diff-nodes.py

#!/usr/bin/python

# Diff two OSM changesets and output a list with
# node id's that are only in the original changeset

import re

def getIDs(f):
  s = set()
  for line in f:
    match = re.search('node id="([0-9]+)"', line)
    if match:
      s.add(match.group(1))
  return s

# construct a set with reverted nodes
with open('reverted-nodes.xml', 'rb') as r:
  revertedNodes = getIDs(r)

# construct a set with the original nodes
with open('original-nodes.xml', 'rb') as o:
  originalNodes = getIDs(o)

# diff the two sets
nodesToRevert = originalNodes - revertedNodes

print '# nodes originally added: ' + str(len(originalNodes))
print '# nodes already reverted: ' + str(len(revertedNodes))
print '# nodes still to be reverted: ' + str(len(nodesToRevert))

with open('changeset.xml', 'w') as f:
  f.write('<osmChange version="0.6" generator="script">\n')
  for node in nodesToRevert:
    f.write('<delete>\n<node id="' + node + '" changeset="35271390" version="1" visible="false" />\n</delete>\n')
  f.write('</osmChange>')