Feeler

Really simple application for running sentiment analysis on arbitrary phrases.

Breakdown

Server

The meat of this application is a web server for running text submitted through a basic text input. It only exposes two routes: / and /sentiments.

The /sentiments route responds to POST requests and receives a s field in its body with the phrase to analyse.

The root route renders a React application with the abovementioned text input. This text input hits the API on every keystroke. Not performant but Go seems to be up to the task of generating sentiments that quickly (thanks, Go!).

Twitter analyser

I've also included a really simple command line application that receives a Twitter handle and an optional tweet limit and scans that user's Twitter feed for overall sentiment.

Usage

Server

You'll need to build the server itself, and then build the Javascript to mount the React application:

go build -o ./bin/app ./cmd/app
yarn install && yarn build

Once it's built, you can run it like any other executable:

./bin/app

Twitter analyser

You'll need to create a developer Twitter account and a Twitter App to use hit the Twitter API. We're using the v2 API here, which is still in beta, but which is significantly easier to use than v1.1.

Once you've created your app, generate a Bearer Token and save it to your .env file. See .env.example for usage.

Then you can build the executable:

go build -o ./bin/analyser ./cmd/analyser

And run it:

./bin/analyser --user=charlesharries

For a list of all available flags, you can run --help on the executable:

./bin/analyser --help

y u use js

Because I'm a lazy son of a gun. I realise that this page tries to download like 164 KB of Javascript. It's a mess. I'm working on server-rendering the sentiment analysis page but Go's templating language isn't very nice (in my opinion).