charmed-analytics-ci 1.2.0

A CLI tool for automating CI tasks across Charmed Operator repositories

charmed-analytics-ci

A CLI tool to automate CI-driven integration of updated rock images into consumer Charmed Operator repositories.

This tool is part of Canonical's Charmed Kubeflow stack and enables automated pull request creation after a rock image is built and published. It eliminates manual effort, reduces human error, and supports scalable, reproducible CI/CD pipelines.

---

✨ Features

• ✅ Automatically clones target charm repositories
• 🔁 Updates image references in YAML or JSON configuration files
• ⚙️ Optionally modifies service-spec fields like user and command
• 🔧 Validates metadata schemas for correctness before modification
• 🤖 Opens pull requests with deterministic branches and templated descriptions
• 🔐 Supports GitHub authentication via token or environment variable
• 🔗 Optionally links back to triggering PR
• 📦 Installable via PyPI and usable from CI pipelines
• 🧪 Supports dry-run mode for previewing changes

---

🚀 Installation

Install from PyPI:

pip install charmed-analytics-ci

Or install for development:

git clone https://github.com/canonical/charmed-analytics-ci.git
cd charmed-analytics-ci
poetry install

---

🧪 CLI Usage

After installing, the CLI provides a single command:

chaci integrate-rock METADATA_FILE BASE_BRANCH ROCK_IMAGE [OPTIONS]

Example:

export GH_TOKEN="ghp_abc123..."  # or pass explicitly with --github-token

chaci integrate-rock rock-ci-metadata.yaml main ghcr.io/canonical/my-rock:1.0.0 --dry-run

Arguments:

Argument	Description
METADATA_FILE	Path to rock-ci-metadata.yaml describing integration targets
BASE_BRANCH	Target branch for PRs (e.g. main or develop)
ROCK_IMAGE	Full rock image string (e.g. ghcr.io/org/my-rock:1.0.0)

Options:

Option	Description
--github-token	Optional. GitHub token. Falls back to $GH_TOKEN environment variable if not provided.
--github-username	Optional. GitHub username. Defaults to "__token__" if not provided.
--clone-dir PATH	Optional. Directory where target repos will be cloned (default: /tmp).
--dry-run	Optional. If set, changes are simulated but not committed or pushed. Logs changes to console.
--triggering-pr URL	Optional. Link to the PR which triggered this run. Included in the PR body if present.

---

📄 rock-ci-metadata.yaml Format

integrations:
  - consumer-repository: canonical/my-charm
    replace-image:
      - file: "metadata.yaml"
        path: "resources.my-rock.upstream-source"
      - file: "src/images.json"
        path: "config.batcher"
    service-spec:
      - file: "service-spec.json"
        user:
          path: "containers[0].user"
          value: "1001"
        command:
          path: "containers[0].command[1]"
          value: "/start"

• All file paths are relative to the repo root
• Paths can use dot and bracket notation for navigating YAML/JSON

---

🧪 Testing

Unit tests

tox -e unit

🔁 Integration tests

⚠️ These tests interact with a real GitHub repository and require a fine-grained GitHub token with appropriate permissions.

Required GitHub token permissions

The token must be a fine-grained personal access token (PAT) with:

• Repository access: Select the repository you're testing against
• Permissions:
  • Contents: Read and write
  • Pull requests: Read and write

These are needed to:

• Clone the repo
• Push branches
• Open and manage pull requests

---

Setup and run:

export CHACI_TEST_TOKEN=<your_token>
export CHACI_TEST_REPO="org/repo-name"
export CHACI_TEST_BASE_BRANCH="target-branch"

tox -e integration

The integration tests will:

• Clone the specified repository
• Create a temporary branch and pull request
• Validate the PR contents
• Clean up the branch and PR after execution

---

🧰 Development & Contributing

This project uses:

• tox for test environments
• pytest for testing
• black + ruff for linting

To run all checks locally:

tox -e lint,unit,integration

---

📁 Project Structure

File	Purpose
rock_integrator.py	Core logic for modifying files with images
git_client.py	Git and GitHub abstraction for PR workflow
rock_metadata_handler.py	Orchestration for multi-repo integration
rock_ci_metadata_models.py	Pydantic model for metadata schema validation
main.py	CLI entrypoint via click
templates/pr_body.md.j2	Jinja2 template for pull request bodies

---

🔒 License

This project is licensed under the Apache 2.0 License.

---

✍️ Authors

Built by the Canonical Charmed Kubeflow team.