def test_no_data(self, _, crawler_task_mock, pipeline):
factories.InstagramProfileFactory()
factories.InstagramProfileFactory(
friends_count=pipeline.DEFAULT_MINIMUM_FRIENDS_COUNT - 1)
profiles_to_use = [
factories.InstagramProfileFactory(
friends_count=pipeline.DEFAULT_MINIMUM_FRIENDS_COUNT),
factories.InstagramProfileFactory(
friends_count=pipeline.DEFAULT_MINIMUM_FRIENDS_COUNT + 1),
]
pipeline.run_pipeline()
assert crawler_task_mock.call_count == len(profiles_to_use)
for run in range(len(profiles_to_use)):
crawler_task_mock.assert_any_call(
queue='test_queue_name',
kwargs={
'klass_name':
'KeywordClassifier',
'profile_id':
profiles_to_use[run].id,
'task_type':
'pipeline',
'route': [
'KeywordClassifier',
'DescriptionLengthClassifier',
'OnlyBloggersProcessor',
]
})
def test_undecided_profiles(self, crawler_task_mock, profile_config,
was_processed):
profile = factories.InstagramProfileFactory(**profile_config)
reprocess_instagram_profiles(friends_lower_bound=50000, period_weeks=0)
assert crawler_task_mock.called is was_processed
assert InstagramProfile.objects.get(
id=profile.id
).reprocess_tries_count == profile.reprocess_tries_count + int(
was_processed)
def test_matching_influencers(self, find_influencers_mock, *args):
url = 'http://google.com'
existing_influencers = {url: debra_factories.InfluencerFactory()}
find_influencers_mock.return_value = existing_influencers, (url, )
instagram_profile = factories.InstagramProfileFactory()
assert create_influencer_from_instagram(instagram_profile.id,
False) == (
False,
existing_influencers,
)
def test_newly_created_profiles(self, crawler_task_mock, date_created,
was_processed):
profile = factories.InstagramProfileFactory(
friends_count=50000,
tags='lifestyle undecided',
reprocess_tries_count=2,
)
profile.date_created = date_created
profile.save()
reprocess_instagram_profiles(friends_lower_bound=50000, period_weeks=2)
assert crawler_task_mock.called is was_processed
assert InstagramProfile.objects.get(
id=profile.id
).reprocess_tries_count == profile.reprocess_tries_count + int(
was_processed)
def instagram_profile(self):
return factories.InstagramProfileFactory(
username='******',
friends_count=51,
followers_count=924,
post_count=38,
last_post_time=datetime(2016, 3, 2, 18, 43),
api_data={
u'biography': u"I don't know what to write",
u'full_name': u'Anonymous',
},
api_data_history=[
{
'2015-10-01': {
u'biography': u"I don't know what to write",
u'full_name': u'Anonymous',
},
},
])
def test_id_as_data(self, _, crawler_task_mock, pipeline, input_type):
instagram_profile = factories.InstagramProfileFactory()
pipeline.run_pipeline(data=input_type(instagram_profile.id))
assert crawler_task_mock.call_count == 1
assert crawler_task_mock.call_args[1] == {
'queue': 'test_queue_name',
'kwargs': {
'klass_name':
'KeywordClassifier',
'profile_id':
instagram_profile.id,
'task_type':
'pipeline',
'route': [
'KeywordClassifier',
'DescriptionLengthClassifier',
'OnlyBloggersProcessor',
]
}
}
def test_create_influencer(self, find_influencers_mock,
create_influencer_platform_mock,
create_platform_for_influencer_mock, *args):
valid_url = 'https://twitter.com/JimCarrey'
find_influencers_mock.return_value = dict(), (valid_url, )
def fake_create_influencer_platform(url, *args, **kwargs):
influencer = debra_factories.InfluencerFactory()
debra_factories.PlatformFactory(influencer=influencer, url=url)
return influencer
create_influencer_platform_mock.side_effect = (
fake_create_influencer_platform)
def fake_create_platform_for_influencer(url, inf, *args, **kwargs):
return debra_factories.PlatformFactory(url=url, influencer=inf)
create_platform_for_influencer_mock.side_effect = (
fake_create_platform_for_influencer)
instagram_profile = factories.InstagramProfileFactory()
result = create_influencer_from_instagram(instagram_profile.id, False)
assert Influencer.objects.count() == 1
assert result == (True, Influencer.objects.all()[0])
def test_profile_has_no_tags(self, processor, tags, expected_result):
profile = factories.InstagramProfileFactory(tags=tags)
assert processor.proceed(profile.id) is expected_result
def test_no_pipeline(self, _, crawler_task_mock, pipeline):
pipeline.PIPELINE_ROUTE = None
instagram_profile = factories.InstagramProfileFactory()
pipeline.run_pipeline(data=instagram_profile.id)
pipeline.run_pipeline()
assert not crawler_task_mock.called
def test_default_api_data_history(self):
instagram_profile = factories.InstagramProfileFactory()
assert instagram_profile.api_data_history == []
def test_get_data_from_api_history(self, api_data_history, keys,
expected_data):
instagram_profile = factories.InstagramProfileFactory(
api_data_history=api_data_history)
assert instagram_profile._get_data_from_api_history(
keys=keys) == expected_data
def test_get_url_from_api(self, api_data, expected_url):
instagram_profile = factories.InstagramProfileFactory(
api_data=api_data)
assert instagram_profile.get_url_from_api() == (expected_url)