fix(opencode): filter empty text/reasoning parts when loading from database - #19311
Closed
erichasinternet wants to merge 1 commit into
Closed
fix(opencode): filter empty text/reasoning parts when loading from database#19311erichasinternet wants to merge 1 commit into
erichasinternet wants to merge 1 commit into
Conversation
Copy link
Copy Markdown
Sorry, something went wrong.
Copy link
Copy Markdown
Contributor
|
The following comment was made by an LLM, it may be inaccurate: The search results show that PR #19311 (the current PR) is related to PR #17742, but they are not duplicates — they are complementary:
According to the PR description, both should be merged as they provide complete coverage: prevention + recovery. No duplicate PRs found |
Sorry, something went wrong.
Copy link
Copy Markdown
Contributor
|
Thanks for updating your PR! It now meets our contributing guidelines. 👍 |
Sorry, something went wrong.
…tabase Empty text and reasoning parts with blank or whitespace-only text can be stored in the database during streaming (from text-start events that receive no deltas, or when streams are interrupted). These empty parts cause permanent ValidationException errors with providers like AWS Bedrock, especially when using LiteLLM proxy (@ai-sdk/openai-compatible). This fix adds defensive filtering at two levels: 1. When hydrating parts from database (filters on load) 2. When converting to model messages (filters during conversion) Both filters use .trim() to catch whitespace-only content. Fixes anomalyco#19309 Complements anomalyco#17742 (prevention) with recovery for existing corruption
erichasinternet
force-pushed
the
fix/filter-empty-parts-on-load
branch
from
a8eaa85 to
203e02f
Compare
March 26, 2026 15:43
6 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue for this PR
Closes #19309
Type of change
What does this PR do?
Adds defensive filtering when loading parts from the database to handle corrupted sessions with empty text/reasoning blocks. This complements PR #17742's prevention approach.
Problem
Users with already-corrupted databases (empty text blocks stored in part table) experience permanent ValidationException errors with AWS Bedrock, especially when using LiteLLM proxy (@ai-sdk/openai-compatible). PR #17742 prevents future corruption but doesn't fix existing corrupted data.
Solution
Filter empty text/reasoning parts at two levels:
Both use .trim() to catch whitespace-only content.
Why this approach?
Relationship to PR #17742
Both PRs should be merged for complete coverage:
How did you verify your code works?
New tests added (5 test cases in message-v2.test.ts):
All existing tests pass: 26 tests, 41 expect() calls, 0 failures
Typecheck passes: bun run typecheck completes with no errors
Real-world validation: Tested with database containing 26 empty parts - sessions load successfully without ValidationException
Screenshots / recordings
N/A - Backend logic change, no UI impact
Checklist