如何在Poetry构建流水线中排除Tensorflow依赖？

Got it, let's fix that slow GitHub Actions build by skipping TensorFlow during your automated tests. Here are two reliable approaches using Poetry's built-in features, depending on your setup:

Approach 1: Use Poetry Dependency Groups (Recommended for Poetry 1.2+) #

Poetry 1.2+ introduced dependency groups, which let you organize dependencies by purpose (like production-only, ML-specific, etc.). This is the cleanest way to separate TensorFlow from your test dependencies.

Step 1: Update pyproject.toml #

Move TensorFlow out of your default dependencies into a dedicated group (e.g., tf or production):

# Remove tensorflow from [tool.poetry.dependencies], then add:
[tool.poetry.group.tf.dependencies]
tensorflow = "^2.15"  # Replace with your actual version

Step 2: Local/Production Installation #

When you need TensorFlow locally or for production builds, install with the group included:

poetry install --with tf

Step 3: Update GitHub Actions Workflow #

In your workflow YAML, install dependencies without the tf group to skip TensorFlow:

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: "3.10"  # Match your project's Python version
      - name: Install Poetry
        run: |
          curl -sSL https://install.python-poetry.org | python3 -
          echo "$HOME/.local/bin" >> $GITHUB_PATH
      - name: Install dependencies (skip TensorFlow)
        run: poetry install --without tf
      - name: Run tests
        run: poetry run pytest

Approach 2: Use Optional Dependencies (Compatible with Older Poetry Versions) #

If you're stuck on Poetry 1.1 or earlier, optional dependencies work great for excluding single large packages like TensorFlow.

Step 1: Update pyproject.toml #

Mark TensorFlow as optional and define an "extra" set for it:

[tool.poetry.dependencies]
python = "^3.10"
tensorflow = { version = "^2.15", optional = true }

[tool.poetry.extras]
tf = ["tensorflow"]

Step 2: Local/Production Installation #

Install with the extra flag when you need TensorFlow:

poetry install --extras tf
# Shorthand: poetry install -E tf

Step 3: Update GitHub Actions Workflow #

Optional dependencies aren't installed by default, so just use the standard install command:

# ... (same checkout/setup steps as above)
- name: Install dependencies (skip TensorFlow)
  run: poetry install
- name: Run tests
  run: poetry run pytest

Critical Notes #

• Test Code Compatibility: Make sure your test suite doesn't hard-import TensorFlow. If you have code that uses TensorFlow conditionally, wrap imports in a try-except block:

  try:
      import tensorflow as tf
  except ImportError:
      tf = None  # Or skip related test cases with pytest.mark.skip
  

• Verify Locally: Test the install command (e.g., poetry install --without tf) on your machine first to confirm TensorFlow isn't installed and your tests still pass.

内容的提问来源于stack exchange，提问作者Daan Klijn