BUG: Failures with ArrowDtype(pa.string_view())

a10y commented 2 weeks ago

Pandas version checks

[X] I have checked that this issue has not already been reported.
[X] I have confirmed this bug exists on the latest version of pandas.
[ ] I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd # pandas 2.2.3 => latest
import pyarrow as pa # pyarrow 17.0.0 => latest

tab = pa.table({"names": pa.array(["a", "b", "c"], type=pa.string_view())})

# this succeeds
tab.to_pandas()

# this throws an exception
tab.to_pandas(types_mapper=pd.ArrowDtype)

Issue Description

Our library is producing Arrow binary view arrays (strings and binary) and we want to allow users to convert into Pandas DataFrame.

We are using the pd.ArrowDtype constructor to allow creating Pandas arrays that are backed with Arrow storage. The example I've attached fails also when you change "c" to None (i.e. problem hits for both nullable and non-nullable types).

Full repro with error message:

>>> import pandas as pd # pandas 2.2.3 => latest
>>> import pyarrow as pa # pyarrow 17.0.0 => latest
>>>
>>> tab = pa.table({"names": pa.array(["a", "b", "c"], type=pa.string_view())})
>>>
>>> # this succeeds
>>> tab.to_pandas()
  names
0     a
1     b
2     c
>>>
>>> # this throws an exception
>>> tab.to_pandas(types_mapper=pd.ArrowDtype)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/core/frame.py", line 1214, in __repr__
    return self.to_string(**repr_params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/util/_decorators.py", line 333, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/core/frame.py", line 1394, in to_string
    return fmt.DataFrameRenderer(formatter).to_string(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/format.py", line 962, in to_string
    string = string_formatter.to_string()
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/string.py", line 29, in to_string
    text = self._get_string_representation()
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/string.py", line 44, in _get_string_representation
    strcols = self._get_strcols()
              ^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/string.py", line 35, in _get_strcols
    strcols = self.fmt.get_strcols()
              ^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/format.py", line 476, in get_strcols
    strcols = self._get_strcols_without_index()
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/format.py", line 729, in _get_strcols_without_index
    str_columns = self._get_formatted_column_labels(self.tr_frame)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/io/formats/format.py", line 809, in _get_formatted_column_labels
    need_leadsp = dict(zip(fmt_columns, map(is_numeric_dtype, dtypes)))
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/core/dtypes/common.py", line 1119, in is_numeric_dtype
    return _is_dtype_type(
           ^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/core/dtypes/common.py", line 1468, in _is_dtype_type
    tipo = pandas_dtype(arr_or_dtype).type
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Volumes/Code/vortex/.venv/lib/python3.11/site-packages/pandas/core/dtypes/dtypes.py", line 2169, in type
    raise NotImplementedError(pa_type)
NotImplementedError: string_view
>>>

This seems like a distinct issue from #59883.

Expected Behavior

I'd expect the throwing example to not throw.

Installed Versions

**Device:** macOS 14.4.1 M2 Max MBP ``` >>> pd.show_versions() INSTALLED VERSIONS ------------------ commit : 0691c5cf90477d3503834d983f69350f250a6ff7 python : 3.11.9 python-bits : 64 OS : Darwin OS-release : 23.4.0 Version : Darwin Kernel Version 23.4.0: Fri Mar 15 00:12:49 PDT 2024; root:xnu-10063.101.17~1/RELEASE_ARM64_T6020 machine : arm64 processor : arm byteorder : little LC_ALL : en_US.UTF-8 LANG : None LOCALE : en_US.UTF-8 pandas : 2.2.3 numpy : 1.26.4 pytz : 2024.1 dateutil : 2.9.0.post0 pip : 24.0 Cython : None sphinx : 8.0.2 IPython : 8.26.0 adbc-driver-postgresql: None adbc-driver-sqlite : None bs4 : 4.12.3 blosc : None bottleneck : None dataframe-api-compat : None fastparquet : None fsspec : None html5lib : None hypothesis : None gcsfs : None jinja2 : 3.1.3 lxml.etree : None matplotlib : 3.9.2 numba : None numexpr : None odfpy : None openpyxl : None pandas_gbq : None psycopg2 : None pymysql : None pyarrow : 17.0.0 pyreadstat : None pytest : 8.1.1 python-calamine : None pyxlsb : None s3fs : None scipy : None sqlalchemy : None tables : None tabulate : None xarray : 2024.7.0 xlrd : None xlsxwriter : None zstandard : None tzdata : 2024.1 qtpy : None pyqt5 : None ```

a10y commented 2 weeks ago

The failure is occurring in the type property accessor on ArrowDtype. I'm not familiar enough with Pandas internals to be sure, but my suspicion is that this if statement should include a check for is_string_view

https://github.com/pandas-dev/pandas/blob/2a10e04a099d5f1633abcdfbb2dd9fdf09142f8d/pandas/core/dtypes/dtypes.py#L2216-L2217

a10y commented 2 weeks ago

I don't have permissions to add but this should include the Arrow label

rhshadrach commented 5 days ago

The proposed fix resolves the OP and doesn't break any tests for me locally. I'm not finding anything on string_view in the code or docs, do we support string views @WillAyd / @jorisvandenbossche?

WillAyd commented 4 days ago

I am not sure if the ArrowDtype has ever fully been scoped out, but as far as I am aware we should allow any Arrow data type to be stored within that container

jorisvandenbossche commented 4 days ago

ArrowDtype is AFAIK indeed quite agnostic and supporting any pyarrow data type to put into it. But then further operations on it rely on pyarrow.compute functions, and not many of those are actually implemented for the newer string_view data type on the pyarrow side.

I don't know if we should warn users about that when the construct a dataframe with string_view.. Or maybe we should actually also consider still by default convert string_view to string, given those usability issues (the question then is mostly how to let the user actually ask for allowing string_view explicitly, if by default we would still convert)

WillAyd commented 3 days ago

Or maybe we should actually also consider still by default convert string_view to string

I'd be hesitant to do this without clarifying how we expect logical types to behave. I think that would also be the exact opposite of what polars does (i.e. they convert string to string_view) so that would lead to some fragmentation in expectations

pandas-dev / pandas