This code loads .osm file and allows to call function on all OSM objects in dataset.

Installation

pip install osm-iterator

Likely pip3 install osm-iterator if pip points to Python2 pip.

It is distributed as an osm_iterator PyPI package.

Usage example

Download data and show it

This usage example includes downloading data using requests library, that you may need to install (also available via pip).

from osm_iterator import osm_iterator
import requests
import os.path

def download_from_overpass(query, output_filepath):
  print(query)
  url = "http://overpass-api.de/api/interpreter"
  r = requests.get(url, params={'data': query})
  result = r.text
  with open(output_filepath, 'w') as file:
      file.write(str(result))

def show_places(element):
    place_tag = element.get_tag_value("place")
    name_tag = element.get_tag_value("name")
    osm_object_url = element.get_link()
    if place_tag != None:
        print(name_tag, "(", place_tag, ") is ", osm_object_url)

filepath = "places_in_Kraków.osm"
query = """
[out:xml][timeout:2500];
area[name='Kraków']->.searchArea;
(
  node["place"](area.searchArea);
  way["place"](area.searchArea);
  relation["place"](area.searchArea);
);
out center;
"""

if os.path.isfile(filepath) == False:
    download_from_overpass(query, filepath)
osm = osm_iterator.Data(filepath)
osm.iterate_over_data(show_places)

Load data only

from osm_iterator import osm_iterator

global osm_object_store
osm_object_store = []

def record_objects(element):
    global osm_object_store
    print(element.element.tag, element.element.attrib['id'])
    osm_object_store.append({"type": element.get_type(), "id": element.get_id()})

filepath = "output.osm"
osm = osm_iterator.Data(filepath)
osm.iterate_over_data(record_objects)
for entry in osm_object_store:
    print(entry)

Running tests

nosetests3 or python3 -m unittest or python3 tests.py

History

Design explanation: this code has deeply suboptimal handling of pretty much everything. For start, all data is loaded into memory and then duplicated in-memory dataset is created.

As result, attempt to process any large datasets will cause issues due to excessive memory consumption.

This situation is consequence of following facts

• This code was written during my first attempt to process OSM data using Python
• API allows (at least in theory) to painlessly switch to real iterator that is not loading all data into memory at once
• So far this was good enough for my purposes so I had no motivation to spend time on improving something that is not a bottleneck

Though, if someone has good ideas for improvements (especially in form of a working code) - comments and pull requests are welcomed.