PDF to TXT (with OCR)

Given one or more PDFs that may include text-as-image content, use OCR (Optical Character Recognition) to convert the content to TXT files (in UTF-8 encoding).

Rationale

A survey of existing PDF-to-TXT solutions found no extant solutions that meet all of the following criteria:

is an offline tool (to keep secure human-subject information)
provides conversion from PDF to TXT (most existing OCR integrations assume an image as input)
supports batch processing of multiple files

Assumptions

This is (currently) a command-line tool, written in Python. Basic familiarity with executing commands in a terminal, as well as directory structure, is assumed.
It is assumed that you have Python version 3.x installed, as well as Pip.
This script relies on an industry-standard OCR library managed by Google, called Tesseract. Since it is written in C++, for Python to be able to use it, it needs to be installed separately (instructions below). Similarly, a PDF-to-image library, Poppler, will need to be installed on Windows and Mac systems.

Windows

Make a new folder on your Desktop called ocr (e.g., C:\Users\mark\Desktop\ocr)
Download and install the Tesseract 4 OCR library from Tesseract at UB Mannheim
The installation should indicate which directory Tesseract-OCR was installed. Most likely, this will either be C:\Program Files (x86)\Tesseract-OCR or C:\Program Files\Tesseract-OCR. Move this folder into your equivalent of C:\Users\mark\Desktop\ocr, so that it is now located at Desktop\ocr\Tesseract-OCR.
Download poppler for Windows.
You may need to install 7Zip to unzip the executable, as well.
Place the unzipped files in Desktop\ocr\poppler-0.68.0_x86).
From your start menu, navigate to Control Panel > System and Security > System > Advanced System Settings
Then click Environment Variables.
In the System Variables window, highlight Path, and click Edit.
Click New to add an additional path.
Paste the full path to the location of Tesseract (e.g., C:\Users\mark\Desktop\\ocr\Tesseract-OCR) and press OK.
Again, click New to add an additional path.
Paste your equivalent of C:\Users\mark\Desktop\ocr\poppler-0.68.0_x86\poppler-0.68.0\bin and press OK.
Press OK on any remaining control panel windows.
Download OCR2Text to Desktop\ocr).
Unzip the project.
Open a cmd.exe terminal, and navigate to the folder via the command line (e.g., cd Desktop\ocr\ocr2text-master)
Run pip install --user --requirement requirements.txt
Optionally, you can check that you set up the PATH variable correctly in steps 6-10 by typing echo %PATH%. The output must include your equivalent of C:\Users\mark\Desktop\ocr\Tesseract-OCR and C:\Users\mark\Desktop\ocr\poppler-0.68.0_x86\poppler-0.68.0\bin for the script to work.

macOS (using python3)

Make a new folder on your Desktop called ocr (i.e., /Users/mark/Desktop/ocr)
Install Tesseract-OCR using either MacPorts (sudo port install tesseract) or Homebrew (brew install tesseract
Install poppler for Mac.
Download this Github project to /Users/mark/Desktop/ocr). git clone ...
Unzip the project.
Open a terminal and navigate to the folder via the command line (e.g., cd /Users/mark/Desktop/ocr/ocr2text)
Run pip3 install --user --requirement requirements.txt

Linux

sudo apt-get install tesseract-ocr
Most distros ship with pdftoppm and pdftocairo. If they are not installed, refer to your package manager to install poppler-utils
Download this Github project.
Unzip the project.
Open a terminal and navigate to the folder
Run pip install --user --requirement requirements.txt

Usage

If you have successfully completed the setup steps and are using Python version 3, usage should now be a breeze:

On the command line, navigate to the directory where you downloaded the script and run:

python3 ocr2text.py

You will see the following:

********************************
*** PDF to TXT file, via OCR ***
********************************

Indicate file or folder of source PDF(s) []:
(Press [Enter] for current working directory)

Enter the full path to the file or directory to convert.

Destination folder for TXT []:
(Press [Enter] for current working directory)

Enter the full path to the directory where the result file(s) should be outputted.

The script will now covert the PDF via OCR into a plaintext file:

Testing the installation

For testing purposes, a test_files directory is included. You can press [Enter] for the source and destination directories & verify that the image.pdf file is converted. It will also be located in the test_files directory:

Converted C:\Users\mark\ocr2text\image.pdf
Percent: [##########] 100%
1 file converted

erikbuunk/ocr2text