fix(ToolCallsAccumulator): fallback to 'unknown_tool' for null tool names and use stable synthetic IDs for streaming tool calls

[chickenlj]

Fixes #1237

[Copilot]

Pull request overview

This PR addresses a streaming-mode tool-calling bug (Issue #1237) where tool-call fragments could end up persisted with a null tool name and/or unstable identifiers, causing downstream conversion to skip ToolUseBlocks and breaking subsequent agent iterations.

Changes:

• Generate stable synthetic tool call IDs during OpenAI-like streaming using toolIndex (e.g., streaming_idx_0) when provider IDs are missing.
• Propagate the resolved tool call ID into fragment ToolUseBlocks (instead of emitting fragments with an empty id).
• Add a null-name fallback in ToolCallsAccumulator (unknown_tool) and extend test coverage for the streaming edge cases.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
agentscope-core/src/main/java/io/agentscope/core/formatter/openai/OpenAIResponseParser.java	Uses stable synthetic IDs (by tool index) and assigns IDs to fragment ToolUseBlocks to enable consistent merging across chunks.
agentscope-core/src/main/java/io/agentscope/core/agent/accumulator/ToolCallsAccumulator.java	Ensures built ToolUseBlocks always have a non-null name by falling back to unknown_tool.
agentscope-core/src/test/java/io/agentscope/core/formatter/openai/OpenAIResponseParserTest.java	Adds a test asserting stable synthetic IDs across streaming chunks by index.
agentscope-core/src/test/java/io/agentscope/core/agent/accumulator/ToolCallsAccumulatorTest.java	Adds a test ensuring fragment-only streams still produce a ToolUseBlock with a non-null name.

[Copilot]

When toolCallId is missing and toolIndex is also null, the fallback "streaming_" + System.currentTimeMillis() will generate a different synthetic ID for each chunk. That makes it impossible for downstream accumulators to reliably merge multiple chunks belonging to the same tool call (and can lead to duplicated/fragmented tool calls). Consider using a deterministic/stable fallback in this branch (e.g., reuse a per-parser last-seen synthetic ID for fragments, or derive one from choice index + position-in-toolCalls list) so repeated chunks map to the same tool call.

[codecov]

Codecov Report

❌ Patch coverage is 83.33333% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...pe/core/formatter/openai/OpenAIResponseParser.java	80.00%	0 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!

[LearningGp]

LGTM

[LearningGp]

[github-actions]

Closing due to inactivity. Feel free to reopen when ready.