perf: 优化较多模组下的搜索表现

[pynickle]

close #1701

[tangge233]

我在寻思要不要给 CE 加入 Rx.NET，这玩意能在部分时候减少许多代码🤔

[ruattd]

不对 再看一眼

请把搜索操作改成纯 Async Task 驱动而不是奇怪的混合，另外用 BeginInvoke 而不是 Invoke

[Copilot]

Pull request overview

This pull request optimizes search performance when dealing with many modules (issue #1701). The changes appear to focus on improving the search algorithm's efficiency in the Search function within ModBase.vb, likely addressing performance bottlenecks that became apparent when searching through large numbers of modules.

Key Changes (inferred from code analysis):

• Implementation of preprocessing and caching strategies in the search algorithm
• Addition of query preprocessing to avoid repeated string operations
• Introduction of absolute match detection for better search result prioritization

Reviewed changes

Copilot reviewed 1 out of 1 changed files in this pull request and generated no comments.

Since I don't have access to the actual diff of changed files in this PR, I cannot provide a detailed file-by-file review. The code I examined appears to be functional and includes what seem to be optimization-related enhancements to the search functionality, particularly in:

Area	Description
Search Algorithm	The Search function in ModBase.vb includes preprocessing steps and caching mechanisms to improve performance with many entries
Favorites Search	The PageDownloadCompFavorites.xaml.vb uses the optimized search function for filtering favorite items

Note: Without access to the actual PR diff showing which lines were added, modified, or removed, I cannot provide specific feedback on the changes made in this pull request. To provide a thorough review, I would need to see the actual diff that shows what code was changed.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ruattd]

哥们要是实在玩不懂 WPF 的异步就算了吧...

[pynickle]

哥们要是实在玩不懂 WPF 的异步就算了吧...

玩不会一点（恼）肯定是 WPF 太垃圾了

[ruattd]

玩不会一点（恼）肯定是 WPF 太垃圾了

WPF 确实垃圾，但是它的单线程多任务模型跟其他语言也没什么区别...
你的代码并没有完全利用上 Dispatcher 多任务的性能优势，反而增加了异步切换上下文的压力，不比完全同步的传统多线程操作性能高多少
回头我看看魔改一下吧....

[AMagicPear]

仅针对这个问题，从操作层面上看，既不需要对模组数量的多少分类讨论，也似乎不需要延时器。一个比较好的解决方案应该是：输入改变时异步启动一个搜索任务，并结束现在正在进行中的上一个搜索任务。当一个搜索任务完成时，更新UI。