What? A self-contained docker system to create a REST api to find QR codes in pdfs. Both file upload and url-based versions available.
Why? Because AFR can't do it directly.
How? Use instructions below to run the unit tests and/or start the container.
Doesn't sometimes detect very small pixel qr-codes, see e.g. ./tests/test_data/tight_codes.pdf for an example where
the system does not (currently) find the qr-code on the first page. Look also at the pagewise counts for ./tests/test_data/Increasing_qr_count.pdf. Large QR-codes seem to be okay, see e.g. ./tests/test_data/LargeQR.pdf.
Run:
docker build -t qrsystem .
docker run -p 5055:5055 qrsystem
Then take your browser to localhost:5055, the fastAPI Swagger docs are waiting for you there together with UI for
endpoint testing.
Alternatively you can just use Postman to call the endpoints at localhost:5055/QR-detection-via-URL/ and localhost:5055/QR-detection-via-file-upload/. Exact input requirements are found from the docs at localhost:5055.
Run:
conda create --name qr_env python=3.9
conda activate qr_env
pip install -r requirements.txt
python -m pytest
We run python -m pytest because plain pytest fails on the current local machine, probably due to
global package installation stuff.
For coverage, run
coverage run -m pytest
coverage report -im
The m flag is needed due to some weird error with coverage and maybe openCV? Without i the system complains about
No source for code: '~/QRDocker/config-3.py'.