Whereabouts API

Archived :(
I've archived this project in order to focus on learning and making new things. The project is no longer receiving bug fixes or security updates. Feel free to fork this repository and use it for your own purposes under GPL-3.0. Cheers.

A GeoJSON Point Feature API, searchable via feature name, radius, or bounding box queries with additional class-based feature filtering supported. Whereabouts is built with Deno, the Hono web framework, and MongoDB.

Response of a geospatial radius query using Ordnance Survey's Open Names dataset.

API Service Endpoints

API responses are not pretty-printed by default. To format JSON responses from any endpoint, add the ?pretty query parameter to your request.

/

Service Root
[ GET ] Returns a list of available service endpoints.

/classes

Class List
[ GET ] Returns a list of dataset classes on which features can be filtered.

/features

Feature Search
[ GET ] Returns GeoJSON features. Supports the following search methods:

• Geospatial > Radius Query
  ?radius=lon,lat,distance
  Accepts a Lon-Lat coordinate pair and a search distance value between 0 and 2000 meters.

  • Results can be filtered by class name using the &class=className query parameter.
  • This endpoint will return a maximum of 1000 features per query. The next 1000 features can be retrieved by providing an &offset=1000 query parameter. The offset value must be a multiple of 1000.

• Geospatial > Bounding Box Query
  ?bounds=1,2,3,4
  Accepts a valid Bounding Box with an area less than 2km².

  • Results can be filtered by class name using the &class=className query parameter.
  • This endpoint will return a maximum of 1000 features per query. The next 1000 features can be retrieved by providing an &offset=1000 query parameter. The offset value must be a multiple of 1000.

• Textual (Full Text Search) > Name Query
  ?text=featureName
  Accepts a partial or full feature name to search by. User input must be between 4 and 20 characters. Outgoing results from MongoDB are subject to secondary filtering via FuseJS.

  • Searches are case-insensitive and span the entire dataset.
  • Results can be filtered by class name using the &class=className query parameter.
  • This endpoint will return a maximum of 1000 features per query. The offset parameter is not supported.

/metadata

Service Metadata
[ GET ] Returns metadata about the Whereabouts API service and the source dataset.

The totalClasses and totalFeatures fields are calculated from the source dataset. The name, license, version, and documentation fields are each optional and can be included by passing a single document (with matching key-value pairs) into a metdata collection within the MongoDB database.

Deployment Guide

The Whereabouts API Server is designed to operate as a Bring-Your-Own-Dataset utility. Providing a dataset consists of point GeoJSON features stored in MongoDB in accordance with the specification below, the server should be able to parse it.

Docker Setup

The server image is published on Docker Hub as abiddiscombe/whereabouts.
The following environment variables are required:

• MONGO_URL
  A valid MongoDB connection string. It should look like: mongodb+srv://uname:pword@example.com.

• MONGO_DATABASE
  The name of the database to connect to. The database collections must be named features and metadata.

• CORS_ORIGIN
  A valid fully-qualified domain name. Supply a wildcard (*) to enable CORS for all origins. Not supplying a value will disable CORS headers.

Database Schema

The MongoDB database must have a collection named features in which each GeoJSON point feature is stored as a unique document. Feature must contain the following properties:

• name - A name string for each feature. Duplicate names are permitted.
• class - A classification string upon which API responses can be filtered.

MongoDB index fields are required to handle geospatial and full-text search queries; the server will otherwise return an error. Setting an index for properties.class is optional, but strongly reccommended because the distinct operation is resource-intensive.

• properties.name - text
• properties.class - regular
• geometry.coordinates - 2Dsphere

The MongoDB database can also optionally contain a collection named metadata. This collection can store a single document containing key-value pairs matching those described in the /metdata endpoint instructions.

Development Notes

The server can be run locally using the following commands. These task entries are pre-configured with the correct permission flags, these include access to the system environment and network. Remember to configure your environment variables as per the instructions.

deno task dev
deno task test

FYI: As of version 2.1.0, a endpoint-based test suite is under development. It's not yet at a point suitable for automated use.