tsia / aip-extract

Extracts AIP VFR document images from the DFS Website and does OCR on it

AIP Extractor

As the DFS now publishes the AIP-VFR online but each page only as an image and doesn't allow to print multiple pages, search the content and so on this tool scrapes the website and creates a nice PDF file.

Features:

• PDF creation: Pull the AIP-VFR and create one single PDF
• OCR: Convert it from pictures to a PDF with searchable text
• Docker: Optionally run the tool as a Docker container

Use standalone

Install the requirements, these are: python3 python3-pip Ghostscript Tesseract jbig2enc pngquant unpaper As well as the python packages with pip install -r requirements.txt to install dependencies. Then run main.py. this will scrape the DFS website, fetch all AIP VFR pages thate are provided as images (!), stick them together into PDF files and run OCR on it.

PDFs will be stored in the output folder.

Example code on Ubuntu:

sudo apt install python3  python3-pip Ghostscript Tesseract, jbig2enc, pngquant, unpaper
pip3 install -r requirements.txt
mkdir output
python3 main.py

Docker container

Clone project and cd into folder. Then build container.

cd aip-extract
docker build -t aip-extract:latest .

Run the container once

docker run -it --rm -v "./output:/home/App/output" --name aip-extract aip-extract:latest

Or on Windows:

docker run -it --rm -v ".\output:/home/App/output" --name aip-extract aip-extract:latest

Please note that neither this repository nor the DockerHub image are actively managed. And sure, using ubuntu:22.10 results in a huge image, so I'm open for suggestions on that side.

Performance

OCR processing runs on my personal machine at approx. 5 pages/second. The complete process for all ~1300 pages therefore takes about 20 to 30 minutes. As the script keeps most of the files in RAM one has to expect a high RAM usage as well as 100% CPU usage during the process.