DagSonar

Deep visibility into your Airflow task changes through AST parsing and tracking

What is DagSonar?

DagSonar is a monitoring tool that provides deep visibility into your Airflow DAG tasks by tracking changes through AST (Abstract Syntax Tree) parsing. It detects modifications in task definitions, external variables, shell scripts, and function calls, ensuring you never miss critical changes to your DAGs.

Key Features

• AST-Based Detection: Tracks changes by parsing the Abstract Syntax Tree of your DAG files
• Task Reference Tracking: Monitors task definitions, external variables, and function calls
• Shell Script Integration: Tracks associated shell scripts referenced in BashOperator tasks
• Change History: Maintains a JSON-based history of all task modifications
• Task Hash Generation: Generates unique hashes for each task state to detect changes
• Support for Multiple DAGs: Track tasks across multiple DAG configurations

Installation

pip install dagsonar

Basic Usage

from pathlib import Path
from dagsonar import TaskTracker, DagConfig

# Initialize the tracker
tracker = TaskTracker(history_file=Path("task_history.json"))

# Define your DAG configuration
config = {
    "tester": DagConfig(
        path=Path("./playground/dag_tester.py"),
        tasks=["task_bash_op"],
    )
}

# Track changes
changes = tracker.track_changes(config, auto_save=False)
print(changes)

Features in Detail

Task Reference Tracking

DagSonar tracks several aspects of your tasks:

• Task content and structure through AST
• External variable references
• Called functions
• Shell scripts referenced in bash tasks
• Task-specific hashes for change detection

Supported Task Types

Currently supports tracking of:

• Function-based task definitions
• BashOperator task instances
• Referenced shell scripts
• External variable dependencies

Configuration

DagConfig

from dagsonar import DagConfig
from pathlib import Path

config = DagConfig(
    path=Path("/path/to/dag.py"),  # Path to DAG file
    tasks=["task1", "task2"]       # Optional: List of specific tasks to track
)

Task History

Task history is stored in JSON format with the following structure:

[
  {
    "dag_id": "example_dag",
    "reference": {
      "dag_id": "example_dag",
      "task_history": [
        {
          "task_id": "task1",
          "content": "<ast_content>",
          "hash": "<computed_hash>",
          "external_variables": [],
          "called_functions": [],
          "shell_scripts": []
        }
      ]
    }
  }
]

Contributing

We welcome contributions! Please check out our Contributing Guide to get started.

Development Setup

1. Clone the repository:

git clone https://github.com/pesnik/dagsonar.git
cd dagsonar

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # Linux/Mac
# or
.\venv\Scripts\activate  # Windows

3. Install development dependencies:

pip install -e ".[dev]"

Running Tests

pytest tests/

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

Acknowledgments

• Apache Airflow community
• All contributors and users providing valuable feedback

Built for the Airflow community