This document describes in detail research into and procedures for downloading building footprint data and converting it to a shapefile. The notes below will be incorporated into a simple web app to provide this functionality.

As xml:

http://overpass-api.de/api/interpreter?data=(node["building"="yes"](-6.185440796831979,106.82374835014343,-6.178966266481431,106.83127999305725);way["building"="yes"](-6.185440796831979,106.82374835014343,-6.178966266481431,106.83127999305725);relation["building"="yes"](-6.185440796831979,106.82374835014343,-6.178966266481431,106.83127999305725););(._;>;);out body;

As json (not used but just here in case of future needs):

http://overpass-api.de/api/interpreter?data=[out:json];(node[%23building%22=%22yes%22](-6.185440796831979,106.82374835014343,-6.178966266481431,106.83127999305725);way[%22building%22=%22yes%22](-6.185440796831979,106.82374835014343,-6.178966266481431,106.83127999305725);relation[%22building%22=%22yes%22](-6.185440796831979,106.82374835014343,-6.178966266481431,106.83127999305725););(._;%3E;);out%20body;

Example python script to retrieve geometries:

import urllib2
params = {'xmin': -6.185440796831979, 'ymin': 106.82374835014343,'xmax': -6.178966266481431, 'ymax': 106.83127999305725}
myOsmXmlUrlPath = ('http://overpass-api.de/api/interpreter?data=(node[%%22building%%22=%%22yes%22](%(xmin)s,%(ymin)s,%(xmax)s,%(ymax)s);way[%22building%22=%22yes%22](%(xmin)s,%(ymin)s,%(xmax)s,%(ymax)s);relation[%22building%22=%22yes%22](%(xmin)s,%(ymin)s,%(xmax)s,%(ymax)s););(._;%3E;);out%20body;' % params)
# Note that osm json is NOT geojson!
myOsmJsonUrlPath = ('http://overpass-api
.de/api/interpreter?data=[out:json];(node[%22building%22=%22yes%22](%(xmin)s,%(ymin)s,%(xmax)s,%(ymax)s);way[%22building%22=%22yes%22](%(xmin)s,%(ymin)s,%(xmax)s,%(ymax)s);relation[%22building%22=%22yes%22](%(xmin)s,%(ymin)s,%(xmax)s,%(ymax)s););(._;%3E;);out%20body;' % params)
myRequest = urllib2.Request(myOsmXmlUrlPath)
try:
    myUrlHandle = urllib2.urlopen(myRequest, timeout=60)
    myFile = file('osm.xml', 'wb')
    myFile.write(myUrlHandle.read())
    myFile.close()
    except urllib2.URLError:
        raise

We want to convert the JOSM preset file (think of it as an xml file that will generate a form in JOSM to ensure the correct attributes are captured) into as shp2pgsql style file (which determines what the schema and rules are for importing osm data into PostgreSQL). To do this we use a converter tool written by Dane Springmeyer:

curl -O https://raw.github.com/hotosm/scripts/master/preset2style.py

For the JOSM preset mentioned above, we want to use one created by Kristy van Puk:

curl -o building-preset.xml -O http://hot-export.geofabrik.de/uploads/preset-19

We need to have this for the JOSM conversion step coming up as it will be merged in as the base style.

curl -O http://svn.openstreetmap.org/applications/utils/export/osm2pgsql/default.style

The file downloaded above misses the group element so you should insert this after the presets opening tag:

<group name="HOT" fr.name="HOT" en.name="HOT">

and this before the presets closing tag:

</group>

Now you can convert it using hte preset2style application we downloaded previously:

python preset2style.py -s default.style --preset building-preset.xml > building.style

You need to have postgresql / postgis installed and osm2pgsql. OSX users can find precompiled binaries for osm2pgsql here: http://cl.ly/201r35290x0e

Save the above python script as fetch_osm_xml.py. To automate the process you can do e.g.:

#!/bin/bash
rm buildings*
rm osm.xml
python fetch_osm_xml.py
DBNAME=$$
createdb -T template_postgis $DBNAME
osm2pgsql -S building.style -d $DBNAME osm.xml
pgsql2shp -f buildings.shp $DBNAME planet_osm_polygon
dropdb $DBNAME

Then save it as a script and run it e.g.:

./fetch_osm_buildings.sh