sumolib is a set of python modules for working with sumo networks, simulation output and other simulation artifacts. For a detailed list of available functions see the pydoc generated documentation. You can browse the code here.

importing sumolib in a script#

To use the library, the <SUMO_HOME>/tools directory must be on the python load path. This is typically done with a stanza like this:

import os, sys
if 'SUMO_HOME' in os.environ:
    tools = os.path.join(os.environ['SUMO_HOME'], 'tools')
    sys.exit("please declare environment variable 'SUMO_HOME'")

loading a network file#

# import the library
import sumolib
# parse the net
net ='')

The following named arguments may be given to the readNet function (i.e. readNet('', withInternal=True)):

  • withPrograms (bool): import all traffic light programs (default False)
  • withLatestPrograms (bool) : import only the last program for each traffic light. This is the program that would be active in sumo by default. (default False)
  • withConnections (bool) : import all connections (default True)
  • withFoes (bool) : import right-of-way information (default True)
  • withInternal (bool) : import internal edges and lanes (default False)
  • withPedestrianConnections (bool) : import connections between sidewalks, crossings (default False)

usage examples#

import a network and retrieve nodes and edges#

# import the library
import sumolib
# parse the net
net ='')

# retrieve the coordinate of a node based on its ID
print net.getNode('myNodeID').getCoord()

# retrieve the successor node ID of an edge
nextNodeID = net.getEdge('myEdgeID').getToNode().getID()

compute the average edge speed in a plain xml edge file#

speedSum = 0.0
edgeCount = 0
for edge in sumolib.xml.parse('myNet.edg.xml', ['edge']):
    speedSum += float(edge.speed)
    edgeCount += 1
avgSpeed = speedSum / edgeCount


This is just a processing example. To compute average travel speeds in a network, process edgeData, tripinfos or summary-output instead.

compute the length of the selected edges#

cumulLength = 0.
for edge in net.getEdges():
    if edge.isSelected():
        cumulLength += edge.getLength()

compute the median speed using the Statistics module#

edgeStats = sumolib.miscutils.Statistics("edge speeds")
for edge in sumolib.xml.parse('myNet.edg.xml', ['edge']):
avgSpeed = edgeStats.median()


Attribute speed is optional in user-generated .edg.xml files but will always be included if that file was written by netconvert or netedit.

locate nearby edges based on the geo-coordinate#

This requires the module pyproj to be installed. For larger networks rtree is also strongly recommended.

net ='')
radius = 0.1
x, y = net.convertLonLat2XY(lon, lat)
edges = net.getNeighboringEdges(x, y, radius)
# pick the closest edge
if len(edges) > 0:
    distancesAndEdges = sorted([(dist, edge) for edge, dist in edges])
    dist, closestEdge = distancesAndEdges[0]

parse all edges in a route file#

for route in sumolib.xml.parse_fast("myRoutes.rou.xml", 'route', ['edges']):
    edge_ids = route.edges.split()
    # do something with the vector of edge ids

parse vehicles and their route edges in a route file#

for vehicle in sumolib.xml.parse("myRoutes.rou.xml", "vehicle"):
    route = vehicle.route[0] # access the first (and only) child element with name 'route'
    edges = route.edges.split()

with automatic data conversions (including depart time as "HH:MM:SS"):

from sumolib.miscutils import parseTime
for vehicle in sumolib.xml.parse("myRoutes.rou.xml", "vehicle", attr_conversions={
            'depart' : parseTime,
            'edges' : (lambda x : x.split())}):
    edges = vehicle.route[0].edges
    if vehicle.depart > 42:

parse all edges in a edge data (meanData) file#

for interval in sumolib.xml.parse("edgedata.xml", "interval"):
    for edge in interval.edge:    
        # do something with the edge attributes i.e. edge.entered

coordinate transformations#

net ='')

# network coordinates (lower left network corner is at x=0, y=0)
x, y = net.convertLonLat2XY(lon, lat)
lon, lat = net.convertXY2LonLat(x, y)

# raw UTM coordinates
x, y = net.convertLonLat2XY(lon, lat, True)
lon, lat = net.convertXY2LonLat(x, y, True)

# lane/offset coordinates
# from lane position to network coordinates
x,y = sumolib.geomhelper.positionAtShapeOffset(net.getLane(laneID).getShape(), lanePos)
# from network coordinates to lane position
lane, d = net.getNeighboringLanes(x, y, radius)[0] (see "locate nearby edges based on the geo-coordinate" above)
lanePos, dist = sumolib.geomhelper.polygonOffsetAndDistanceToPoint((x,y), lane.getShape())

see also TraCI/Interfacing_TraCI_from_Python#coordinate_transformations

Manipulating and writing xml#

with open("patched.nod.xml", 'w') as outf:
    # setting attrs is optional, it results in a cleaner patch file
    attrs = {'node': ['id', 'x', 'y']}  # other attrs are not needed for patching
    # parse always returns a generator but there is only one root element
    nodes = list(sumolib.xml.parse('plain.nod.xml', 'nodes', attrs))[0]
    for node in nodes.node:
        node.addChild("param", { "key": "origPos", "value" : "%s %s" % (node.x, node.y) } )
        node.x = float(node.x) + random.randint(-20, 20)
        node.y = float(node.y) + random.randint(-20, 20)

Further Examples#

The files in the test subfolders of <SUMO_HOME>/tests/tools/sumolib provide additional examples for sumolib use.