amauryval / osmgt

Python library to extract roads network on Open Street Map (and POIs) in order to create a graph with graph-tools and/or computing shortest paths and isochrones

OsmGT

This package will not be maintain: a new package OsmRx based on rustworkx has been built and is currently developed!

OpenStreetMap (OSM) network analysis based on graph-tools (GT):

• load data from a location name or a bounding box (roads and pois)
• graph creation (and topology processing)
• isochrone builder
• shortest path

Demo

To play with the notebook on binder:

If you do not want to wait (OsmGt docker building takes time here on myBinder), you can find the html version of the jupyter notebook : click here to get the result

Documentation

Open the wiki

Releases

• 0.8.12:

Code cleaning...

• 0.8.4:

First usable version

How to install it

Only on Linux

conda install -c amauryval osmgt

How to run the dockerfile

docker build -t osmgt . && docker run -p 8888:8888 osmgt:latest

osmgt's Issues

reduce tests duration

Use spatial pandas

split input area to get more input data

Use gridding func from geolib

TypeError: GeometryCollection geom type not compatible (l155 - osmgt/geometry/geom_helpers.py in convert_to_polygon(polygon_geom))

Exception returned due to GeometryCollection type.

Workflow used:

from osmgt import OsmGt
from shapely.geometry import Point
import geopandas as gpd

pt = Point(3.930914582537432, 43.612687207093416)
source_nodes = gpd.GeoDataFrame(index=[0], crs='epsg:4326', geometry=[pt])

isochrones_polygon_values = {
    250: "#d9ef8b",
    500: "#fee08b",
    750: "#f46d43"
}

isochrones_lines_values = {
    250: "#668100",
    500: "#e2a803",
    750: "#962603"
}

trip_speed = 3.6  # km/h

location_points = list(source_nodes["geometry"])
# location_points.extend(location_points)

isochrones_polygons, isochrones_lines = OsmGt.isochrone_distances_from_nodes(
    location_points,
    list(isochrones_polygon_values.keys()),
    trip_speed,
    mode="pedestrian"
)

GeometryCollection should be supported ! (extract [multi]polygon(s) parts ?)