Merge pull request #54 from tomalrussell/feature/improve_split_multil…

…inestrings

Improve split_multilinestrings

src/snkit/network.py

@@ -186,50 +186,24 @@ def _set_precision(geom):
 
 
 def split_multilinestrings(network):
-    """Create multiple edges from any MultiLineString edge
-
-    Ensures that edge geometries are all LineStrings, duplicates attributes over any
-    created multi-edges.
     """
-    simple_edge_attrs = []
-    simple_edge_geoms = []
-    edges = network.edges
-
-    # # Commented out idea about dealing with multi-part MultiLineStrings separately
-    # is_multi_single_part = edges.geometry.apply(lambda geom: geom.geom_type == 'MultiLineString' and len(geom) == 1)
-    # is_multi_multi_part = edges.geometry.apply(lambda geom: geom.geom_type == 'MultiLineString' and len(geom) > 1)
-    # is_not_multi = edges.geometry.apply(lambda geom: geom.geom_type != 'MultiLineString')
-
-    # not_multi_edges = edges[is_not_multi].copy()
-    # multi_single_part_edges = edges[is_multi_single_part].copy()
-    # multi_single_part_edges.geometry = multi_single_part_edges.geometry.apply(lambda geom: next(geom))
-    # edges = edges[is_multi_multi_part].copy()
-
-    for edge in tqdm(
-        edges.itertuples(index=False), desc="split_multi", total=len(edges)
-    ):
-        if edge.geometry.geom_type == "MultiLineString":
-            edge_parts = list(edge.geometry)
-        else:
-            edge_parts = [edge.geometry]
-
-        for part in edge_parts:
-            simple_edge_geoms.append(part)
+    Create multiple edges from any MultiLineString edge
 
-        attrs = GeoDataFrame([edge] * len(edge_parts))
-        simple_edge_attrs.append(attrs)
+    Ensures that edge geometries are all LineStrings, duplicates attributes
+    over any created multi-edges.
+    """
 
-    simple_edge_geoms = GeoDataFrame(simple_edge_geoms, columns=["geometry"])
-    edges = (
-        pandas.concat(simple_edge_attrs, axis=0)
-        .reset_index(drop=True)
-        .drop("geometry", axis=1)
-    )
-    edges = pandas.concat([edges, simple_edge_geoms], axis=1)
+    edges = network.edges
+    geom_col: str = geometry_column_name(edges)
+    split_edges = edges.explode(column=geom_col, ignore_index=True)
 
-    # edges = pandas.concat([edges, multi_single_part_edges, not_multi_edges], axis=0).reset_index(drop=True)
+    geo_types = set(split_edges.geom_type)
+    if geo_types != {'LineString'}:
+        raise ValueError(
+            f"exploded edges are of type(s) {geo_types} but should only be LineString"
+        )
 
-    return Network(nodes=network.nodes, edges=edges)
+    return Network(nodes=network.nodes, edges=split_edges)
 
 
 def merge_multilinestring(geom):

tests/test_init.py

@@ -9,7 +9,7 @@
 
 from pandas.testing import assert_frame_equal
 from pytest import fixture, mark
-from shapely.geometry import Point, LineString, MultiPoint
+from shapely.geometry import Point, LineString, MultiPoint, MultiLineString
 
 try:
     import networkx as nx
@@ -229,6 +229,52 @@ def test_split_at_nodes(unsplit, split):
     assert_frame_equal(split.edges, actual.edges)
 
 
+def test_split_multilinestrings():
+    """Explode multilinestrings into linestrings"""
+
+    # point coordinates comprising three linestrings
+    mls_coords = [
+        (
+            (0, 0),
+            (0, 1),
+        ),
+        (
+            (1, 1),
+            (2, 2),
+            (2, 1),
+        ),
+        (
+            (0, 1),
+            (-1, 1),
+            (-1, 2),
+            (-1, 0),
+        ),
+    ]
+    # point coordsinate comprising a single linestring
+    ls_coords = [(4, 0), (4, 1), (4, 2)]
+
+    # make input network edges
+    edges = GeoDataFrame(
+        {
+            "data": ["MLS", "LS"],
+            "geometry": [MultiLineString(mls_coords), LineString(ls_coords)],
+        }
+    )
+    multi_network = snkit.network.Network(edges=edges)
+    # check we have two rows (multilinestring and linestring) in the input network
+    assert len(multi_network.edges) == 2
+
+    # split and check we have four rows (the resulting linestrings) in the output network
+    split_network = snkit.network.split_multilinestrings(multi_network)
+    assert len(split_network.edges) == 4
+
+    # check data is replicated from multilinestring to child linestrings
+    assert list(split_network.edges["data"].values) == ["MLS"] * 3 + ["LS"]
+
+    # check everything is reindexed
+    assert list(split_network.edges.index.values) == list(range(4))
+
+
 def test_split_line():
     """Definitively split line at a point"""
     line = LineString(