A Digital Humanities Exercise: Batch converting folders of Arabic PDFs to plaintext files.
This exercise uses a Jupyter Notebook to demonstrates converting Arabic PDFs to plain text files. Rather than install locally, see this notebook in action in your browser at MyBinder. Once the server is launched (it takes awhile, ignore any warnings), choose acabic-ocr.ipynb from the navigation menu on the left. To run code, click your mouse inside a cell of code, then click the play button to run that cell.
Open your prompt, navigate to your desire parent folder and enter
git clone https://github.com/thePortus/arabic-ocr.git
cd arabic-ocr
pip install -r requirements.txt
You also need to install poppler, a software library that powers the PDF reader.
- On OSX machines, open your terminal and enter
brew install poppler - For Linux machines, open your terminal and enter
sudo apt-get install poppler-utils. - Windows is a bit harder as
popplerwasn't originally meant for it, but you can find a Windows downloadable installer here along with some discussion from a forum about how to set it up here.
To launch the notebook from your terminal just enter jupyter notebook
Once the interface launches in your webbrowser, select arabic-ocr.ipynb and get started.