pysilu 0.1.6

A simple AI programming language

Silu Programming Language

斯路编程语言

A simple AI-first interpreted programming language with Python-like syntax, designed as a foundation for advanced AI reasoning capabilities.

Vision & AI Roadmap

Silu aims to become a platform for developing AI systems that can:

• Combine rigorous logical reasoning with probabilistic reasoning
• Achieve self-bootstrapping and self-correction capabilities
• Conduct autonomous hypothesis testing and experimentation
• Discover inconsistencies and drive curiosity-based learning

Development Stages

• 2025-2028: Logic mastery (CFG, data flow, SMT verification)
• 2028-2032: Logic + probabilistic reasoning fusion
• 2032-2036: Self-bootstrapping and self-correction
• 2036-2040: Hypothesis-driven experimentation
• 2040+: Inconsistency detection and curiosity-driven knowledge creation

See AI Development Plan for detailed roadmap and benchmarks.

Quick Start

Installation

# Using pip
pip install -e .

# Using uv (recommended for development)
uv sync && uv pip install -e .

Basic Usage

# Run a Silu program
silu hello.si

# Check if Python code is compatible with Silu
silu check program.py

# Interactive execution
silu interpret --source "print('Hello, World!')"

# Generate LLVM IR for native compilation
silu ir hello.si --output hello.ir.json
silu llvm hello.ir.json --output hello.ll

Compilation to Native Code

Silu now supports compilation to native machine code via LLVM IR:

# Complete compilation workflow
silu ir program.si --output program.ir.json  # Generate Silu IR
silu llvm program.ir.json --output program.ll # Convert to LLVM IR
llc program.ll -o program.s                   # Compile to assembly
clang program.s -o program                    # Link to executable
./program                                     # Run native executable

Hello World Example

Create hello.si:

# Variables and basic operations
name = "Silu"
version = 1.0

print("Hello from", name, version)

# Functions
def greet(person):
    return "Welcome to " + person + "!"

message = greet("Silu Programming")
print(message)

Run it:

silu hello.si

Key Features

• Simple Syntax: Python-like syntax that's easy to learn
• Multiple Execution Modes:
  • Direct interpretation
  • Python compatibility checking
  • IR (Intermediate Representation) generation and execution
  • Symbolic execution for program analysis
• Rich Data Types: int, float, str, bool, bytes, lists, dictionaries
• Control Flow: if/elif/else, while loops, for loops with range()
• Advanced For Loops: Support for tuple unpacking (for key, value in dict.items())
• Functions: User-defined functions with parameters, return values, and closures
• Built-in Functions: print(), type(), isinstance(), type conversions, len()
• Dictionary Operations: Full dictionary support with .keys(), .values(), .items() methods

Advanced Usage

Check Python Compatibility

# Check if a Python file can be interpreted by Silu
silu check program.py
# Output: program.py ok (or program.py bad: specific error reason)

# Examples of error messages:
# program.py bad: syntax error: expected ':' at line 4
# program.py bad: name error: Name 'undefined_var' is not defined  
# program.py bad: runtime error: division by zero
# program.py bad: type error: can only concatenate str (not "int") to str
# program.py bad: empty file
# program.py bad: file not found

Generate Intermediate Representation

silu ir program.si --format json --output program.ir

Symbolic Execution Analysis

silu symbolic program.si

Execute IR Directly

silu exec program.ir

LLVM IR Generation and Native Compilation

# Generate LLVM IR from Silu IR
silu llvm program.ir.json --output program.ll

# Complete compilation pipeline
silu ir program.si --output program.ir.json    # Generate Silu IR
silu llvm program.ir.json --output program.ll  # Convert to LLVM IR
llc program.ll -o program.s                    # Compile to assembly
clang program.s -o program                     # Link to executable

# Run demo workflow
python demo_silu_to_llvm.py --save-files

Documentation

• AI Development Plan - Long-term AI reasoning roadmap and benchmarks
• CLI Usage Guide - Complete command-line reference
• Language Features - Syntax and examples
• Development Guide - Setup, testing, and contributing
• Symbolic Execution - Program analysis capabilities
• IR Design - Intermediate representation format
• LLVM IR Generation - Native code compilation via LLVM
• Interpreter Specification - Core implementation details

Examples

Demo Programs

The repository includes comprehensive demo programs showcasing all language features:

# Run the main demo (variables, functions, control flow)
silu demo.si

# Complex programming patterns
silu complex_example.si

# Symbolic execution analysis
silu symbolic symbolic_demo.si

Performance Comparison

Compare different execution modes:

# Interpretation vs IR execution
time silu interpret demo.si
silu ir demo.si --output demo.ir.json --format json
time silu exec demo.ir.json

Contributing

We welcome contributions! See the Development Guide for setup instructions and coding guidelines.

License

MIT License