[ARROW-5161] [Python] Cannot convert struct type from Pandas object column

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 0.13.0
• Fix Version/s: 0.15.0
• Component/s: Python
• External issue URL: https://github.com/apache/arrow/issues/21640
• Assignee: Unassigned
• Reporter: Antoine Pitrou
• Created: 11/Apr/19 15:20
• Updated: 11/Jan/23 07:38
• Resolved: 06/Sep/19 19:24

Description

As reported on https://github.com/apache/arrow/issues/4045. Interestingly, the datatype is inferred correctly.

>>> df = pd.DataFrame({'col': [{'ints': 5, 'strs': 'foo'}, {'ints': 6, 'strs': 'bar'}]})                                                                             
>>> df                                                                                                                                                               
                          col
0  {'ints': 5, 'strs': 'foo'}
1  {'ints': 6, 'strs': 'bar'}
>>> pa.Table.from_pandas(df)                                                                                                                                         
Traceback (most recent call last):
  File "<ipython-input-16-6eac5d0eec08>", line 1, in <module>
    pa.Table.from_pandas(df)
  File "pyarrow/table.pxi", line 1139, in pyarrow.lib.Table.from_pandas
    names, arrays, metadata = dataframe_to_arrays(
  File "/home/antoine/arrow/dev/python/pyarrow/pandas_compat.py", line 480, in dataframe_to_arrays
    types)
  File "/home/antoine/arrow/dev/python/pyarrow/pandas_compat.py", line 209, in construct_metadata
    field_name=sanitized_name)
  File "/home/antoine/arrow/dev/python/pyarrow/pandas_compat.py", line 151, in get_column_metadata
    logical_type = get_logical_type(arrow_type)
  File "/home/antoine/arrow/dev/python/pyarrow/pandas_compat.py", line 79, in get_logical_type
    raise NotImplementedError(str(arrow_type))
NotImplementedError: struct<ints: int64, strs: string>