gorm文章、标签多对多create实现过程记录（包含踩坑）

基于《Go语言编程实战》第二章课后作业，做出tag和article的多对多关系映射

书中所写一对多感觉不准确，关系应该说多对多的关系。书中已经提前建立了blog_article_tag表，本文选择将表删掉，通过gorm标签建立多对多映射关系，自动生成数据表

首先在字段中添加多对多映射的部分：

type Article struct {
	*Model
	Title         string `json:"title"`
	Desc          string `json:"desc"`
	Content       string `json:"content" gorm:"type:longtext"`
	CoverImageUrl string `json:"cover_image_url"`
	State         uint8  `json:"state"`
	Tags          []Tag  `json:"tags" gorm:"many2many:blog_tag_article"`
}

type Tag struct {
	Model
	Name     string    `json:"name"`
	State    uint8     `json:"state"`
	Articles []Article `json:"articles" gorm:"many2many:blog_tag_article"`
}

在model的struct定义里，向两个struct里写上包含对方的struct列表。在后面gorm中声明“many2many:blog_tag_article”，blog_tag_article即将要建立的连接表的名称。

完成上述步骤后，在数据库初始化代码中加入db.AutoMigrate(&Article{},&Tag{})。AutoMigrate会根据struct声明的字段构造对应的数据库表，Article中指定了gorm:"type:longtext"可以直接设定数据库表中对应字段的类型，如果不设定则默认根据struct对应类型映射

更改请求接口struct部分代码：

type ArticleCreateRequest struct {
	Title    string `form:"title" binding:"required,min=1,max=100"`
	Desc     string `form:"desc" binding:"required,min=2,max=255"`
	Content  string `form:"content" binding:"required,min=1"`
	CreateBy string `form:"create_by" binding:"required,min=2,max=100"`
	State    uint8  `form:"state,default=1" binding:"oneof=0 1"`
	Tags     string `form:"tags" binding:"required,max=255"`
}

这里采用的方式是，在最后增加Tags字段，以string的方式传入，传入格式为“1-2-3-4”（代表标签id为1，2，3，4），在services层使用split处理为列表

func (svc *Service) CreateArticle(param *ArticleCreateRequest) error {
	tagList := strings.Split(param.Tags, "-")
	return svc.dao.CreateArticle(param.Title, param.Desc, param.Content, param.CreateBy, param.State, tagList)
}

更改dao层创建文章代码：

func (d *Dao) CreateArticle(title string, desc string, content string, create_by string, state uint8, tagListStr []string) error {
	var tagList []model.Tag
	var tag model.Tag
	for _, tagName := range tagListStr {
		tagid := convert.SrcTo(tagName).MustInt()
		//queryDB := d.engine.Where("is_del=?",0).Session()
		//d.engine.Where("id=?", tagid).Find(&tag)
		d.engine.Debug().Raw("select * from blog_tag where id = ?", tagid).Scan(&tag)
		tagList = append(tagList, tag)
	}
	for _, tagName := range tagList {
		fmt.Println(tagName.ID)
	}

	article := model.Article{
		Title:   title,
		Desc:    desc,
		Content: content,
		State:   state,
		Model:   &model.Model{CreateBy: create_by},
		Tags:    tagList,
	}
	return article.Create(d.engine)
}

坑：

一：理论上，在automigrate中已经建立了多对多关系，所以在创建新的article的时候，在对应的Tags字段输入对应的tag结构体即可自动插入文章表、标签表、以及标签文章关联表相关信息。但是如果按照标签名称插入：

article := model.Article{
		Title:   title,
		Desc:    desc,
		Content: content,
		State:   state,
		Model:   &model.Model{CreateBy: create_by},
		Tags:    model.Tag{Name:tag1}
	}

由于Name不是主键，所以如果文章1和文章2都有tag1，会造成在tag表中创建两个tag1，没有实现真正的多对多关系

于是选择使用标签主键进行映射，但是又产生问题，通过多对多映射生成的标签很多字段默认为空或者0，导致state默认为0，检索的时候没有办法检索到。

整个项目实现的思路应该为：服务端向客户提供文章标签，客户选择文章标签，在代码内部处理文章标签部分为对应的请求，我们这里只是模拟内部请求（感觉涉及到前端接口之类的问题了），最重要的是：tag表中的tags应该预先建好，通过tag的api建立

mysql> select * from blog_tag;
+----+------------+-------------+------------+-------------+------------+--------+--------+-------+
| id | create_by  | modified_by | created_on | modified_on | deleted_on | is_del | name   | state |
+----+------------+-------------+------------+-------------+------------+--------+--------+-------+
|  1 |            |             |          0 |  1685956876 |          0 |      0 |        |     0 |
|  2 |            |             |          0 |  1685956876 |          0 |      0 |        |     0 |
|  3 |            |             | 1685952118 |  1685956876 |          0 |      0 |        |     0 |
|  4 | temptation |             | 1685953063 |  1685956876 |          0 |      0 | tag666 |     1 |
|  5 | temptation |             | 1685956337 |  1685956876 |          0 |      0 | tag5   |     1 |
+----+------------+-------------+------------+-------------+------------+--------+--------+-------+

1、2、3是多对多关系中生成的，字段很多为0，4、5是通过api生成的，可用

再解释一下上面所说的“1、2、3是多对多关系中生成的”，一开始实现方式是在Article中生成新的tag实体插入taglist中进行建库，这样就自动生成了id123的条目，没有创建时间和state=1，并且无法建立关联。因为article实体中加入的Tags必须是从数据库中查询得到的对象，即已经存在于数据库中的条目。所以采用的方法是按照输入的tagid，从tag表中查到对应的条目插入taglist中，再随着article的建立构造映射关系

