Transform variable relationships into interactive web applications with real-time updates.

from rh import MeshBuilder

# Define relationships between variables
mesh_spec = {
    "temp_fahrenheit": ["temp_celsius"],
    "temp_kelvin": ["temp_celsius"],
}

# Define how to compute each relationship
functions_spec = {
    "temp_fahrenheit": "return temp_celsius * 9/5 + 32;",
    "temp_kelvin": "return temp_celsius + 273.15;",
}

# Set initial values
initial_values = {
    "temp_celsius": 20.0
}

# Create and build the app
builder = MeshBuilder(mesh_spec, functions_spec, initial_values)
app_path = builder.build_app(title="Temperature Converter")

# Serve it locally
builder.serve(port=8080)

• Bidirectional Dependencies: Variables can depend on each other cyclically
• Real-time Updates: Changes propagate instantly through the mesh
• Convention over Configuration: Smart defaults based on variable names
• Type Inference: Automatic UI widget selection from initial values
• Zero External Dependencies: Works with Python stdlib only

Variable names automatically determine UI behavior:

initial_values = {
    "slider_opacity": 50,        # → Range slider (0-100)
    "readonly_result": 0,        # → Read-only display
    "hidden_internal": 10,       # → Hidden field
    "color_theme": "#ff0000",    # → Color picker
    "date_created": "2023-01-01" # → Date input
}

# Physics calculator with custom field overrides
mesh_spec = {
    "kinetic_energy": ["mass", "velocity"],
    "momentum": ["mass", "velocity"],
    "total_energy": ["kinetic_energy", "potential_energy"]
}

functions_spec = {
    "kinetic_energy": "return 0.5 * mass * velocity * velocity;",
    "momentum": "return mass * velocity;",
    "total_energy": "return kinetic_energy + potential_energy;"
}

field_overrides = {
    "mass": {
        "title": "Mass (kg)",
        "minimum": 0.1,
        "maximum": 1000,
        "ui:help": "Object mass in kilograms"
    }
}

builder = MeshBuilder(mesh_spec, functions_spec, 
                     initial_values={"mass": 10, "velocity": 5, "potential_energy": 100},
                     field_overrides=field_overrides)

Run the test suite:

# Run pytest discovery from the project root (recommended)
python -m pytest

Try the examples:

python demo.py      # Multiple example apps
python example.py   # Simple temperature converter with server
python persistent_apps_example.py  # Shows app directory management

RH automatically manages app storage for persistent applications:

builder = MeshBuilder(mesh_spec, functions_spec, initial_values)

# App name inferred from title, stored in RH_APP_FOLDER
app_path = builder.build_app(title="Temperature Converter")
# Creates: ~/.rh/apps/temperature_converter/index.html

# Explicit app name
app_path = builder.build_app(title="My App", app_name="custom_name")
# Creates: ~/.rh/apps/custom_name/index.html

from rh.util import RH_APP_FOLDER, get_app_directory

# Check current app folder location
print(f"Apps stored in: {RH_APP_FOLDER}")

# Get path for specific app
app_dir = get_app_directory("my_calculator")

Control where RH stores apps by setting environment variables:

# Custom app folder location
export RH_APP_FOLDER="/path/to/my/apps"

# Custom local data folder (apps will be in $RH_LOCAL_DATA_FOLDER/apps)
export RH_LOCAL_DATA_FOLDER="/path/to/my/data"

For full control over output location:

builder = MeshBuilder(mesh_spec, functions_spec, initial_values)
builder.output_dir = "/path/to/specific/location"
app_path = builder.build_app(title="My App")

This section communicates the design principles and architectural decisions that guide the development of RH, helping both users understand the framework's approach and contributors align with the project's vision. Contributors are welcome!!!

🧩 Declarative over Imperative

• Users describe what relationships exist between variables, not how to update the UI
• The framework handles the complex orchestration of updates, event handling, and DOM manipulation
• Mental model: "I have variables that relate to each other" → "I get a working interactive app"

🔄 Convention over Configuration

• Smart defaults based on naming patterns (slider_*, readonly_*, hidden_*)
• Type inference from initial values (float → number input, bool → checkbox)
• Zero-config working apps, with escape hatches for customization when needed

⚡ Functional Programming & Immutability

• Pure functions with no side effects in core logic
• Immutable configuration objects generated once and used everywhere
• Predictable behavior through referential transparency
• Data transformations rather than object mutations

📐 Clean Architecture

┌─────────────────┐
│   MeshBuilder   │  ← Facade/Interface Layer
│    (Facade)     │
├─────────────────┤
│ Generators      │  ← Application Logic
│ • HTML          │
│ • RJSF Schema   │
├─────────────────┤
│ Core Logic      │  ← Business Logic
│ • Type Inference│
│ • Propagation   │
│ • Validation    │
├─────────────────┤
│ Infrastructure  │  ← Framework/Tools
│ • HTTP Server   │
│ • File I/O      │
│ • Templates     │
└─────────────────┘

🔧 Separation of Concerns

• Specification Layer: Parse and validate user input (mesh specs, functions)
• Configuration Layer: Transform specs into explicit, normalized config
• Generation Layer: Create output artifacts (HTML, JS, schemas) from config
• Runtime Layer: Serve applications and handle deployment

🎯 Single Source of Truth (SSOT)

• The generate_config() method produces the canonical representation
• All downstream components (HTML generator, RJSF schemas, JS functions) consume only this config
• No scattered state or multiple sources of truth

🔌 Dependency Injection & Plugin Architecture

• Pluggable components for different output formats
• Registry pattern for optional tool detection (Jinja2, esbuild, etc.)
• Interface-based design allowing custom generators and processors

📚 Zero Dependencies by Design

• Core functionality works with Python stdlib only
• Optional enhancements auto-detected and registered
• Reduces deployment complexity and increases reliability

🧪 Test-Driven Development

• Tests written first to clarify requirements and edge cases
• Comprehensive test coverage (24+ tests) ensuring behavioral correctness
• Integration tests verify end-to-end functionality

📖 Documentation as Code

• README examples are tested as part of the test suite
• Docstrings include type hints and behavioral descriptions
• Self-documenting code through clear naming and structure

🌱 Incremental Complexity

• Simple use cases work with minimal code
• Advanced features available through progressive disclosure
• Each abstraction level serves a clear purpose

🤝 What We Welcome

• New Generators: Support for other UI frameworks (Vue, Angular, Svelte)
• Input Parsers: YAML/TOML mesh specs, Excel formula parsing, natural language
• Enhanced Conventions: More naming patterns, automatic grouping, layout hints
• Performance Optimizations: Faster propagation algorithms, caching strategies
• Developer Experience: Better error messages, debugging tools, IDE integration

🎨 Code Style Expectations

• Functional first: Prefer pure functions and data transformations
• Type hints: All public APIs should be fully annotated
• Docstrings: Include behavior descriptions and simple doctests where applicable
• Modular design: Single responsibility principle, composable components
• Immutable data: Avoid mutation, prefer transformation and copying

🏗️ Architectural Consistency

• New features should extend existing patterns rather than introduce new paradigms
• Maintain the declarative user interface - complexity hidden behind simple APIs
• Follow the SSOT principle - all generators work from the same configuration
• Preserve zero-dependency core with optional enhancements

🔍 Testing Philosophy

• Write tests first to clarify the expected behavior
• Test both happy paths and edge cases
• Include integration tests for user-facing workflows
• Ensure examples in documentation actually work

This framework embodies the principle that complexity should be in the implementation, not the interface. Users describe simple relationships; the framework handles the complexity of turning those into rich, interactive applications.

MIT License - see LICENSE file for details.