waldiez 0.1.15

waldiez

Waldiez

Translate a Waldiez flow:

To a python script or a jupyter notebook with the corresponding ag2 agents and chats.

Features

• Convert .waldiez flows to .py or .ipynb
• Run a .waldiez flow
• Provide a custom IOSStream to handle input and output.

Installation

On PyPI:

python -m pip install waldiez

From the repository:

python -m pip install git+https://github.com/waldiez/waldiez.git

Usage

CLI

# Convert a Waldiez flow to a python script or a jupyter notebook
waldiez convert --file /path/to/a/flow.waldiez --output /path/to/an/output/flow[.py|.ipynb]
# Convert and run the script, optionally force generation if the output file already exists
waldiez run --file /path/to/a/flow.waldiez --output /path/to/an/output/flow[.py] [--force]

Using docker/podman

CONTAINER_COMMAND=docker # or podman
# pull the image
$CONTAINER_COMMAND pull waldiez/waldiez
# Convert a Waldiez flow to a python script or a jupyter notebook
$CONTAINER_COMMAND run \
  --rm \
  -v /path/to/a/flow.waldiez:/flow.waldiez \
  -v /path/to/an/output:/output \
  waldiez/waldiez convert --file /flow.waldiez --output /output/flow[.py|.ipynb] [--force]

# with selinux and/or podman, you might get permission (or file not found) errors, so you can try:
$CONTAINER_COMMAND run \
  --rm \
  -v /path/to/a/flow.waldiez:/flow.waldiez \
  -v /path/to/an/output:/output \
  --userns=keep-id \
  --security-opt label=disable \
  waldiez/waldiez convert --file /flow.waldiez --output /output/flow[.py|.ipynb] [--force]

# Convert and run the script
$CONTAINER_COMMAND run \
  --rm \
  -v /path/to/a/flow.waldiez:/flow.waldiez \
  -v /path/to/an/output:/output \
  waldiez/waldiez run --file /flow.waldiez --output /output/output[.py]

UI

For creating-only (no exporting or running) waldiez flows, you can use the playground at https://waldiez.github.io.
The repo for the js library is here.
We are currently working on waldiez-studio to provide a visual interface for creating and running Waldiez flows (you can find more here).
Until then, you can use our Jupyter or the VSCode extension to create and run Waldiez flows.

As a library

Export a flow

# Export a Waldiez flow to a python script or a jupyter notebook
from waldiez import WaldiezExporter
flow_path = "/path/to/a/flow.waldiez"
output_path = "/path/to/an/output.py"  # or .ipynb
exporter = WaldiezExporter.load(flow_path)
exporter.export(output_path)

Run a flow

# Run a flow
from waldiez import WaldiezRunner
flow_path = "/path/to/a/flow.waldiez"
output_path = "/path/to/an/output.py"
runner = WaldiezRunner.load(flow_path)
runner.run(output_path=output_path)

Run a flow with a custom IOStream

# Run the flow with a custom IOStream
# In case the standard 'input' and 'print' functions cannot be used
import time
import threading

from typing import Any

from waldiez import WaldiezRunner
from waldiez.io import WaldiezIOStream

flow_path = "/path/to/a/flow.waldiez"
output_path = "/path/to/an/output.py"


def custom_print_function(*args: Any, sep: str = " ", **kwargs: Any) -> None:
    """Custom print function."""
    print(*args, sep=sep, **kwargs)


# Custom input handler
class InputProcessorWrapper:
    """Wrapper input processor.
    
    To manage the interaction between the custom input processor and IOStream.
    """

    def __init__(self):
        self.stream = None  # Placeholder for the WaldiezIOStream instance
        self.lock = threading.Lock()  # Ensure thread-safe operations

    def custom_input_processor(self, prompt: str) -> None:
        """Simulate external input and send it back to the IOStream."""
        def external_input_simulation():
            with self.lock:  # Ensure thread-safe access
                time.sleep(2)  # Simulate delay for network input
                if self.stream:
                    self.stream.set_input("Simulated external input")
                else:
                    raise RuntimeError("Stream reference not set!")

        threading.Thread(target=external_input_simulation, daemon=True).start()

    def set_stream(self, stream: "WaldiezIOStream"):
        """Set the WaldiezIOStream instance."""
        with self.lock:  # Ensure thread-safe setting of the stream
            self.stream = stream

processor_wrapper = InputProcessorWrapper()

stream = WaldiezIOStream(
    input_timeout=30,
    print_function=
    on_prompt_input=processor_wrapper.custom_input_processor,
)

# Link the processor wrapper to the WaldiezIOStream instance
processor_wrapper.set_stream(custom_stream)

with WaldiezIOStream.set_default(io_stream):
    runner = WaldiezRunner.load(flow_path)
    runner.run(stream=io_stream, output_path=output_path)

Tools

• ag2 (formerly AutoGen)
• juptytext
• pydantic
• typer

Known Conflicts

• autogen-agentchat: This package conflicts with ag2 / pyautogen. Ensure that autogen-agentchat is uninstalled before installing waldiez. If you have already installed autogen-agentchat, you can uninstall it with the following command:

  pip uninstall autogen-agentchat -y
  

  If already installed waldiez you might need to reinstall it after uninstalling autogen-agentchat:

  pip install --force --no-cache waldiez pyautogen
  

License

This project is licensed under the MIT License - see the LICENSE file for details.