Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Enable decimal support in parquet writer #7673

Merged
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions python/cudf/cudf/_lib/cpp/io/parquet.pxd
Original file line number Diff line number Diff line change
Expand Up @@ -70,6 +70,7 @@ cdef extern from "cudf/io/parquet.hpp" namespace "cudf::io" nogil:
column_in_metadata& set_nullability(bool nullable)
column_in_metadata& set_list_column_as_map()
column_in_metadata& set_int96_timestamps(bool req)
column_in_metadata& set_decimal_precision(uint8_t precision)
column_in_metadata& child(size_type i)

cdef cppclass table_input_metadata:
Expand Down
15 changes: 9 additions & 6 deletions python/cudf/cudf/_lib/parquet.pyx
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,8 @@ from cudf.utils.dtypes import (
np_to_pa_dtype,
is_categorical_dtype,
is_list_dtype,
is_struct_dtype
is_struct_dtype,
is_decimal_dtype,
)

from cudf._lib.utils cimport get_column_names
Expand Down Expand Up @@ -310,7 +311,7 @@ cpdef write_parquet(

for i, name in enumerate(table._column_names, num_index_cols_meta):
tbl_meta.get().column_metadata[i].set_name(name.encode())
_set_col_children_names(
_set_col_metadata(
table[name]._column, tbl_meta.get().column_metadata[i]
)

Expand Down Expand Up @@ -448,7 +449,7 @@ cdef class ParquetWriter:

for i, name in enumerate(table._column_names, num_index_cols_meta):
self.tbl_meta.get().column_metadata[i].set_name(name.encode())
_set_col_children_names(
_set_col_metadata(
table[name]._column, self.tbl_meta.get().column_metadata[i]
)

Expand Down Expand Up @@ -546,14 +547,16 @@ cdef Column _update_column_struct_field_names(
col.set_base_children(tuple(children))
return col

cdef _set_col_children_names(Column col, column_in_metadata& col_meta):
cdef _set_col_metadata(Column col, column_in_metadata& col_meta):
if is_struct_dtype(col):
for i, (child_col, name) in enumerate(
zip(col.children, list(col.dtype.fields))
):
col_meta.child(i).set_name(name.encode())
_set_col_children_names(child_col, col_meta.child(i))
_set_col_metadata(child_col, col_meta.child(i))
elif is_list_dtype(col):
_set_col_children_names(col.children[1], col_meta.child(1))
_set_col_metadata(col.children[1], col_meta.child(1))
else:
if is_decimal_dtype(col):
col_meta.set_decimal_precision(col.dtype.precision)
return
7 changes: 6 additions & 1 deletion python/cudf/cudf/_lib/utils.pyx
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@ from cudf.utils.dtypes import (
is_categorical_dtype,
is_list_dtype,
is_struct_dtype,
is_decimal_dtype,
)


Expand Down Expand Up @@ -80,7 +81,11 @@ cpdef generate_pandas_metadata(Table table, index):
"'category' column dtypes are currently not "
+ "supported by the gpu accelerated parquet writer"
)
elif is_list_dtype(col) or is_struct_dtype(col):
elif (
is_list_dtype(col)
or is_struct_dtype(col)
or is_decimal_dtype(col)
):
types.append(col.dtype.to_arrow())
else:
types.append(np_to_pa_dtype(col.dtype))
Expand Down
15 changes: 15 additions & 0 deletions python/cudf/cudf/tests/test_parquet.py
Original file line number Diff line number Diff line change
Expand Up @@ -1920,3 +1920,18 @@ def test_parquet_writer_nested(tmpdir, data):

got = pd.read_parquet(fname)
assert_eq(expect, got)


def test_parquet_writer_decimal(tmpdir):
from cudf.core.dtypes import Decimal64Dtype

gdf = cudf.DataFrame({"val": [0.00, 0.01, 0.02]})

gdf["dec_val"] = gdf["val"].astype(Decimal64Dtype(7, 2))

fname = tmpdir.join("test_parquet_writer_decimal.parquet")
gdf.to_parquet(fname)
assert os.path.exists(fname)

got = pd.read_parquet(fname)
assert_eq(gdf, got)