performance(to_geodataframe()): update GeoDataFrame construction routine (#2691)

Author: jlarsen-usgs

flopy/discretization/grid.py

@@ -601,12 +601,37 @@ def to_geodataframe(self, features, featuretype="Polygon"):
         -------
             GeoDataFrame
         """
-        from ..utils.geospatial_utils import GeoSpatialCollection
+        from ..utils.utl_import import import_optional_dependency
+
+        gpd = import_optional_dependency("geopandas")
+        shp_geom = import_optional_dependency("shapely.geometry")
+
+        if featuretype.lower() == "polygon":
+            cache_index = "grid_polygons"
+            if (
+                cache_index not in self._cache_dict
+                or self._cache_dict[cache_index].out_of_date
+            ):
+                geoms = [shp_geom.Polygon(i[0]) for i in features]
+                self._cache_dict[cache_index] = CachedData(geoms)
+            else:
+                geoms = self._cache_dict[cache_index].data_nocopy
+            gdf = gpd.GeoDataFrame(data=None, geometry=geoms)
+
+        elif featuretype.lower() == "linestring":
+            cache_index = "grid_linestrings"
+            if (
+                cache_index not in self._cache_dict
+                or self._cache_dict[cache_index].out_of_date
+            ):
+                geoms = [shp_geom.LineString(i) for i in features]
+                self._cache_dict[cache_index] = CachedData(geoms)
+            else:
+                geoms = self._cache_dict[cache_index].data_nocopy
+            gdf = gpd.GeoDataFrame(data=None, geometry=geoms)
+        else:
+            raise NotImplementedError(f"{featuretype} is not currently supported")
 
-        gc = GeoSpatialCollection(
-            features, shapetype=[featuretype for _ in range(len(features))]
-        )
-        gdf = gc.geodataframe
         gdf["node"] = gdf.index + 1
         if self.crs is not None:
             gdf = gdf.set_crs(crs=self.crs)

flopy/discretization/structuredgrid.py

@@ -767,7 +767,16 @@ def to_geodataframe(self):
         -------
             GeoDataFrame
         """
-        polys = [[list(zip(*i))] for i in zip(*self.cross_section_vertices)]
+        cache_index = "gdf_polys"
+        if (
+            cache_index not in self._cache_dict
+            or self._cache_dict[cache_index].out_of_date
+        ):
+            polys = [[list(zip(*i))] for i in zip(*self.cross_section_vertices)]
+            self._cache_dict[cache_index] = CachedData(polys)
+        else:
+            polys = self._cache_dict[cache_index].data_nocopy
+
         gdf = super().to_geodataframe(polys)
         gdf["row"] = sorted(list(range(1, self.nrow + 1)) * self.ncol)
         gdf["col"] = list(range(1, self.ncol + 1)) * self.nrow

flopy/discretization/unstructuredgrid.py

@@ -599,7 +599,16 @@ def to_geodataframe(self):
         -------
             GeoDataFrame
         """
-        polys = [[self.get_cell_vertices(nn)] for nn in range(self.nnodes)]
+        cache_index = "gdf_polys"
+        if (
+            cache_index not in self._cache_dict
+            or self._cache_dict[cache_index].out_of_date
+        ):
+            polys = [[self.get_cell_vertices(nn)] for nn in range(self.nnodes)]
+            self._cache_dict[cache_index] = CachedData(polys)
+        else:
+            polys = self._cache_dict[cache_index].data_nocopy
+
         gdf = super().to_geodataframe(polys)
         if self.nlay > 1:
             lays = []

flopy/discretization/vertexgrid.py

@@ -308,11 +308,20 @@ def to_geodataframe(self):
         -------
             GeoDataFrame
         """
-        cells = [[self.get_cell_vertices(nn)] for nn in range(self.ncpl)]
+        cache_index = "gdf_polys"
+        if (
+            cache_index not in self._cache_dict
+            or self._cache_dict[cache_index].out_of_date
+        ):
+            polys = [[self.get_cell_vertices(nn)] for nn in range(self.ncpl)]
+            self._cache_dict[cache_index] = CachedData(polys)
+        else:
+            polys = self._cache_dict[cache_index].data_nocopy
+
         featuretype = "Polygon"
         if self._cell1d is not None:
             featuretype = "multilinestring"
-        gdf = super().to_geodataframe(cells, featuretype)
+        gdf = super().to_geodataframe(polys, featuretype)
         if self.idomain is not None:
             active = np.sum(
                 self.idomain.reshape(