def test_generate_scorable_job_report_prefer_job_breakdown_false(
mock_generate_child_claims, mock_prefer_job_breakdown,
mock_fetch_job_report):
mock_prefer_job_breakdown.side_effect = [True, False]
mock_fetch_job_report.return_value = sentinel.job_report
mock_generate_child_claims.return_value = [
Mock(
entity_id='ad_id',
entity_type=Entity.Ad,
report_type=ReportType.lifetime,
job_signature=JobSignature('job_id'),
)
]
claim = Mock(
entity_id='adset_id',
entity_type=Entity.AdSet,
report_type=ReportType.lifetime,
job_signature=JobSignature('job_id'),
is_divisible=True,
)
result = list(generate_scorable(claim))
assert len(result) == 1
assert result[0].entity_id == 'ad_id'
def test_day_metrics_per_entity_under_ad_account_not_divisible(
mock_iter_reality_per_ad_account):
reality_claim = RealityClaim(
ad_account_id='ad-account-id',
entity_id='ad-account-id',
entity_type=Entity.AdAccount,
timezone='America/Los_Angeles',
)
mock_iter_reality_per_ad_account.return_value = [
RealityClaim(
entity_type=Entity.Ad,
campaign_id='campaign-1',
adset_id='adset-1',
entity_id='ad-1',
bol=datetime(2019, 1, 1, 12, 0),
eol=datetime(2019, 1, 2, 12, 0),
),
RealityClaim(entity_type=Entity.Ad,
entity_id='ad-2',
bol=datetime(2019, 1, 3, 12, 0),
eol=datetime(2019, 1, 3, 12, 0)),
]
result = list(
day_metrics_per_ads_under_ad_account([ReportType.day], reality_claim))
assert result == [
ExpectationClaim(
'ad-account-id',
Entity.AdAccount,
ReportType.day,
Entity.Ad,
JobSignature('fb|ad-account-id|||day|A|2019-01-01'),
ad_account_id='ad-account-id',
timezone='America/Los_Angeles',
range_start=date(2019, 1, 1),
),
ExpectationClaim(
'ad-account-id',
Entity.AdAccount,
ReportType.day,
Entity.Ad,
JobSignature('fb|ad-account-id|||day|A|2019-01-02'),
ad_account_id='ad-account-id',
timezone='America/Los_Angeles',
range_start=date(2019, 1, 2),
),
ExpectationClaim(
'ad-account-id',
Entity.AdAccount,
ReportType.day,
Entity.Ad,
JobSignature('fb|ad-account-id|||day|A|2019-01-03'),
ad_account_id='ad-account-id',
timezone='America/Los_Angeles',
range_start=date(2019, 1, 3),
),
]
def test_assign_score(report_type, historical_ratio, skew_ratio, score):
claim = ScorableClaim('A1', Entity.Ad, report_type, Entity.Ad, JobSignature('jobid'), None)
with patch.object(ScoreCalculator, 'historical_ratio', return_value=historical_ratio), \
patch.object(ScoreCalculator, 'skew_ratio', return_value=skew_ratio):
result = ScoreCalculator.assign_score(claim)
assert result == pytest.approx(score, abs=0.1)
def lifetime_metrics_per_entity_under_ad_account(
entity_type: str, reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
"""Generate ad-account level expectation claims for lifetime."""
if not reality_claim.timezone:
return
# TODO: Remove once all entities have parent ids
# Divide tasks only if parent levels are defined for all ads
is_dividing_possible = True
root_node = EntityNode(reality_claim.entity_id, reality_claim.entity_type)
for child_claim in iter_reality_per_ad_account_claim(
reality_claim, entity_types=[entity_type]):
is_dividing_possible = is_dividing_possible and child_claim.all_parent_ids_set
new_node = EntityNode(child_claim.entity_id, child_claim.entity_type)
root_node.add_node(new_node, path=child_claim.parent_entity_ids)
logger.warning(
f'[dividing-possible] Ad Account {reality_claim.ad_account_id} Dividing possible: {is_dividing_possible}'
)
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.lifetime,
entity_type,
JobSignature(
generate_id(ad_account_id=reality_claim.ad_account_id,
report_type=ReportType.lifetime,
report_variant=entity_type)),
ad_account_id=reality_claim.ad_account_id,
entity_hierarchy=root_node if is_dividing_possible else None,
timezone=reality_claim.timezone,
)
def test_lifetime_score(dt, expected_score):
signature = JobSignature('jobid')
claim = ScorableClaim('A1', Entity.Ad, ReportType.lifetime, Entity.Ad, signature, None)
with patch.object(ScoreSkewHandlers, 'get_now', return_value=dt) as mm:
score = ScoreSkewHandlers.lifetime_skew(claim=claim)
assert mm.called
assert score == pytest.approx(expected_score, abs=0.01)
def test_generate_child_claims():
entity_hierarchy = EntityNode('adset-id', Entity.AdSet)
ad1, ad2 = EntityNode('ad-id1', Entity.Ad), EntityNode('ad-id2', Entity.Ad)
entity_hierarchy.add_node(ad1)
entity_hierarchy.add_node(ad2)
claim = ExpectationClaim(
'adset-id',
Entity.AdSet,
ReportType.lifetime,
Entity.Ad,
JobSignature('fb|ad-account-id|A|adset-id|lifetime|A|2019-02-20'),
ad_account_id='ad-account-id',
timezone='timezone',
entity_hierarchy=entity_hierarchy,
range_start=date(2019, 2, 20),
)
result = list(generate_child_claims(claim))
assert result == [
ExpectationClaim(
'ad-id1',
Entity.Ad,
ReportType.lifetime,
Entity.Ad,
