Go语言ORM框架 自动生成CRUD代码，附库源码

A safer orm base on GORM, aims to be developer friendly.

Overview

• CRUD or DIY query method code generation
• Auto migration from database to code
• Transactions, Nested Transactions, Save Point, RollbackTo to Saved Point
• Competely compatible with GORM
• Developer Friendly
• Multiple Generate modes

Contents

• GORM/GEN
• Overview
• Contents
• Installation
• Quick start
• Project Directory
• API Examples
• Generate
• Generate Model
• Field Expression
• Create Field
• CRUD API
• Create
• Create record
• Create record with selected fields
• Batch Insert
• Query
• Retrieving a single object
• Retrieving objects with primary key
• Retrieving all objects
• Conditions
• String Conditions
• Inline Condition
• Not Conditions
• Or Conditions
• Group Conditions
• Selecting Specific Fields
• Tuple Query
• JSON Query
• Order
• Limit & Offset
• Group By & Having
• Distinct
• Joins
• SubQuery
• From SubQuery
• Update from SubQuery
• Transaction
• Nested Transactions
• Transactions by manual
• SavePoint/RollbackTo
• Advanced Query
• Iteration
• FindInBatches
• Pluck
• Scopes
• Count
• FirstOrInit
• FirstOrCreate
• Association
• Relation
• Relate to exist model
• Relate to table in database
• Relate Config
• Operation
• Skip Auto Create/Update
• Find Associations
• Append Associations
• Replace Associations
• Delete Associations
• Clear Associations
• Count Associations
• Delete with Select
• Preloading
• Preload
• Preload All
• Preload with conditions
• Nested Preloading
• Update
• Update single column
• Updates multiple columns
• Update selected fields
• Delete
• Delete record
• Delete with primary key
• Batch Delete
• Soft Delete
• Find soft deleted records
• Delete permanently
• DIY method
• Method interface
• Syntax of template
• placeholder
• template
• If clause
• Where clause
• Set clause
• Method interface example
• Smart select fields
• Advanced Topics
• Hints
• Maintainers
• Contributing
• License

Installation

To install Gen package, you need to install Go and set your Go workspace first.

1.The first need Go installed(version 1.14+ is required), then you can use the below Go command to install Gen.

go get -u gorm.io/gen

2.Import it in your code:

import "gorm.io/gen"

Quick start

Emphasis: All use cases in this doc are generated under WithContext mode. And if you generate code under WithoutContext mode, please remove WithContext(ctx) before you call any query method, it helps you make code more concise.

# assume the following code in generate.go file$ cat generate.go

package mainimport "gorm.io/gen"// generate codefunc main() {    // specify the output directory (default: "./query")    // ### if you want to query without context constrain, set mode gen.WithoutContext ###    g := gen.NewGenerator(gen.Config{        OutPath: "../dal/query",        /* Mode: gen.WithoutContext|gen.WithDefaultQuery*/        //if you want the nullable field generation property to be pointer type, set FieldNullable true        /* FieldNullable: true,*/        //if you want to generate index tags from database, set FieldWithIndexTag true        /* FieldWithIndexTag: true,*/        //if you want to generate type tags from database, set FieldWithTypeTag true        /* FieldWithTypeTag: true,*/        //if you need unit tests for query code, set WithUnitTest true        /* WithUnitTest: true, */    })      // reuse the database connection in Project or create a connection here    // if you want to use GenerateModel/GenerateModelAs, UseDB is necessray or it will panic    // db, _ := gorm.Open(mysql.Open("root:@(127.0.0.1:3306)/demo?charset=utf8mb4&parseTime=True&loc=Local"))    g.UseDB(db)      // apply basic crud api on structs or table models which is specified by table name with function    // GenerateModel/GenerateModelAs. And generator will generate table models' code when calling Excute.    g.ApplyBasic(model.User{}, g.GenerateModel("company"), g.GenerateModelAs("people", "Person", gen.FieldIgnore("address")))        // apply diy interfaces on structs or table models    g.ApplyInterface(func(method model.Method) {}, model.User{}, g.GenerateModel("company"))    // execute the action of code generation    g.Execute()}

Generate Mode:

• gen.WithoutContext generate code without WithContext contrain
• gen.WithDefaultQuery generate code with a default global variable Q as a singleton

Project Directory

Here is a template for best practices:

demo├── cmd│   └── generate│       └── generate.go # execute it will generate codes├── dal│   ├── dal.go # create connections with database server here│   ├── model│   │   ├── method.go # DIY method interfaces│   │   └── model.go  # store struct which corresponding to the database table│   └── query  # generated code's directory|       ├── user.gen.go # generated code for user│       └── gen.go # generated code├── biz│   └── query.go # call function in dal/gorm_generated.go and query databases├── config│   └── config.go # DSN for database server├── generate.sh # a shell to execute cmd/generate├── go.mod├── go.sum└── main.go

API Examples

Generate

Generate Model

// generate a model struct map to table `people` in databaseg.GenerateModel("people")// generate a struct and specify struct's nameg.GenerateModelAs("people", "People")// add option to ignore fieldg.GenerateModel("people", gen.FieldIgnore("address"), gen.FieldType("id", "int64"))

