||||
|
||||
if urls_with_contact_link == urls:
|
||||
score = self.max_score
|
||||
value = True
|
||||
|
||||
return {
|
||||
'type': self.rating_type,
|
||||
'value': value,
|
||||
'score': score,
|
||||
'max_score': self.max_score,
|
||||
}
|
59
rating/social_media_links.py
Normal file
59
rating/social_media_links.py
Normal file
|
@ -0,0 +1,59 @@
|
|||
"""
|
||||
Checks whether the pages have a link to social media profiles.
|
||||
"""
|
||||
|
||||
from rating.abstract_rater import AbstractRater
|
||||
from urllib.parse import urlparse
|
||||
import logging
|
||||
|
||||
class Rater(AbstractRater):
|
||||
|
||||
rating_type = 'boolean'
|
||||
default_value = False
|
||||
depends_on_checks = ['hyperlinks']
|
||||
max_score = 1
|
||||
|
||||
def __init__(self, check_results):
|
||||
super().__init__(check_results)
|
||||
|
||||
def rate(self):
|
||||
value = self.default_value
|
||||
score = 0
|
||||
|
||||
urls = 0
|
||||
urls_with_social_media_links = 0
|
||||
|
||||
for url in self.check_results['hyperlinks']:
|
||||
|
||||
urls += 1
|
||||
|
||||
for link in self.check_results['hyperlinks'][url]['links']:
|
||||
|
||||
if link['href'] is None:
|
||||
continue
|
||||
|
||||
# only process absolute links
|
||||
if not (link['href'].startswith('http:') or link['href'].startswith('https:')):
|
||||
continue
|
||||
|
||||
parsed = urlparse(link['href'])
|
||||
if ("facebook.com" in parsed.hostname or
|
||||
"twitter.com" in parsed.hostname or
|
||||
"instagram.com" in parsed.hostname or
|
||||
"plus.google.com" in parsed.hostname):
|
||||
logging.debug("Found social media link on %s: %s" % (url, link['href']))
|
||||
urls_with_social_media_links += 1
|
||||
|
||||
# make sure we only count 1 for this url
|
||||
break
|
||||
|
||||
if urls_with_social_media_links == urls:
|
||||
score = self.max_score
|
||||
value = True
|
||||
|
||||
return {
|
||||
'type': self.rating_type,
|
||||
'value': value,
|
||||
'score': score,
|
||||
'max_score': self.max_score,
|
||||
}
|
Loading…
Reference in a new issue