Mistral adapter error handling: Improve object validation and error messages for nested content

[ingo-eichhorst-ionos]

Before submitting your bug report

• I believe this is a bug. I'll try to join the Continue Discord for questions
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS: MacOS 15.2 (24C101) - M1 (ARM)
- Continue version: 0.9.251
- IDE version: VS Code - Version: 1.96.2 (Universal)
- Model: GPT-4o

- config.json:
  
  "models": [
    {
      "title": "codestral",
      "model": "codestral-latest",
      "provider": "mistral",
      "apiKey": "random_string"
     }  ...
  ]

Description

The Mistral adapter for Codestral (And properly other models as well) is not displaying a proper error.
Instead the not very helpful "Cannot read properties of undefined (reading 'content')" is displayed. IMHO the code should handle deep nested objects more gracefully and e.g. display a JSON stringify of the object if the object nesting is different than expected.

To reproduce

1. A mistral model like in the config above with a wrong API-Key.
2. Send any chat request

Log output

##### Completion options #####
{
  "contextLength": 32768,
  "model": "codestral-latest",
  "maxTokens": 4096
}

##### Request options #####
{}

##### Prompt #####
<system>


<user>

[Patrick-Erichsen]

This was a separate issue with our error logging that I believe has been resolved in latest pre-release. Thanks for the +1 on it though!

[AfterStories]

@Patrick-Erichsen

I encountered this problem again today, and it happened on the computer where the plugin was installed for the first time.

When I used Run Continue to run the plugin locally, it worked well. It was packaged into an installation package and ran well after installation in my conputer. This may be related to the session history that already exists locally.

But I sent the installation package to my colleague, who installed it in IntelliJ for the first time. When he sent the problem, the plug-in window flashed for a while, and no content was displayed. It looks like the streaming request was stopped and not completed.，If sent text multiple times, an error message that the content could not be read would show.

Strangely, when I cloned the latest main code test again in another directory and built it as an installation package, the stream request call was not invalid (the UI flickered and nothing was displayed), and it runs normally now.

But I don’t think it’s because the problem was fixed in the latest code. The previous fix for this problem was #3725

When I pulled the code, the code was #3790

That is to say, my current code already includes the changes of #3725

[AfterStories]

@Patrick-Erichsen
It looks like you guys tried to fix similar issue, I think my current error may have a different reason

After the first request stream was invalid for unknown reasons (nothing was displayed, the window flickered), and then I sent text again, the reading content error message appeared. （ I'll try to record a video later ）

It seems that something strange happened when I asked the question for the first time. After that, plugin want to update this history record, but it could not be updated :

In the plugin window after installation, I found the dev tools tool and saw the following error message when the reading content error:

According to the log information, I found the code:
gui\src\redux\slices\sessionSlice.ts

[AfterStories]

In fact, I had this problem in the earlier 0.83 version: the stream request seemed to be stopped instantly (the UI flickered and nothing was displayed)

Then I changed the directory to clone the code again, and ran it again, and there was no problem. I think this may be related to the running/packaging environment, and it seems not to be a problem with the code itself.

I think I should create a new issue to record this problem detail , because the root cause of the problem may not be that the content cannot be read, but it is just an incidental issue

[ingo-eichhorst-ionos]

This was a separate issue with our error logging that I believe has been resolved in latest pre-release. Thanks for the +1 on it though!

I tested and the issue no longer occurs. I get a very nice error message now 😃

[ingo-eichhorst-ionos]

@Patrick-Erichsen Thanks for your support!

@AfterStories Can you please open your comments in another issue, so we keep things nice and clean?