Field Generate Options

FieldNew         // create new fieldFieldIgnore      // ignore fieldFieldIgnoreReg   // ignore field (match with regexp)FieldRename      // rename field in structFieldType        // specify field typeFieldTypeReg     // specify field type (match with regexp)FieldTag         // specify gorm and json tagFieldJSONTag     // specify json tagFieldGORMTag     // specify gorm tagFieldNewTag      // append new tagFieldTrimPrefix  // trim column prefixFieldTrimSuffix  // trim column suffixFieldAddPrefix   // add prefix to struct member's nameFieldAddSuffix   // add suffix to struct member's nameFieldRelate      // specify relationship with other tablesFieldRelateModel // specify relationship with exist models

Field Expression

Create Field

Actually, you're not supposed to create a new field variable, cause it will be accomplished in generated code.

Create field examples:

import "gorm.io/gen/field"// create a new generic field map to `generic_a`a := field.NewField("table_name", "generic_a")// create a field map to `id`i := field.NewInt("user", "id")// create a field map to `address`s := field.NewString("user", "address")// create a field map to `create_time`t := field.NewTime("user", "create_time")

CRUD API

Here is a basic struct user and struct DB.

// generated code// generated code// generated codepackage queryimport "gorm.io/gen"// struct map to table `users` type user struct {    gen.DO    ID       field.Uint    Name     field.String    Age      field.Int    Address  field.Field    Birthday field.Time}// struct collectiontype DB struct {    db       *gorm.DB    User     *user}

Create

Create record

// u refer to query.useruser := model.User{Name: "Modi", Age: 18, Birthday: time.Now()}u := query.Use(db).Usererr := u.WithContext(ctx).Create(&user) // pass pointer of data to Createerr // returns error

Create record with selected fields

Create a record and assign a value to the fields specified.

u := query.Use(db).Useru.WithContext(ctx).Select(u.Name, u.Age).Create(&user)// INSERT INTO `users` (`name`,`age`) VALUES ("modi", 18)

Create a record and ignore the values for fields passed to omit

u := query.Use(db).Useru.WithContext(ctx).Omit(u.Name, u.Age).Create(&user)// INSERT INTO `users` (`Address`, `Birthday`) VALUES ("2021-08-17 20:54:12.000", 18)

Batch Insert

To efficiently insert large number of records, pass a slice to the Create method. GORM will generate a single SQL statement to insert all the data and backfill primary key values.

var users = <>model.User{{Name: "modi"}, {Name: "zhangqiang"}, {Name: "songyuan"}}query.Use(db).User.WithContext(ctx).Create(&users)for _, user := range users {    user.ID // 1,2,3}

You can specify batch size when creating with CreateInBatches, e.g:

var users = <>User{{Name: "modi_1"}, ...., {Name: "modi_10000"}}// batch size 100query.Use(db).User.WithContext(ctx).CreateInBatches(users, 100)

It will works if you set CreateBatchSize in gorm.Config / gorm.Session

db, err := gorm.Open(sqlite.Open("gorm.db"), &gorm.Config{    CreateBatchSize: 1000,})// ORdb = db.Session(&gorm.Session{CreateBatchSize: 1000})u := query.NewUser(db)var users = <>User{{Name: "modi_1"}, ...., {Name: "modi_5000"}}u.WithContext(ctx).Create(&users)// INSERT INTO users xxx (5 batches)

Query

Retrieving a single object

Generated code provides First, Take, Last methods to retrieve a single object from the database, it adds LIMIT 1 condition when querying the database, and it will return the error ErrRecordNotFound if no record is found.

u := query.Use(db).User// Get the first record ordered by primary keyuser, err := u.WithContext(ctx).First()// SELECT * FROM users ORDER BY id LIMIT 1;// Get one record, no specified orderuser, err := u.WithContext(ctx).Take()// SELECT * FROM users LIMIT 1;// Get last record, ordered by primary key descuser, err := u.WithContext(ctx).Last()// SELECT * FROM users ORDER BY id DESC LIMIT 1;// check error ErrRecordNotFounderrors.Is(err, gorm.ErrRecordNotFound)

Retrieving objects with primary key

u := query.Use(db).Useruser, err := u.WithContext(ctx).Where(u.ID.Eq(10)).First()// SELECT * FROM users WHERE id = 10;users, err := u.WithContext(ctx).Where(u.ID.In(1,2,3)).Find()// SELECT * FROM users WHERE id IN (1,2,3);

If the primary key is a string (for example, like a uuid), the query will be written as follows:

user, err := u.WithContext(ctx).Where(u.ID.Eq("1b74413f-f3b8-409f-ac47-e8c062e3472a")).First()// SELECT * FROM users WHERE id = "1b74413f-f3b8-409f-ac47-e8c062e3472a";

Retrieving all objects

u := query.Use(db).User// Get all recordsusers, err := u.WithContext(ctx).Find()// SELECT * FROM users;

