def test_risk_id():
''' draws can accept risk_id or rei_id '''
df = draws({'risk_ids': [84]},
source='risk',
location_ids=[1],
draw_type='rr')
return df
def test_big_hdf():
''' some epi draws are stored as one big hdf '''
df = draws({'rei_ids': [94]},
source='risk',
location_ids=101,
year_ids=1990,
sex_ids=1,
draw_type='exposure')
return df
def test_exposure_draws():
df = draws({'rei_ids': [84]},
source='risk',
year_ids=[1990],
location_ids=[102],
age_group_ids=[6],
sex_ids=[2],
draw_type='exposure')
# there should be one row per categorical risk
assert len(df) == len(
df.modelable_entity_id.unique()), ('more than one row per ME returned')
return df
def test_string_format():
''' some identifier variables were returning as strings'''
df1 = draws({'rei_ids': [114]},
source='risk',
location_ids=101,
year_ids=1990,
sex_ids=1,
draw_type='rr',
verbose=True)
df2 = draws({'modelable_entity_ids': [1449]},
source='epi',
location_ids=[132],
verbose=True)
for df in [df1, df2]:
for col in df.columns:
if col.endswith('id'):
assert df[col].dtype != 'O'
return df1, df2
def test_rr_draws():
df = draws({'rei_ids': [128]},
source='risk',
year_ids=[1990],
location_ids=[102],
age_group_ids=[15],
sex_ids=[2],
draw_type='rr')
# there should be one row per categorical risk
assert len(df) == len(df.parameter.unique()), ('more than one row per '
'category returned')
return df
def test_como_version():
''' version should override status'''
df = draws({'sequela_ids': [1]},
source='como',
location_id=[1],
sex_id=[1],
year_id=[1990],
verbose=True,
age_group_id=[12],
metric_id=[1],
status='best',
version=86)
return df
def test_missing_files():
''' some risk exposure h5 files were missing because save results
called h5 conversion from csv and that occasionally fails. If get_draws
gets a file not found error, it should try to fall back on reading
the original csv'''
df = draws({'modelable_entity_ids': [8946]},
location_ids=[44553],
sex_ids=[2],
year_ids=[2010],
status=56300,
verbose=True,
source='epi')
return df
def test_como_gopher_interface():
'''instead of calling como.draws, call gopher.draws for como results'''
df = draws({'sequela_ids': [1]},
source='como',
location_id=[1],
sex_id=[1],
year_id=[1990],
verbose=True,
age_group_id=[12],
metric_id=[1],
status='latest')
assert not df.empty
assert len(df.groupby(['sequela_id',
'measure_id'])) == len(df.measure_id.unique())
return df
def test_risk_dups():
''' some risk exposure draws had duplicate cat4 rows '''
df = draws({'rei_ids': [94]},
source='risk',
location_ids=101,
year_ids=1990,
sex_ids=1,
age_group_ids=4,
draw_type='exposure')
assert ~df.duplicated(
subset=[col for col in df.columns if 'draw' not in col]).any(), (
'duplicates found')
return df
status = args['status']
args.pop('status')
# convert kwargs from a list of single key dicts to one dict with
# multiple keys, if any specified from get_draws.ado
for d in args.pop('kwargs'):
for k, v in d.iteritems():
args[k] = v
# Get draws
try:
df = draws(gbd_id_dict,
measure_ids=args.pop('measure_ids'),
location_ids=args.pop('location_ids'),
year_ids=args.pop('year_ids'),
age_group_ids=args.pop('age_group_ids'),
sex_ids=args.pop('sex_ids'),
status=status,
source=args.pop('source'),
include_risks=args.pop('include_risks'),
**args)
except Exception as e:
# catch all exceptions, because we need to write something to stdout
# no matter what error. Get_draws.ado creates a pipe and reads from
# it -- if nothing is written to the pipe, stata hangs
print "Encountered error while reading draws: {}".format(e)
raise
# stream results to sys.stdout for get_draws.ado to read in
# Use a dct because stata is faster at reading those
to_dct(df=df, fname=sys.stdout, include_header=True)