def test_interaction_coi_extraction_when_interaction_does_not_have_a_note(
add_objects, add_side_effect, mock_datetime
):
# Test mapping
add_objects(
Interactions,
[
{
'datahub_interaction_id': '18c3e449-ce56-45b4-93d3-362960622ca2',
'datahub_company_id': 'df14625d-c89b-4efd-bc5a-9b46f68b71bb',
'notes': None,
}
],
)
output = [()]
add_side_effect(output)
mock_datetime.datetime.now.return_value = datetime.datetime(2020, 1, 1)
mapper.analyse_interactions()
expected_rows = []
assert rows_equal_query_results(
flask_app.dbi,
expected_rows,
f'SELECT * FROM "{mapper.output_schema}"."{mapper.output_table}"',
)
session = flask_app.db.session
interaction_ids = session.query(InteractionsAnalysedInteractionIdLog).all()
assert len(interaction_ids) == 1
assert str(interaction_ids[0].datahub_interaction_id) == '18c3e449-ce56-45b4-93d3-362960622ca2'
assert interaction_ids[0].analysed_at == datetime.datetime(2020, 1, 1)
def test_interaction_coi_extraction(add_objects, add_side_effect, mock_datetime):
# Test mapping
add_objects(
Interactions,
[
{
'datahub_interaction_id': '52552367-436f-4a5d-84a2-dbf4ffeddb76',
'datahub_company_id': 'df14625d-c89b-4efd-bc5a-9b46f68b71bb',
'notes': 'Google exports to Brussels',
}
],
)
output = [[('Brussels', 'Belgium', 'exported', 'GPE', ['export'], False)]]
add_side_effect(output)
mock_datetime.datetime.now.return_value = datetime.datetime(2020, 1, 1)
mapper.analyse_interactions()
expected_rows = [
(
1,
'52552367-436f-4a5d-84a2-dbf4ffeddb76',
'Brussels',
'Belgium',
'exported',
'GPE',
['export'],
False,
)
]
assert rows_equal_query_results(
flask_app.dbi,
expected_rows,
f'SELECT * FROM "{mapper.output_schema}"."{mapper.output_table}"',
)
session = flask_app.db.session
interaction_ids = session.query(InteractionsAnalysedInteractionIdLog).all()
assert len(interaction_ids) == 1
assert str(interaction_ids[0].datahub_interaction_id) == '52552367-436f-4a5d-84a2-dbf4ffeddb76'
assert interaction_ids[0].analysed_at == datetime.datetime(2020, 1, 1)
def test_company_matching(
mocker,
app,
add_datahub_company,
add_datahub_contact,
add_export_wins,
add_countries_and_sectors_of_interest,
):
add_datahub_company([
{
'company_name': 'company_name1',
'datahub_company_id': 'f17645ba-c38e-4215-b09e-47451edb9125',
'companies_house_id': '90821308',
'sector': 'Food',
'reference_code': 'ORG-01233',
'postcode': 'GS3 3PL',
'modified_on': '2018-01-01 00:00:00',
},
{
'company_name': 'company_name2',
'datahub_company_id': '5965b49e-dae5-4a06-81fa-c894e652cd52',
'companies_house_id': '92890123',
'sector': 'Retail',
'reference_code': 'ORG-01244',
'postcode': 'GH3 3PL',
'modified_on': '2017-01-01 00:00:00',
},
])
add_datahub_contact([{
'datahub_contact_id': '93041a63-57d3-4b6b-a83d-172ba854f770',
'datahub_company_id': '5965b49e-dae5-4a06-81fa-c894e652cd52',
'email': '*****@*****.**',
}])
add_export_wins([
{
'export_wins_company_id': 'ORG-01244',
'contact_email_address': '*****@*****.**',
'sector': 'Aerospace',
'company_name': 'company_name2',
'country': 'China',
'export_wins_id': 'ffa75985-7bc0-4e9f-8d58-28a7f234b7fc',
'created_on': '2011-10-10 12:12:12',
'date_won': '2010-11-11 12:12:12',
},
])
add_countries_and_sectors_of_interest(
[
{
'service_company_id': 'f17645ba-c38e-4215-b09e-47451edb9125',
