fix relay stream handling and key query logs

[mumingluan]

⚠️ 提交说明 / PR Notice

Important

• 请提供人工撰写的简洁摘要，避免直接粘贴未经整理的 AI 输出。

📝 变更描述 / Description

将流式响应的语义有效性与终止帧完整性分开处理：已经产生有效内容但缺少终止事件的请求继续结算，同时在流状态中保留异常；真正空回和不完整工具参数仍按错误处理。Claude 服务端工具调用现在会按 block index 聚合 input_json_delta，避免 Web Search 和 Code Execution 在首个参数分片处被错误截断。

Xuancat 密钥查询同时展示消费日志和错误日志，并显示状态及完整错误详情。MY_CHANGES_BACKUP.md 已同步记录这些保留差异。

🚀 变更类型 / Type of change

• 🐛 Bug 修复 (Bug fix) - 请关联对应 Issue，避免将设计取舍、理解偏差或预期不一致直接归类为 bug
• ✨ 新功能 (New feature) - 重大特性建议先通过 Issue 沟通
• ⚡ 性能优化 / 重构 (Refactor)
• 📝 文档更新 (Documentation)

🔗 关联任务 / Related Issue

• 无公开 Issue；问题由本地代理链路抓包复现。

✅ 提交前检查项 / Checklist

• 人工确认: 我已亲自整理并撰写此描述，没有直接粘贴未经处理的 AI 输出。
• 非重复提交: 我已搜索现有的 Issues 与 PRs，确认不是重复提交。
• Bug fix 说明: 该修复基于可复现的 Claude SSE 服务端工具流截断问题。
• 变更理解: 我已理解这些更改的工作原理及可能影响。
• 范围聚焦: 本 PR 未包含部署二进制、构建产物或前端备份目录。
• 本地验证: 已在本地运行并通过测试或手动验证，维护者可以据此复核结果。
• 安全合规: 代码中无敏感凭据，且符合项目代码规范。

📸 运行证明 / Proof of Work

go test ./relay/...
PASS

bun test src/features/xuancat-pages/__tests__/key-query-logs.test.ts
1 pass, 0 fail

bun run typecheck
PASS

[Copilot]

Pull request overview

This PR adjusts relay stream validation to decouple “semantic output exists” from “terminal frame received”, allowing billing/usage settlement to proceed for non-empty streams that end abnormally while still recording stream anomalies. It also updates the Xuancat key-query UI to include error logs with status + full details.

Changes:

• Split stream validation into ValidateOutput() vs terminal checks, and update OpenAI/Responses/Claude/Gemini stream handlers to record missing-terminal as a soft stream error instead of failing the request when semantic output exists.
• Fix Claude streamed server tool calls by treating server_tool_use like tool_use and associating input_json_delta by block index.
• Enhance Xuancat key query logs display (include error logs, show status badge and error details), with a small shared isKeyQueryLog() helper + tests.

Reviewed changes

Copilot reviewed 14 out of 14 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
web/src/features/xuancat-pages/utils.ts	Adds isKeyQueryLog() helper to centralize which log types appear in key query.
web/src/features/xuancat-pages/types.ts	Extends UsageLog to include content for displaying log details.
web/src/features/xuancat-pages/components/key-query-dialog.tsx	Shows success/error status and full content details; includes error logs in the list.
web/src/features/xuancat-pages/tests/key-query-logs.test.ts	Regression test for key-query log type filtering (consume + error, excludes unrelated).
web/src/features/system-settings/request-limits/token-rate-limit-section.tsx	Import reordering/formatting cleanup.
web/src/features/system-settings/request-limits/token-daily-rate-limit-section.tsx	Import reordering/formatting cleanup.
relay/responsevalidator/validator.go	Introduces ValidateOutput() and updates OpenAI/Claude/Responses stream semantics; includes server_tool_use tool tracking.
relay/responsevalidator/validator_test.go	Adds tests for output-vs-terminal separation and server tool streamed arguments.
relay/channel/openai/relay-openai.go	Accepts non-empty unterminated streams, records StreamStatus soft error instead of failing.
relay/channel/openai/relay_responses.go	Same as above for Responses streaming.
relay/channel/openai/chat_stream_test.go	End-to-end regression tests for billing semantic output without terminal and still rejecting empty streams.
relay/channel/gemini/relay-gemini.go	Records missing terminal finish reason as StreamStatus error rather than failing when output exists.
relay/channel/claude/relay-claude.go	Uses ValidateOutput() for semantic validation and records missing message_stop as StreamStatus error.
MY_CHANGES_BACKUP.md	Updates internal change log to reflect the preserved behavioral differences.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.