Conditions

String Conditions

u := query.Use(db).User// Get first matched recorduser, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).First()// SELECT * FROM users WHERE name = 'modi' ORDER BY id LIMIT 1;// Get all matched recordsusers, err := u.WithContext(ctx).Where(u.Name.Neq("modi")).Find()// SELECT * FROM users WHERE name <> 'modi';// INusers, err := u.WithContext(ctx).Where(u.Name.In("modi", "zhangqiang")).Find()// SELECT * FROM users WHERE name IN ('modi','zhangqiang');// LIKEusers, err := u.WithContext(ctx).Where(u.Name.Like("%modi%")).Find()// SELECT * FROM users WHERE name LIKE '%modi%';// ANDusers, err := u.WithContext(ctx).Where(u.Name.Eq("modi"), u.Age.Gte(17)).Find()// SELECT * FROM users WHERE name = 'modi' AND age >= 17;// Timeusers, err := u.WithContext(ctx).Where(u.Birthday.Gt(birthTime).Find()// SELECT * FROM users WHERE birthday > '2000-01-01 00:00:00';// BETWEENusers, err := u.WithContext(ctx).Where(u.Birthday.Between(lastWeek, today)).Find()// SELECT * FROM users WHERE birthday BETWEEN '2000-01-01 00:00:00' AND '2000-01-08 00:00:00';

Inline Condition

u := query.Use(db).User// Get by primary key if it were a non-integer typeuser, err := u.WithContext(ctx).Where(u.ID.Eq("string_primary_key")).First()// SELECT * FROM users WHERE id = 'string_primary_key';// Plain SQLusers, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Find()// SELECT * FROM users WHERE name = "modi";users, err := u.WithContext(ctx).Where(u.Name.Neq("modi"), u.Age.Gt(17)).Find()// SELECT * FROM users WHERE name <> "modi" AND age > 17;

Not Conditions

Build NOT conditions, works similar to Where

u := query.Use(db).Useruser, err := u.WithContext(ctx).Not(u.Name.Eq("modi")).First()// SELECT * FROM users WHERE NOT name = "modi" ORDER BY id LIMIT 1;// Not Inusers, err := u.WithContext(ctx).Not(u.Name.In("modi", "zhangqiang")).Find()// SELECT * FROM users WHERE name NOT IN ("modi", "zhangqiang");// Not In slice of primary keysuser, err := u.WithContext(ctx).Not(u.ID.In(1,2,3)).First()// SELECT * FROM users WHERE id NOT IN (1,2,3) ORDER BY id LIMIT 1;

Or Conditions

u := query.Use(db).Userusers, err := u.WithContext(ctx).Where(u.Role.Eq("admin")).Or(u.Role.Eq("super_admin")).Find()// SELECT * FROM users WHERE role = 'admin' OR role = 'super_admin';

Group Conditions

Easier to write complicated SQL query with Group Conditions

p := query.Use(db).Pizzapizzas, err := p.WithContext(ctx).Where(    p.WithContext(ctx).Where(p.Pizza.Eq("pepperoni")).        Where(p.Where(p.Size.Eq("small")).Or(p.Size.Eq("medium"))),).Or(    p.WithContext(ctx).Where(p.Pizza.Eq("hawaiian")).Where(p.Size.Eq("xlarge")),).Find()// SELECT * FROM `pizzas` WHERE (pizza = "pepperoni" AND (size = "small" OR size = "medium")) OR (pizza = "hawaiian" AND size = "xlarge")

Selecting Specific Fields

Select allows you to specify the fields that you want to retrieve from database. Otherwise, GORM will select all fields by default.

u := query.Use(db).Userusers, err := u.WithContext(ctx).Select(u.Name, u.Age).Find()// SELECT name, age FROM users;u.WithContext(ctx).Select(u.Age.Avg()).Rows()// SELECT Avg(age) FROM users;

Tuple Query

u := query.Use(db).Userusers, err := u.WithContext(ctx).Where(u.Columns(u.ID, u.Name).In(field.Values(<><>inferface{}{{1, "modi"}, {2, "zhangqiang"}}))).Find()// SELECT * FROM `users` WHERE (`id`, `name`) IN ((1,'humodi'),(2,'tom'));

JSON Query

u := query.Use(db).Userusers, err := u.WithContext(ctx).Where(gen.Cond(datatypes.JSONQuery("attributes").HasKey("role"))...).Find()// SELECT * FROM `users` WHERE JSON_EXTRACT(`attributes`,'$.role') IS NOT NULL;

Order

Specify order when retrieving records from the database

u := query.Use(db).Userusers, err := u.WithContext(ctx).Order(u.Age.Desc(), u.Name).Find()// SELECT * FROM users ORDER BY age DESC, name;// Multiple ordersusers, err := u.WithContext(ctx).Order(u.Age.Desc()).Order(u.Name).Find()// SELECT * FROM users ORDER BY age DESC, name;

Limit & Offset

Limit specify the max number of records to retrieve Offset specify the number of records to skip before starting to return the records

u := query.Use(db).Userurers, err := u.WithContext(ctx).Limit(3).Find()// SELECT * FROM users LIMIT 3;// Cancel limit condition with -1users, err := u.WithContext(ctx).Limit(10).Limit(-1).Find()// SELECT * FROM users;users, err := u.WithContext(ctx).Offset(3).Find()// SELECT * FROM users OFFSET 3;users, err := u.WithContext(ctx).Limit(10).Offset(5).Find()// SELECT * FROM users OFFSET 5 LIMIT 10;// Cancel offset condition with -1users, err := u.WithContext(ctx).Offset(10).Offset(-1).Find()// SELECT * FROM users;

