def test_make_meta_backends(index):
dtypes = ["int8", "int32", "int64", "float64"]
df = cudf.DataFrame(
{dt: np.arange(start=0, stop=3, dtype=dt)
for dt in dtypes})
df["strings"] = ["cat", "dog", "fish"]
df["cats"] = df["strings"].astype("category")
df["time_s"] = np.array(["2018-10-07", "2018-10-08", "2018-10-09"],
dtype="datetime64[s]")
df["time_ms"] = df["time_s"].astype("datetime64[ms]")
df["time_ns"] = df["time_s"].astype("datetime64[ns]")
df = df.set_index(index)
# Check "empty" metadata types
chk_meta = make_meta(df)
dd.assert_eq(chk_meta.dtypes, df.dtypes)
# Check "non-empty" metadata types
chk_meta_nonempty = meta_nonempty(df)
dd.assert_eq(chk_meta.dtypes, chk_meta_nonempty.dtypes)
# Check dask code path if not MultiIndex
if not isinstance(df.index, cudf.MultiIndex):
ddf = dgd.from_cudf(df, npartitions=1)
# Check "empty" metadata types
dd.assert_eq(ddf._meta.dtypes, df.dtypes)
# Check "non-empty" metadata types
dd.assert_eq(ddf._meta.dtypes, ddf._meta_nonempty.dtypes)
def make_meta_object(x, index=None):
"""Create an empty cudf object containing the desired metadata.
Parameters
----------
x : dict, tuple, list, cudf.Series, cudf.DataFrame, cudf.Index,
dtype, scalar
To create a DataFrame, provide a `dict` mapping of `{name: dtype}`, or
an iterable of `(name, dtype)` tuples. To create a `Series`, provide a
tuple of `(name, dtype)`. If a cudf object, names, dtypes, and index
should match the desired output. If a dtype or scalar, a scalar of the
same dtype is returned.
index : cudf.Index, optional
Any cudf index to use in the metadata. If none provided, a
`RangeIndex` will be used.
Examples
--------
>>> make_meta([('a', 'i8'), ('b', 'O')])
Empty DataFrame
Columns: [a, b]
Index: []
>>> make_meta(('a', 'f8'))
Series([], Name: a, dtype: float64)
>>> make_meta('i8')
1
"""
if hasattr(x, "_meta"):
return x._meta
elif is_arraylike(x) and x.shape:
return x[:0]
if index is not None:
index = make_meta(index)
if isinstance(x, dict):
return cudf.DataFrame(
{c: _empty_series(c, d, index=index)
for (c, d) in x.items()},
index=index,
)
if isinstance(x, tuple) and len(x) == 2:
return _empty_series(x[0], x[1], index=index)
elif isinstance(x, (list, tuple)):
if not all(isinstance(i, tuple) and len(i) == 2 for i in x):
raise ValueError(
f"Expected iterable of tuples of (name, dtype), got {x}")
return cudf.DataFrame(
{c: _empty_series(c, d, index=index)
for (c, d) in x},
columns=[c for c, d in x],
index=index,
)
elif not hasattr(x, "dtype") and x is not None:
# could be a string, a dtype object, or a python type. Skip `None`,
# because it is implictly converted to `dtype('f8')`, which we don't
# want here.
try:
dtype = np.dtype(x)
return _scalar_from_dtype(dtype)
except Exception:
# Continue on to next check
pass
if is_scalar(x):
return _nonempty_scalar(x)
raise TypeError(f"Don't know how to create metadata from {x}")