fertiscan-pipeline

This repository contains the core analysis pipeline for FertiScan. It is designed to be used as a standalone Python package that can be integrated with other projects, such as the fertiscan-backend.

Setup for Development

Prerequisites

• Python 3.8+
• pip
• virtualenv
• Azure Document Intelligence and OpenAI API keys

Installation

To install the package directly from GitHub:

Direct Installation using pip

Run the following command in your terminal:

pip install git+https://github.com/ai-cfia/fertiscan-pipeline.git@main

Installation via requirements.txt

Add the following line to your requirements.txt file:

git+https://github.com/ai-cfia/fertiscan-pipeline.git@main

Then, install the dependencies with:

pip install -r requirements.txt

Environment Variables

Create a .env file and set the necessary environment variables:

AZURE_API_ENDPOINT=your_azure_form_recognizer_endpoint
AZURE_API_KEY=your_azure_form_recognizer_key
AZURE_OPENAI_API_ENDPOINT=your_azure_openai_endpoint
AZURE_OPENAI_API_KEY=your_azure_openai_key
AZURE_OPENAI_DEPLOYMENT=your_azure_openai_deployment

Packaging and release workflow

The pipeline triggers on PRs to check code quality, markdown, repository standards, and ensures that the version in pyproject.toml is bumped. When a PR is merged, the workflow automatically creates a release based on the version in pyproject.toml. The latest releases and changelogs are available here.

To use this package in other projects, add it to your requirements.txt (e.g., in the fertiscan-backend):

git+https://github.com/ai-cfia/[email protected]

Where vX.X.X is the version from the release page.