Group By & Having

u := query.Use(db).Usertype Result struct {    Date  time.Time    Total int}var result Resulterr := u.WithContext(ctx).Select(u.Name, u.Age.Sum().As("total")).Where(u.Name.Like("%modi%")).Group(u.Name).Scan(&result)// SELECT name, sum(age) as total FROM `users` WHERE name LIKE "%modi%" GROUP BY `name`err := u.WithContext(ctx).Select(u.Name, u.Age.Sum().As("total")).Group(u.Name).Having(u.Name.Eq("group")).Scan(&result)// SELECT name, sum(age) as total FROM `users` GROUP BY `name` HAVING name = "group"rows, err := u.WithContext(ctx).Select(u.Birthday.As("date"), u.Age.Sum().As("total")).Group(u.Birthday).Rows()for rows.Next() {  ...}o := query.Use(db).Orderrows, err := o.WithContext(ctx).Select(o.CreateAt.Date().As("date"), o.Amount.Sum().As("total")).Group(o.CreateAt.Date()).Having(u.Amount.Sum().Gt(100)).Rows()for rows.Next() {  ...}var results <>Resulto.WithContext(ctx).Select(o.CreateAt.Date().As("date"), o.WithContext(ctx).Amount.Sum().As("total")).Group(o.CreateAt.Date()).Having(u.Amount.Sum().Gt(100)).Scan(&results)

Distinct

Selecting distinct values from the model

u := query.Use(db).Userusers, err := u.WithContext(ctx).Distinct(u.Name, u.Age).Order(u.Name, u.Age.Desc()).Find()

Distinct works with Pluck and Count too

Joins

Specify Joins conditions

u := query.Use(db).Usere := query.Use(db).Emailc := query.Use(db).CreditCardtype Result struct {    Name  string    Email string}var result Resulterr := u.WithContext(ctx).Select(u.Name, e.Email).LeftJoin(e, e.UserID.EqCol(u.ID)).Scan(&result)// SELECT users.name, emails.email FROM `users` left join emails on emails.user_id = users.idrows, err := u.WithContext(ctx).Select(u.Name, e.Email).LeftJoin(e, e.UserID.EqCol(u.ID)).Rows()for rows.Next() {  ...}var results <>Resulterr := u.WithContext(ctx).Select(u.Name, e.Email).LeftJoin(e, e.UserID.EqCol(u.ID)).Scan(&results)// multiple joins with parameterusers := u.WithContext(ctx).Join(e, e.UserID.EqCol(u.id), e.Email.Eq("modi@example.org")).Join(c, c.UserID.EqCol(u.ID)).Where(c.Number.Eq("411111111111")).Find()

SubQuery

A subquery can be nested within a query, GEN can generate subquery when using a Dao object as param

o := query.Use(db).Orderu := query.Use(db).Userorders, err := o.WithContext(ctx).Where(u.Columns(o.Amount).Gt(o.WithContext(ctx).Select(o.Amount.Avg())).Find()// SELECT * FROM "orders" WHERE amount > (SELECT AVG(amount) FROM "orders");subQuery := u.WithContext(ctx).Select(u.Age.Avg()).Where(u.Name.Like("name%"))users, err := u.WithContext(ctx).Select(u.Age.Avg().As("avgage")).Group(u.Name).Having(u.Columns(u.Age.Avg()).Gt(subQuery).Find()// SELECT AVG(age) as avgage FROM `users` GROUP BY `name` HAVING AVG(age) > (SELECT AVG(age) FROM `users` WHERE name LIKE "name%")

From SubQuery

GORM allows you using subquery in FROM clause with method Table, for example:

u := query.Use(db).Userp := query.Use(db).Petusers, err := gen.Table(u.WithContext(ctx).Select(u.Name, u.Age).As("u")).Where(u.Age.Eq(18)).Find()// SELECT * FROM (SELECT `name`,`age` FROM `users`) as u WHERE `age` = 18subQuery1 := u.WithContext(ctx).Select(u.Name)subQuery2 := p.WithContext(ctx).Select(p.Name)users, err := gen.Table(subQuery1.As("u"), subQuery2.As("p")).Find()db.Table("(?) as u, (?) as p", subQuery1, subQuery2).Find(&User{})// SELECT * FROM (SELECT `name` FROM `users`) as u, (SELECT `name` FROM `pets`) as p

Update from SubQuery

Update a table by using SubQuery

u := query.Use(db).Userc := query.Use(db).Companyu.WithContext(ctx).Update(u.CompanyName, c.Select(c.Name).Where(c.ID.EqCol(u.CompanyID)))// UPDATE "users" SET "company_name" = (SELECT name FROM companies WHERE companies.id = users.company_id);u.WithContext(ctx).Where(u.Name.Eq("modi")).Update(u.CompanyName, c.Select(c.Name).Where(c.ID.EqCol(u.CompanyID)))

