Visualization Frameworks Integration

Batuta provides ecosystem visualization for Python data visualization and ML demo frameworks, showing how they map to sovereign Rust replacements. The batuta viz command displays framework hierarchies and PAIML replacement mappings.

Core Principle

Python visualization frameworks are replaced by sovereign Rust alternatives. No Python runtime dependencies are permitted in the PAIML stack. Python code is transpiled to Rust via Depyler.

Framework Replacement Matrix

Python Framework	PAIML Replacement	Migration Path
Gradio	Presentar	Depyler transpilation
Streamlit	Presentar	Depyler transpilation
Panel	Trueno-Viz	Depyler transpilation
Dash	Presentar + Trueno-Viz	Depyler transpilation
Matplotlib	Trueno-Viz	Direct API mapping
Plotly	Trueno-Viz	Direct API mapping

Toyota Way Principles

Principle	Application
Genchi Genbutsu	Direct visualization enables first-hand observation
Poka-Yoke	Python interpreter eliminated from production
Heijunka	Frame-rate limiting prevents GPU saturation
Jidoka	Explicit component trees for predictable rendering
Muda	Signal-based rendering eliminates wasted computation
Kanban	Visual data flow with explicit signal graphs

CLI Usage

View All Frameworks

batuta viz tree

Output:

VISUALIZATION FRAMEWORKS ECOSYSTEM
==================================

GRADIO (Python) → Presentar (Rust)
├── Interface
│   └── Interface → Presentar::QuickApp
├── Blocks
│   └── Blocks → Presentar::Layout
├── Components
│   ├── Image → Trueno-Viz::ImageView
│   ├── Audio → Presentar::AudioPlayer
│   ├── Chatbot → Realizar + Presentar
│   └── DataFrame → Trueno-Viz::DataGrid
└── Deployment
    └── HuggingFace Spaces → Batuta deploy

STREAMLIT (Python) → Presentar (Rust)
├── Widgets
│   ├── Input → Presentar::Widgets
│   └── Display → Presentar + Trueno-Viz
├── Caching
│   ├── @st.cache_data → Trueno::TensorCache
│   └── session_state → Presentar::State
└── Deployment
    └── Streamlit Cloud → Batuta deploy
...

Filter by Framework

batuta viz tree --framework gradio
batuta viz tree --framework streamlit
batuta viz tree --framework panel
batuta viz tree --framework dash

View PAIML Replacement Mappings

batuta viz tree --integration

Output:

PAIML REPLACEMENTS FOR PYTHON VIZ
=================================

UI FRAMEWORKS
├── [REP] Presentar::QuickApp ← gr.Interface
├── [REP] Presentar::Layout ← gr.Blocks
├── [REP] Presentar::App ← dash.Dash
├── [REP] Presentar::Layout ← st.columns/sidebar

VISUALIZATION
├── [REP] Trueno-Viz::Chart ← dcc.Graph
├── [REP] Trueno-Viz::Chart ← st.plotly_chart
├── [REP] Trueno-Viz::DataGrid ← st.dataframe
├── [REP] Trueno-Viz::GPURaster ← datashader

COMPONENTS
├── [REP] Presentar::TextInput ← st.text_input
├── [REP] Presentar::Slider ← st.slider
├── [REP] Trueno-Viz::ImageView ← gr.Image

STATE & CACHING
├── [REP] Presentar::State ← st.session_state
├── [REP] Trueno::TensorCache ← @st.cache_data
├── [REP] Presentar::on_event ← @callback

DEPLOYMENT
├── [REP] Batuta deploy ← HuggingFace Spaces
├── [REP] Batuta deploy ← Streamlit Cloud
├── [REP] Batuta deploy ← Dash Enterprise

Legend: [REP]=Replaces (Python eliminated)

Summary: 21 Python components replaced by sovereign Rust alternatives
         Zero Python dependencies in production

JSON Output

batuta viz tree --format json
batuta viz tree --framework streamlit --format json
batuta viz tree --integration --format json

Why Replace Python Frameworks?

Gradio → Presentar

Problems with Gradio:

Python server restarts on every interaction
~2s cold start time
~100ms interaction latency
No offline capability

Presentar Benefits:

Persistent state with sub-millisecond updates
~50ms cold start
~16ms interaction latency (60fps)
WebAssembly deployment for edge/offline

Streamlit → Presentar

Problems with Streamlit:

Full script reruns on each interaction (Muda)
~3s cold start, ~200ms latency
~8MB bundle size
~200MB memory usage

Presentar Benefits:

Signal-based reactivity (minimal DOM updates)
Compile-time type checking
~500KB bundle size
~20MB memory usage

Panel → Trueno-Viz

Problems with Panel:

6+ HoloViz dependencies (Panel, HoloViews, Datashader, Bokeh, Param, Colorcet)
WebGL rendering (older API)
Python GIL contention

Trueno-Viz Benefits:

Single unified library
Native WebGPU rendering
Rust memory safety for big data
Billion-point rendering capability

Dash → Presentar + Trueno-Viz

Problems with Dash:

Callback spaghetti (invisible data dependencies)
Large Plotly.js bundle
WebGL performance limits

Presentar + Trueno-Viz Benefits:

Explicit signal graph (debuggable)
Smaller WASM bundle
WebGPU for maximum performance

Performance Comparison

Metric	Gradio	Streamlit	Dash	Presentar
Cold start	~2s	~3s	~1s	~50ms
Interaction	~100ms	~200ms	~80ms	~16ms
Bundle size	~5MB	~8MB	~3MB	~500KB
Memory	~150MB	~200MB	~100MB	~20MB
GPU	No	No	WebGL	WebGPU
Offline	No	No	No	Yes
WASM	No	No	No	Yes

Component Mapping Reference

Gradio Components

Gradio	Presentar/Trueno-Viz
`gr.Interface`	`Presentar::QuickApp`
`gr.Blocks`	`Presentar::Layout`
`gr.Image`	`Trueno-Viz::ImageView`
`gr.Audio`	`Presentar::AudioPlayer`
`gr.Chatbot`	`Realizar + Presentar`
`gr.DataFrame`	`Trueno-Viz::DataGrid`

Streamlit Components

Streamlit	Presentar/Trueno-Viz
`st.write`	`Presentar::Text`
`st.dataframe`	`Trueno-Viz::DataGrid`
`st.plotly_chart`	`Trueno-Viz::Chart`
`st.text_input`	`Presentar::TextInput`
`st.slider`	`Presentar::Slider`
`st.selectbox`	`Presentar::Select`
`st.session_state`	`Presentar::State`
`@st.cache_data`	`Trueno::TensorCache`

Dash Components

Dash	Presentar/Trueno-Viz
`dash.Dash`	`Presentar::App`
`dcc.Graph`	`Trueno-Viz::Chart`
`dcc.Input`	`Presentar::TextInput`
`dash_table`	`Trueno-Viz::DataGrid`
`@callback`	`Presentar::on_event`

The Batuta Book