PDF OCR via React, Django REST Framework, and Heroku, Part 4: OCR our Documents

Now that we have our front and back ends working at the bare minimum level, we need to go back and update the back end so it will actually run OCR on the file we upload to it. After all, there’s no point in just uploading our own documents to ourselves locally! This is going to be a big change in terms of functionality more than lines of code, which is why I’m combining it with preparing for deployment of our apps.

Download and Install Prerequisites

If for any reason you didn’t install the following packages in Step 1, go back and get them added to Poetry, as these packages are going to do the heavy lifting of our OCR app:

OCRmyPDF pytesseract ghostscript

Update views.py

So now we are going back to our tutorial_backend directory, into our ocr/views.py file. You’ll add the following to your imports:

import ocrmypdf
from subprocess import Popen

Now we will add two lines to our PostViews class. We’ll add the following two lines right underneath the line reading :

if posts_serializer.is_valid():

uploaded = posts_serializer.save()
process = Popen(['ocrmypdf', uploaded.file.path, 'output.pdf'])

Your views.py file should now look like this:

from django.shortcuts import render
from rest_framework.views import APIView
from rest_framework.parsers import MultiPartParser, FormParser
from rest_framework.response import Response
from rest_framework import status
import ocrmypdf
from subprocess import Popen
from .serializers import FileSerializer
from .models import Post
# Create your views here.
class PostViews(APIView):
    parser_classes = (MultiPartParser, FormParser)
    def get(self,request, *args, **kwargs):
        posts = Post.objects.all()
        serializer = FileSerializer(posts, many=True)
        return Response(serializer.data)
    def post(self, request, *args, **kwargs):
        posts_serializer = FileSerializer(data=request.data)
        if posts_serializer.is_valid():
            uploaded = posts_serializer.save()
            process = Popen(['ocrmypdf', uploaded.file.path, 'output.pdf'])
            return Response(posts_serializer.data, status=status.HTTP_201_CREATED)
        else:
            print('error', posts_serializer.errors)
            return Response(posts_serializer.errors,    status=status.HTTP_400_BAD_REQUEST)

That’s it. Now start your Django server and upload a file through the front end. To make sure I wasn’t using a PDF that wasn’t already OCR’d, I converted a photo of something with clear, plain text to PDF. Whatever you want to use, just make sure you can’t already search for text in the file to be uploaded.

In your terminal, you should the OCRmyPDF working on your file.

In the terminal, you should see OCRmyPDF working on the file.

Your OCR’d output file should be put inside your main project directory. Open it in a PDF reader and search for a word. Clearly, those two lines allow the above three imports to work their magic on a PDF and OCR it’s contents. To better explain this magic, we need to look at the subprocess module we’re importing here.

The subprocess module allows you to call other programs from within another Python program. In our program, the uploaded variable stores the value of posts_serializer . That value then gets passed on to the the subprocess Popen module, which uses this data to as part of the arguments needed to call the OCRmyPDF program. We then return the response as normal.

Now that our apps work and actually does something, we can create a new GitHub branch and push our work.

Now that we have our front end working and hooked up to working OCR engine, we can not move onto deploying our apps to Heroku!