Transaction

To perform a set of operations within a transaction, the general flow is as below.

q := query.Use(db)q.Transaction(func(tx *query.Query) error {  if _, err := tx.User.WithContext(ctx).Where(tx.User.ID.Eq(100)).Delete(); err != nil {    return err  }  if _, err := tx.Article.WithContext(ctx).Create(&model.User{Name:"modi"}); err != nil {    return err  }  return nil})

Nested Transactions

GEN supports nested transactions, you can rollback a subset of operations performed within the scope of a larger transaction, for example:

q := query.Use(db)q.Transaction(func(tx *query.Query) error {  tx.User.WithContext(ctx).Create(&user1)  tx.Transaction(func(tx2 *query.Query) error {    tx2.User.WithContext(ctx).Create(&user2)    return errors.New("rollback user2") // Rollback user2  })  tx.Transaction(func(tx2 *query.Query) error {    tx2.User.WithContext(ctx).Create(&user3)    return nil  })  return nil})// Commit user1, user3

Transactions by manual

q := query.Use(db)// begin a transactiontx := q.Begin()// do some database operations in the transaction (use 'tx' from this point, not 'db')tx.User.WithContext(ctx).Create(...)// ...// rollback the transaction in case of errortx.Rollback()// Or commit the transactiontx.Commit()

For example:

q := query.Use(db)func doSomething(ctx context.Context, users ...*model.User) (err error) {    tx := q.Begin()    defer func() {        if recover() != nil || err != nil {            _ = tx.Rollback()        }    }()    err = tx.User.WithContext(ctx).Create(users...)    if err != nil {        return    }    return tx.Commit()}

SavePoint/RollbackTo

GEN provides SavePoint, RollbackTo to save points and roll back to a savepoint, for example:

tx := q.Begin()txCtx = tx.WithContext(ctx)txCtx.User.Create(&user1)tx.SavePoint("sp1")txCtx.Create(&user2)tx.RollbackTo("sp1") // Rollback user2tx.Commit() // Commit user1

Advanced Query

Iteration

GEN supports iterating through Rows

u := query.Use(db).Userdo := u.WithContext(ctx)rows, err := do.Where(u.Name.Eq("modi")).Rows()defer rows.Close()for rows.Next() {    var user User    // ScanRows is a method of `gorm.DB`, it can be used to scan a row into a struct    do.ScanRows(rows, &user)    // do something}

FindInBatches

Query and process records in batch

u := query.Use(db).User// batch size 100err := u.WithContext(ctx).Where(u.ID.Gt(9)).FindInBatches(&results, 100, func(tx gen.Dao, batch int) error {    for _, result := range results {      // batch processing found records    }      // build a new `u` to use it's api    // queryUsery := query.NewUser(tx.UnderlyingDB())    tx.Save(&results)    batch // Batch 1, 2, 3    // returns error will stop future batches    return nil})

Pluck

Query single column from database and scan into a slice, if you want to query multiple columns, use Select with Scan instead

u := query.Use(db).Uservar ages <>int64u.WithContext(ctx).Pluck(u.Age, &ages)var names <>stringu.WithContext(ctx).Pluck(u.Name, &names)// Distinct Plucku.WithContext(ctx).Distinct().Pluck(u.Name, &names)// SELECT DISTINCT `name` FROM `users`// Requesting more than one column, use `Scan` or `Find` like this:db.WithContext(ctx).Select(u.Name, u.Age).Scan(&users)users, err := db.Select(u.Name, u.Age).Find()

Scopes

Scopes allows you to specify commonly-used queries which can be referenced as method calls

o := query.Use(db).Orderfunc AmountGreaterThan1000(tx gen.Dao) gen.Dao {    return tx.Where(o.Amount.Gt(1000))}func PaidWithCreditCard(tx gen.Dao) gen.Dao {    return tx.Where(o.PayModeSign.Eq("C"))}func PaidWithCod(tx gen.Dao) gen.Dao {    return tx.Where(o.PayModeSign.Eq("C"))}func OrderStatus(status <>string) func (tx gen.Dao) gen.Dao {    return func (tx gen.Dao) gen.Dao {      return tx.Where(o.Status.In(status...))    }}orders, err := o.WithContext(ctx).Scopes(AmountGreaterThan1000, PaidWithCreditCard).Find()// Find all credit card orders and amount greater than 1000orders, err := o.WithContext(ctx).Scopes(AmountGreaterThan1000, PaidWithCod).Find()// Find all COD orders and amount greater than 1000orders, err := o.WithContext(ctx).Scopes(AmountGreaterThan1000, OrderStatus(<>string{"paid", "shipped"})).Find()// Find all paid, shipped orders that amount greater than 1000

Count

Get matched records count

u := query.Use(db).Usercount, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Or(u.Name.Eq("zhangqiang")).Count()// SELECT count(1) FROM users WHERE name = 'modi' OR name = 'zhangqiang'count, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Count()// SELECT count(1) FROM users WHERE name = 'modi'; (count)// Count with Distinctu.WithContext(ctx).Distinct(u.Name).Count()// SELECT COUNT(DISTINCT(`name`)) FROM `users`

