Skip to content

Latest commit

 

History

History
90 lines (63 loc) · 2.98 KB

README.md

File metadata and controls

90 lines (63 loc) · 2.98 KB
Raw

Location Service

This is a micro-service intended to be used in service-oriented back-ends. The main purpose is to help users find all other zip codes within a given radius of a supplied zip code.

So for example, if we wanted all cities within a 3 km radius of the 95833 zip code, we'd hit the service with:

GET localhost:3000/?zip=95833&radius=3000

and then we'd receive a JSON response from the service like:

[
    {
        "city": "Sacramento",
        "latitude": 38.642805,
        "longitude": -121.50466,
        "state": "CA",
        "zip": "95834"
    },
    {
        "city": "Sacramento",
        "latitude": 38.616891,
        "longitude": -121.49623,
        "state": "CA",
        "zip": "95833"
    }
]

The project is fueled by Microsoft's open-source project, the cpprestsdk.

Building

You will need CMake 3.6 and above to build this project.

Windows

The easiest way to build this project on Windows is to download vcpkg.

vcpkg install cpprestsdk:x64-windows boost:x64-windows catch:x64-windows

This will generate a CMake toolchain file that's located in \vcpkg\scripts\buildsystem\vcpkg.cmake.

You can then use CMake to generate the solution and then use MSBuild to build it. See build.bat for a working example.

You can also simply build this using VS2017 by opening the containing folder once you've downloaded the repo.

Linux

cpprestsdk is also available on Ubuntu as well among othe select distros like Arch.

For Ubuntu, simply use:

sudo apt-get install libcpprest-dev

You will also need Boost and Catch so you'll need to see how your specific distro wants you to build Boost. Catch is header-only so it simply just needs to exist on your computer and in a location where CMake can find it.

Using the service

Once you've build the service, it's simple to use it!

Simply query the service with /?zip=...&radius=... and it'll retrieve all cities within the supplied radius (radius is in meters).\

Type exit into the console to gracefully close the server.

The Algorithm

Unfortunately, geolocation is a tricky problem.

The algorithm that we use makes a few assumptions.

First, we're limited to our dataset. It's somewhat difficult to get a hold of this kind of data for free so we use a random CSV that may be outdated.

Nevertheless, we parse it and then convert longitude-latitude to Cartesian space by approximating the Earth as a perfect sphere. We don't need a more robust model as we're interested in the relative positions between zip codes and not the exact distances between them.

Using this, we create a simple rtree which maps points to the data of each city (name, zip, state, etc.).

We use Boost.Spirit to parse the CSV and generate our point data. We use Boost.Geometry to create the rtree and the required queries. Boost.Variant is used in conjunction with the cpprestsdk to handle the HTTP requests from clients.