Merge pull request #18 from UCL/chore/pandas3-hardening

Updated pandas syntax for pandas3 compatability

Author: o-smith

docs/build/html/.doctrees/environment.pickle

@@ -1,7 +1,7 @@
 ###############################################################################
 # NOTE: This file has been auto-generated by poetry2conda
 #       poetry2conda version = 0.3.0
-#       date: Tue Mar 24 13:04:56 2026
+#       date: Tue Mar 24 15:49:13 2026
 ###############################################################################
 # If you want to change the contents of this file, you should probably change
 # the pyproject.toml file and then use poetry2conda again to update this file.
@@ -12,7 +12,7 @@ dependencies:
   - python>=3.11,<3.14
   - shapely>=2.0,<3.0
   - conda-forge::eppy-core>=0.5.63,<0.6.0
-  - pandas>=2.2,<3.0
+  - pandas>=2.2,<4.0
   - numpy>=1.26,<3.0
   - matplotlib>=3.8,<4.0
   - conda-forge::geopandas>=1.0,<2.0

docs/build/html/.doctrees/simstock.doctree

@@ -14,7 +14,7 @@ packages = [{include = "simstock", from = "src"}]
 python = ">=3.11,<3.14"
 shapely = ">=2.0,<3.0"
 eppy = ">=0.5.63,<0.6.0"
-pandas = ">=2.2,<3.0"
+pandas = ">=2.2,<4.0"
 numpy = ">=1.26,<3.0"
 matplotlib = ">=3.8,<4.0"
 geopandas = ">=1.0,<2.0"

environment.yaml

@@ -928,7 +928,11 @@ def polygon_topology(self, **kwargs) -> None:
         # Create column named ``touching``. Each element
         # is an empty list
         length_range = range(self._df.__len__())
-        self._df['touching'] = pd.Series([] for _ in length_range)
+        self._df["touching"] = pd.Series(
+            [[] for _ in length_range],
+            index=self._df.index,
+            dtype=object,
+        )
 
         # Iterate over all possible pairs of polygons in the data
         for i, j in itertools.combinations(length_range, 2):
@@ -943,8 +947,8 @@ def polygon_topology(self, **kwargs) -> None:
                 # also add i's osbg to j's 
                 # ``touching`` column
                 if algs._is_touching(poly_i, poly_j):
-                    self._df['touching'][i].append(osgb_j)
-                    self._df['touching'][j].append(osgb_i)
+                    self._df.at[i, "touching"].append(osgb_j)
+                    self._df.at[j, "touching"].append(osgb_i)
 
             # If polygon i intersects polygon j, then the
             # _is_touching function will throw a ValueError.
@@ -993,15 +997,19 @@ def _polygon_within_hole(self) -> None:
         # Create column named ``poly_within_hole``. Each element
         # is an empty list
         length_range = range(self._df.__len__())
-        self._df['poly_within_hole'] = pd.Series([] for _ in length_range)
+        self._df["poly_within_hole"] = pd.Series(
+            [[] for _ in length_range],
+            index=self._df.index,
+            dtype=object,
+        )
 
         # Iterate over the cartesian product of polygons
         for i, j in itertools.product(length_range, length_range):
             if i != j and self._df['interiors'][i]:
 
                 # If polygon i has interiors, then check if polygon j
                 # touches polygon i
-                touches = self._df['osgb'][j] in self._df['touching'][i]
+                touches = self._df.at[j, "osgb"] in self._df.at[i, "touching"]
 
                 # Now iterate over the interiors of polygon i
                 for item in self._df['polygon'][i].interiors:
@@ -1010,10 +1018,8 @@ def _polygon_within_hole(self) -> None:
                     # If the interior both touchs and contains j, then we know
                     # j exists within a hole inside i
                     # We make a note of this in the `poly_within_hole` column
-                    if item_poly.contains(self._df['polygon'][j]) and touches:
-                        self._df['poly_within_hole'][i].append(
-                            self._df['osgb'][j]
-                            )
+                    if item_poly.contains(self._df.at[j, "polygon"]) and touches:
+                        self._df.at[i, "poly_within_hole"].append(self._df.at[j, "osgb"])
 
     def _polygon_buffer(self) -> None:
         for i, polygon in enumerate(self._df['polygon']):

poetry.lock

@@ -16,8 +16,7 @@
     LineString,
     MultiPolygon
 )
-from pandas.core.series import Series
-from pandas.core.frame import DataFrame
+from pandas import Series, DataFrame
 from simstock.base import SimstockDataframe
 
 

pyproject.toml

@@ -106,6 +106,56 @@ def test_simstockdf_invalid_df_geoms(self) -> None:
         """
         with self.assertRaises(TypeError):
             SimstockDataframe(self.df_invalid_geoms)
+            
+    def test_touching_lists_are_independent(self) -> None:
+        """ 
+        Regression test to check that the lists in the ``touching`` column are independent 
+        of each other after upgrading to later pandas version syntax.
+        """
+        df = pd.DataFrame({
+            "id": ["a", "b"],
+            "osgb": ["osgb_a", "osgb_b"],
+            "polygon": [
+                "POLYGON ((0 0, 0 1, 1 1, 1 0, 0 0))",
+                "POLYGON ((3 0, 3 1, 4 1, 4 0, 3 0))",
+            ],
+            "shading": [False, False],
+            "height": [5.0, 5.0],
+            "wwr": [0.2, 0.2],
+            "nofloors": [2, 2],
+            "construction": ["x", "x"],
+        })
+        sdf = SimstockDataframe(df)
+        sdf.polygon_topology()
+
+        self.assertIsNot(sdf._df.at[0, "touching"], sdf._df.at[1, "touching"])
+        sdf._df.at[0, "touching"].append("probe")
+        self.assertEqual(sdf._df.at[0, "touching"], ["probe"])
+        self.assertEqual(sdf._df.at[1, "touching"], [])
+
+    def test_osgb_string_dtype_in_topology(self) -> None:
+        """ 
+        Regression test to check that the osgb column is still of string dtype
+        after upgrading to later pandas version syntax in the polygon_topology method.
+        """
+        df = pd.DataFrame({
+            "id": ["a", "b"],
+            "osgb": pd.Series(["osgb_a", "osgb_b"], dtype="string"),
+            "polygon": [
+                "POLYGON ((0 0, 0 1, 1 1, 1 0, 0 0))",
+                "POLYGON ((1 0, 1 1, 2 1, 2 0, 1 0))",
+            ],
+            "shading": [False, False],
+            "height": [5.0, 5.0],
+            "wwr": [0.2, 0.2],
+            "nofloors": [2, 2],
+            "construction": ["x", "x"],
+        })
+        sdf = SimstockDataframe(df)
+        sdf.polygon_topology()
+
+        self.assertIn("osgb_b", sdf._df.at[0, "touching"])
+        self.assertIn("osgb_a", sdf._df.at[1, "touching"])
 
     def test_simstockdf_invalid_df_osgbnames(self) -> None:
         """