[
https://issues.apache.org/jira/browse/ARROW-11388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17272711#comment-17272711
]
Andy Douglas edited comment on ARROW-11388 at 1/27/21, 9:36 AM:
----------------------------------------------------------------
Thank for your response [~jorisvandenbossche], that makes sense and fits with
what I'm seeing.
Basically I have a small python library that wraps pyarrow datasets to provide
a convenient method for accessing multiple datasets and exposing via pandas.
One of the things I want to be able to do is define schemas for all datasets
upfront in something like a YAML file. The schema can then be applied/checked
consistently on write/read avoiding issues like numerical columns being typed
based on contents and therefore sometimes ending up as integers and other times
floats. I initially tried to do this using pyarrow schemas however (as you
mention above) the schema alone is not enough to restore a pandas dataframe
which contains both index and timezone info.
Do you have any suggestions for how I would handle the above? Would you suggest
doing the schema checking within the library and not passing the schema
parameter on pyarrow dataset read/write calls?
Separately, I also see an issue on write for indexed pandas dataframes where
the index column is duplicated in the pandas metadata without the timezone
information being added. I'll raise a separate issue for this.
was (Author: andydoug):
That's for your response [~jorisvandenbossche], that makes sense and fits with
what I'm seeing.
Basically I have a small python library that wraps pyarrow datasets to provide
a convenient method for accessing multiple datasets and exposing via pandas.
One of the things I want to be able to do is define schemas for all datasets
upfront in something like a YAML file. The schema can then be applied/checked
consistently on write/read avoiding issues like numerical columns being typed
based on contents and therefore sometimes ending up as integers and other times
floats. I initially tried to do this using pyarrow schemas however (as you
mention above) the schema alone is not enough to restore a pandas dataframe
which contains both index and timezone info.
Do you have any suggestions for how I would handle the above? Would you suggest
doing the schema checking within the library and not passing the schema
parameter on pyarrow dataset read/write calls?
Separately, I also see an issue on write for indexed pandas dataframes where
the index column is duplicated in the pandas metadata without the timezone
information being added. I'll raise a separate issue for this.
> [Python] Dataset Timezone Handling
> ----------------------------------
>
> Key: ARROW-11388
> URL: https://issues.apache.org/jira/browse/ARROW-11388
> Project: Apache Arrow
> Issue Type: Bug
> Components: Python
> Affects Versions: 2.0.0, 3.0.0
> Reporter: Andy Douglas
> Priority: Minor
>
> I'm trying to write a pandas dataframe with a datetimeindex with timezone
> information to a pyarrow dataset but the timezone information doesn't seem to
> be written (apart from in the pandas metadata)
>
> For example
>
> {code:java}
> import os
> import pandas as pd
> import numpy as np
> import pyarrow as pa
> import pyarrow.parquet as pq
> import pyarrow.dataset as ds
> from pathlib import Path
> # I've tried with both v2.0 and v3.0 today
> print(pa.__version__)
> # create dummy dataframe with datetime index containing tz info
> df = pd.DataFrame(
> dict(
> timestamp=pd.date_range("2021-01-01", freq="1T", periods=100,
> tz="US/Eastern"),
> x=np.arange(100),
> )
> ).set_index("timestamp")
> test_dir = Path("test_dir")
> table = pa.Table.from_pandas(df)
> schema = table.schema
> print(schema)
> print(schema.pandas_metadata)
> # warning - creates dir in cwd
> pq.write_to_dataset(table, test_dir)
> # timestamp column is us and UTC
> print(pq.ParquetFile(test_dir / os.listdir(test_dir)[0]).read())
> # create dataset using schema from earlier
> dataset = ds.dataset(test_dir, format="parquet", schema=schema)
> # doesn't work
> dataset.to_table()
> {code}
>
>
> Is this a bug or am I missing something?
> Thanks
> Andy
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
