This repository is part of the Pelias project. Pelias is an open-source, open-data geocoder originally sponsored by Mapzen. Our official user documentation is here.
** This service is deprecated and will no longer be maintained **
Module that provides a web service to aid language-agnostic importers with creating documents for insertion into an Elasticsearch index queryable by the Pelias API. Who's on First data is required in order to populate the documents' administrative hierarchy.
$ git clone git@github.com:pelias/document-service.git
$ cd document-service
$ npm install
$ npm start /path/to/whosonfirst/data
For ease of use, Who's on First data can be downloaded using scripts provided by the Pelias Who's on First module.
The pelias-document-service
npm module can be found here:
https://npmjs.org/package/pelias-document-service
To start the document service, type: npm start <path to Who's on First data>
. By default, the service runs on port 5000 but can be overridden in the PORT
environmental variable.
GET
requests are made to the /synthesize
endpoint in the format: http://localhost:5000/synthesize/<source>/<layer>
.
source
is the name of the source of the data that can be used to filter and is used in the synthesized document. For example, data imported from OpenAddresses would use openaddresses
for source
.
layer
is the type of data that this document represents. Currently, the only valid values for layer
are address
, street
, and venue
.
The following parameters are supported for the service:
name | required | description |
---|---|---|
id |
yes | a unique identifier for reference in Elasticsearch |
lon |
yes | longitude of the record |
lat |
yes | latitude of the record |
name |
yes | a textual name of the record such as the name of a business (for venues) or house_number + street (for addresses), used by the Pelias API to create result labels |
house_number |
|
house number of an address or venue |
street |
|
street of an address or venue |
postcode |
no | postcode of an address or venue |
GET
requests to the /synthesize
endpoint return a content-type application/json
response ready to be sent to Elasticsearch for create/update and queryable by the Pelias API. The output for OpenAddresses 30 W 26th St
is:
{
"name": {
"default": "30 W 26th St"
},
"phrase": {
"default": "30 W 26th St"
},
"parent": {
"locality": [
"New York"
],
"locality_id": [
"85977539"
],
"locality_a": [
null
],
"neighbourhood": [
"Flatiron District"
],
"neighbourhood_id": [
"85869245"
],
"neighbourhood_a": [
null
],
"county": [
"New York County"
],
"county_id": [
"102081863"
],
"county_a": [
null
],
"borough": [
"Manhattan"
],
"borough_id": [
"421205771"
],
"borough_a": [
null
],
"region": [
"New York"
],
"region_id": [
"85688543"
],
"region_a": [
"NY"
],
"country": [
"United States"
],
"country_id": [
"85633793"
],
"country_a": [
"USA"
]
},
"address_parts": {
"number": "30",
"street": "W 26th St",
"zip": "10010"
},
"center_point": {
"lon": -73.990409,
"lat": 40.74427
},
"source": "openaddresses",
"layer": "address",
"source_id": "6364a510f0268d6f"
}
There are 3 types of documents that can be synthesized, each corresponding to the layer
value of the request path:
- venue
- address
- street
Venue documents are synthesized by calling the /synthesize/<source>/venue
endpoint. Each venue has a lat
, lon
, id
, name
, and optional house_number
, street
, and postcode
. name
is typically the name of the business or point-of-interest, such as "New York Bakery" or "Yellowstone National Park". house_number
and street
are optional since in some cases this information is either not applicable (as in the case of national parks or water features which are defined as polygons) or confidential (such as women's shelters or other cases where point accuracy is to be purposely obscured).
Example (data from OpenStreetMap): http://localhost:5000/synthesize/openstreetmap/venue?id=264768896&lon=-73.989642&lat40.74101&name=Flatiron+Building&house_number=175&street=5th+Avenue&postcode=10010
Address documents can be synthesized by calling the /synthesize/<source>/address
endpoint. Each address has a lat
, lon
, id
, name
, house_number
, street
, and optional postcode
. The name
value is typically just the formatted address, which can be number-prefixed, as in "30 West 26th Street, New York, NY", or -postfixed, as in "Rigaer Straße 11, Berlin, Germany", but can be anything. The document service makes no judgements on what the value of name should be; its value is determined by the caller.
Example (data from OpenAddresses): http://localhost:5000/synthesize/openaddresses/address?id=6364a510f0268d6f&lon=-73.9904095&lat=40.74427&name=30+W+26th+St&house_number=30&street=W+26th+St&postcode=10010
Street documents are synthesized using the /synthesize/<source>/street
endpoint. Each street has a lat
, lon
, id
, name
, street
, and optional postcode
. If a street is entirely contained within a single postcode, it should be supplied if available. Typically, the name
value should be the same as the street
value but there are no restrictions placed upon this condition.
Example (data from OpenStreetMap): http://localhost:5000/synthesize/openaddresses/address?id=10540891&lon=-73.935546&lat=40.813082&name=Madison+Avenue&street=Madison+Avenue
The /synthesize
endpoint returns an HTTP status code 400 is returned with an error message under any of the following conditions:
lat
value is not parseable as a finite numberlon
value is not parseable as a finite numberid
value is emptyname
value is emptyaddress
layer-specific:house_number
value is emptystreet
value is empty
street
layer-specific:house_number
value is non-empty
venue
layer-specific:house_number
value is non-empty andstreet
value is empty
The /synthesize
endpoint returns an HTTP status code 500 is returned when a error occurs when performing administrative hierarchy lookup.
As this service currently only looks up the administrative hierarchy and formats the request parameters into JSON ready to be sent to Elasticsearch, an importer only needs to be able to make HTTP GET requests and either an Elasticsearch library or the ability to make HTTP POST requests (to index documents into Elasticsearch in lieu of a library).
Example importers have been written in a variety of languages: