Skip to content

Commit

Permalink
Merge pull request #67 from golnazads/master
Browse files Browse the repository at this point in the history 
bibcode verification, needs &
  • Loading branch information
@golnazads
golnazads authored May 19, 2021
2 parents 2b7f65f + 454ec69 commit af94cc0
Show file tree
Hide file tree
Showing 3 changed files with 44 additions and 19 deletions.
6 changes: 4 additions & 2 deletions config.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
LOGGING_LEVEL = 'DEBUG'

# This the URL to resolver_service api
GATEWAY_RESOLVER_SERVICE_URL = 'https://api.adsabs.harvard.edu/v1/resolver/%s'
GATEWAY_RESOLVER_SERVICE_URL = 'https://dev.adsabs.harvard.edu/v1/resolver/%s'

# This is a URL to adsws account info service
GATEWAY_SERVICE_ACCOUNT_INFO_URL = ''
Expand All @@ -14,4 +14,6 @@
REDIS_URL = "redis://localhost:6379/0"
REDIS_NAME_PREFIX = "link_gateway_"
# save to cache for a week
REDIS_EXPIRATION_TIME = 604800
REDIS_EXPIRATION_TIME = 604800

GATEWAY_SERVICE_REFERRED_DOMAIN = 'adsabs.harvard.edu'
24 changes: 12 additions & 12 deletions resolverway/tests/unittests/test_resolver_gateway.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ def test_route(self):
Tests for the existence of a /link_gateway route, and that it returns
properly formatted JSON data when the URL is supplied
"""
r= self.client.get('/link_gateway/1987gady.book.....B/ABSTRACT/https://ui.adsabs.harvard.edu/#abs/1987gady.book.....B/ABSTRACT')
r= self.client.get('/link_gateway/1987gady.book.....B/ABSTRACT/https://ui.adsabs.harvard.edu/abs/1987gady.book.....B/ABSTRACT')
self.assertEqual(r.status_code, 302)

def test_route_error_invalid_url(self):
Expand All @@ -44,7 +44,7 @@ def test_single_link(self):
"""
the_json = {"action": "redirect",
"link": "http://archive.stsci.edu/mastbibref.php?bibcode=2013MNRAS.435.1904M",
"service": "https://ui.adsabs.harvard.edu/#abs/2013MNRAS.435.1904/ESOURCE"}
"service": "https://ui.adsabs.harvard.edu/abs/2013MNRAS.435.1904/ESOURCE"}
r = LinkRequest('1987gady.book.....B', 'ABSTRACT', '').process_resolver_response(the_json, True)
self.assertEqual(r[1], 302)

Expand Down Expand Up @@ -72,13 +72,13 @@ def test_identification_link(self):
"""
the_json = {"action": "redirect",
"link": "http://dx.doi.org/10.1088/2041-8205/713/2/L10",
"service": "https://ui.adsabs.harvard.edu/#abs/2010ApJ...713L.103B/DOI:10.1088/2041-8205/713/2/L103"}
"service": "https://ui.adsabs.harvard.edu/abs/2010ApJ...713L.103B/DOI:10.1088/2041-8205/713/2/L103"}
r = LinkRequest('2010ApJ...713L.103B', 'DOI', '10.1088,2041-8205,713,2,L10').process_resolver_response(the_json, True)
self.assertEqual(r[1], 302)

the_json = {"action": "redirect",
"link": "http://arxiv.org/abs/1803.03598",
"service": "https://ui.adsabs.harvard.edu/#abs/2018arXiv180303598K/ARXIV:1803.03598"}
"service": "https://ui.adsabs.harvard.edu/abs/2018arXiv180303598K/ARXIV:1803.03598"}
r = LinkRequest('2018arXiv180303598K', 'ARXIV', '1803.03598').process_resolver_response(the_json, False)
self.assertEqual(r[1], 302)

Expand All @@ -89,7 +89,7 @@ def test_action_error(self):
"""
the_json = {"action": "redirecterror",
"link": "http://archive.stsci.edu/mastbibref.php?bibcode=2013MNRAS.435.1904M",
"service": "https://ui.adsabs.harvard.edu/#abs/2013MNRAS.435.1904/ESOURCE"}
"service": "https://ui.adsabs.harvard.edu/abs/2013MNRAS.435.1904/ESOURCE"}
r = LinkRequest('1987gady.book.....B', 'ABSTRACT', '').process_resolver_response(the_json, True)
self.assertEqual(r[1], 400)

