Many-to-Many & Pivot

Many-to-many relations use a join table. The join table can be just a table name, or it can be modeled when it has business fields.

Relation definition

func (article Article) Tags() oro.Relation {
    return oro.ManyToMany(article, "Tags", "Tag").
        JoinTable("article_tags").
        ForeignKey("ArticleID").
        RelatedKey("TagID")
}

Pivot model

type ArticleTag struct {
    oro.Model
    ArticleID uint64
    TagID     uint64
    Position  int
}

func (ArticleTag) Define(s *oro.SchemaBuilder) {
    s.Table("article_tags")
    s.Field("ArticleID").UnsignedBigInt().Index()
    s.Field("TagID").UnsignedBigInt().Index()
    s.Field("Position").Int().Default(0)
    s.Unique("uk_article_tag", "ArticleID", "TagID")
}

Using a model keeps pivot fields visible and avoids special cases for schema sync.

Attach and sync

err := db.Relation(article.Tags()).Attach(ctx, tag, oro.Map{"position": 10})
err = db.Relation(article.Tags()).Sync(ctx, []*Tag{tagA, tagB})

Read pivot data

When pivot fields matter, query the pivot model directly:

links, err := db.Use[ArticleTag]().Where("ArticleID", article.ID).Get(ctx)

This keeps pivot data explicit instead of hiding it inside related models.