fix: Improve stream error handling and parsing

[qnixsynapse]

Description

This PR significantly enhances the robustness and reliability of stream processing within the llamacpp-extension. Previously, certain errors transmitted via the streaming API from the llamacpp server might not have been correctly interpreted or propagated, leading to silent failures or unexpected behavior in the client.

Changes

• Enhanced Error Handling: Introduced explicit handling for stream lines that begin with error:. These lines are now parsed as JSON, and the contained error message is used to throw a proper JavaScript Error, ensuring that server-side errors are immediately recognized and dealt with.
• Centralized JSON Parsing: The logic for parsing JSON from both data: and error: prefixed lines has been unified. This ensures consistency and simplifies the code.
• Improved Error Propagation: Any errors encountered during the JSON parsing process, whether due to malformed data or server-originated errors, are now re-thrown. This guarantees that the asynchronous iterator consuming the stream will correctly terminate with an error, preventing the application from continuing with incomplete or corrupted data.
• Removed Debug Log: A superfluous console.log statement related to is_process_running has been removed, cleaning up the console output.

Why these changes?

• To fix commonly occuring out of context error dialog in Jan. In llamacpp extension, previously only a vague message was shown. Now the FE catches the errors and displays it properly.

Further work:

• update on FE side to fix the "increase context size" button

---

Important

Improves error handling and JSON parsing in llamacpp-extension's handleStreamingResponse function, ensuring robust error propagation and removing a debug log.

• Error Handling:
  • handleStreamingResponse in index.ts now explicitly handles lines starting with error: by parsing them as JSON and throwing a JavaScript Error.
  • Errors during JSON parsing are re-thrown to terminate the async iterator with an error.
• JSON Parsing:
  • Unified JSON parsing logic for data: and error: lines in handleStreamingResponse.
• Misc:
  • Removed console.log related to is_process_running in index.ts.

^{This description was created by for fa7fe8b. You can customize this summary. It will automatically update as commits are pushed.}

[ellipsis-dev]

Caution

Changes requested ❌

Reviewed everything up to 49af64f in 2 minutes and 22 seconds. Click for details.

• Reviewed 52 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 1 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. extensions/llamacpp-extension/src/index.ts:974

• Draft comment:
  Consider wrapping JSON.parse in the error branch with try-catch to provide clearer error context if malformed JSON is received.
• Reason this comment was not posted:
  Comment looked like it was already resolved.

Workflow ID: wflow_RpOAUJItZZktvF3a

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[github-actions]

Barecheck - Code coverage report

Total: 35.1%

Your code coverage diff: -0.01% ▾

✅ All code changes are covered

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed fa7fe8b in 2 minutes and 29 seconds. Click for details.

• Reviewed 36 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 3 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. extensions/llamacpp-extension/src/index.ts:875

• Draft comment:
  Removed the console.log(sInfo) call. This cleanup helps avoid leaking session info. If logging is needed, consider using a proper debug-level logger.
• Reason this comment was not posted:
  Comment looked like it was already resolved.

2. extensions/llamacpp-extension/src/index.ts:969

• Draft comment:
  Removed the console.log(trimmedLine) and console.log(jsonStr) debug statements in the streaming handler. This cleans up the output; ensure that any needed debugging info uses a proper logging strategy.
• Reason this comment was not posted:
  Comment did not seem useful. Confidence is useful = 0% <= threshold 50% The comment is purely informative, explaining what was done without suggesting any changes or asking for confirmation. It doesn't align with the rules for good comments, as it doesn't provide a specific suggestion or ask for a test.

3. extensions/llamacpp-extension/src/index.ts:976

• Draft comment:
  The new else branch that throws 'Malformed chunk' improves robustness by handling unexpected streamed data. Consider including the actual chunk data (with caution) for better diagnostics if appropriate.
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 20% vs. threshold = 50% The suggestion to include chunk data could help with debugging, but there are risks: 1) The chunk could contain sensitive data 2) The chunk could be very large 3) The error is for an edge case that "should not normally reach here". The current simple error is likely sufficient since this is an unexpected edge case. The comment correctly identifies an opportunity for better error diagnostics. Including the problematic data could help track down issues if they occur in production. While better diagnostics are good, the risks of exposing potentially sensitive or large chunks of data outweigh the debugging benefits for this edge case that should rarely occur. The comment should be deleted. The current simple error message is appropriate for this edge case, and including chunk data could expose sensitive information.

Workflow ID: wflow_ua94S8uAeUxnpzPN

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

[louis-menlo]

LGTM