/GeocoderService

Primary LanguagePython

Geocoding HTTP Service

This repository contains a simple Python 3 implementation of a RESTful HTTP API which provides a geocoding service using multiple geocoding service providers (Here and Google) in the backend. This implementation uses only standard Python 3 modules.

Usage

Requirements

Standard Python 3 environment.

First Time

Before using the server, you must acquire API keys for both backend services. Once the credentials are acquired, they should be stored in geocoding/handlers.py

HERE_APP_ID = "YOUR_APP_ID"
HERE_APP_CODE = "YOUR_APP_CODE"
GOOGLE_APP_ID = "YOUR_APP_KEY"

Google

You can start here to get credentials: https://developers.google.com/maps/documentation/geocoding/start

Here

You can start here to get credentials: https://developer.here.com/documentation/geocoder/topics/quick-start.html

Starting the server

To start the server invoke the following command from the root of this repository:

./server

By default the server is bound to the wildcard network address (bound to all adapters) and listening on port 8888.

This can be modified by changing the server script:

BIND_ADDRESS = ''
BIND_PORT = 8888

API Usage

HTTP GET Endpoints

All endpoints return JSON of the following format:

{"location": {"Latitude": 40.71455, "Longitude": -74.00714}}

If nothing was found, the result would look like this:

{"location": {}}

The user must pass in the search text as a query string in the following format: ?searchtext="new york city"

  • /geocode - Attempt to geocode using Here and then Google as a fallback
  • /geocode/here - Attempt to geocode using Here
  • /geocode/google - Attempt to geocode using Google

Example: http://localhost:8888/geocode?searchtext=%22new%20york%20city%22

Geocoding Clients

There's also a standalone test client for both services, which can be invoked as follows

Google

./google_geocode API_KEY "Search Query"

Here

./here_geocode API_ID APP_CODE "Search Query"

Future Improvements

  • Endpoint routing should use regular expressions
  • Print statements could be implemented as system log messages
  • API could support a JSON format for carrying error messages
  • Unit tests should be written to test implementation
  • Pass in listening port to server script as command line argument
  • Use separate configuration file for API keys