'company_match_id': None,
'country': 'country1',
'sector': 'sector1',
'type': 'interested',
'service': 'datahub',
'source': 'source1',
'source_id': 'f17645ba-c38e-4215-b09e-47451edb9125',
'timestamp': '2009-10-10T12:12:12',
},
{
'service_company_id': '5965b49e-dae5-4a06-81fa-c894e652cd52',
'company_match_id': None,
'country': 'country2',
'sector': None,
'type': 'exported',
'service': 'datahub',
'source': 'source2',
'source_id': '5965b49e-dae5-4a06-81fa-c894e652cd52',
'timestamp': '2009-10-10T13:00:00',
},
{
'service_company_id': 'ORG-01244',
'company_match_id': None,
'country': 'China',
'sector': 'Aerospace',
'type': 'exported',
'service': 'export_wins',
'source': 'export_wins',
'source_id': 'ffa75985-7bc0-4e9f-8d58-28a7f234b7fc',
'timestamp': '2010-11-11 12:12:12',
},
],
temp=True,
)
_api_mock(
mocker,
{
"matches": [
{
"id": 'f17645ba-c38e-4215-b09e-47451edb9125',
"match_id": 1,
"similarity": "101010",
},
{
"id": '5965b49e-dae5-4a06-81fa-c894e652cd52',
"match_id": 2,
"similarity": "101110",
},
{
"id": 'ffa75985-7bc0-4e9f-8d58-28a7f234b7fc',
"match_id": 2,
"similarity": "001110",
},
]
},
200,
)
app.config['cms']['base_url'] = 'https://test.com'
task = Task()
task()
expected_rows = [
(
1,
'f17645ba-c38e-4215-b09e-47451edb9125',
1,
'country1',
'sector1',
'interested',
'datahub',
'source1',
'f17645ba-c38e-4215-b09e-47451edb9125',
'2009-10-10 12:12:12',
),
(
2,
'5965b49e-dae5-4a06-81fa-c894e652cd52',
2,
'country2',
None,
'exported',
'datahub',
'source2',
'5965b49e-dae5-4a06-81fa-c894e652cd52',
'2009-10-10 13:00:00',
),
(
3,
'ORG-01244',
2,
'China',
'Aerospace',
'exported',
'export_wins',
'export_wins',
'ffa75985-7bc0-4e9f-8d58-28a7f234b7fc',
'2010-11-11 12:12:12',
),
]
assert rows_equal_query_results(
flask_app.dbi,
expected_rows,
f'SELECT * FROM {CountriesAndSectorsInterest.get_fq_table_name()}',
)
def test_interaction_already_seen(add_objects, add_side_effect):
add_objects(
Interactions,
[
{
'id': 1,
'datahub_interaction_id': '52552367-436f-4a5d-84a2-dbf4ffeddb76',
'datahub_company_id': 'df14625d-c89b-4efd-bc5a-9b46f68b71bb',
'notes': 'Google exports to Brussels',
},
{
'id': 2,
'datahub_interaction_id': '52552367-436f-4a5d-84a2-dbf4ffeddb77',
'datahub_company_id': 'df14625d-c89b-4efd-bc5a-9b46f68b71bb',
'notes': 'nothing mentioning a location',
},
{
'id': 3,
'datahub_interaction_id': '52552367-436f-4a5d-84a2-dbf4ffeddb78',
'datahub_company_id': 'df14625d-c89b-4efd-bc5a-9b46f68b71bb',
'notes': 'China',
},
],
)
add_objects(
InteractionsAnalysedInteractionIdLog,
[{'datahub_interaction_id': '52552367-436f-4a5d-84a2-dbf4ffeddb76'}],
)
outputs = [
[], # no country information
[('China', 'China', 'exported', 'GPE', ['export'], False)],
]
add_side_effect(outputs)
mapper.analyse_interactions()
expected_rows = [
(
'52552367-436f-4a5d-84a2-dbf4ffeddb78',
'China',
'China',
'exported',
'GPE',
['export'],
False,
),
]
assert rows_equal_query_results(
flask_app.dbi,
expected_rows,
f'SELECT datahub_interaction_id,place,'
f'standardized_place,action,type,context,negation'
f' FROM "{mapper.output_schema}"."{mapper.output_table}"',
)
session = flask_app.db.session
interaction_ids = session.query(InteractionsAnalysedInteractionIdLog).all()
assert len(interaction_ids) == 3
def test_country_mapping(
