Metadata-Version: 2.4
Name: tavor
Version: 1.4.1
Summary: Python SDK for Tavor cloud sandboxes
Project-URL: Homepage, https://github.com/tavor-dev/sdk-py
Project-URL: Issues, https://github.com/tavor-dev/sdk-py/issues
Project-URL: Documentation, https://tavor.dev
Author-email: Tavor <pkg@tavor.dev>
License: MIT
License-File: LICENSE
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Requires-Python: >=3.9
Requires-Dist: requests>=2.28.0
Requires-Dist: typing-extensions>=4.0.0; python_version < '3.10'
Provides-Extra: async
Requires-Dist: aiohttp>=3.8.0; extra == 'async'
Provides-Extra: dev
Requires-Dist: aiohttp>=3.8.0; extra == 'dev'
Requires-Dist: black>=23.0.0; extra == 'dev'
Requires-Dist: mypy>=1.0.0; extra == 'dev'
Requires-Dist: pytest-asyncio>=0.21.0; extra == 'dev'
Requires-Dist: pytest-cov>=4.0.0; extra == 'dev'
Requires-Dist: pytest>=7.0.0; extra == 'dev'
Requires-Dist: ruff>=0.1.0; extra == 'dev'
Requires-Dist: types-requests>=2.28.0; extra == 'dev'
Description-Content-Type: text/markdown

# Tavor Python SDK

Official Python SDK for [Tavor](https://tavor.dev), the cloud sandbox platform that provides secure, isolated execution environments.

## Installation

```bash
pip install tavor
```

For async support:

```bash
pip install tavor[async]
```

## Quick Start

```python
from tavor import Tavor

# Initialize the client
tavor = Tavor(api_key="sk-tavor-...")

# Create and use a sandbox
with tavor.box() as box:
    result = box.run("echo 'Hello from Tavor!'")
    print(result.stdout)  # "Hello from Tavor!"
```

## Features

- **Simple API**: Intuitive interface for creating and managing boxes
- **Automatic cleanup**: Boxes are automatically cleaned up when done
- **Streaming output**: Real-time command output streaming
- **Async support**: Full async/await support for concurrent operations
- **Type hints**: Complete type annotations for better IDE support
- **Error handling**: Comprehensive error handling with specific exception types

## Usage Examples

### Basic Command Execution

```python
from tavor import Tavor

tavor = Tavor(api_key="sk-tavor-...")

with tavor.box() as box:
    # Run simple commands
    result = box.run("pwd")
    print(f"Current directory: {result.stdout}")

    # Install packages
    box.run("pip install numpy pandas")

    # Run Python code
    result = box.run("python -c 'import numpy as np; print(np.array([1,2,3]))'")
    print(result.stdout)
```

### Error Handling

```python
from tavor import Tavor, CommandTimeoutError, TavorError

try:
    with tavor.box() as box:
        # This will timeout
        result = box.run("sleep 120", timeout=5)
except CommandTimeoutError as e:
    print(f"Command timed out: {e}")
except TavorError as e:
    print(f"Error: {e}")
```

### Streaming Output

```python
with tavor.box() as box:
    # Stream output as it's generated
    result = box.run(
        "for i in {1..5}; do echo \"Line $i\"; sleep 1; done",
        on_stdout=lambda line: print(f"[LIVE] {line}", end=""),
        on_stderr=lambda line: print(f"[ERROR] {line}", end="")
    )
```

### Custom Box Configuration

```python
from tavor import Tavor, BoxConfig

config = BoxConfig(
    cpu=2,
    mib_ram=2048,
    timeout=7200,  # 2 hours
    metadata={"project": "data-analysis"}
)

with tavor.box(config=config) as box:
    # Run resource-intensive tasks
    result = box.run("python train_model.py")
```

### Web Support

Boxes can expose services to the internet via public URLs. Each box gets a unique hostname, and you can access specific ports using the `get_public_url()` method:

```python
with tavor.box() as box:
    # Start a web server on port 8080
    box.run("python -m http.server 8080 &")

    # Wait for server to start
    box.run("sleep 2")

    # Get the public URL for port 8080
    public_url = box.get_public_url(8080)
    print(f"Access your server at: {public_url}")
    # Output: https://8080-uuid.tavor.app

    # The service is now accessible from anywhere on the internet
    box.run(f"curl {public_url}")
```

This feature is useful for:

- Testing webhooks and callbacks
- Sharing development servers temporarily
- Demonstrating web applications
- Running services that need to be accessible from external systems

