networkx-osm import open street map data as a networkx graph

gistfile1.py

"""
Read graphs in Open Street Maps osm format

Based on osm.py from brianw's osmgeocode
http://github.com/brianw/osmgeocode, which is based on osm.py from
comes from Graphserver:
http://github.com/bmander/graphserver/tree/master and is copyright (c)
2007, Brandon Martin-Anderson under the BSD License
"""


import xml.sax
import copy
import networkx

def download_osm(left,bottom,right,top):
    """ Return a filehandle to the downloaded data."""
    from urllib import urlopen
    fp = urlopen( "http://api.openstreetmap.org/api/0.6/map?bbox=%f,%f,%f,%f"%(left,bottom,right,top) )
    return fp

def read_osm(filename_or_stream, only_roads=True):
    """Read graph in OSM format from file specified by name or by stream object.

    Parameters
    ----------
    filename_or_stream : filename or stream object

    Returns
    -------
    G : Graph

    Examples
    --------
    >>> G=nx.read_osm(nx.download_osm(-122.33,47.60,-122.31,47.61))
    >>> plot([G.node[n]['data'].lat for n in G], [G.node[n]['data'].lon for n in G], ',')

    """
    osm = OSM(filename_or_stream)
    G = networkx.Graph()

    for w in osm.ways.itervalues():
        if only_roads and 'highway' not in w.tags:
            continue
        G.add_path(w.nds, id=w.id, data=w)
    for n_id in G.nodes_iter():
        n = osm.nodes[n_id]
        G.node[n_id] = dict(data=n)
    return G
        
    
class Node:
    def __init__(self, id, lon, lat):
        self.id = id
        self.lon = lon
        self.lat = lat
        self.tags = {}
        
class Way:
    def __init__(self, id, osm):
        self.osm = osm
        self.id = id
        self.nds = []
        self.tags = {}
        
    def split(self, dividers):
        # slice the node-array using this nifty recursive function
        def slice_array(ar, dividers):
            for i in range(1,len(ar)-1):
                if dividers[ar[i]]>1:
                    #print "slice at %s"%ar[i]
                    left = ar[:i+1]
                    right = ar[i:]
                    
                    rightsliced = slice_array(right, dividers)
                    
                    return [left]+rightsliced
            return [ar]
            
        slices = slice_array(self.nds, dividers)
        
        # create a way object for each node-array slice
        ret = []
        i=0
        for slice in slices:
            littleway = copy.copy( self )
            littleway.id += "-%d"%i
            littleway.nds = slice
            ret.append( littleway )
            i += 1
            
        return ret
        
        

class OSM:
    def __init__(self, filename_or_stream):
        """ File can be either a filename or stream/file object."""
        nodes = {}
        ways = {}
        
        superself = self
        
        class OSMHandler(xml.sax.ContentHandler):
            @classmethod
            def setDocumentLocator(self,loc):
                pass
            
            @classmethod
            def startDocument(self):
                pass
                
            @classmethod
            def endDocument(self):
                pass
                
            @classmethod
            def startElement(self, name, attrs):
                if name=='node':
                    self.currElem = Node(attrs['id'], float(attrs['lon']), float(attrs['lat']))
                elif name=='way':
                    self.currElem = Way(attrs['id'], superself)
                elif name=='tag':
                    self.currElem.tags[attrs['k']] = attrs['v']
                elif name=='nd':
                    self.currElem.nds.append( attrs['ref'] )
                
            @classmethod
            def endElement(self,name):
                if name=='node':
                    nodes[self.currElem.id] = self.currElem
                elif name=='way':
                    ways[self.currElem.id] = self.currElem
                
            @classmethod
            def characters(self, chars):
                pass

        xml.sax.parse(filename_or_stream, OSMHandler)
        
        self.nodes = nodes
        self.ways = ways
            
        #count times each node is used
        node_histogram = dict.fromkeys( self.nodes.keys(), 0 )
        for way in self.ways.values():
            if len(way.nds) < 2:       #if a way has only one node, delete it out of the osm collection
                del self.ways[way.id]
            else:
                for node in way.nds:
                    node_histogram[node] += 1
        
        #use that histogram to split all ways, replacing the member set of ways
        new_ways = {}
        for id, way in self.ways.iteritems():
            split_ways = way.split(node_histogram)
            for split_way in split_ways:
                new_ways[split_way.id] = split_way
        self.ways = new_ways

The api for OSM has been updated since I used this last, 0.5 has been advanced to 0.6

Fortunately, the way to get map data has not changed: http://wiki.openstreetmap.org/wiki/API_changes_between_v0.5_and_v0.6

So making this work again was as simple as updating the url. It should work for you now. Please let the NetworkX developers know that you would like this included in their package: https://networkx.lanl.gov/trac/ticket/309

Hi I revised your very helpful code above in order to also write each node's lat/lon coordinates into the resulting graph, see https://gist.github.com/rajanski/ccf65d4f5106c2cdc70e

The question is how can i further expand your script to also write the tags of the ways as an attribute to the edges of the graph?

I'll answer myself:

https://gist.github.com/rajanski/ccf65d4f5106c2cdc70e#file-gistfile1-py-L46

Hi as asked here https://help.openstreetmap.org/questions/19213/how-can-i-convert-an-osm-xml-file-into-a-graph-representation as well, I wonder why we need to split all ways (section betweeen lines 144 and 159) ?
If i outcomment that section and run the script I get a fine graph anyway...

See edited version enabling directionality https://gist.github.com/rajanski/ccf65d4f5106c2cdc70e

Hi !
Thanks for your code. I used it on a scholar project.

Here is the list of things I added

• Python3.6 compatibility
• Cache for avoiding to download again the same osm tiles
• Distance computation to estimate length of each ways (useful to compute the shortest path)
• Directional Graph (to consider oneway road on the shortest path computation)

My edited version : https://gist.github.com/Tofull/49fbb9f3661e376d2fe08c2e9d64320e