GiTaDi-CrEaTe opened a new issue, #50312:
URL: https://github.com/apache/arrow/issues/50312
### Describe the bug, including details regarding any error messages,
version, and platform.
### Description of the bug
When writing `uuid.UUID` objects to Parquet using PyArrow, the data is
correctly stored as a `FIXED_LEN_BYTE_ARRAY`. However, when reading this data
back on **Python 3.14 / Nightly builds**, PyArrow fails to cast the 16 bytes
back into Python `uuid.UUID` objects, instead returning raw `bytes`.
This works perfectly on Python 3.13 and below, but introduces a regression
on other builds.
**To Reproduce:**
```python
import pyarrow as pa
import pyarrow.parquet as pq
import pandas as pd
import uuid
# Create a simple table with a UUID
original_uuid = uuid.uuid4()
df = pd.DataFrame({"id": [original_uuid]})
table = pa.Table.from_pandas(df)
# Write to parquet and read back
pq.write_table(table, "test_uuid.parquet")
read_table = pq.read_table("test_uuid.parquet")
result_df = read_table.to_pandas()
# On Python 3.13, this is a uuid.UUID object.
# On Python 3.14 / Nightly, this is raw bytes.
print(type(result_df.loc[0, "id"]))
---
This was discovered while adding upstream UUID Parquet tests to the pandas
test suite (pandas-dev/pandas#65647).
### Component(s)
Python, Parquet
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
