fix(formatter): preserve reasoning_content from ThinkingBlock in DashScopeMessageConverter

[uuuyuqi]

Description

When Qwen3 and similar models operate in thinking mode, assistant messages contain reasoning_content (chain-of-thought reasoning). AgentScope correctly parses this into ThinkingBlock internally via DashScopeResponseParser.

However, DashScopeMessageConverter did not preserve this reasoning_content when converting internal Msg objects back to DashScopeMessage DTOs for subsequent API calls (i.e., as conversation history). The reasoning context was silently dropped, causing the model to lose its thinking history in multi-turn conversations.

Root Cause

In DashScopeMessageConverter.convertToSimpleContent(), there was no logic to extract ThinkingBlock from the internal Msg and set it as reasoning_content on the outgoing DashScopeMessage.

Fix

For assistant messages, extract the first ThinkingBlock and set its content as reasoning_content on the DashScopeMessage.Builder:

ThinkingBlock thinkingBlock = msg.getFirstContentBlock(ThinkingBlock.class);
if (thinkingBlock != null) {
    String thinking = thinkingBlock.getThinking();
    if (thinking != null && !thinking.isEmpty()) {
        builder.reasoningContent(thinking);
    }
}

Changed files:

• DashScopeMessageConverter.convertToSimpleContent() — extract and preserve reasoning_content from ThinkingBlock

Close #1284

Checklist

• Code has been formatted with mvn spotless:apply
• All tests are passing (mvn test)
• Javadoc comments are complete and follow project conventions
• Related documentation has been updated (e.g. links, examples, etc.)
• Code is ready for review

Made with Cursor

[codecov]

Codecov Report

❌ Patch coverage is 80.00000% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...formatter/dashscope/DashScopeMessageConverter.java	80.00%	0 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!