FirstOrInit

Get first matched record or initialize a new instance with given conditions

u := query.Use(db).User// User not found, initialize it with give conditionsuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).FirstOrInit()// user -> User{Name: "non_existing"}// Found user with `name` = `modi`user, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).FirstOrInit()// user -> User{ID: 1, Name: "modi", Age: 17}

initialize struct with more attributes if record not found, those Attrs won’t be used to build SQL query

u := query.Use(db).User// User not found, initialize it with give conditions and Attrsuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).Attrs(u.Age.Value(20)).FirstOrInit()// SELECT * FROM USERS WHERE name = 'non_existing' ORDER BY id LIMIT 1;// user -> User{Name: "non_existing", Age: 20}// User not found, initialize it with give conditions and Attrsuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).Attrs(u.Age.Value(20)).FirstOrInit()// SELECT * FROM USERS WHERE name = 'non_existing' ORDER BY id LIMIT 1;// user -> User{Name: "non_existing", Age: 20}// Found user with `name` = `modi`, attributes will be ignoreduser, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Attrs(u.Age.Value(20)).FirstOrInit()// SELECT * FROM USERS WHERE name = modi' ORDER BY id LIMIT 1;// user -> User{ID: 1, Name: "modi", Age: 17}

Assign attributes to struct regardless it is found or not, those attributes won’t be used to build SQL query and the final data won’t be saved into database

// User not found, initialize it with give conditions and Assign attributesuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).Assign(u.Age.Value(20)).FirstOrInit()// user -> User{Name: "non_existing", Age: 20}// Found user with `name` = `modi`, update it with Assign attributesuser, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Assign(u.Age.Value(20)).FirstOrInit()// SELECT * FROM USERS WHERE name = modi' ORDER BY id LIMIT 1;// user -> User{ID: 111, Name: "modi", Age: 20}

FirstOrCreate

Get first matched record or create a new one with given conditions

u := query.Use(db).User// User not found, create a new record with give conditionsuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).FirstOrCreate()// INSERT INTO "users" (name) VALUES ("non_existing");// user -> User{ID: 112, Name: "non_existing"}// Found user with `name` = `modi`user, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).FirstOrCreate()// user -> User{ID: 111, Name: "modi", "Age": 18}

Create struct with more attributes if record not found, those Attrs won’t be used to build SQL query

u := query.Use(db).User// User not found, create it with give conditions and Attrsuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).Attrs(u.Age.Value(20)).FirstOrCreate()// SELECT * FROM users WHERE name = 'non_existing' ORDER BY id LIMIT 1;// INSERT INTO "users" (name, age) VALUES ("non_existing", 20);// user -> User{ID: 112, Name: "non_existing", Age: 20}// Found user with `name` = `modi`, attributes will be ignoreduser, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Attrs(u.Age.Value(20)).FirstOrCreate()// SELECT * FROM users WHERE name = 'modi' ORDER BY id LIMIT 1;// user -> User{ID: 111, Name: "modi", Age: 18}

Assign attributes to the record regardless it is found or not and save them back to the database.

u := query.Use(db).User// User not found, initialize it with give conditions and Assign attributesuser, err := u.WithContext(ctx).Where(u.Name.Eq("non_existing")).Assign(u.Age.Value(20)).FirstOrCreate()// SELECT * FROM users WHERE name = 'non_existing' ORDER BY id LIMIT 1;// INSERT INTO "users" (name, age) VALUES ("non_existing", 20);// user -> User{ID: 112, Name: "non_existing", Age: 20}// Found user with `name` = `modi`, update it with Assign attributesuser, err := u.WithContext(ctx).Where(u.Name.Eq("modi")).Assign(u.Age.Value(20)).FirstOrCreate(&user)// SELECT * FROM users WHERE name = 'modi' ORDER BY id LIMIT 1;// UPDATE users SET age=20 WHERE id = 111;// user -> User{ID: 111, Name: "modi", Age: 20}

Association

GEN will auto-save associations as GORM do. The relationships (BelongsTo/HasOne/HasMany/Many2Many) reuse GORM's tag. This feature only support exist model for now.

Relation

There are 4 kind of relationship.

const (    HasOne    RelationshipType = RelationshipType(schema.HasOne)    // HasOneRel has one relationship    HasMany   RelationshipType = RelationshipType(schema.HasMany)   // HasManyRel has many relationships    BelongsTo RelationshipType = RelationshipType(schema.BelongsTo) // BelongsToRel belongs to relationship    Many2Many RelationshipType = RelationshipType(schema.Many2Many) // Many2ManyRel many to many relationship)

Relate to exist model

package model// exist modeltype Customer struct {    gorm.Model    CreditCards <>CreditCard `gorm:"foreignKey:CustomerRefer"`}type CreditCard struct {    gorm.Model    Number        string    CustomerRefer uint}

GEN will detect model's associations:

// specify modelg.ApplyBasic(model.Customer{}, model.CreditCard{})// assoications will be detected and converted to code package querytype customer struct {    ...    CreditCards customerHasManyCreditCards}type creditCard struct{    ...}

