streamshape 0.1.0

A unified LLM response parser with streaming support for multiple providers

Unified LLM Interface

A Python package that provides a simple, consistent interface for interacting with multiple Large Language Model (LLM) providers. Write once, run anywhere - switch between OpenAI, Anthropic, Google, and other providers without changing your code.

Key Features

✨ 5 Output Modes

• Normal text generation
• Streaming text
• Function calling (tool use)
• Structured output (validated Pydantic objects)
• Streaming structured output

🔌 6 Supported Providers

• OpenAI (GPT-4, GPT-3.5)
• Anthropic (Claude)
• Google (Gemini)
• OpenRouter (100+ models)
• xAI (Grok)
• Any OpenAI-compatible endpoint (Ollama, LM Studio, etc.)

🛡️ Type-Safe & Validated

• Full type hints for IDE autocomplete
• Pydantic validation for structured outputs
• Clear, consistent error messages

🎯 Simple & Consistent

• Same API across all providers
• Built on battle-tested libraries (LiteLLM, Pydantic)
• Minimal code to get started

Quick Start

Installation

pip install -r requirements.txt

Or install from source:

git clone <repository-url>
cd output_parsar
pip install -e .

Basic Usage

from src import OpenAI
from pydantic import BaseModel

# Initialize provider
client = OpenAI(api_key="your-api-key")

# 1. Generate text
response = client.generate(
    model="gpt-4",
    system_prompt="You are a helpful assistant.",
    user_prompt="What is Python?"
)
print(response["data"])

# 2. Stream text
for chunk in client.stream(
    model="gpt-4",
    system_prompt="You are a storyteller.",
    user_prompt="Tell me a short story"
):
    print(chunk["data"], end="", flush=True)

# 3. Structured output
class Recipe(BaseModel):
    name: str
    ingredients: list[str]
    steps: list[str]

result = client.structured_output(
    model="gpt-4",
    system_prompt="You are a chef.",
    user_prompt="Give me 2 simple pasta recipes",
    output_schema=Recipe
)

for recipe in result["data"]:
    print(f"Recipe: {recipe.name}")
    print(f"Ingredients: {', '.join(recipe.ingredients)}")

Switch Providers Easily

# Just change the import - same API!

from src import OpenAI
client = OpenAI(api_key="sk-...")

# Or use Anthropic
from src import Anthropic
client = Anthropic(api_key="sk-ant-...")

# Or Google
from src import Google
client = Google(api_key="...")

# Same code works for all providers!

Documentation

📚 Complete documentation available in the /docs folder:

• Quick Start Guide - Get up and running in 5 minutes
• Installation Guide - Detailed installation instructions
• Usage Guide - Comprehensive examples for all methods
• API Reference - Complete API documentation
• Provider Configuration - Provider-specific setup
• Examples - Real-world usage examples
• Error Handling - Understanding and handling errors
• Best Practices - Tips and recommendations

Five Ways to Use LLMs

1. Generate (Complete Text)

response = client.generate(
    model="gpt-4",
    system_prompt="You are a helpful assistant.",
    user_prompt="Explain quantum computing in simple terms",
    temperature=0.7
)
print(response["data"])

2. Stream (Real-time Text)

for chunk in client.stream(
    model="gpt-4",
    system_prompt="You are a helpful assistant.",
    user_prompt="Write a short story",
    temperature=0.7
):
    print(chunk["data"], end="", flush=True)

3. Tool Call (Function Calling)

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Get current weather",
        "parameters": {
            "type": "object",
            "properties": {
                "location": {"type": "string"}
            }
        }
    }
}]

result = client.tool_call(
    model="gpt-4",
    system_prompt="You are a helpful assistant.",
    user_prompt="What's the weather in Paris?",
    tools=tools
)

print(result["data"]["tool_name"])    # "get_weather"
print(result["data"]["arguments"])    # {"location": "Paris"}

4. Structured Output (Complete)

from pydantic import BaseModel

class Task(BaseModel):
    title: str
    priority: str
    estimated_hours: int

result = client.structured_output(
    model="gpt-4",
    system_prompt="You are a project manager.",
    user_prompt="Create 5 tasks for building a website",
    output_schema=Task
)

for task in result["data"]:
    print(f"Task: {task.title} (Priority: {task.priority})")

5. Structured Streaming Output

for task in client.structured_streaming_output(
    model="gpt-4",
    system_prompt="You are a project manager.",
    user_prompt="Create 10 tasks for building a mobile app",
    output_schema=Task
):
    print(f"New task: {task.title} (Priority: {task.priority})")

Real-World Examples

Check out the /real_use directory for complete working examples:

• All Methods Usage - Demonstrates all 5 output modes
• Structured Output - Complete structured output examples
• Streaming Structured Output - Real-time structured data

Testing

Run the test suite to verify everything works:

# Run all tests
python -m pytest tests/

# Run specific test files
python -m pytest tests/test_normalizer.py
python -m pytest tests/test_base_provider.py
python -m pytest tests/test_streaming_structered_output.py

Tests use mock payloads and Hypothesis for property-based testing, so no API calls are made during testing.

Environment Variables

Store your API keys securely using environment variables:

# .env file
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GOOGLE_API_KEY=...
OPENROUTER_API_KEY=sk-or-...
XAI_API_KEY=...

Load them in your code:

import os
from dotenv import load_dotenv
from src import OpenAI

load_dotenv()
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

Project Structure

output_parsar/
├── src/                          # Main package source code
│   ├── base.py                   # Base provider class
│   ├── providers.py              # Provider implementations
│   ├── exceptions.py             # Custom exceptions
│   ├── litellm_integration.py    # LiteLLM integration
│   ├── parser_integration.py     # Parser integration
│   └── streaming_structured_output_parser/  # Streaming parser
├── docs/                         # Complete documentation
├── tests/                        # Test suite
├── real_use/                     # Real-world examples
├── dummy/                        # Demo scripts
└── requirements.txt              # Dependencies

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is open source and available under the MIT License.

Support

• Check the documentation for detailed guides
• Review examples for common use cases
• See error handling for troubleshooting
• Open an issue on GitHub for bugs or feature requests