[GH-2013] Geopandas.Series: Implement crs property and set_crs

[petern48]

Did you read the Contributor Guide?

• Yes, I have read the Contributor Rules and Contributor Development Guide

Is this PR related to a ticket?

• Yes, and the PR name follows the format [GH-XXX] my subject.
• Closes Geopandas.GeoSeries: Implement crs property and set_crs #2013

What changes were proposed in this PR?

Implement crs property and set_crs()

How was this patch tested?

Added new tests

Did this PR include necessary documentation updates?

• Yes, I have updated the documentation.

[petern48]

inplace is not yet working. I tried fixing it a few ways (e.g modifying the internal frame or anchor), but it didn't work. I prefer to just move on for now since it's not the most important feature. Implementing inplace for all functions should be the same.

EDIT: nvm i got it

[petern48]

Ready for review @zhangfengcdt. I'll also start properly doing draft PRs in the future and open for review when they're ready.

[zhangfengcdt]

looks good, just one inline comment

[Copilot]

Pull Request Overview

This PR implements a new crs property and a set_crs method for GeoSeries, and adds corresponding tests.

• Updates tests in both test_match_geopandas_series.py and test_geoseries.py to validate CRS handling.
• Implements CRS assignment, error handling, and internal processing changes in geoseries.py.

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
python/tests/geopandas/test_match_geopandas_series.py	Added tests comparing crs behavior against geopandas.GeoSeries
python/tests/geopandas/test_geoseries.py	Introduced tests for set_crs functionality including error cases
python/sedona/geopandas/geoseries.py	Implemented crs property, its setter, and set_crs with improved SQL alias and error handling

Comments suppressed due to low confidence (1)

python/sedona/geopandas/geoseries.py:378

• Ensure that BinaryType is imported from pyspark.sql.types to make the code self-contained and clearer.

            if isinstance(data_type, BinaryType):

[jiayuasu]

Note that to_pandas() function will collect the distributed Sedona DF to the master node. This will likely cause OOM on the master. The best way to do this is via Spark's take() which only takes the first value of the distributed DF without collecting the entire DF.

In addition, this function assumes all records in the DF has the same CRS/SRID which is not true for Sedona in many cases. It is worth mentioning this in the docs.

[zhangfengcdt]

You can also do tmp_df._to_spark_pandas_df() that returns the pspd.DataFrame (pyspark).

[petern48]

Thanks. Yeah I remember realizing that, but I forgot to circle back to figure out how to remove the to_pandas.

Also added the doc change. Slight nit about your wording: all CRS have to be the same in the Series, not the DF. Different series are still allowed to have different CRS in geopandas.

[zhangfengcdt]

I think we should overwrite the two functions to_pandas() and to_geopandas in GeoSeries to disable end users from performing these operations directly. We can still keep internal versions of them _to_pandas and _to_geopandas if needed by the implementation itself.

In PySpark, it keeps the to_pandas() that warn users that

        log_advice(
            "`to_pandas` loads all data into the driver's memory. "
            "It should only be used if the resulting pandas Series is expected to be small."
        )

And we should at least do this in to_geopandas() as well if we don't like the idea of disabling it.

[petern48]

I personally prefer keeping the public method, so I added the warning message to to_geopandas().

[jiayuasu]

@petern48 can you import pyproj only when it is needed?

[petern48]

Yep, just did. Didn't realize I had a top level import