I am trying to deploy a Databricks app that uses PyMUPDF library to extract text from PDF files. Under the hood, it tries to use an OS dependency called 'tesseract-ocr'. Please note that this is not the pytesseract library, but a dependency that need to be installed on the system. My understanding is that when deploying a Databricks App, it takes your source code, requirements.txt, and app.yaml to construct a databricks-specific Docker container after which it exposes your app (e.g. Streamlit). I tried to install the required dependency like so in the app.yaml: command: - bash - -c - | apt-get update apt-get install -y tesseract-ocr streamlit run app.py However, I get a no access error. Using sudo apt also doesn't work. Is there any other way to install tesseract-ocr within a Databricks App? Please note that installing it on a cluster won't work AFAIK, as a Databricks App runs in isolation. I was hoping I could simply provide a Dockerfile that the app deployment process could use, which would resolve this case. This does not seem to be possible. I would find it cumbersome to create a separate API just to make calls for OCR stuff. Any suggestions how to resolve this? Continue reading...
