Gorm Make LIKE query using struct fields

Your Question

I want to run a LIKE query using gorm. Following the documentation, the only way is to use raw SQL query as explained in this page: https://gorm.io/docs/query.html --> db.Where("name LIKE ?", "%jin%").Find(&users)

The problem is that using this command I need to use real database field names. So, the advantage to use an ORM is erased (different DB could use different naming conventions, like uppercase or other).

How can I run LIKE queries using struct fields and no real database field names?

Comment From: muharamdani

I have the same question, for now how you handle it? @lgtti

Comment From: lgtti

Hi,

i have mapped db fields with struct fields using this:

type BaseDAO struct {
    FieldCache map[string]string
}

func NewBaseDAO(model interface{}) BaseDAO {
    dao := BaseDAO{
        FieldCache: make(map[string]string),
    }

    s, err := schema.Parse(model, &sync.Map{}, schema.NamingStrategy{})
    if err != nil {
        panic("Failed to setup basedao field name cache")
    }

    for _, field := range s.Fields {
        dao.FieldCache[field.Name] = field.DBName
    }

    return dao
}

func (dao BaseDAO) GetByFilter(ctx fw.Context, output interface{}, filters map[string]interface{}) error {
    db := ctx.DB().(*gorm.DB)
    for k, v := range filters {
        dbName := dao.FieldCache[k]
        switch v.(type) {
        case []string:
            db = db.Where(dbName+" IN ?", v)
        case []uint:
            db = db.Where(dbName+" IN ?", v)
        case []int:
            db = db.Where(dbName+" IN ?", v)
        case int:
            db = db.Where(dbName+" = ?", v)
        case uint:
            db = db.Where(dbName+" = ?", v)
        case time.Time:
            db = db.Where(dbName+" = ", v)
        default:
            db = db.Where(dbName+" LIKE ?", v)
        }
    }
    return db.Find(output).Error
}

You can invoke NewBaseDAO with: xx := NewBaseDAO(myEntity{}) and use the struct field names in the GetByFilter input map

PS: actually i have tested it only for postgres. I hope it works for every db :)