Always write fid index of Node and Edge tables (#995)

We are using `GeoDataFrame.to_file` to write these tables to GeoPackage.
By default GeoPandas only writes the index if it has a name. We don't
want to rely on users naming indices, so this sets the name and ensure
it is written.

I ran into this since I noticed the `model.network.edge.df.index` (by
default 0-based) did not correspond to the 1-based fid column that
GeoPandas writes.


https://geopandas.org/en/stable/docs/reference/api/geopandas.GeoDataFrame.to_file.html

core/test/validation_test.jl

@@ -250,7 +250,7 @@ end
     @test logger.logs[3].kwargs[:control_state] == ""
     @test logger.logs[4].level == Error
     @test logger.logs[4].message == "Cannot connect a basin to a fractional_flow."
-    @test logger.logs[4].kwargs[:edge_id] == 7
+    @test logger.logs[4].kwargs[:edge_id] == 6
     @test logger.logs[4].kwargs[:id_src] == NodeID(2)
     @test logger.logs[4].kwargs[:id_dst] == NodeID(8)
 end
@@ -358,10 +358,10 @@ end
     @test length(logger.logs) == 2
     @test logger.logs[1].level == Error
     @test logger.logs[1].message ==
-          "Invalid edge type 'foo' for edge #1 from node #1 to node #2."
+          "Invalid edge type 'foo' for edge #0 from node #1 to node #2."
     @test logger.logs[2].level == Error
     @test logger.logs[2].message ==
-          "Invalid edge type 'bar' for edge #2 from node #2 to node #3."
+          "Invalid edge type 'bar' for edge #1 from node #2 to node #3."
 end
 
 @testitem "Subgrid validation" begin

python/ribasim/ribasim/input_base.py

@@ -374,8 +374,9 @@ def _write_table(self, path: FilePath) -> None:
 
         gdf = gpd.GeoDataFrame(data=self.df)
         gdf = gdf.set_geometry("geometry")
+        gdf.index.name = "fid"
 
-        gdf.to_file(path, layer=self.tablename(), driver="GPKG")
+        gdf.to_file(path, layer=self.tablename(), driver="GPKG", index=True)
 
     def sort(self):
         self.df.sort_index(inplace=True)

python/ribasim/tests/test_model.py

@@ -136,10 +136,33 @@ def test_plot(discrete_control_of_pid_control):
 
 def test_write_adds_fid_in_tables(basic, tmp_path):
     model_orig = basic
+    # for node an explicit index was provided
+    nrow = len(model_orig.network.node.df)
+    assert model_orig.network.node.df.index.name == "fid"
+    assert model_orig.network.node.df.index.equals(
+        pd.RangeIndex(start=1, stop=nrow + 1)
+    )
+    # for edge no index was provided, but it still needs to write it to file
+    nrow = len(model_orig.network.edge.df)
+    assert model_orig.network.edge.df.index.name is None
+    assert model_orig.network.edge.df.index.equals(pd.RangeIndex(start=0, stop=nrow))
+
     model_orig.write(tmp_path / "basic/ribasim.toml")
     with connect(tmp_path / "basic/database.gpkg") as connection:
         query = f"select * from {esc_id('Basin / profile')}"
-        df = pd.read_sql_query(query, connection, parse_dates=["time"])
+        df = pd.read_sql_query(query, connection)
         assert "fid" in df.columns
-        fids = df.get("fid")
+        fids = df["fid"]
         assert fids.equals(pd.Series(range(1, len(fids) + 1)))
+
+        query = "select fid from Node"
+        df = pd.read_sql_query(query, connection)
+        assert "fid" in df.columns
+        fids = df["fid"]
+        assert fids.equals(pd.Series(range(1, len(fids) + 1)))
+
+        query = "select fid from Edge"
+        df = pd.read_sql_query(query, connection)
+        assert "fid" in df.columns
+        fids = df["fid"]
+        assert fids.equals(pd.Series(range(0, len(fids))))