def test_encode_df():
labels = {
'int': 7,
'str': 'wassup?',
}
df = pd.read_csv('{}/weather.csv'.format(here))
df['STATION_CAT'] = df['STATION'].astype('category')
df['WDF2_F'] = df['WDF2'].astype(np.float)
msg = pbutils.df2msg(df, labels)
names = [col.name for col in msg.columns]
assert set(names) == set(df.columns), 'columns mismatch'
assert not msg.indices, 'has index'
assert pbutils.pb2py(msg.labels) == labels, 'lables mismatch'
# Now with index
index_name = 'DATE'
df = df.set_index(index_name)
msg = pbutils.df2msg(df, None)
names = [col.name for col in msg.columns]
assert set(names) == set(df.columns), 'columns mismatch'
assert msg.indices, 'no index'
assert msg.indices[0].name == index_name, 'bad index name'
def test_index_cols():
cols = list('abcdef')
size = 10
df = pd.DataFrame({col: np.random.rand(size) for col in cols})
index_cols = np.random.choice(cols, size=2)
cols = set(col for col in cols if col not in index_cols)
msg = pbutils.df2msg(df, index_cols=index_cols)
assert set(col.name for col in msg.columns) == cols, 'bad columns'
assert set(col.name for col in msg.indices) == set(index_cols), \
'bad indices'
def test_multi_index():
tuples = [('bar', 'one'), ('bar', 'two'), ('baz', 'one'), ('baz', 'two'),
('foo', 'one'), ('foo', 'two'), ('qux', 'one'), ('qux', 'two')]
index = pd.MultiIndex.from_tuples(tuples, names=['first', 'second'])
df = pd.DataFrame(index=index)
df['x'] = range(len(df))
data = pbutils.df2msg(df).SerializeToString()
msg = fpb.Frame.FromString(data)
for col in msg.indices:
values = col.strings
assert len(values) == len(df), 'bad index length'
def _read(self, *args, **kw):
io = BytesIO()
for df in self.data:
data = df2msg(df, None).SerializeToString()
io.write(struct.pack(http.header_fmt, len(data)))
io.write(data)
io.seek(0, 0)
class Response:
raw = io
ok = True
return Response
def test_encode_df():
labels = {
'int': 7,
'str': 'wassup?',
}
df = pd.read_csv('{}/weather.csv'.format(here))
msg = pbutils.df2msg(df, labels)
names = [col.name for col in msg.columns]
assert set(names) == set(df.columns), 'columns mismatch'
assert not msg.indices, 'has index'
assert pbutils.pb2py(msg.labels) == labels, 'lables mismatch'
# Now with index
index_name = 'DATE'
df.index = df.pop(index_name)
msg = pbutils.df2msg(df, None)
names = [col.name for col in msg.columns]
assert set(names) == set(df.columns), 'columns mismatch'
assert msg.indices, 'no index'
assert msg.indices[0].name == index_name, 'bad index name'