refactor(cmd/gf): refactor command gf gen service with AST

[oldme-git]

Review Dear：
本次 PR 对 gf gen service 进行了部分重构，即使用 AST 替换 Regex 来解析 logic 文件。因为gf gen是一个比较重要功能，反馈的问题也比较多，所以此次提交比较重要！

性能方面

相较于多次正则不同，AST 可一次循环解析出所有需要的语法，例如 import, func, param, comment，所以性能得到了略微的提升。
以下是对1000个文件，若干方法的测试结果，单位秒，测试环境：go1.22.1 windows11/AMD Ryzen 5500U/16G memory
AST 三次：

4.0956484
3.862473
3.8385975

Regex 三次：

4.1110877
4.0835742
4.2559844

测试文件和方法：click me

改动简述

• 增加更为复杂的测试用例
• 增加 genservice_ast_parse.go。用来解析源文件；
• 增加 genservice_ast_utils.go。用来从 AST 原始节点中获取 gf-cli 所需的数据；
• 清理掉 genservice_calculate.go 中用于正则匹配源文件的方法；
• 调整生成逻辑 genservice_generate.go，使用 AST 获取的规范数据来生成 service;
• CGenService.Service 方法的部分重构，用以适应新的逻辑

fixed #3067
fixed #3492
fixed #2310
fixed #3104
fixed #3434
fixed #3474

[gqcn]

这块逻辑比较复杂，需要花点时间review。

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

This logic is relatively complex and will take some time to review.

[gqcn]

@oldme-git 我在项目中执行了一下，遇到生成方法重复的问题。我把文件作为附件上传，你可以尝试一下，建议加上一个类似的单例在提交代码中。

打包的logic最小文件：logic.zip

[oldme-git]

@oldme-git 我在项目中执行了一下，遇到生成方法重复的问题。我把文件作为附件上传，你可以尝试一下，建议加上一个类似的单例在提交代码中。

打包的logic最小文件：logic.zip

@gqcn

单测已经加上了。生成文件的那一块也整理了一下，去掉了因为正则不准确而又多添加的矫正代码。另外，在生成文件的地方进行了两个优化：

• 使用协程
• 使用 bytes.Buffer 替代字符串拼接，以提高性能。

最终性能得到了约 30% 的提升，以下是对 3000 个文件，若干方法测试的结果，单位秒，测试环境：go1.22.1 windows11/i7-13700k/64G memory

正则
38.7244941
33.9636121
45.8428419

AST
26.774159
21.3870076
31.7287746

在测试中，大部分的时间都用来 gofmt 生成好的文件。如果想获得更好的性能，可以对 gofmt 这一步进行优化，不过当前，也足够使用了

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@oldme-git I executed it in the project and encountered the problem of duplicate generation methods. I uploaded the file as an attachment. You can try it. It is recommended to add a similar singleton in the submission code.

Packaged logic minimum file: logic.zip

Single test has been added. The generated file has also been tidied up, and the correction code added due to inaccurate regular expressions has been removed. In addition, two optimizations have been made where the file is generated:

• Use coroutines
• Use bytes.Buffer instead of string concatenation to improve performance.

The final performance was improved by about 30%. The following are the results of testing 3000 files with several methods, unit seconds, test environment: go1.22.1 windows11/i7-13700k/64G memory

regular
38.7244941
33.9636121
45.8428419

AST
26.774159
21.3870076
31.7287746

In testing, most of the time is spent generating good files with gofmt. If you want to get better performance, you can optimize the gofmt step, but currently, it is enough.

[gqcn]

@oldme-git 有点小细节需要特殊处理一下，就是在logic中涉及到import service的调用中，需要做下替换。参考以下报错。这个是logic模块代码，通用分布式锁，不包含业务逻辑：
locker.zip

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@oldme-git There are some small details that need special treatment, that is, the calls to import service in logic need to be replaced. Refer to the following error report. This is the logic module code, a universal distributed lock, which does not contain business logic:
locker.zip

[oldme-git]

@oldme-git 有点小细节需要特殊处理一下，就是在logic中涉及到import service的调用中，需要做下替换。参考以下报错。这个是logic模块代码，通用分布式锁，不包含业务逻辑： locker.zip

可以了，强哥。

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@oldme-git There are some small details that need special treatment, that is, the calls to import service in logic need to be replaced. Refer to the following error report. This is the logic module code, a universal distributed lock, which does not contain business logic: locker.zip

That’s it, Brother Qiang.