sayou-visualizer 0.0.10

Visualizer components for the Sayou Data Platform

sayou-visualizer

The Interactive Observability Engine for Sayou Fabric.

sayou-visualizer provides a transparent layer to monitor and visualize the execution flow of Sayou components. By attaching to any pipeline, it transforms invisible execution logs into intuitive, interactive HTML Knowledge Graphs.

It separates the logic of Observation (Tracer) from Presentation (Renderer), allowing you to debug complex pipelines and visualize data lineage without modifying your core logic.

💡 Core Philosophy

"Trace the Process, Render the Insight."

Observability should not be an afterthought. We decouple the responsibility into two roles:

1. Tracer (Recorder): The "Camera". It silently observes events (on_start, on_finish) from the pipeline via the Callback system and builds an internal graph structure.

2. Renderer (Painter): The "Canvas". It takes the recorded graph and generates human-readable artifacts (e.g., Interactive HTML, Static Images).

This separation enables a Non-intrusive Monitoring experience, where visualization is just a plug-and-play feature.

📦 Installation

pip install sayou-visualizer

⚡ Quick Start

The VisualizerPipeline acts as a facade, easily attaching to other pipelines to generate reports.

from sayou.connector.pipeline import ConnectorPipeline
from sayou.visualizer.pipeline import VisualizerPipeline

def main():
    connector = ConnectorPipeline()

    viz = VisualizerPipeline()
    viz.attach_to(connector)

    print("🚀 Running Pipeline...")
    iterator = connector.run("http://example.com")
    for packet in iterator:
        print(f"Processed: {packet.task.uri}")

    print("🎨 Generating Report...")
    viz.report("examples/pipeline_flow.html")

if __name__ == "__main__":
    main()

🔑 Key Concepts

Tracers

• GraphTracer: Listens to pipeline events and constructs a NetworkX Directed Acyclic Graph (DAG) in real-time. It distinguishes between Components (Generator/Fetcher) and Data (Tasks).

Renderers

• PyVisRenderer: Converts the internal graph into an interactive HTML file powered by Vis.js. Features physics-based layout and modern dark UI.
• RichRenderer (Planned): Displays a tree-structure summary directly in the console using the Rich library.

🤝 Contributing

We welcome contributions for new Renderers (e.g., MatplotlibRenderer, StreamlitRenderer) or specialized Tracers for new components!

📜 License

Apache 2.0 License © 2025 Sayouzone