def test_count_availability_reset(self, mock_fn):
self.resource.extras = {'check:status': 200, 'check:available': True,
'check:date': datetime.now(),
'check:count-availability': 2}
check_res = {'check:status': 200, 'check:available': False,
'check:date': datetime.now()}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
check_resource(self.resource)
self.assertEqual(self.resource.extras['check:count-availability'], 1)
def test_count_availability_reset(self, mock_fn):
self.resource.extras = {'check:status': 200, 'check:available': True,
'check:date': datetime.now(),
'check:count-availability': 2}
check_res = {'check:status': 200, 'check:available': False,
'check:date': datetime.now()}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
check_resource(self.resource)
self.assertEquals(self.resource.extras['check:count-availability'], 1)
def test_check_resource_ignored_domain(self):
self.resource.extras = {}
self.resource.url = 'http://example-ignore.com/url'
self.resource.save()
res = check_resource(self.resource)
self.assertEquals(res.get('check:status'), 204)
self.assertEquals(res.get('check:available'), True)
def test_check_resource_linkchecker_check_error(self, mock_fn):
class DummyLinkchecker:
def check(self, _):
return {'check:error': 'ERROR'}
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEquals(res, ({'error': 'ERROR'}, 500))
def test_check_resource_ignored_domain(self):
self.resource.extras = {}
self.resource.url = 'http://example-ignore.com/url'
self.resource.save()
res = check_resource(self.resource)
self.assertEqual(res.get('check:status'), 204)
self.assertEqual(res.get('check:available'), True)
def test_check_resource_linkchecker_check_error(self, mock_fn):
class DummyLinkchecker:
def check(self, _):
return {'check:error': 'ERROR'}
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEqual(res, ({'error': 'ERROR'}, 500))
def test_check_resource_creates_no_activity(activity_app, mocker):
resource = ResourceFactory()
dataset = DatasetFactory(resources=[resource])
user = UserFactory()
login_user(user)
check_res = {'check:status': 200, 'check:available': True,
'check:date': datetime.now()}
class DummyLinkchecker:
def check(self, _):
return check_res
mocker.patch('udata.linkchecker.checker.get_linkchecker',
return_value=DummyLinkchecker)
check_resource(resource)
activities = Activity.objects.filter(related_to=dataset)
assert len(activities) == 0
def test_check_resource_linkchecker_no_status(self, mock_fn):
class DummyLinkchecker:
def check(self, _):
return {'check:available': True}
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEquals(res,
({'error': 'No status in response from linkchecker'},
503))
def test_check_resource_creates_no_activity(activity_app, mocker):
resource = ResourceFactory()
dataset = DatasetFactory(resources=[resource])
user = UserFactory()
login_user(user)
check_res = {'check:status': 200, 'check:available': True,
'check:date': datetime.now()}
class DummyLinkchecker:
def check(self, _):
return check_res
mocker.patch('udata.linkchecker.checker.get_linkchecker',
return_value=DummyLinkchecker)
check_resource(resource)
activities = Activity.objects.filter(related_to=dataset)
assert len(activities) == 0
def test_check_resource_linkchecker_no_status(self, mock_fn):
class DummyLinkchecker:
def check(self, _):
return {'check:available': True}
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEqual(res,
({'error': 'No status in response from linkchecker'},
503))
def test_check_resource_filter_result(self, mock_fn):
check_res = {'check:status': 200, 'dummy': 'dummy'}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEquals(res, check_res)
self.assertNotIn('dummy', self.resource.extras)
def test_check_resource_filter_result(self, mock_fn):
check_res = {'check:status': 200, 'dummy': 'dummy'}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEqual(res, check_res)
self.assertNotIn('dummy', self.resource.extras)
def test_check_resource_linkchecker_ok(self, mock_fn):
check_res = {'check:status': 200, 'check:available': True,
'check:date': datetime.now()}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEquals(res, check_res)
check_res.update({'check:count-availability': 1})
self.assertEquals(self.resource.extras, check_res)
def test_check_resource_linkchecker_ok(self, mock_fn):
check_res = {'check:status': 200, 'check:available': True,
'check:date': datetime.now()}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
self.assertEqual(res, check_res)
check_res.update({'check:count-availability': 1})
self.assertEqual(self.resource.extras, check_res)
def test_unvalid_cache(self, mock_fn):
self.resource.extras = {
'check:date': datetime.now() - timedelta(seconds=3600),
'check:status': 42
}
check_res = {
'check:status': 200,
'check:available': True,
'check:date': datetime.now()
}
class DummyLinkchecker:
def check(self, _):
return check_res
mock_fn.return_value = DummyLinkchecker
res = check_resource(self.resource)
# we get the result from DummyLinkchecker and not from cache
self.assertEquals(res, check_res)
def get(self, dataset, rid):
'''Checks that a resource's URL exists and returns metadata.'''
