def test_lookup_generic_handle_missing(self):
data = pd.DataFrame.from_records(
[
{
'sid': 0,
'file_name': 'real',
'company_name': 'real',
'start_date_nano': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
{
'sid': 1,
'file_name': 'also_real',
'company_name': 'also_real',
'start_date_nano': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose end date is before our query date. We should
# still correctly find it.
{
'sid': 2,
'file_name': 'real_but_old',
'company_name': 'real_but_old',
'start_date_nano': pd.Timestamp('2002-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2003-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose start_date is **after** our query date. We should
# **not** find it.
{
'sid': 3,
'file_name': 'real_but_in_the_future',
'company_name': 'real_but_in_the_future',
'start_date_nano': pd.Timestamp('2014-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2020-1-1', tz='UTC'),
'exchange': 'THE FUTURE',
},
]
)
finder = AssetFinder(data)
results, missing = finder.lookup_generic(
['real', 1, 'fake', 'real_but_old', 'real_but_in_the_future'],
pd.Timestamp('2013-02-01', tz='UTC'),
)
self.assertEqual(len(results), 3)
self.assertEqual(results[0].symbol, 'real')
self.assertEqual(results[0].sid, 0)
self.assertEqual(results[1].symbol, 'also_real')
self.assertEqual(results[1].sid, 1)
self.assertEqual(results[2].symbol, 'real_but_old')
self.assertEqual(results[2].sid, 2)
self.assertEqual(len(missing), 2)
self.assertEqual(missing[0], 'fake')
self.assertEqual(missing[1], 'real_but_in_the_future')
def test_lookup_generic_handle_missing(self):
data = pd.DataFrame.from_records(
[
# Sids that will be found when we do lookups.
{
'sid': 0,
'file_name': 'real',
'company_name': 'real',
'start_date_nano': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
{
'sid': 1,
'file_name': 'also_real',
'company_name': 'also_real',
'start_date_nano': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose end date is before our query date. We should
# still correctly find it.
{
'sid': 2,
'file_name': 'real_but_old',
'company_name': 'real_but_old',
'start_date_nano': pd.Timestamp('2002-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2003-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose end date is before our query date. We should
# still correctly find it.
{
'sid': 3,
'file_name': 'real_but_in_the_future',
'company_name': 'real_but_in_the_future',
'start_date_nano': pd.Timestamp('2014-1-1', tz='UTC'),
'end_date_nano': pd.Timestamp('2020-1-1', tz='UTC'),
'exchange': 'THE FUTURE',
},
]
)
finder = AssetFinder(data)
results, missing = finder.lookup_generic(
['real', 1, 'fake', 'real_but_old', 'real_but_in_the_future'],
pd.Timestamp('2013-02-01', tz='UTC'),
)
self.assertEqual(len(results), 3)
self.assertEqual(results[0].symbol, 'real')
self.assertEqual(results[0].sid, 0)
self.assertEqual(results[1].symbol, 'also_real')
self.assertEqual(results[1].sid, 1)
self.assertEqual(len(missing), 2)
self.assertEqual(missing[0], 'fake')
self.assertEqual(missing[1], 'real_but_in_the_future')
def test_lookup_generic_handle_missing(self):
data = pd.DataFrame.from_records(
[
{
'sid': 0,
'symbol': 'real',
'start_date': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
{
'sid': 1,
'symbol': 'also_real',
'start_date': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose end date is before our query date. We should
# still correctly find it.
{
'sid': 2,
'symbol': 'real_but_old',
'start_date': pd.Timestamp('2002-1-1', tz='UTC'),
'end_date': pd.Timestamp('2003-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose start_date is **after** our query date. We should
