daggr 0.2.0

A Python package

DAG-based Gradio workflows!

daggr is a Python library for building AI workflows that connect Gradio Spaces, ML models, and custom functions.

Installation

pip install daggr

Usage

Define nodes with inputs and outputs dicts. Inputs can be Gradio components (for UI), port references (for connections), or fixed values.

Basic Example

import gradio as gr
from daggr import Graph, FnNode, GradioNode

host_voice = GradioNode(
    space_or_url="Qwen/Qwen3-TTS",
    api_name="/generate_voice_design",
    inputs={
        "voice_description": gr.Textbox(label="Host Voice", value="Professional British voice..."),
        "language": "auto",
        "text": "Hello, welcome to the show!",
    },
    outputs={
        "audio": gr.Audio(label="Host Voice"),
        "status": gr.Text(visible=False),
    },
)

guest_voice = GradioNode(
    space_or_url="Qwen/Qwen3-TTS",
    api_name="/generate_voice_design",
    inputs={
        "voice_description": gr.Textbox(label="Guest Voice", value="Friendly American voice..."),
        "language": "auto",
        "text": "Thanks for having me!",
    },
    outputs={
        "audio": gr.Audio(label="Guest Voice"),
        "status": gr.Text(visible=False),
    },
)

def generate_dialogue(topic: str, host_voice: str, guest_voice: str) -> dict:
    return {"dialogue": f"Topic: {topic}", "metadata": {"host": host_voice, "guest": guest_voice}}

dialogue = FnNode(
    fn=generate_dialogue,
    inputs={
        "topic": gr.Textbox(label="Topic", value="AI in healthcare"),
        "host_voice": host_voice.audio,
        "guest_voice": guest_voice.audio,
    },
    outputs={
        "dialogue": gr.JSON(label="Dialogue"),
        "metadata": gr.Markdown(label="Metadata"),
    },
)

graph = Graph(name="Podcast Generator", nodes=[host_voice, guest_voice, dialogue])
graph.launch()

Input Types

• Gradio component: Creates a UI input (e.g., gr.Textbox(label="Topic"))
• Port reference: Connects to another node's output (e.g., other_node.output_name)
• Fixed value: Constant value, no UI (e.g., "auto")

Node Types

• GradioNode: Calls a Gradio Space API
• FnNode: Runs a Python function
• InferenceNode: Calls HuggingFace Inference API

Development

pip install -e ".[dev]"

Code Formatting

ruff check --fix --select I && ruff format

License

MIT License