This library is provided as-is, right now. It lacks proper documentation which will probably make it hard for you to use unless you are willing to poke and around and investigate things on your own.
There is already a handy import.py script for importing a directory full of files (like say, maybe... whosonfirst-data) and basically all it is doing is this:
import mapzen.whosonfirst.spatial
import mapzen.whosonfirst.utils
dsn = "..."
idx = mapzen.whosonfirst.spatial.index(dsn)
source = os.path.asbpath(source)
crawl = mapzen.whosonfirst.utils.crawl(source, inflate=True)
for feature in crawl:
db.import_feature(feature)
Find one or more WOF places (optionally of a specific placetype) that contain a given point.
import mapzen.whosonfirst.spatial
dsn = "..."
db = mapzen.whosonfirst.spatial.query(dsn)
lat = 37.763251
lon = -122.424002
placetype = 'neighbourhood'
rsp = db.get_by_latlon(lat, lon, placetype=placetype)
for feature in rsp:
props = feature['properties']
print "%s %s" % (props['wof:id'], props['wof:name'])
For example:
$> whosonfirst.py -p neighbourhood -l 37.763251,-122.424002
85865951 Mission Dolores
Find one or more WOF places (optionally of a specific placetype) that intersect a given bounding box.
import mapzen.whosonfirst.spatial
dsn = "..."
db = mapzen.whosonfirst.spatial.query(dsn)
bbox = '37.763251,-122.424002,37.768476,-122.417865'
swlat, swlon, nelat, nelon = bbox.split(",")
placetype = 'neighbourhood'
rsp = db.get_by_extent(swlat, swlon, nelat, nelon, placetype=placetype)
for feature in rsp:
props = feature['properties']
print "%s %s" % (props['wof:id'], props['wof:name'])
For example:
$> whosonfirst.py -p neighbourhood -b 37.763251,-122.424002,37.768476,-122.417865
85834637 Mission
85887467 The Hub
85865951 Mission Dolores
This just requires writing some hoop-jumping around Migurski's whereami.py code to convert a Z/X/Y tile address in to a bounding box and then invoke get_by_extent. In the meantime you could just write some hoop-jumping code of your around Migurski's whereami.py code... to convert a Z/X/Y tile address in to a bounding box.
See also: ops-mapzen-whosonfirst-spatial for a fancier gunicorn/nginx version of this.
On the server-side:
$> export WOF_LOOKUP_DSN='...'
$> ./scripts/server.py -v
INFO:werkzeug: * Running on http://127.0.0.1:8888/
INFO:werkzeug:127.0.0.1 - - [23/Jul/2015 20:39:13] "GET /?bbox=37.763251,-122.424002,37.768476,-122.417865&placetype=country HTTP/1.1" 200 -
On the client-side:
$> curl 'http://localhost:8888/?bbox=37.763251,-122.424002,37.768476,-122.417865&placetype=country'
{
"features": [
{
"id": 85633793,
"properties": {
"geom:area": 9833516741647.34,
"geom:latitude": 45.964509,
"geom:longitude": -113.2686,
"iso:country": "US",
"name:chi_p": [
"\u7f8e\u570b"
],
"name:chi_v": [
"\u7f8e\u5229\u575a\u5408\u8846\u56fd",
"\u7f8e\u5229\u5805\u5408\u8846\u56fd",
"\u7f8e\u56fd"
],
"name:cze_p": [
"Spojen\u00e9 St\u00e1ty Americk\u00e9"
],
"name:cze_v": [
"Spojen\u00e9 st\u00e1ty"
],
"name:dan_v": [
"Amerikas Forenede Stater"
],
"name:dut_p": [
"Verenigde Staten"
],
"name:dut_v": [
"VS",
"Verenigde Staten van Amerika"
],
"name:eng_a": [
"US",
"USA"
],
# and so on...
"wof:id": 85633793,
"wof:lastmodified": 1437610336,
"wof:name": "United States",
"wof:parent_id": 102191575,
"wof:path": "856/337/93/85633793.geojson",
"wof:placetype": "country",
"wof:superseded_by": [],
"wof:supersedes": []
},
"type": "Feature"
}
],
"type": "FeatureCollection"
}
$> sudo apt-get install postgresql-9.3 postgresql-client postgis postgresql-9.3-postgis-scripts python-psycopg2
$> sudo su -m postgres
$> createdb whosonfirst
$> psql -d whosonfirst -c "CREATE EXTENSION postgis;"
$> psql -d whosonfirst -c "CREATE EXTENSION postgis_topology;"
$> psql -d whosonfirst
gazetteer=# CREATE TABLE whosonfirst (id BIGINT PRIMARY KEY, placetype VARCHAR, properties TEXT, geom GEOGRAPHY(MULTIPOLYGON, 4326), centroid GEOGRAPHY(POINT, 4326));
CREATE INDEX by_geom ON whosonfirst USING GIST(geom);
CREATE INDEX by_placetype ON whosonfirst (placetype);
VACUUM ANALYZE;
-
This schema assumes
MULTIPOLYGONgeometries so 1) the import tools will convert single geometries to... mutli-geometries and 2) it will not work with point data (aka venues) – this is no longer true but I haven't updated the code... -
As I write this GeoJSON
geometryelements are not included in any responses by default. This is by design for performance reasons. It will be possible
This is not meant for fulltext search. No. There will be a separate py-mapzen-whosonfirst-search package that will make happy with Elasticsearch and together they will make beautiful search. This is not that package.