fix: 修复 And 识别器引用多结果 ROI 时 Or 子识别未按候选框短路的问题 fix #1313

[overflow65537]

Summary by Sourcery

改进复合识别逻辑，以正确处理带有 ROI 依赖的 AND/OR 组合，以及推测性子识别结果缓存。

Bug Fixes:

• 修复 AND 识别器：当某个子识别返回多个候选 ROI 时，后续的 OR 子识别仍能正确进行短路，而不会将失败分支的结果泄漏到运行时缓存中。

Enhancements:

• 抽取通用的子识别执行逻辑为可复用的辅助函数，由 AND 和 OR 组合器共同使用。
• 引入对识别参数的递归分析，以检测依赖特定 ROI 目标的嵌套 AND/OR 子识别，从而在 AND 分支中更好地指导候选框的传递。
• 添加带显式刷新机制的推测性子识别结果缓冲，避免将失败 AND 分支的中间结果写入正式缓存，同时仍能登记成功子节点的详细结果。

Original summary in English

Summary by Sourcery

Improve composite recognition logic to correctly handle AND/OR combinations with ROI dependencies and speculative sub-recognition caching.

Bug Fixes:

• Fix AND recognizer so that when a sub-recognition returns multiple candidate ROIs, subsequent OR sub-recognitions still short-circuit correctly without leaking failed branch results into the runtime cache.

Enhancements:

• Factor out shared sub-recognition execution into a reusable helper used by both AND and OR combinators.
• Introduce recursive analysis of recognition parameters to detect nested AND/OR sub-recognitions that depend on specific ROI targets, guiding candidate box propagation in AND branches.
• Add speculative sub-recognition result buffering with explicit flushing to avoid committing intermediate results from failed AND branches while still registering successful sub-node details.

Bug 修复：

• 确保 And 识别器在某个子识别返回多个候选 ROI 时，能够正确评估各分支，同时在包含 Or 子识别和 ROI 依赖关系的情况下，仍然保持短路求值行为。

增强功能：

• 提取通用的子识别调用逻辑到一个共享的辅助方法中，由 And 和 Or 组合器共同使用。
• 引入对识别参数的递归分析，以检测嵌套的 And/Or 子识别之间的 ROI 目标依赖关系，从而防止错误的候选框传播。

Original summary in English

Summary by Sourcery

改进复合识别逻辑，以正确处理带有 ROI 依赖的 AND/OR 组合，以及推测性子识别结果缓存。

Bug Fixes:

• 修复 AND 识别器：当某个子识别返回多个候选 ROI 时，后续的 OR 子识别仍能正确进行短路，而不会将失败分支的结果泄漏到运行时缓存中。

Enhancements:

• 抽取通用的子识别执行逻辑为可复用的辅助函数，由 AND 和 OR 组合器共同使用。
• 引入对识别参数的递归分析，以检测依赖特定 ROI 目标的嵌套 AND/OR 子识别，从而在 AND 分支中更好地指导候选框的传递。
• 添加带显式刷新机制的推测性子识别结果缓冲，避免将失败 AND 分支的中间结果写入正式缓存，同时仍能登记成功子节点的详细结果。

Original summary in English

Summary by Sourcery

Improve composite recognition logic to correctly handle AND/OR combinations with ROI dependencies and speculative sub-recognition caching.

Bug Fixes:

• Fix AND recognizer so that when a sub-recognition returns multiple candidate ROIs, subsequent OR sub-recognitions still short-circuit correctly without leaking failed branch results into the runtime cache.

Enhancements:

• Factor out shared sub-recognition execution into a reusable helper used by both AND and OR combinators.
• Introduce recursive analysis of recognition parameters to detect nested AND/OR sub-recognitions that depend on specific ROI targets, guiding candidate box propagation in AND branches.
• Add speculative sub-recognition result buffering with explicit flushing to avoid committing intermediate results from failed AND branches while still registering successful sub-node details.

[sourcery-ai]

你好——我已经查看了你的修改，看起来很棒！

---

Sourcery 对开源项目免费——如果你喜欢我们的代码审查，请考虑分享 ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{帮我变得更有用！请在每条评论上点 👍 或 👎，我会根据你的反馈改进之后的审查。}

Original comment in English

Hey - I've reviewed your changes and they look great!

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[sourcery-ai]

Hey - 我在这里给出了一些总体反馈：

• 在 and_ 中围绕 cache_sub_results_/pending_sub_results_ 的猜测性缓存逻辑，在多个返回路径上都涉及手动快照和对 resize 的调用；建议把这部分封装进一个 RAII 辅助工具中（例如，一个在作用域结束时自动回滚或提交的 guard），以减少将来新增早返回路径时忘记恢复状态的风险。
• 在 recognition_param_uses_roi_target 中，requires { value.roi_target; } 这个检查有些脆弱，它依赖于参数变体上一个不明显的隐式接口；你可能需要改成对已知、具有 roi_target 的参数类型进行显式处理，从而让依赖分析更加健壮，也更易于维护。

给 AI Agent 的提示

Please address the comments from this code review:

## Overall Comments
- The speculative caching logic around `cache_sub_results_`/`pending_sub_results_` in `and_` involves manual snapshots and `resize` calls on multiple return paths; consider encapsulating this in an RAII helper (e.g., a guard that automatically rolls back or flushes) to reduce the risk of future early-return paths forgetting to restore state.
- In `recognition_param_uses_roi_target`, the `requires { value.roi_target; }` check is somewhat brittle and relies on a non-obvious implicit interface of param variants; you may want to replace this with explicit handling of the known param types that have `roi_target` to make the dependency analysis more robust and easier to maintain.

---

Sourcery 对开源项目是免费的——如果你觉得我们的评审有帮助，欢迎分享 ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{帮我变得更有用！请在每条评论上点 👍 或 👎，我会根据你的反馈改进后续的评审。}

Original comment in English

Hey - I've left some high level feedback:

• The speculative caching logic around cache_sub_results_/pending_sub_results_ in and_ involves manual snapshots and resize calls on multiple return paths; consider encapsulating this in an RAII helper (e.g., a guard that automatically rolls back or flushes) to reduce the risk of future early-return paths forgetting to restore state.
• In recognition_param_uses_roi_target, the requires { value.roi_target; } check is somewhat brittle and relies on a non-obvious implicit interface of param variants; you may want to replace this with explicit handling of the known param types that have roi_target to make the dependency analysis more robust and easier to maintain.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- The speculative caching logic around `cache_sub_results_`/`pending_sub_results_` in `and_` involves manual snapshots and `resize` calls on multiple return paths; consider encapsulating this in an RAII helper (e.g., a guard that automatically rolls back or flushes) to reduce the risk of future early-return paths forgetting to restore state.
- In `recognition_param_uses_roi_target`, the `requires { value.roi_target; }` check is somewhat brittle and relies on a non-obvious implicit interface of param variants; you may want to replace this with explicit handling of the known param types that have `roi_target` to make the dependency analysis more robust and easier to maintain.

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}