[do not merge] Demonstrate registering committed models via GitHub Actions

.github/workflows/register_verta_model.yaml

@@ -0,0 +1,42 @@
+name: Register Verta Model Version
+
+on:
+  push:
+    paths:
+      - github-actions/register-verta-model/models/*.pkl
+    branches:
+      - workflow1
+
+jobs:
+  Register-Verta-Model-Version:
+    runs-on: ubuntu-latest
+    defaults:
+      run:
+        working-directory: github-actions/register-verta-model
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          fetch-depth: 0 # fetch full history, in case push is multiple commits
+
+      - name: Set up Python 3.10
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.10"
+          cache: "pip"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install -r requirements.txt
+
+      - name: Register model version
+        env:
+          VERTA_HOST: dev.verta.ai
+          VERTA_EMAIL: [email protected]
+          VERTA_DEV_KEY: ${{ secrets.TEST_ACCOUNT_DEV_KEY }}
+          REGISTERED_MODEL_NAME: From GitHub Action
+          REQUIREMENTS_FILEPATH: requirements.txt
+        run: |
+          export MODEL_FILEPATHS=$(git diff --diff-filter=AR --relative --name-only ${{ github.event.before }} ${{ github.event.after }} | grep '\.pkl$')
+          python register_model_version.py

github-actions/register-verta-model/register_model_version.py

@@ -0,0 +1,34 @@
+# -*- coding: utf-8 -*-
+
+"""This script takes model pickle files and registers them to Verta."""
+
+import os
+
+import cloudpickle
+from verta import Client
+from verta.environment import Python
+
+
+MODEL_FILEPATHS = os.environ["MODEL_FILEPATHS"].splitlines()
+REGISTERED_MODEL_NAME = os.environ["REGISTERED_MODEL_NAME"]
+REQUIREMENTS_FILEPATH = os.environ["REQUIREMENTS_FILEPATH"]
+
+
+if __name__ == "__main__":
+    client = Client()
+    reg_model = client.get_or_create_registered_model(REGISTERED_MODEL_NAME)
+
+    for model_filepath in MODEL_FILEPATHS:
+        with open(model_filepath, "rb") as f:
+            model_cls = cloudpickle.load(f)
+
+        requirements = Python.read_pip_file(
+            os.path.join(os.path.dirname(__file__), REQUIREMENTS_FILEPATH),
+        )
+
+        print(f'Registering model "{model_filepath}"')
+        model_ver = reg_model.create_standard_model(
+            name=os.path.basename(model_filepath).replace(":", "_").replace(".", "_"),
+            model_cls=model_cls,
+            environment=Python(requirements),
+        )

github-actions/register-verta-model/requirements.txt

@@ -0,0 +1 @@
+verta==0.22.2

github-actions/register-verta-model/save_model_file.py

@@ -0,0 +1,33 @@
+# -*- coding: utf-8 -*-
+
+"""This script simply saves a pickled model class to disk.
+
+This is for demonstration purposes—meant to simulate a data scientist checking
+a model into the repo.
+
+"""
+
+from datetime import datetime
+import os
+
+import cloudpickle
+from verta.registry import VertaModelBase, verify_io
+
+
+MODEL_DIR = os.path.join(os.path.dirname(__file__), "models")
+
+
+class Model(VertaModelBase):
+    def __init__(self, artifacts=None):
+        pass
+
+    @verify_io
+    def predict(self, input):
+        return input
+
+
+if __name__ == "__main__":
+    model_filename = f"{datetime.now().isoformat()}.pkl"
+    model_filepath = os.path.join(MODEL_DIR, model_filename)
+    with open(model_filepath, "wb") as f:
+        cloudpickle.dump(Model, f)