def merge_lines(df, by):
"""Use GEOS line merge to merge MultiLineStrings into LineStrings (where possible).
This uses aggregate_lines first to aggregate lines to MultiLineStrings.
WARNING: this can be a bit slow.
Parameters
----------
df : GeoDataFrame
by : string or list-like
field(s) to aggregate by
Returns
-------
GeoDataFrame of LineStrings or MultiLinestrings (if required)
"""
agg = aggregate_lines(df, by)
agg["geometry"] = pg.line_merge(agg.geometry.values.data)
geom_type = pg.get_type_id(agg["geometry"].values.data)
ix = geom_type == 5
if ix.sum() > 0:
agg.loc[~ix, "geometry"] = pg.multilinestrings(
agg.loc[~ix].geometry.values.data, np.arange((~ix).sum()))
return agg
def test_line_merge_geom_array():
actual = pygeos.line_merge([line_string, multi_line_string])
assert pygeos.equals(actual[0], line_string)
assert pygeos.equals(actual[1], multi_line_string)
def remove_false_nodes(gdf):
"""
Clean topology of existing LineString geometry by removal of nodes of degree 2.
Parameters
----------
gdf : GeoDataFrame, GeoSeries, array of pygeos geometries
(Multi)LineString data of street network
Returns
-------
gdf : GeoDataFrame, GeoSeries
See also
--------
momepy.extend_lines
momepy.close_gaps
"""
if isinstance(gdf, (gpd.GeoDataFrame, gpd.GeoSeries)):
# explode to avoid MultiLineStrings
# double reset index due to the bug in GeoPandas explode
df = gdf.reset_index(drop=True).explode().reset_index(drop=True)
# get underlying pygeos geometry
geom = df.geometry.values.data
else:
geom = gdf
# extract array of coordinates and number per geometry
coords = pygeos.get_coordinates(geom)
indices = pygeos.get_num_coordinates(geom)
# generate a list of start and end coordinates and create point geometries
edges = [0]
i = 0
for ind in indices:
ix = i + ind
edges.append(ix - 1)
edges.append(ix)
i = ix
edges = edges[:-1]
points = pygeos.points(np.unique(coords[edges], axis=0))
# query LineString geometry to identify points intersecting 2 geometries
tree = pygeos.STRtree(geom)
inp, res = tree.query_bulk(points, predicate="intersects")
unique, counts = np.unique(inp, return_counts=True)
merge = res[np.isin(inp, unique[counts == 2])]
if len(merge) > 0:
# filter duplications and create a dictionary with indication of components to
# be merged together
dups = [
item for item, count in collections.Counter(merge).items()
if count > 1
]
split = np.split(merge, len(merge) / 2)
components = {}
for i, a in enumerate(split):
if a[0] in dups or a[1] in dups:
if a[0] in components.keys():
i = components[a[0]]
elif a[1] in components.keys():
i = components[a[1]]
components[a[0]] = i
components[a[1]] = i
# iterate through components and create new geometries
new = []
for c in set(components.values()):
keys = []
for item in components.items():
if item[1] == c:
keys.append(item[0])
new.append(pygeos.line_merge(pygeos.union_all(geom[keys])))
# remove incorrect geometries and append fixed versions
df = df.drop(merge)
final = gpd.GeoSeries(new).explode().reset_index(drop=True)
if isinstance(gdf, gpd.GeoDataFrame):
return df.append(
gpd.GeoDataFrame({df.geometry.name: final},
geometry=df.geometry.name),
ignore_index=True,
)
return df.append(final, ignore_index=True)