You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
Peter Johnson f1c5e8d9a8
feat(dedupe): treat all non-canonical layers and analogous to a venue, prefer non-canonical records
6 years ago
.circleci Update Docker version 6 years ago
bin chore(tests): Update test running script with latest info 6 years ago
controller fix(confidence): Update query type names in confidence score code 6 years ago
helper feat(dedupe): treat all non-canonical layers and analogous to a venue, prefer non-canonical records 6 years ago
middleware feat(dedupe): treat all non-canonical layers and analogous to a venue, prefer non-canonical records 6 years ago
public Add a link to the OSM Geocoding Guidelines 7 years ago
query feat(query): Modify custom boosts feature to use function_score queries 6 years ago
routes [Fix] Always apply geometric filter for placeholder 7 years ago
sanitizer feat(log): add structured data to address parser logs 6 years ago
service feat(log): Use structured logs for place endpoint 6 years ago
test feat(dedupe): treat all non-canonical layers and analogous to a venue, prefer non-canonical records 6 years ago
.dockerignore Make .dockerignore node-specific 7 years ago
.gitignore Add pids directory to gitignore 9 years ago
.jshintignore remove code-climate hooks 10 years ago
.jshintrc tell eshint we allow ECMA6 functions 8 years ago
.npmrc disable package-lock 7 years ago
.travis.yml Only run semantic-release on production branch 6 years ago
Dockerfile feat(docker): Run API processes using pelias user 6 years ago
LICENSE Update LICENSE 7 years ago
README.md Add Configurable Boosts readme section 6 years ago
app.js Add default route redirecting to /v1 7 years ago
index.js feat(log): replace console logs with pelias-logger 6 years ago
package.json Require microservice-wrapper version with response time 6 years ago
schema.js Use `pelias.json` to inform boost configuration 6 years ago

README.md

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.

Pelias API Server

This is the API server for the Pelias project. It's the service that runs to process user HTTP requests and return results as GeoJSON by querying Elasticsearch and the other Pelias services.

NPM

Gitter

Documentation

Full documentation for the Pelias API lives in the pelias/documentation repository.

Install Dependencies

Note: Pelias requires Node.js v6 or newer

npm install

scripts

The API ships with several convenience commands (runnable via npm):

  • npm start: start the server
  • npm test: run unit tests
  • npm run ciao: run functional tests (this requires that the server be running)
  • npm run docs: generate API documentation
  • npm run coverage: generate code coverage reports
  • npm run config: dump the configuration to the command line, which is useful for debugging configuration issues

Configuration via pelias-config

The API recognizes the following properties under the top-level api key in your pelias.json config file:

parameter required default description
services no service definitions for point-in-polygon, libpostal, placeholder, and interpolation services. If missing (which is not recommended), the services will not be called.
defaultParameters.focus.point.lon
defaultParameters.focus.point.lat
no default coordinates for focus point
targets.layers_by_source
targets.source_aliases
targets.layer_aliases
no custom values for which sources and layers the API accepts (more info).
customBoosts no {} Allows configuring boosts for specific sources and layers, in order to influence result order. See Configurable Boosts below for details
indexName no pelias name of the Elasticsearch index to be used when building queries
attributionURL no (autodetected) The full URL to use for the attribution link returned in all Pelias responses. Pelias will attempt to autodetect this host, but it will often be correct if, for example, there is a proxy between Pelias and its users. This parameter allows setting a specific URL to avoid any such issues
accessLog no name of the format to use for access logs; may be any one of the predefined values in the morgan package. Defaults to "common"; if set to false, or an otherwise falsy value, disables access-logging entirely.
relativeScores no true if set to true, confidence scores will be normalized, realistically at this point setting this to false is not tested or desirable

A good starting configuration file includes this section (fill in the service and Elasticsearch hosts as needed):

{
  "esclient": {
    "hosts": [{
      "host": "elasticsearch"
    }]
  },
  "api": {
    "services": {
      "placeholder": {
        "url": "http://placeholder:4100"
      },
      "libpostal": {
        "url": "http://libpostal:8080"
      },
      "pip": {
        "url": "http://pip-service:4200",
        "timeout": 1000,
        "retries": 2
      },
      "interpolation": {
        "url": "http://interpolation:4300"
      }
    }
  },
  "logger": {
    "level": "debug"
  }
}

The timeout and retry values, as show in in the pip service section, are optional but configurable for all services (see pelias/microservice-wrapper for more details).

Custom Boosts

The customBoosts config section allows influencing the sorting of results returned from most Pelias queries. Every Pelias record has a source and layer value, and this section allows prioritizing certain sources and layers.

First, keep in mind:

  1. This will not affect all Pelias queries. In particular, when using the /v1/search endpoint, queries for administrative areas (cities, countries, etc) will likely not be affected
  2. Custom boosts allow influencing results, but not completely controlling them. Very good matches that aren't in a boosted source or layer may still be returned first.

The basic form of the configuration looks like this:

{
  "api":
    "customBoosts": {
      "layer": {
        "layername": 5,
        "layername2": 3
      },
      "source": {
        "sourcename": 5
      }
    }
  }
}

There are subsections for both layer and source, and each subsection must be an object. Keys in those objects represent the sources and layers to be boosted, and the value associated with those keys must be a numeric value.

Boost values are essentially multipliers, so values greater than 1 will cause a source or layer to be returned more often, and higher in results. Boosts of the value 1 are the same as no boost, and boosts between 0 and 1 will de-prioritize matching records.

Recommended boost values are between 1 and 5. Higher boosts are likely to cause unexpected impact without really improving results much.

Configuration via Environment variable

Most Pelias configuration is done through pelias-config, however the API has additional environment variables that affect its operation:

environment variable default description
HOST undefined The network interface the Pelias API will bind to. Defaults to whatever the current Node.js default is, which is currently to listen on all interfaces. See the Node.js Net documentation for more info.
PORT 3100 The TCP port the Pelias API will use for incoming network connections.

Contributing

Please fork and pull request against upstream master on a feature branch. Pretty please; provide unit tests and script fixtures in the test directory.

Unit tests

You can run the unit test suite using the command:

$ npm test

HTTP tests

We have another set of tests which are used to test the HTTP API layer, these tests send expected HTTP requests and then assert that the responses coming back have the correct geoJSON format and HTTP status codes.

You can run the HTTP test suite using the command:

$ npm run ciao

Note: some of the tests in this suite fail when no data is present in the index, there is a small set of test documents provided in ./test/ciao_test_data which can be inserted in order to avoid these errors.

To inject dummy data in to your local index:

$ node test/ciao_test_data.js

You can confirm the dummy data has been inserted with the command:

$ curl localhost:9200/pelias/_count?pretty
{
  "count" : 9,
  ...
}

Continuous Integration

Travis tests every release against all supported Node.js versions.

Build Status

Versioning

We rely on semantic-release and Greenkeeper to maintain our module and dependency versions.

Greenkeeper badge