Expand All @@ -106,7 +106,7 @@ def test_with_header_info(self):
:return:
"""
header = {'Referer': 'https://www.google.com/'}
r = self.client.get('/link_gateway/1987gady.book.....B/ABSTRACT/https://ui.adsabs.harvard.edu/#abs/1987gady.book.....B/ABSTRACT', headers=header)
r = self.client.get('/link_gateway/1987gady.book.....B/ABSTRACT/https://ui.adsabs.harvard.edu/abs/1987gady.book.....B/ABSTRACT', headers=header)
self.assertEqual(r.status_code, 302)

def test_redis_available(self):
Expand All @@ -133,12 +133,12 @@ def test_redis_put_get(self):

# verify that when the same session id is passed as cookie, the entry was fetched from redis
self.client.set_cookie('/','session','key1')
r = self.client.get('/link_gateway/2018AAS...23130709A/ABSTRACT/https://ui.adsabs.harvard.edu/#abs/2018AAS...23130709A/ABSTRACT', headers={'x-real-ip': '0.0.0.0'})
r = self.client.get('/link_gateway/2018AAS...23130709A/ABSTRACT/https://ui.adsabs.harvard.edu/abs/2018AAS...23130709A/ABSTRACT', headers={'x-real-ip': '0.0.0.0'})
self.assertEqual(r.headers['user_id'], 'ec43c30b9a81ed89765a2b8a04cac38925058eeacd5b5264389b1d4a7df2b28c')

# verify that when no cookie is send, session_id is None
self.client.cookie_jar.clear()
r = self.client.get('/link_gateway/2018AAS...23130709A/ABSTRACT/https://ui.adsabs.harvard.edu/#abs/2018AAS...23130709A/ABSTRACT')
r = self.client.get('/link_gateway/2018AAS...23130709A/ABSTRACT/https://ui.adsabs.harvard.edu/abs/2018AAS...23130709A/ABSTRACT')
self.assertEqual(r.headers['user_id'], 'None')

def test_redis_exception(self):
Expand All @@ -149,7 +149,7 @@ def test_redis_exception(self):
self.current_app.extensions['redis']._redis_client.connection_pool.max_connections = 0
self.current_app.extensions['redis']._redis_client.connection_pool.disconnect()
header = {'cookie': 'session=key1', 'x-real-ip': '0.0.0.0'}
r = self.client.get('/link_gateway/2018AAS...23130709A/ABSTRACT/https://ui.adsabs.harvard.edu/#abs/2018AAS...23130709A/ABSTRACT', headers=header)
r = self.client.get('/link_gateway/2018AAS...23130709A/ABSTRACT/https://ui.adsabs.harvard.edu/abs/2018AAS...23130709A/ABSTRACT', headers=header)
self.assertEqual(r.headers['user_id'], 'None')

def test_adsws_call(self):
Expand All @@ -158,7 +158,7 @@ def test_adsws_call(self):
:return:
"""
# verify when None is send in for session, account is None
account = LinkRequest('2018AAS...23130709A', 'ABSTRACT', 'https://ui.adsabs.harvard.edu/#abs/2018AAS...23130709A/ABSTRACT').get_user_info_from_adsws(None)
account = LinkRequest('2018AAS...23130709A', 'ABSTRACT', 'https://ui.adsabs.harvard.edu/abs/2018AAS...23130709A/ABSTRACT').get_user_info_from_adsws(None)
self.assertEqual(account, None)