Relate to table in database

The association have to be speified by gen.FieldRelate

card := g.GenerateModel("credit_cards")customer := g.GenerateModel("customers", gen.FieldRelate(field.HasMany, "CreditCards", b,     &field.RelateConfig{        // RelateSlice: true,        GORMTag: "foreignKey:CustomerRefer",    }),)g.ApplyBasic(card, custormer)

GEN will generate models with associated field:

// customerstype Customer struct {    ID          int64          `gorm:"column:id;type:bigint(20) unsigned;primaryKey" json:"id"`    CreatedAt   time.Time      `gorm:"column:created_at;type:datetime(3)" json:"created_at"`    UpdatedAt   time.Time      `gorm:"column:updated_at;type:datetime(3)" json:"updated_at"`    DeletedAt   gorm.DeletedAt `gorm:"column:deleted_at;type:datetime(3)" json:"deleted_at"`    CreditCards <>CreditCard   `gorm:"foreignKey:CustomerRefer" json:"credit_cards"`}// credit_cardstype CreditCard struct {    ID            int64          `gorm:"column:id;type:bigint(20) unsigned;primaryKey" json:"id"`    CreatedAt     time.Time      `gorm:"column:created_at;type:datetime(3)" json:"created_at"`    UpdatedAt     time.Time      `gorm:"column:updated_at;type:datetime(3)" json:"updated_at"`    DeletedAt     gorm.DeletedAt `gorm:"column:deleted_at;type:datetime(3)" json:"deleted_at"`    CustomerRefer int64          `gorm:"column:customer_refer;type:bigint(20) unsigned" json:"customer_refer"`}

If associated model already exists, gen.FieldRelateModel can help you build associations between them.

customer := g.GenerateModel("customers", gen.FieldRelateModel(field.HasMany, "CreditCards", model.CreditCard{},     &field.RelateConfig{        // RelateSlice: true,        GORMTag: "foreignKey:CustomerRefer",    }),)g.ApplyBasic(custormer)

Relate Config

type RelateConfig struct {    // specify field's type    RelatePointer      bool // ex: CreditCard  *CreditCard    RelateSlice        bool // ex: CreditCards <>CreditCard    RelateSlicePointer bool // ex: CreditCards <>*CreditCard    JSONTag      string // related field's JSON tag    GORMTag      string // related field's GORM tag    NewTag       string // related field's new tag    OverwriteTag string // related field's tag}

Operation

Skip Auto Create/Update

user := model.User{  Name:            "modi",  BillingAddress:  Address{Address1: "Billing Address - Address 1"},  ShippingAddress: Address{Address1: "Shipping Address - Address 1"},  Emails:          <>Email{    {Email: "modi@example.com"},    {Email: "modi-2@example.com"},  },  Languages:       <>Language{    {Name: "ZH"},    {Name: "EN"},  },}u := query.Use(db).Useru.WithContext(ctx).Select(u.Name).Create(&user)// INSERT INTO "users" (name) VALUES ("jinzhu", 1, 2);u.WithContext(ctx).Omit(u.BillingAddress.Field()).Create(&user)// Skip create BillingAddress when creating a useru.WithContext(ctx).Omit(u.BillingAddress.Field("Address1")).Create(&user)// Skip create BillingAddress.Address1 when creating a useru.WithContext(ctx).Omit(field.AssociationFields).Create(&user)// Skip all associations when creating a user

Method Field will join a serious field name with ''.", for example: u.BillingAddress.Field("Address1", "Street") equals to BillingAddress.Address1.Street

Find Associations

Find matched associations

u := query.Use(db).Userlanguages, err = u.Languages.Model(&user).Find()

Find associations with conditions

q := query.Use(db)u := q.Userlanguages, err = u.Languages.Where(q.Language.Name.In(<>string{"ZH","EN"})).Model(&user).Find()

Append Associations

Append new associations for many to many, has many, replace current association for has one, belongs to

u := query.Use(db).Useru.Languages.Model(&user).Append(&languageZH, &languageEN)u.Languages.Model(&user).Append(&Language{Name: "DE"})u.CreditCards.Model(&user).Append(&CreditCard{Number: "411111111111"})

Replace Associations

Replace current associations with new ones

u.Languages.Model(&user).Replace(&languageZH, &languageEN)

Delete Associations

Remove the relationship between source & arguments if exists, only delete the reference, won’t delete those objects from DB.

u := query.Use(db).Useru.Languages.Model(&user).Delete(&languageZH, &languageEN)u.Languages.Model(&user).Delete(<>*Language{&languageZH, &languageEN}...)

Clear Associations

Remove all reference between source & association, won’t delete those associations

u.Languages.Model(&user).Clear()

Count Associations

Return the count of current associations

u.Languages.Model(&user).Count()

Delete with Select

You are allowed to delete selected has one/has many/many2many relations with Select when deleting records, for example:

u := query.Use(db).User// delete user's account when deleting useru.Select(u.Account).Delete(&user)// delete user's Orders, CreditCards relations when deleting userdb.Select(u.Orders.Field(), u.CreditCards.Field()).Delete(&user)// delete user's has one/many/many2many relations when deleting userdb.Select(field.AssociationsFields).Delete(&user)

Preloading

This feature only support exist model for now.