# **not** find it.
{
'sid': 3,
'symbol': 'real_but_in_the_future',
'start_date': pd.Timestamp('2014-1-1', tz='UTC'),
'end_date': pd.Timestamp('2020-1-1', tz='UTC'),
'exchange': 'THE FUTURE',
},
]
)
self.env.write_data(equities_df=data)
finder = AssetFinder(self.env.engine)
results, missing = finder.lookup_generic(
['REAL', 1, 'FAKE', 'REAL_BUT_OLD', 'REAL_BUT_IN_THE_FUTURE'],
pd.Timestamp('2013-02-01', tz='UTC'),
)
self.assertEqual(len(results), 3)
self.assertEqual(results[0].symbol, 'REAL')
self.assertEqual(results[0].sid, 0)
self.assertEqual(results[1].symbol, 'ALSO_REAL')
self.assertEqual(results[1].sid, 1)
self.assertEqual(results[2].symbol, 'REAL_BUT_OLD')
self.assertEqual(results[2].sid, 2)
self.assertEqual(len(missing), 2)
self.assertEqual(missing[0], 'FAKE')
self.assertEqual(missing[1], 'REAL_BUT_IN_THE_FUTURE')
def test_lookup_generic_handle_missing(self):
data = pd.DataFrame.from_records(
[
{
"sid": 0,
"symbol": "real",
"start_date": pd.Timestamp("2013-1-1", tz="UTC"),
"end_date": pd.Timestamp("2014-1-1", tz="UTC"),
"exchange": "",
},
{
"sid": 1,
"symbol": "also_real",
"start_date": pd.Timestamp("2013-1-1", tz="UTC"),
"end_date": pd.Timestamp("2014-1-1", tz="UTC"),
"exchange": "",
},
# Sid whose end date is before our query date. We should
# still correctly find it.
{
"sid": 2,
"symbol": "real_but_old",
"start_date": pd.Timestamp("2002-1-1", tz="UTC"),
"end_date": pd.Timestamp("2003-1-1", tz="UTC"),
"exchange": "",
},
# Sid whose start_date is **after** our query date. We should
# **not** find it.
{
"sid": 3,
"symbol": "real_but_in_the_future",
"start_date": pd.Timestamp("2014-1-1", tz="UTC"),
"end_date": pd.Timestamp("2020-1-1", tz="UTC"),
"exchange": "THE FUTURE",
},
]
)
self.env.write_data(equities_df=data)
finder = AssetFinder(self.env.engine)
results, missing = finder.lookup_generic(
["REAL", 1, "FAKE", "REAL_BUT_OLD", "REAL_BUT_IN_THE_FUTURE"], pd.Timestamp("2013-02-01", tz="UTC")
)
self.assertEqual(len(results), 3)
self.assertEqual(results[0].symbol, "REAL")
self.assertEqual(results[0].sid, 0)
self.assertEqual(results[1].symbol, "ALSO_REAL")
self.assertEqual(results[1].sid, 1)
self.assertEqual(results[2].symbol, "REAL_BUT_OLD")
self.assertEqual(results[2].sid, 2)
self.assertEqual(len(missing), 2)
self.assertEqual(missing[0], "FAKE")
self.assertEqual(missing[1], "REAL_BUT_IN_THE_FUTURE")
def test_lookup_generic_handle_missing(self):
data = pd.DataFrame.from_records([
{
'sid': 0,
'symbol': 'real',
'start_date': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
{
'sid': 1,
'symbol': 'also_real',
'start_date': pd.Timestamp('2013-1-1', tz='UTC'),
'end_date': pd.Timestamp('2014-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose end date is before our query date. We should
# still correctly find it.
{
'sid': 2,
'symbol': 'real_but_old',
'start_date': pd.Timestamp('2002-1-1', tz='UTC'),
'end_date': pd.Timestamp('2003-1-1', tz='UTC'),
'exchange': '',
},
# Sid whose start_date is **after** our query date. We should
# **not** find it.
{
'sid': 3,
'symbol': 'real_but_in_the_future',
'start_date': pd.Timestamp('2014-1-1', tz='UTC'),
'end_date': pd.Timestamp('2020-1-1', tz='UTC'),
'exchange': 'THE FUTURE',
},
])
self.env.write_data(equities_df=data)
finder = AssetFinder(self.env.engine)
results, missing = finder.lookup_generic(
['real', 1, 'fake', 'real_but_old', 'real_but_in_the_future'],
pd.Timestamp('2013-02-01', tz='UTC'),
)
self.assertEqual(len(results), 3)
self.assertEqual(results[0].symbol, 'REAL')
self.assertEqual(results[0].sid, 0)
self.assertEqual(results[1].symbol, 'ALSO_REAL')
self.assertEqual(results[1].sid, 1)
self.assertEqual(results[2].symbol, 'REAL_BUT_OLD')
self.assertEqual(results[2].sid, 2)
self.assertEqual(len(missing), 2)
self.assertEqual(missing[0], 'fake')
self.assertEqual(missing[1], 'real_but_in_the_future')
