vizzpy 0.8.0

Interactive Python call tree visualizer — upload a project, explore its call graph in the browser

VizzPy

Understand any Python codebase in seconds — no instrumentation, no code changes, no runtime required.

VizzPy statically analyzes Python source files and renders an interactive call graph showing which functions call which, grouped by module hierarchy. Works on code you can't or don't want to run.

Common uses:

• Drop it in CI to auto-generate an always-fresh architecture diagram committed alongside the code
• Use interactive mode during code reviews to see exactly what changed and what it touches
• Onboard to an unfamiliar codebase without reading every file

Features

• Interactive web UI — zoom, pan, drag nodes, collapse subtrees, dark mode, minimap
• Single-click node focus — click any node to highlight it (amber), its upstream callers (orange), and downstream callees (violet); click again or the background to clear; toggle Immediate / Deep trace to see one hop or the full call chain in both directions
• Hierarchical module grouping — app.services.orders is visually nested inside app.services inside app; depth-coded with distinct colors (blue → green → amber), consistent across all three output formats
• Two granularity levels — toggle between function-level and module-level graphs in the UI or via --level
• Three output formats — interactive browser UI, Mermaid markdown (CI-friendly, zero extra deps), and SVG (print-quality via Graphviz)
• Cross-module resolution — follows imports across files to draw edges between modules
• External libraries separated — stdlib and third-party calls shown in a distinct muted section; your code stays readable
• Pre-load a project — --serve --project ./myapp renders the graph before the browser opens
• Archive upload — drag-and-drop .zip, .tar.gz, .egg, or .whl in the web UI
• Docstring tooltips — hover any node to see the fully-qualified name and docstring (interactive and SVG)

Installation

Install only what you need:

Goal	Command
Mermaid headless (CI-friendly, zero extra deps)	pip install vizzpy
SVG headless	pip install 'vizzpy[svg]'
Interactive web UI	pip install 'vizzpy[serve]'
Everything	pip install 'vizzpy[all]'

SVG rendering additionally requires the Graphviz system package:

brew install graphviz   # macOS
apt install graphviz    # Debian/Ubuntu

Usage

Web UI

vizzpy --serve
# open http://127.0.0.1:8000

# pre-load a local project — graph renders immediately on page open
vizzpy --serve --project ./myproject

Upload a .zip, .tar.gz, .egg, or .whl, then click Analyze.

• Scroll / pinch to zoom · Drag background to pan · Drag nodes to rearrange
• Single-click a node to highlight it and its callers (orange) / callees (violet); click again or the background to clear
• Immediate / Deep trace button — switch between one-hop and full upstream + downstream chain highlighting
• Double-click a node to collapse/expand its downstream call subtree
• Hover a node to see its docstring
• Function view / Module view button — toggle granularity without re-uploading

Headless rendering

# Mermaid (default) — no extra dependencies
vizzpy --headless --project ./myproject                          # → myproject_call_tree.md
vizzpy --headless --project ./myproject --level module           # module-level graph
vizzpy --headless --project ./myproject --level both             # both files in one run

# SVG — requires Graphviz
vizzpy --headless --project ./myproject --format svg
vizzpy --headless --project ./myproject --format svg --level both

# Layout engine (Mermaid only)
vizzpy --headless --project ./myproject --layout elk     # default: better for nested subgraphs
vizzpy --headless --project ./myproject --layout dagre   # classic Mermaid layout

# Custom output path
vizzpy --headless --project ./myproject --format mermaid --output graph.md

Options

vizzpy --serve   [--host HOST] [--port PORT] [--project PROJECT_PATH]
vizzpy --headless --project PROJECT_PATH
                  [--format svg|mermaid]
                  [--level function|module|both]
                  [--layout elk|dagre]
                  [--output OUTPUT]

How it works

VizzPy uses Python's ast module to do a two-pass analysis of your source files:

1. Scope pass — collects every function and class method with its qualified name (module.ClassName.method)
2. Edge pass — walks each file again, resolves self.method(), cls.method(), imported names, and direct calls to project-internal functions

Results are rendered with dagre-d3 (vendored, works offline). Calls to external libraries are shown as muted leaf nodes — their internals are not traced.

Why static analysis instead of runtime tracing? Static analysis covers all code paths (not just exercised ones), works on untrusted or incomplete code, and adds zero dependencies to the analyzed project.

Development

git clone https://github.com/atulsaurav/vizzpy
cd vizzpy
pip install -e ".[dev]"

pytest                                          # run tests
pytest --cov=vizzpy --cov-report=term-missing   # with coverage

License

Apache-2.0