JobSignature('fb|ad-account-id|A|ad-id1|lifetime|A|2019-02-20'),
ad_account_id='ad-account-id',
timezone='timezone',
entity_hierarchy=ad1,
range_start=date(2019, 2, 20),
),
ExpectationClaim(
'ad-id2',
Entity.Ad,
ReportType.lifetime,
Entity.Ad,
JobSignature('fb|ad-account-id|A|ad-id2|lifetime|A|2019-02-20'),
ad_account_id='ad-account-id',
timezone='timezone',
entity_hierarchy=ad2,
range_start=date(2019, 2, 20),
),
]
def test_lifetime_metrics_per_entity_under_ad_account_is_divisible(
mock_iter_reality_per_ad_account):
reality_claim = RealityClaim(
ad_account_id='ad-account-id',
entity_id='ad-account-id',
entity_type=Entity.AdAccount,
timezone='America/Los_Angeles',
)
mock_iter_reality_per_ad_account.return_value = [
RealityClaim(
entity_type=Entity.Ad,
campaign_id='campaign-1',
adset_id='adset-1',
entity_id='ad-1',
bol=datetime(2019, 1, 1, 12, 0),
eol=datetime(2019, 1, 1, 12, 0),
)
]
result = list(
lifetime_metrics_per_entity_under_ad_account(Entity.Ad, reality_claim))
assert result == [
ExpectationClaim(
'ad-account-id',
Entity.AdAccount,
ReportType.lifetime,
Entity.Ad,
JobSignature('fb|ad-account-id|||lifetime|A'),
ad_account_id='ad-account-id',
timezone='America/Los_Angeles',
entity_hierarchy=EntityNode(
'ad-account-id',
Entity.AdAccount,
children=[
EntityNode(
'campaign-1',
Entity.Campaign,
children=[
EntityNode(
'adset-1',
Entity.AdSet,
children=[EntityNode('ad-1', Entity.Ad)])
],
)
],
),
)
]
def test_generate_scorable_job_report_none(mock_generate_child_claims,
mock_prefer_job_breakdown,
mock_fetch_job_report):
mock_fetch_job_report.return_value = None
claim = Mock(
entity_id='entity_id',
entity_type=Entity.Ad,
report_type=ReportType.lifetime,
job_signature=JobSignature('job_id'),
is_divisible=True,
)
result = list(generate_scorable(claim))
assert len(result) == 1
assert not mock_prefer_job_breakdown.called
assert not mock_generate_child_claims.called
def page_entity(
reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
assert reality_claim.entity_type == Entity.Page, 'Page expectation should be triggered only by page reality claims'
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.entity,
Entity.Page,
JobSignature(
generate_id(
ad_account_id=reality_claim.ad_account_id,
entity_id=reality_claim.entity_id,
report_type=ReportType.entity,
report_variant=Entity.Page,
)),
ad_account_id=reality_claim.ad_account_id,
)
def entities_per_page(
entity_type: str, reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
"""
Generates "fetch EntityType entities metadata per given Page" job call sig
"""
assert entity_type in Entity.NON_AA_SCOPED
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.entity,
entity_type,
JobSignature(
generate_id(ad_account_id=reality_claim.ad_account_id,
report_type=ReportType.entity,
report_variant=entity_type)),
ad_account_id=reality_claim.ad_account_id,
)
def lifetime_page_metrics_per_entity(
entity_type: str, reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
assert entity_type in Entity.ALL
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.lifetime,
entity_type,
JobSignature(
generate_id(
ad_account_id=reality_claim.ad_account_id,
entity_type=reality_claim.entity_type,
entity_id=reality_claim.entity_id,
report_type=ReportType.lifetime,
report_variant=entity_type,
)),
ad_account_id=reality_claim.ad_account_id,
)
def ad_account_entity(
reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
assert (
reality_claim.entity_type == Entity.AdAccount
), 'Ad account expectation should be triggered only by ad account reality claims'
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.entity,
Entity.AdAccount,
JobSignature(
generate_id(
ad_account_id=reality_claim.ad_account_id,
entity_id=reality_claim.entity_id,
report_type=ReportType.entity,
report_variant=Entity.AdAccount,
)),
ad_account_id=reality_claim.ad_account_id,
timezone=reality_claim.timezone,
)
def generate_child_claims(
claim: ExpectationClaim) -> Generator[ExpectationClaim, None, None]:
for child_entity_node in claim.entity_hierarchy.children:
yield ExpectationClaim(
child_entity_node.entity_id,
child_entity_node.entity_type,
claim.report_type,
claim.report_variant,
JobSignature(
generate_id(
ad_account_id=claim.ad_account_id,
range_start=claim.range_start,
report_type=claim.report_type,
report_variant=claim.report_variant,
entity_id=child_entity_node.entity_id,
entity_type=child_entity_node.entity_type,
)),
ad_account_id=claim.ad_account_id,
timezone=claim.timezone,
entity_hierarchy=child_entity_node,
range_start=claim.range_start,
)
def entities_per_ad_account(
entity_type: str, reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
"""
Generates "fetch EntityType entities metadata per given AA" job call sig
"""