add_datahub_company_export_country,
add_datahub_company_export_country_history,
add_datahub_omis,
add_export_wins,
add_country_territory_registry,
):
add_datahub_company_export_country([
{
'company_export_country_id':
'9c83fe1f-24c9-4f44-8d04-3a6b82d6ca34',
'company_id': '25262ffe-e062-49af-a620-a84d4f3feb8b',
'country': 'afganistan',
'country_iso_alpha2_code': 'AF',
'created_on': '2020-01-01',
'modified_on': '2020-01-02',
'status': 'currently_exporting',
},
{
'company_export_country_id':
'8272b09f-f778-417f-a613-daae5c8a50c3',
'company_id': '25262ffe-e062-49af-a620-a84d4f3feb8b',
'country': 'netherlands antilles',
'country_iso_alpha2_code': 'AN',
'created_on': '2020-02-01',
'modified_on': '2020-02-02',
'status': 'future_interest',
},
])
add_datahub_company_export_country_history([{
'company_id': 'd584c5e2-ef16-4aba-91d4-71949078831f',
'country': 'Andorra',
'country_iso_alpha2_code': 'AD',
'history_date': '2020-01-01',
'history_id': 'd413d348-8828-48a7-88a1-1e8a1d7a2f1f',
'history_type': 'insert',
'status': 'future_interest',
}])
add_datahub_omis([{
'company_id': '7ff93060-1f05-4bbc-a58a-65ae7d7e8ba8',
'created_date': '2009-10-10',
'datahub_omis_order_id': 'a0e38b4f-f4c9-4ebc-b196-208972268efb',
'market': 'usa',
'sector': 'Aerospace',
}])
add_export_wins([
{
'export_wins_company_id': '12345678',
'contact_email_address': '*****@*****.**',
'sector': 'Aerospace',
'company_name': 'Test Company',
'country': 'uae',
'export_wins_id': 'ffa75985-7bc0-4e9f-8d58-28a7f234b7fc',
'created_on': '2009-10-10 12:12:12',
'date_won': '2010-11-11 12:12:12',
},
{
'export_wins_company_id': '12345678',
'contact_email_address': '*****@*****.**',
'sector': 'Aerospace',
'company_name': 'Company testing',
'country': 'unknown',
'export_wins_id': 'ffa75985-7bc0-4e9f-8d58-28a7f234b7fc',
'created_on': '2009-10-10 12:12:12',
'date_won': '2010-11-11 12:12:12',
},
])
# Populate registry
countries = [
('AF', 'Afghanistan'),
('AL', 'Albania'),
('DZ', 'Algeria'),
('AD', 'Andorra'),
('AO', 'Angola'),
('AG', 'Antigua and Barbuda'),
('BE', 'Belgium'),
('CL', 'Chile'),
('FR', 'France'),
('NZ', 'New Zealand'),
('CA', 'Canada'),
('US', 'United States'),
('CN', 'China'),
('DE', 'Germany'),
('BS', 'The Bahamas'),
('IN', 'India'),
('JP', 'Japan'),
('US', 'United States'),
('AU', 'Australia'),
('NZ', 'New Zealand'),
('CA', 'Canada'),
('AE', 'United Arab Emirates'),
]
country_territory_entries = []
for iso_alpha2_code, country in countries:
entry = {'country_iso_alpha2_code': iso_alpha2_code, 'name': country}
country_territory_entries.append(entry)
add_country_territory_registry(country_territory_entries)
# Standardise
mapper.standardise_countries()
# TODO: uncomment once export wins is included
expected_rows = [
(1, 'afganistan', 'Afghanistan', 91),
(2, 'Andorra', 'Andorra', 100),
(3, 'netherlands antilles', 'Bonaire', 100),
(4, 'netherlands antilles', 'Saint Eustatius', 100),
(5, 'netherlands antilles', 'Saba', 100),
(6, 'netherlands antilles', 'Curaçao', 100),
(7, 'netherlands antilles', 'Sint Maarten (Dutch part)', 100),
(8, 'uae', 'United Arab Emirates', 100),
(9, 'usa', 'United States', 100),
]
assert rows_equal_query_results(
flask_app.dbi,
expected_rows,
f'SELECT * FROM "{mapper.output_schema}"."{mapper.output_table}"',
)