### Port Exposing

You can dynamically expose ports from inside the sandbox to random external ports. This is useful when you need to access services running inside the sandbox but don't know the port in advance or need multiple services:

```python
with tavor.box() as box:
    # Start a service on port 3000 inside the sandbox
    box.run("python -m http.server 3000 &")

    # Give the server a moment to start
    box.run("sleep 2")

    # Expose the internal port 3000 to an external port
    exposed_port = box.expose_port(3000)

    print(f"Internal port {exposed_port.target_port} is now accessible on external port {exposed_port.proxy_port}")
    print(f"Port mapping expires at: {exposed_port.expires_at}")

    # You can now access the service using the proxy_port
    # For example: http://sandbox-hostname:proxy_port
```

The `expose_port` method returns an `ExposedPort` object with:

- `target_port` - The port inside the sandbox (what you requested)
- `proxy_port` - The external port assigned by the system
- `expires_at` - When this port mapping will expire

### Async Operations

```python
import asyncio
from tavor import AsyncTavor

async def run_parallel_tasks():
    async with AsyncTavor(api_key="sk-tavor-...") as tavor:
        async with tavor.box() as box:
            # Run multiple commands in parallel
            results = await asyncio.gather(
                box.run("task1.py"),
                box.run("task2.py"),
                box.run("task3.py")
            )

            for result in results:
                print(result.stdout)

asyncio.run(run_parallel_tasks())
```

### Manual Box Management

```python
# Create a box without automatic cleanup
box = tavor.create_box()

try:
    # Wait for box to be ready
    box.wait_until_ready()

    # Run commands
    result = box.run("echo 'Hello'")
    print(result.stdout)

    # Check box status
    print(f"Box status: {box.status}")
finally:
    # Don't forget to clean up!
    box.stop()
```

## API Reference

### Client Classes

- `Tavor`: Main client for synchronous operations
- `AsyncTavor`: Client for async operations (requires `pip install tavor[async]`)

### Configuration

- `BoxConfig`: Configuration for box creation
  - `cpu`: Number of CPUs to allocate (default: 1)
  - `mib_ram`: MiB RAM to allocate (default: 1024)
  - `timeout`: Maximum lifetime in seconds (default: 3600, or TAVOR_BOX_TIMEOUT env var)
  - `metadata`: Custom metadata dictionary

### Models

- `Box`: Represents a box instance
  - `hostname`: Public hostname for web access (e.g., `uuid.tavor.app`)
  - `get_public_url(port)`: Get the public URL for accessing a specific port
- `CommandResult`: Result of command execution
  - `stdout`: Command output
  - `stderr`: Error output
  - `exit_code`: Process exit code
  - `status`: Command status (QUEUED, RUNNING, DONE, FAILED, ERROR)
- `ExposedPort`: Result of exposing a port
  - `proxy_port`: External port assigned by the system
  - `target_port`: Port inside the sandbox
  - `expires_at`: When this port mapping expires

### Exceptions

- `TavorError`: Base exception for all SDK errors
- `APIError`: Base for API-related errors
- `AuthenticationError`: Invalid API key
- `BoxNotFoundError`: Box doesn't exist
- `CommandTimeoutError`: Command execution timeout
- `ValidationError`: Invalid request parameters

## Environment Variables

The SDK supports the following environment variables:

- `TAVOR_API_KEY`: Your API key (alternative to passing it in code)
- `TAVOR_BASE_URL`: API base URL (default: <https://api.tavor.dev>)
- `TAVOR_CPU`: CPU to allocate to the sandbox (default: 1)
- `TAVOR_MIB_RAM`: MiB RAM to allocate to the sandbox (default: 1024)
- `TAVOR_BOX_TIMEOUT`: Default box timeout in seconds (default: 3600)

Example:

```bash
export TAVOR_API_KEY="sk-tavor-..."
export TAVOR_BASE_URL="https://api.tavor.dev"
export TAVOR_BOX_TIMEOUT="7200"
export TAVOR_CPU="1"
export TAVOR_MIB_RAM="1024"

# Now you can initialize without parameters
python -c "from tavor import Tavor; client = Tavor()"
```

## Requirements

- Python 3.9+
- `requests` library
- `aiohttp` (optional, for async support)

## Support

- Documentation: <https://tavor.dev>
- Issues: <https://github.com/tavor-dev/sdk-py/issues>

## License

MIT License - see LICENSE file for details.
