Streamlit App

Run locally

• Install requirements:

  pip install -r requirements.txt

• Start streamlit web-app locally by running the following command:

  streamlit run app.py

• Note: This app uses "tesseract" for OCR. You need to install the executable as well as the python package in order to run it successfully.

• Note: In order to convert a PDF to an image with "image2pdf" before sending it to tesseract you also need to install poppler which can not be installed with pip

Dockerize application

• Build the docker image

  x86:

  docker build -t my-streamlit-app .

  Apple silicon:

  docker build --platform linux/amd64 -t my-streamlit-app .

• Run the docker container (optional)

  docker run -p 8080:8080 my-streamlit-app

  Note: If you want to run the docker container locally on an Apple silicon Mac you should not specify the platform to built it natively for ARM to massively increase the performance of the application

Deploy to GCP using Cloud Run

For more detailed instructions check out this tutorial.

• Tag the docker image

  docker tag my-streamlit-app europe-west3-docker.pkg.dev/propane-nomad-396712/cloud-run/my-streamlit-app

• Push the tagged image to Artifact Registry

  docker push europe-west3-docker.pkg.dev/propane-nomad-396712/cloud-run/my-streamlit-app

• Deploy the application using gcloud

  gcloud run deploy my-streamlit-app \
  --image europe-west3-docker.pkg.dev/propane-nomad-396712/cloud-run/my-streamlit-app \
  --platform managed \
  --allow-unauthenticated \
  --memory=8Gi \
  --cpu=4

• You can retrieve the url at any time with this command

  gcloud run services describe my-streamlit-app --platform managed --region europe-west3 --format 'value(status.url)'