veritag: rough and ready OSM (node) data verification tool

sample-output.csv

veritag.py

#!/usr/bin/python
# veritag.py - rough and ready OSM (node) data verification tool.
# 2012-05-10 Andrew Chadwick a.t.chadwick at gmail.
# https://creativecommons.org/publicdomain/zero/1.0/
# 
# Generates a random sampling of OSM data (nodes only right now), then fires up
# Bing Maps showing aerial imagery in a running Firefox session and prompts for
# classification of each of the things seen. Should work well with the windows
# side by side.  For the classifications, make up your own strings, or use
# numbers.
# 
# Input can be overpass API dumps (for example).
# 
# Output is a CSV file.

from xml.dom import pulldom
import random

sample_size = 100
bing_uri_pattern = 'http://www.bing.com/maps/?v=2&cp=%0.5f~%0.5f&lvl=20&sty=a&eo=0'


# TODO: expand to ways and closed ways. I guess for a closed way you'd want to
# show the centroid of the convex hull, or the centre of the bbox or something.


def load_osm_file(filename, nodes):
    events = pulldom.parse(filename)
    for event, domnode in events:
        if event == "START_ELEMENT" and domnode.tagName == "node":
            # events.expandNode(node)  # as necessary
            lat = domnode.getAttribute("lat")
            lon = domnode.getAttribute("lon")
            id = domnode.getAttribute("id")
            nodes.append((id, lat, lon))


def browse_uri(uri):
    os.system("firefox '%s'" % (uri,))



if __name__ == '__main__':
    import os, sys
    nodes = []

    usage = "%s OUTPUT.csv INPUT1.osm [INPUT2.osm ...]"
    script = sys.argv[0]
    if len(sys.argv) <= 2:
        print usage % (script,)
        sys.exit(1)
    outfile = sys.argv[1]
    infiles = sys.argv[2:]

    print "Will write classifications to %s" % (outfile,)
    for f in infiles:
        print "Loading %s..." % (f,)
        load_osm_file(f, nodes)
    print "Loaded all files."
    print "Writing classifications to %s" % (outfile,)
    out_fp = open(outfile, "w")
    s_num = 1
    for id, lat, lon in random.sample(nodes, sample_size):
        id = int(id)
        lat, lon = float(lat), float(lon)
        uri = bing_uri_pattern % (float(lat), float(lon))
        browse_uri(uri)
        cls = raw_input("Classification for id=%s (%d/%d)? " \
          % (id, s_num, sample_size))
        print >>out_fp, "%d,%0.4f,%0.4f,\"%s\"" % (id, lat, lon, cls)
        out_fp.flush()
        s_num += 1
    out_fp.close()
    print "%s written." % (outfile,)