def test_base_query():
query = SnapshotQuery(VALID_WHERE_STATEMENT)
assert query.get_base_query() == {
'query': {
'where': VALID_WHERE_STATEMENT
}
}
def test_extraction_query():
query = SnapshotQuery(VALID_WHERE_STATEMENT)
print(query.get_extraction_query())
assert query.get_extraction_query() == {
'query': {
'where': "publication_datetime >= '2021-01-01'",
'format': 'avro'
}
}
def test_analytics_query():
query = SnapshotQuery(VALID_WHERE_STATEMENT)
assert query.get_analytics_query() == {
'query': {
'where': "publication_datetime >= '2021-01-01'",
'frequency': 'MONTH',
'date_field': 'publication_datetime',
'group_by_source_code': False,
'top': 10
}
}
def test_create_snapshot_query_parameter():
q = SnapshotQuery(VALID_WHERE_STATEMENT)
s = Snapshot(query=q)
assert s.api_user.api_key == ENVIRONMENT_USER_KEY
assert s.query.get_base_query() == {
'query': {
'where': VALID_WHERE_STATEMENT
}
}
from factiva.news.snapshot import Snapshot, SnapshotQuery
querystr = "publication_datetime >= '2018-01-01 00:00:00' AND publication_datetime <= '2018-01-10 00:00:00' AND LOWER(language_code) = 'en'"
st1 = Snapshot(query=querystr)
print(st1.query.get_explain_query())
print(st1.query.get_analytics_query())
print(st1.query.get_extract_query())
query1 = SnapshotQuery(where=querystr,
limit=1000,
file_format='avro',
frequency='YEAR',
date_field='publication_datetime',
group_by_source_code=True,
top=25)
st2 = Snapshot(query=query1)
print(st2.query.get_explain_query())
print(st2.query.get_analytics_query())
print(st2.query.get_extract_query())
def test_invalid_field():
with pytest.raises(
ValueError,
match=r'Value of file_format is not within the allowed options'):
query = SnapshotQuery(VALID_WHERE_STATEMENT, file_format='TXT')