# verify when successfully user info is set, True is returned
Expand All @@ -168,7 +168,7 @@ def test_adsws_call(self):
mock_response.json.return_value = {"source": "session:key1", "hashed_client_id": "013c1b1280353b3319133b9c528fb29ba998ae3b7af9b669166a786bc6796c9d", "anonymous": True, "hashed_user_id": "ec43c30b9a81ed89765a2b8a04cac38925058eeacd5b5264389b1d4a7df2b28c"}
self.client.set_cookie('/', 'session', 'key1')

status = LinkRequest('2018AAS...23130709A', 'ABSTRACT', 'https://ui.adsabs.harvard.edu/#abs/2018AAS...23130709A/ABSTRACT').set_user_info(get_mock)
status = LinkRequest('2018AAS...23130709A', 'ABSTRACT', 'https://ui.adsabs.harvard.edu/abs/2018AAS...23130709A/ABSTRACT').set_user_info(get_mock)
self.assertEqual(status, True)

# verify when successfully user info is read, dict of user is returned
Expand All @@ -178,7 +178,7 @@ def test_adsws_call(self):
mock_response.json.return_value = {"source": "session:key1", "hashed_client_id": "013c1b1280353b3319133b9c528fb29ba998ae3b7af9b669166a786bc6796c9d", "anonymous": True, "hashed_user_id": "ec43c30b9a81ed89765a2b8a04cac38925058eeacd5b5264389b1d4a7df2b28c"}
self.client.set_cookie('/', 'session', 'client_id')

account = LinkRequest('2018AAS...23130709A', 'ABSTRACT', 'https://ui.adsabs.harvard.edu/#abs/2018AAS...23130709A/ABSTRACT').get_user_info_from_adsws('key1')
account = LinkRequest('2018AAS...23130709A', 'ABSTRACT', 'https://ui.adsabs.harvard.edu/abs/2018AAS...23130709A/ABSTRACT').get_user_info_from_adsws('key1')
self.assertEqual(account['hashed_client_id'], "013c1b1280353b3319133b9c528fb29ba998ae3b7af9b669166a786bc6796c9d")
self.assertEqual(account['hashed_user_id'], "ec43c30b9a81ed89765a2b8a04cac38925058eeacd5b5264389b1d4a7df2b28c")

Expand Down
33 changes: 28 additions & 5 deletions resolverway/views.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ class LinkRequest(object):
user_id = None
referrer = None

re_ads_link = re.compile(r"^(/abs/[12][09]\d\d[A-Za-z\.]{5}[A-Za-z0-9\.]{9}[A-Z]/abstract)$")
re_ads_link = re.compile(r"^(/abs/[12][09]\d\d[A-Za-z&\.]{5}[A-Za-z0-9\.]{9}[A-Z]/abstract)$")

def __init__(self, bibcode, link_type, url=None, id=None):
self.bibcode = bibcode
Expand Down Expand Up @@ -174,15 +174,38 @@ def verify_link_type(self):
response = self.get_request_to_service('check_link_type' + '/' + self.link_type)
return response.status_code == 200

def verify_url(self):
def verify_url(self, referrer):
"""
:param referrer:
:return:
"""
# if in-house link detected
match = self.re_ads_link.match(self.url)
if match:
return True

referred_ads = current_app.config['GATEWAY_SERVICE_REFERRED_DOMAIN']
url = urllib.parse.urlparse(self.url)
match = self.re_ads_link.match(self.url)
return all([url.scheme, url.netloc]) or match
if all([url.scheme, url.netloc]):
try:
# if full url in-house link detected
if url.netloc.split('.',1)[-1] == referred_ads:
return True
except:
try:
# if a valid outside link, see if originated from ads
referrer_url = urllib.parse.urlparse(referrer)
if all([referrer_url.scheme, referrer_url.netloc]):
if referrer_url.netloc.split('.',1)[-1] == referred_ads:
return True
except:
# TODO: check back with resolver servcie, make sure it is from there
pass
pass

# for now do not redirect if outside link and did not originate from BBB
return False

def process_request(self):
"""
Expand Down Expand Up @@ -210,7 +233,7 @@ def process_request(self):
# make sure link_type is valid
if self.verify_link_type():
# make sure we have a valid url to redirect to
if self.verify_url():
if self.verify_url(self.referrer):
current_app.logger.debug('received to redirect to %s' %(self.url))
if log_the_click:
log_request(self.bibcode, self.user_id, self.link_type, self.url, self.referrer, self.client_id, self.real_ip, self.user_agent)
Expand Down

0 comments on commit af94cc0

Please sign in to comment.