Issue #714 CsvJobDatabase: workaround GeoPandas error with column named "crs"

Author: soxofaan

CHANGELOG.md

@@ -21,6 +21,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- `CsvJobDatabase`: workaround GeoPandas issue when there is a column named "crs" ([#714](https://github.com/Open-EO/openeo-python-client/issues/714))
+
 
 ## [0.37.0] - 2025-01-21 - "SAP10" release
 

openeo/extra/job_management/__init__.py

@@ -904,7 +904,8 @@ def read(self) -> pd.DataFrame:
             import geopandas
 
             # `df.to_csv()` in `persist()` has encoded geometries as WKT, so we decode that here.
-            df = geopandas.GeoDataFrame(df, geometry=geopandas.GeoSeries.from_wkt(df["geometry"]))
+            df.geometry = geopandas.GeoSeries.from_wkt(df["geometry"])
+            df = geopandas.GeoDataFrame(df)
         return df
 
     def persist(self, df: pd.DataFrame):

setup.py

@@ -25,7 +25,7 @@
     "urllib3<2.3.0",  # httpretty doesn't work properly with urllib3>=2.3.0. See #700 and https://github.com/gabrielfalcao/HTTPretty/issues/484
     "netCDF4>=1.7.0",
     "matplotlib",  # TODO: eliminate matplotlib as test dependency
-    "geopandas",
+    "geopandas>0.13.2",
     "flake8>=5.0.0",
     "time_machine",
     "pyproj>=3.2.0",  # Pyproj is an optional, best-effort runtime dependency

tests/extra/job_management/test_job_management.py

@@ -977,6 +977,26 @@ def test_initialize_from_df_on_exists_skip(self, tmp_path):
         )
         assert set(db.read()["some_number"]) == {1, 2, 3}
 
+    def test_read_with_crs_column(self, tmp_path):
+        """
+        Having a column named "crs" can cause obscure error messages when creating a GeoPandas dataframe
+        https://github.com/Open-EO/openeo-python-client/issues/714
+        """
+        source_df = pd.DataFrame(
+            {
+                "crs": [1234],
+                "geometry": ["Point(2 3)"],
+            }
+        )
+        path = tmp_path / "jobs.csv"
+        source_df.to_csv(path, index=False)
+        result_df = CsvJobDatabase(path).read()
+        assert isinstance(result_df, geopandas.GeoDataFrame)
+        assert result_df.to_dict(orient="list") == {
+            "crs": [1234],
+            "geometry": [shapely.geometry.Point(2, 3)],
+        }
+
 
 class TestParquetJobDatabase: