Add support for returning multiple messages from gr.ChatInterface chat function

[abidlabs]

This PR firstly allows returning a list from gr.ChatInterface, which gets rendered as multiple assistant messages. Secondly, it allows files to be returned from a chat function just by returning a tuple, which is not only simpler but also more consistent with how files are actually stored in the history. Also improves test coverage for API use.

See demo/chatinterface_echo_multimodal/run.py that demonstrates both of these changes.

[gradio-pr-bot]

🪼 branch checks and previews

•	Name	Status	URL
	Spaces	ready!	Spaces preview
	Website	ready!	Website preview
🦄	Changes	detected!	Details

---

Install Gradio from this PR

pip install https://gradio-pypi-previews.s3.amazonaws.com/f8a50cffcf0d957c7347e1efee3644b95bddbe04/gradio-5.9.1-py3-none-any.whl

Install Gradio Python Client from this PR

pip install "gradio-client @ git+https://github.com/gradio-app/gradio@f8a50cffcf0d957c7347e1efee3644b95bddbe04#subdirectory=client/python"

Install Gradio JS Client from this PR

npm install https://gradio-npm-previews.s3.amazonaws.com/f8a50cffcf0d957c7347e1efee3644b95bddbe04/gradio-client-1.8.0.tgz

Use Lite from this PR

<script type="module" src="https://gradio-lite-previews.s3.amazonaws.com/f8a50cffcf0d957c7347e1efee3644b95bddbe04/dist/lite.js""></script>

[gradio-pr-bot]

🦄 change detected

This Pull Request includes changes to the following packages.

Package	Version
gradio	minor

• Maintainers can select this checkbox to manually select packages to update.

With the following changelog entry.

Add support for returning multiple messages from gr.ChatInterface chat function

Maintainers or the PR author can modify the PR title to modify this entry.

Something isn't right?

• Maintainers can change the version label to modify the version bump.
• If the bot has failed to detect any changes, or if this pull request needs to update multiple packages to different versions or requires a more comprehensive changelog entry, maintainers can update the changelog file directly.

[aliabd]

The changes lgtm @abidlabs. theres just two things:

• I think you need to fix the end of the docstring for the fn parameter to reflect the changes.
• the $code_chatinterface_options breaks the page rendering because it contains

example_code = """
Here's the code I generated:

```python
def greet(x):
    return f"Hello, {x}!"
\```

Is this correct?
"""

the ```python is breaking the page. I've seen this before and because i didn't want to change the demo, my suggestion is to instead copy/paste the code here and just remove the backticks/code syntax.

[abidlabs]

Thanks @aliabd, addressed both of the comments above

[freddyaboulton]

Nice PR @abidlabs !

One comment is that instead of supporting tuples as files we should point developers towards using a gradio component like gr.File or gr.Image.

I tried converting the demo you wrote to the tuples case and it's resulting in an error because it's treating (None, "You wrote...") as a file. I guess we could support returning a list of lists but that conflicts with the parameter name type="tuples".

import gradio as gr

def echo_multimodal(message, history):
    response = []
    response.append((None, "You wrote: '" + message["text"] + "' and uploaded:"))
    for file in message.get("files", []):
        response.append((None, gr.File(value=file)))
    return response

demo = gr.ChatInterface(
    echo_multimodal,
    multimodal=True,
    type="tuples",
    textbox=gr.MultimodalTextbox(file_count="multiple"),
)

if __name__ == "__main__":
    demo.launch()

So I think it would break the ambiguity to use a gradio component.

[abidlabs]

Ah I see okay yeah let's avoid that for now. I'll revert the docs changes around the file changes

[abidlabs]

Should be ok now @freddyaboulton

[freddyaboulton]

We should add this logic to _stream_fn to support streaming. Otherwise LGTM!

[abidlabs]

Nice catch thank you!

[abidlabs]

Did some more refactoring to avoid duplication

[abidlabs]

And of course my refactor broke something

[abidlabs]

Thanks for the feedback @freddyaboulton @aliabd!