坑二：本来是通过for循环，db.engine.Where.First直接进行的查询，但是由于查询条件污染，后面的查询会带有前面所有查询成功的where条件。官方文档提供了新建Session的方法，但是不知道为什么db里没有对应的Session函数。

解决办法：直接使用原始sql语句进行查询：

for _, tagName := range tagListStr {
		tagid := convert.SrcTo(tagName).MustInt()
		//queryDB := d.engine.Where("is_del=?",0).Session()
		//d.engine.Where("id=?", tagid).Find(&tag)
		d.engine.Debug().Raw("select * from blog_tag where id = ?", tagid).Scan(&tag)
		tagList = append(tagList, tag)
	}

坑三：如果文章带有1，2，3，4四个标签，最后得到的映射关系只有4（最后一个）

+------------+--------+
| article_id | tag_id |
+------------+--------+
|         11 |      5 |
|         12 |      1 |
|         12 |      2 |
|         12 |      3 |
|         12 |      4 |
|         12 |      5 |

这里就是插入文章11的时候，只建立起来和tag5的映射。

原因是tag中的model一开始是*Model，可能因为指针传递等问题，导致最后所有的tag都成为了id为5的tag，将model.go里的tag中的*Model改为Model即可（即指针传递改为值传递），问题解决

请求格式 :   curl -X POST 'http://127.0.0.1:8080/api/v1/articles' -F "title={title}" -F "create_by={creater}" xxxxxxxxx.  -F "tags=1-2-3-4-5"（即给这个文章添加12345标签）

同时 更改get部分代码，加入预加载部分：

func (a Article) Get(db *gorm.DB) (*Article, error) {
	article := &Article{}
	db = db.Where("is_del=?", 0)

	if err := db.Preload("Tags").First(article, a.ID).Error; err != nil {
		return nil, err
	}
	return article, nil
}

func (t Tag) List(db *gorm.DB, pageOffset, pageSize int) ([]*Tag, error) {
	var tags []*Tag
	var err error
	if pageOffset >= 0 && pageSize > 0 {
		db = db.Offset(pageOffset).Limit(pageSize)
	}
	if t.Name != "" {
		db = db.Where("name = ?", t.Name)
	}
	db = db.Where("state = ?", t.State)
	if err = db.Preload("Articles").Where("is_del = ?", 0).Find(&tags).Error; err != nil {
		return nil, err
	}

	return tags, nil
}

因为默认查询情况下，struct list部分的属性默认是空的，所以需要查询时候Preload(struct list字段名)，在查询的时候将其关联的信息也查询回来

查询效果：

root@iZ0jl9uy0ja8cnm78j65glZ:~/project/go_blog/go-programming-tour-book/blog-service# curl -X GET 'http://127.0.0.1:8080/api/v1/articles/12'
{"id":12,"create_by":"temptation","modified_by":"","created_on":1685956876,"modified_on":1685956876,"deleted_on":0,"is_del":0,"title":"testtitle12","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":[{"id":1,"create_by":"","modified_by":"","created_on":0,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"","state":0,"articles":null},{"id":2,"create_by":"","modified_by":"","created_on":0,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"","state":0,"articles":null},{"id":3,"create_by":"","modified_by":"","created_on":1685952118,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"","state":0,"articles":null},{"id":4,"create_by":"temptation","modified_by":"","created_on":1685953063,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"tag666","state":1,"articles":null},{"id":5,"create_by":"temptation","modified_by":"","created_on":1685956337,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"tag5","state":1,"articles":null}]}

反向查询效果：

curl -X GET 'http://127.0.0.1:8080/api/v1/tags'
{"list":[{"id":4,"create_by":"temptation","modified_by":"","created_on":1685953063,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"tag666","state":1,"articles":[{"id":6,"create_by":"temptation","modified_by":"","created_on":1685953162,"modified_on":1685953162,"deleted_on":0,"is_del":0,"title":"testtitle6","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":null},{"id":9,"create_by":"temptation","modified_by":"","created_on":1685955812,"modified_on":1685955812,"deleted_on":0,"is_del":0,"title":"testtitle9","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":null},{"id":10,"create_by":"temptation","modified_by":"","created_on":1685956211,"modified_on":1685956211,"deleted_on":0,"is_del":0,"title":"testtitle10","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":null},{"id":12,"create_by":"temptation","modified_by":"","created_on":1685956876,"modified_on":1685956876,"deleted_on":0,"is_del":0,"title":"testtitle12","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":null}]},{"id":5,"create_by":"temptation","modified_by":"","created_on":1685956337,"modified_on":1685956876,"deleted_on":0,"is_del":0,"name":"tag5","state":1,"articles":[{"id":11,"create_by":"temptation","modified_by":"","created_on":1685956361,"modified_on":1685956361,"deleted_on":0,"is_del":0,"title":"testtitle11","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":null},{"id":12,"create_by":"temptation","modified_by":"","created_on":1685956876,"modified_on":1685956876,"deleted_on":0,"is_del":0,"title":"testtitle12","desc":"testdesc","content":"testcontent","cover_image_url":"","state":1,"tags":null}]}],"pager":{"page":1,"page_size":10,"total_rows":2}}

可以通过tag查询到article，也可以通过article查询到tag