# Mental Note:
# This job signature generator is designed to be parked
# *under AdAccount job signatures generators inventory*
# In other words, NOT under EntityType inventory (where it would be called
# for each EntityType).
# Unlike metrics report types,
# We don't have an effective "fetch single EntityType entity data per EntityType ID" task (yet).
# So, instead of generating many job signatures per EntityTypes,
# we create only one per-parent-AA, and making that
# into "normative_job_signature" per AA level.
# When we have a need to have atomic per-EntityType entity data collection celery task,
# atomic per-C entity data job signature would go into normative column
# on ExpectationClaim for each and separate EntityType and per-parent-AA
# job signature will go into "effective_job_signatures" list on those claims,
# AND this function must move from AA-level to EntityType-level signature
# generators inventory.
assert entity_type in Entity.ALL
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.entity,
entity_type,
JobSignature(
generate_id(ad_account_id=reality_claim.ad_account_id,
report_type=ReportType.entity,
report_variant=entity_type)),
ad_account_id=reality_claim.ad_account_id,
timezone=reality_claim.timezone,
)
def pages_per_scope(
reality_claim: RealityClaim
) -> Generator[ExpectationClaim, None, None]:
"""
Generates "fetch Pages active entity metadata per given scope" job ID
To be used by Scope-level RealityClaim / ExpectationClaim.
"""
yield ExpectationClaim(
reality_claim.entity_id,
reality_claim.entity_type,
ReportType.import_pages,
Entity.Page,
JobSignature(
generate_id(
namespace=config.application.UNIVERSAL_ID_SYSTEM_NAMESPACE,
# Note absence of value for Page
# This is "all Pages per scope X" job.
entity_id=reality_claim.entity_id,
entity_type=reality_claim.entity_type,
report_type=ReportType.import_pages,
report_variant=Entity.Page,
)),
)
def test_persister_saves_job_scope_auxiliary_data_to_data_flower(self):
# There is a need to save some context data that does not fit on JobIS
# Persister should store that on the Data Flower.
sweep_id = random.gen_string_id()
entity_id = random.gen_string_id()
ad_account_id = random.gen_string_id()
job_id = generate_id(ad_account_id=ad_account_id,
report_type=ReportType.lifetime,
report_variant=Entity.Campaign)
prioritized_iter = [
PrioritizationClaim(
entity_id,
Entity.Campaign,
ReportType.lifetime,
JobSignature(job_id),
100,
ad_account_id=ad_account_id,
timezone='Europe/London',
# TODO: contemplate moving auxiliary data formation to
# place where JobSignatures are generated and use that
# data for Data Flower (as it was originally intended
# but not implemented because saving each job's data
# individually to Data Flower was too slow)
)
]
persisted = persister.iter_persist_prioritized(sweep_id,
prioritized_iter)
cnt = 0
for item in persisted:
cnt += 1
# just need to spin the generator
# so it does all the saving it needs to do per item
assert cnt == 1
# Now, finally, the testing:
jobs_queued_actual = []
with SortedJobsQueue(sweep_id).JobsReader() as jobs_iter:
for job_id, job_scope_data, score in jobs_iter:
jobs_queued_actual.append((job_id, job_scope_data, score))
jobs_queued_should_be = [(
job_id,
# Contents of this dict is what we are testing here
dict(
# comes from Persister code
# manually peeled off *Claim and injected into Data Flower
ad_account_timezone_name='Europe/London'),
100,
)]
assert jobs_queued_actual == jobs_queued_should_be
# And, another way of looking at it
# looper.iter_tasks preassembles JobScope and should apply aux data to it.
job_scope = None
cnt = 0
for celery_task, job_scope, job_context, score in TaskProducer(
sweep_id).iter_tasks():
cnt += 1
# this just needs to spin once
assert cnt == 1
job_scope_should_be = JobScope(
sweep_id=sweep_id,
namespace='fb',
ad_account_id=ad_account_id,
report_type=ReportType.lifetime,
report_variant=Entity.Campaign,
# \/ This is what we are testing \/
# comes from Persister code
# manually peeled off *Claim and injected into Data Flower
ad_account_timezone_name='Europe/London',
score=100,
)
assert job_scope.to_dict() == job_scope_should_be.to_dict()
