One of the projects we’re working on at the moment relies heavily on the use of file attachments. Not a problem, LDC Via does that! But we also needed to be able to render some preview text from the contents of the files, so if a PDF, Word document, Excel sheet, or even an image has some text in it, we wanted to be able to show a brief summary of the contents to the user.

Enter the Textract plugin. This is a Python utility which can dig into the contents of files and extract them. It’s also really simple to use in your local environment: you just need to make sure Python’s installed and npm install does the rest for you.

But then we get to the real world. A dev machine is one thing, but when your app needs to be deployed to Heroku things get a little more complex. Typically, your app will be of one type: node.js, Python, Ruby or whatever. But we now need to have an app that contains two different code-bases. Luckily it’s a relatively simple procedure to set it up.

First, you’ll want to configure your project. In our case today, we’re dealing with a node.js app, so we already have our package.json file which defines all our dependencies. However, because we also want to add some Python dependencies, we need to add a new file to the project, which is called requirements.txt:

# This file contains all Python dependencies that are required
# by the Textract package in order for it to properly work.

#pdfminer.six <-- go back to this after the shebang fix is released (see

Next, we need to add another file called Procfile. This is basically an instruction file for Heroku that tells it what to do when it’s starting your application. Normally, when you only have a single code-base you don’t need the file as Heroku can make some assumptions, but with two different strands of code we need to give some guidance. So, in our case the file will simply look like this:

web: npm start

Once you’ve committed these files to your code repository… nothing will happen. We’ve made all the code changes we need to, but now we need to go to Heroku. You can obviously do all this with the CLI, but for pretty picture’s sake, you’ll want to go to your application settings page and scroll down to the Buildpacks section:


Click the Add Buildpack button and for this case, choose “Python” from the list of options.

Once that’s added, we can simply re-deploy the application in the normal way and Textract will magically work!