fix(database/gdb): performance improvement in fields grouping when in…

[wanghaolong613]

fix #3906

[Copilot]

Pull Request Overview

This PR improves performance in the database fields grouping logic during batch insert operations by reducing redundant operations. The optimization specifically addresses issue #3906 by caching field combinations and minimizing calls to the expensive fieldsToSequence method.

Key changes:

• Optimizes field grouping logic by pre-sorting keys and caching field combinations
• Reduces redundant fieldsToSequence calls by grouping items with identical field sets
• Adds benchmark test to measure batch insert performance improvements

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
database/gdb/gdb_core.go	Refactors DoInsert method to cache field combinations and reduce redundant operations
contrib/drivers/mysql/mysql_z_unit_core_bench_test.go	Adds benchmark test for batch insert operations to validate performance improvements

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[hailaz]

@wanghaolonggit 做了一点点修改，复查一下

[wanghaolong613]

@wanghaolonggit 做了一点点修改，复查一下

好

[wanghaolong613]

@wanghaolonggit 做了一点点修改，复查一下

@hailaz 修改的特别好，我另外加了一个测试用例

[wanghaolong613]

我还有个小问题，是否要保证最后执行插入的一条，是数据列表的最后一条数据。 LastInsertId() 查询到的 ID是最后一条数据的ID

[hailaz]

我还有个小问题，是否要保证最后执行插入的一条，是数据列表的最后一条数据。 LastInsertId() 查询到的 ID是最后一条数据的ID

底层分组之后的map是随机顺序的，看起来目前没法保证哪个是最后一条。

[wanghaolong613]

我还有个小问题，是否要保证最后执行插入的一条，是数据列表的最后一条数据。 LastInsertId() 查询到的 ID是最后一条数据的ID

底层分组之后的map是随机顺序的，看起来目前没法保证哪个是最后一条。

是的 确实目前保证不了