How to use the SA lib

This lib source code is in the deepomatic-workflows-sa-lib repository: https://github.com/Deepomatic/deepomatic-workflows-sa-lib

How to use the lib

If you want your workflow to use that lib, you'll have to specify the sa_lib key in your in specs.yaml.

# specs.yaml
version: 2
sa_lib: <version>
...

With that done, sa_lib will be made available for standard import:

from deepomatic.workflows.sa import <...>
...

Version can be:

• local: will use the version of that lib you have locally installed. Useful for local debuging/developping. To use it, first install the library in your local env. and change your package version in specs.yaml to sa_lib: local

  git clone [e-mail]:Deepomatic/deepomatic-workflows-sa-lib.git
  cd deepomatic-workflows-sa-lib
  pip install --editable ./

• <MAJOR.MINOR.BUGFIX>: workflow server will download the corresponding sa lib.

  

How do you release a new version ?

• work on a branch different from master: you can test how your library behave with the local option of the workflow server

• make a Pull Request to validate your work:

  • CI/CD will automatically be launch and validate code (lint + test)

  • Have someone review your code

• merge the PR once it's validated

• then, on your shell:

  git checkout master
  git pull origin master
  git tag -a <X.Y.Z> -m "version <X.Y.Z>" #ex 0.0.4
  git push origin tag <X.Y.Z>

• a github action will automatically package your library and made a <X.Y.Z> version available as a release.

• you can then specify your library on the workflow server side in the specs.yaml

What's in that lib ?

This is not at all the complete doc of the lib SA which is in the SA's ownership. It's purpose is just to introduce it, but don't be scared to dig in the code, it's yours !

Go take a look here !

Google OCR

I recommend using the GoogleOCRProcessor class to use the Google vision OCR:

from lib.google_ocr_processing import GoogleOCRProcessor

image_ocr = GoogleOCRProcessor(image)

# Access the predicted text
image_ocr.text

GoogleOCRProcessor is a wrapper around the google_ocr function implemented in deepomatic-workflows. It gives various methods to efficiently use google predictions.