def test_read_issue_375_snappy(self):
assert_arrow_table_equals(
parquet.read(self._testPath("fastparquet-issue-375-snappy.par")),
{
"A": ["A" * 32760] * 10,
"__index_level_0__": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
},
)
def _test_read_write_table(self, table, expected=None):
table = arrow_table(table).table
if expected is None:
expected = table
else:
expected = arrow_table(expected).table
parquet.write(self.temp_path, table)
result = parquet.read(self.temp_path)
assert_arrow_table_equals(result, table)
def test_read_issue_375_uncompressed(self):
# https://github.com/dask/fastparquet/issues/375
# large dictionary written by pyarrow.parquet.
assert_arrow_table_equals(
parquet.read(self._testPath("fastparquet-issue-375.par")),
{
"A": ["A" * 32755] * 10,
"__index_level_0__": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
},
)
def test_read_issue_361(self):
# https://github.com/dask/fastparquet/issues/361
# IndexError loading zero-column dataframe
assert_arrow_table_equals(
parquet.read(self._testPath("fastparquet-issue-361.par")), {})