frameplot 0.5.0

Turn Python-defined pipeline graphs into presentation-ready SVG and PNG diagrams.

frameplot

Turn Python-defined pipeline graphs into presentation-ready SVG and PNG diagrams.

한국어 README

frameplot is a compact Python library for rendering left-to-right pipeline diagrams with clean defaults. Define nodes, edges, groups, and optional detail panels in plain Python, then export polished SVG for documentation or high-resolution PNG for slides and papers.

Theme Gallery

All built-in presets stay on a white canvas. The same hero pipeline is rendered below once per theme so you can compare them directly.

Soft Retro	Retro	Pastel

Dark	Cyberpunk	Monochrome

Why frameplot?

• Clean and Professional: Left-to-right architecture diagrams with modern defaults.
• Diagram as Code: Define your pipeline in Python, get deterministic SVG/PNG outputs.
• Detail Panels: Unique feature to expand a summary node into a lower inset mini-graph for deep dives.
• Deep Customization: Fine-tune typography, spacing, colors, and corner radii via Theme.
• White-Canvas Themes: Built-in presets stay presentation-friendly on white backgrounds.
• Presentation Ready: High-quality SVG for web/docs and PNG for slides or papers.

Install

python -m pip install frameplot

PNG export depends on CairoSVG and may require Cairo or libffi packages from the host OS.

Quickstart

from frameplot import Edge, Group, Node, Pipeline

pipeline = Pipeline(
    nodes=[
        Node("start", "Start", "Receive request"),
        Node("fetch", "Fetch Data", "Load source tables"),
        Node("retry", "Retry", "Loop on transient failure", fill="#FFF2CC"),
        Node("done", "Done", "Return result", fill="#D9EAD3"),
    ],
    edges=[
        Edge("e1", "start", "fetch"),
        Edge("e2", "fetch", "retry", dashed=True),
        Edge("e3", "retry", "fetch", color="#C0504D"),
        Edge("e4", "fetch", "done"),
    ],
    groups=[
        Group("g1", "Execution", ["start", "fetch", "retry"], edge_ids=["e2"]),
    ],
)

svg = pipeline.to_svg()
pipeline.save_svg("pipeline.svg")
pipeline.save_png("pipeline.png")

Public API

Top-level imports are the supported public API:

• Node(id, title, subtitle=None, fill=None, stroke=None, text_color=None, metadata=None, width=None, height=None)
• Edge(id, source, target, color=None, dashed=False, metadata=None)
• Group(id, label, node_ids, edge_ids=(), stroke=None, fill=None, metadata=None)
• DetailPanel(id, focus_node_id, label, nodes, edges, groups=(), stroke=None, fill=None, metadata=None)
• Theme(...)
• Pipeline(nodes, edges, groups=(), detail_panel=None, theme=None)

Pipeline exposes:

• to_svg() -> str
• save_svg(path) -> None
• to_png_bytes(scale=4.0) -> bytes
• save_png(path, scale=4.0) -> None

Advanced Example: Multi-cloud Data Pipeline

The hero image at the top and the theme gallery above are generated from examples/theme_heroes.py, using the shared pipeline definition in examples/hero_pipeline.py. Together they showcase:

• Complex Routing: Seamlessly connecting AWS (S3/Lambda) to GCP (Pub/Sub/Dataflow) services.
• Contextual Details: Using a DetailPanel to explain the internal Spark Job Pipeline of the "Dataflow" node.
• Soft Retro Styling: Applying the built-in Theme.soft_retro() preset on a white canvas.

Design Notes

• Layout is intentionally left-to-right in v0.x.
• Edge labels are not supported yet.
• Groups stay visual overlays, and routes leaving or re-entering grouped nodes bend outside grouped areas.
• Detail panels render as separate lower insets attached to a focus node in the main flow.

Development

python -m venv .venv
source .venv/bin/activate
python -m pip install -e '.[dev]'
python -m pytest -q

Release publishing is automated through GitHub Actions and PyPI Trusted Publishing. Bump the version in pyproject.toml, create a tag like v0.4.0, and push the tag to trigger a release from .github/workflows/workflow.yml.