Parse a osm.pbf file (e.g. from https://download.geofabrik.de/) and output a csv file listing all populated places in it

placenames.py

#!/usr/bin/env python3
import osmium
import sys
import csv
from shapely.geometry import shape


PLACE_TYPES = [
    "city",
    "town",
    "village",
    "hamlet",
    "isolated_dwelling",  # These are often farms and such. Might want to skip.
]


def get_location(obj):
    cnt = shape(obj.__geo_interface__["geometry"]).centroid
    return {"lat": cnt.y, "lon": cnt.x}


def main(osmfile):
    out_csv = csv.DictWriter(
        sys.stdout, ["place", "lat", "lon", "name"], extrasaction="ignore"
    )
    out_csv.writeheader()
    processor = (
        osmium.FileProcessor(osmfile)
        .with_filter(osmium.filter.KeyFilter("name"))
        .with_filter(osmium.filter.TagFilter(*[("place", v) for v in PLACE_TYPES]))
        .with_filter(osmium.filter.GeoInterfaceFilter())
        .with_locations()
        .with_areas()
    )
    for obj in processor:
        tags = {k: v for k, v in obj.tags}
        tags.update(get_location(obj))
        out_csv.writerow(tags)


if __name__ == "__main__":
    if len(sys.argv) != 2:
        print("Usage: python %s <osmfile>" % sys.argv[0])
        sys.exit(1)

    exit(main(sys.argv[1]))