resource = self.get_resource_or_404(dataset, rid)
return check_resource(resource)
def get(self, dataset, rid):
'''Checks that a resource's URL exists and returns metadata.'''
resource = self.get_resource_or_404(dataset, rid)
return check_resource(resource)
def test_check_resource_linkchecker_no_check(self):
self.resource.extras['check:checker'] = 'no_check'
self.resource.save()
res = check_resource(self.resource)
self.assertEquals(res.get('check:status'), 204)
self.assertEquals(res.get('check:available'), True)
def test_check_resource_linkchecker_in_resource(self, mock_fn):
self.resource.extras['check:checker'] = 'another_linkchecker'
self.resource.save()
check_resource(self.resource)
args, kwargs = mock_fn.call_args
self.assertEquals(args, ('another_linkchecker', ))
def test_check_resource_no_linkchecker(self, mock_fn):
mock_fn.return_value = None
res = check_resource(self.resource)
self.assertEquals(res, ({'error': 'No linkchecker configured.'}, 503))
def test_check_resource_no_linkchecker(self, mock_fn):
mock_fn.return_value = None
res = check_resource(self.resource)
self.assertEqual(res, ({'error': 'No linkchecker configured.'}, 503))
def check_resources(self, number=5000):
'''Check <number> of URLs that have not been (recently) checked'''
if not current_app.config.get('LINKCHECKING_ENABLED'):
log.error('Link checking is disabled.')
return
base_pipeline = [
{
'$match': {
'resources': {
'$gt': []
}
}
},
{
'$project': {
'resources._id': True,
'resources.extras.check:date': True
}
},
{
'$unwind': '$resources'
},
]
# unchecked resources
pipeline = base_pipeline + [{
'$match': {
'resources.extras.check:date': {
'$eq': None
}
}
}, {
'$limit': number
}]
resources = list(Dataset.objects.aggregate(*pipeline))
# not recently checked resources
slots_left = number - len(resources)
if slots_left:
pipeline = base_pipeline + [{
'$match': {
'resources.extras.check:date': {
'$ne': None
}
}
}, {
'$sort': {
'resources.extras.check:date': 1
}
}, {
'$limit': slots_left
}]
resources += list(Dataset.objects.aggregate(*pipeline))
nb_resources = len(resources)
log.info('Checking %s resources...', nb_resources)
resource_check_result = []
for idx, dataset_resource in enumerate(resources):
dataset_obj = Dataset.objects.get(id=dataset_resource['_id'])
resource_id = dataset_resource['resources']['_id']
rid = uuid.UUID(resource_id)
resource_obj = get_by(dataset_obj.resources, 'id', rid)
log.info('Checking resource %s (%s/%s)', resource_id, idx + 1,
nb_resources)
if resource_obj.need_check():
result = check_resource(resource_obj)
#log.info(resource_obj.url)
#log.info(result)
if not result['check:available']:
data = {
'dataset': dataset_obj,
'resource': resource_obj,
'status': result['check:status']
}
resource_check_result.append(data)
else:
log.info("--> Skipping this resource, cache is fresh enough.")
#Group resources by dataset and send email
if resource_check_result:
resource_check_result.sort(key=lambda item: item['dataset'].id)
resource_groups = groupby(resource_check_result,
lambda item: item['dataset'])
admin_role = Role.objects.filter(name='admin').first()
recipients = [
user.email for user in User.objects.filter(roles=admin_role).all()
]
subject = 'Relatório de verificação de links do dados.gov.'
context = {
'subject': subject,
'resources': resource_groups,
'server': current_app.config.get('SERVER_NAME')
}
msg = Message(subject=subject,
sender='*****@*****.**',
recipients=recipients)
#msg.body = theme.render('mail/link_checker_warning.txt', **context)
msg.html = theme.render('mail/link_checker_warning.html', **context)
mail = current_app.extensions.get('mail')
try:
mail.send(msg)
except:
pass
log.info('Done.')
def test_check_resource_linkchecker_in_resource(self, mock_fn):
self.resource.extras['check:checker'] = 'another_linkchecker'
self.resource.save()
check_resource(self.resource)
args, kwargs = mock_fn.call_args
self.assertEqual(args, ('another_linkchecker', ))
def test_check_resource_linkchecker_no_check(self):
self.resource.extras['check:checker'] = 'no_check'
self.resource.save()
res = check_resource(self.resource)
self.assertEqual(res.get('check:status'), 204)
self.assertEqual(res.get('check:available'), True)