I am trying to process a .osm file in python with ogr, but I have problems accessing the lat and lon attributes of the point layer - they are always null.

I want to do spatial filtering to find the closest neighbor in a certain distance, e.g. like:

sql_lyr = osmData.ExecuteSQL("SELECT * FROM points WHERE lat BETWEEN ... AND ... AND long BETWEEN ... AND ...") 

(Inspired by the answer to this question: Python: how to speed up spatial search (nearest point)? )

When I call DumpReadable on a feature in the point layer, I get the following:

  osm_id (String) = 392633
  name (String) = (null)
  lon (String) = (null)
  lat (String) = (null)
  all_tags (String) = (null)
  POINT (16.3540923 48.2041024)

So I see that the lat and lon are read in for the POINT Geometry, but I have no idea how to use this information in a SQL query.

Is there any way I can access the lat and lon via the POINT Geometry in a query, or can I do any further configuration to the OSMdriver so that lat and lon are read in as fields?

I modified osmconf.ini to enable all_tags, to report all nodes, I tried enabling all common attributes and I also tried adding lat and long to the list of keys like this (even though I think this is pointless, because lat long are attributes of the node tag itself and not really keys...):

# keys to report as OGR fields attributes=name,barrier,highway,ref,address,is_in,place,man_made,lon,lat
  • Lat and long are NOT tags in the OSM data model. They are properties of a point, like timestamp and user_id. Same as the points that form a way are not tags of that way.
    – AndreJ
    Commented Sep 7, 2016 at 17:46

1 Answer 1


If you want to do it with SQL you can use the SQLite dialect http://www.gdal.org/ogr_sql_sqlite.html and if your GDAL is compiled with the most recent SpatiaLite version then you can use all these functions: https://www.gaia-gis.it/gaia-sins/spatialite-sql-latest.html

Here is a test query that works for me with ogrinfo

ogrinfo -dialect sqlite -sql "select * from points where (ST_X(geometry) between 10.1 and 10.2) AND (ST_Y(geometry) between 53.4 and 53.5)" hamburg-latest.osm.pbf

However, GDAL/OGR has a native method for spatial filters and I believe that layer.SetSpatialFilter would work for you as in the example in http://pcjericks.github.io/py-gdalogr-cookbook/vector_layers.html#spatial-filter

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.