Preload

GEN allows eager loading relations in other SQL with Preload, for example:

type User struct {  gorm.Model  Username string  Orders   <>Order}type Order struct {  gorm.Model  UserID uint  Price  float64}q := query.Use(db)u := q.Usero := q.Order// Preload Orders when find usersusers, err := u.WithContext(ctx).Preload(u.Orders).Find()// SELECT * FROM users;// SELECT * FROM orders WHERE user_id IN (1,2,3,4);users, err := u.WithContext(ctx).Preload(u.Orders).Preload(u.Profile).Preload(u.Role).Find()// SELECT * FROM users;// SELECT * FROM orders WHERE user_id IN (1,2,3,4); // has many// SELECT * FROM profiles WHERE user_id IN (1,2,3,4); // has one// SELECT * FROM roles WHERE id IN (4,5,6); // belongs to

Preload All

clause.Associations can work with Preload similar like Select when creating/updating, you can use it to Preload all associations, for example:

type User struct {  gorm.Model  Name       string  CompanyID  uint  Company    Company  Role       Role  Orders     <>Order}users, err := u.WithContext(ctx).Preload(field.Associations).Find()

clause.Associations won’t preload nested associations, but you can use it with Nested Preloading together, e.g:

users, err := u.WithContext(ctx).Preload(u.Orders.OrderItems.Product).Find()

Preload with conditions

GORM allows Preload associations with conditions, it works similar to Inline Conditions.

q := query.Use(db)u := q.Usero := q.Order// Preload Orders with conditionsusers, err := u.WithContext(ctx).Preload(u.Orders.On(o.State.NotIn("cancelled")).Find()// SELECT * FROM users;// SELECT * FROM orders WHERE user_id IN (1,2,3,4) AND state NOT IN ('cancelled');users, err := u.WithContext(ctx).Where(u.State.Eq("active")).Preload(u.Orders.On(o.State.NotIn("cancelled")).Find()// SELECT * FROM users WHERE state = 'active';// SELECT * FROM orders WHERE user_id IN (1,2) AND state NOT IN ('cancelled');users, err := u.WithContext(ctx).Preload(u.Orders.Order(o.ID.Desc(), o.CreateTime).Find()// SELECT * FROM users;// SELECT * FROM orders WHERE user_id IN (1,2) Order By id DESC, create_time;users, err := u.WithContext(ctx).Preload(u.Orders.On(o.State.Eq("on")).Order(o.ID.Desc()).Find()// SELECT * FROM users;// SELECT * FROM orders WHERE user_id IN (1,2) AND state = "on" Order By id DESC;users, err := u.WithContext(ctx).Preload(u.Orders.Clauses(hints.UseIndex("idx_order_id"))).Find()// SELECT * FROM users;// SELECT * FROM orders WHERE user_id IN (1,2) USE INDEX (`idx_order_id`);

Nested Preloading

GEN supports nested preloading, for example:

db.Preload(u.Orders.OrderItems.Product).Preload(u.CreditCard).Find(&users)// Customize Preload conditions for `Orders`// And GEN won't preload unmatched order's OrderItems thendb.Preload(u.Orders.On(o.State.Eq("paid"))).Preload(u.Orders.OrderItems).Find(&users)

Update

Update single column

When updating a single column with Update, it needs to have any conditions or it will raise error ErrMissingWhereClause, for example:

u := query.Use(db).User// Update with conditionsu.WithContext(ctx).Where(u.Activate.Is(true)).Update(u.Name, "hello")// UPDATE users SET name='hello', updated_at='2013-11-17 21:34:10' WHERE active=true;// Update with conditionsu.WithContext(ctx).Where(u.Activate.Is(true)).Update(u.Age, u.Age.Add(1))// oru.WithContext(ctx).Where(u.Activate.Is(true)).UpdateSimple(u.Age.Add(1))// UPDATE users SET age=age+1, updated_at='2013-11-17 21:34:10' WHERE active=true;u.WithContext(ctx).Where(u.Activate.Is(true)).UpdateSimple(u.Age.Zero())// UPDATE users SET age=0, updated_at='2013-11-17 21:34:10' WHERE active=true;

Updates multiple columns

Updates supports update with struct or mapinterface{}, when updating with struct it will only update non-zero fields by default

u := query.Use(db).User// Update attributes with `map`u.WithContext(ctx).Where(u.ID.Eq(111)).Updates(mapinterface{}{"name": "hello", "age": 18, "active": false})// UPDATE users SET name='hello', age=18, active=false, updated_at='2013-11-17 21:34:10' WHERE id=111;// Update attributes with `struct`u.WithContext(ctx).Where(u.ID.Eq(111)).Updates(model.User{Name: "hello", Age: 18, Active: false})// UPDATE users SET name='hello', age=18, active=false, updated_at='2013-11-17 21:34:10' WHERE id=111;// Update with expressionu.WithContext(ctx).Where(u.ID.Eq(111)).UpdateSimple(u.Age.Add(1), u.Number.Add(1))// UPDATE users SET age=age+1,number=number+1, updated_at='2013-11-17 21:34:10' WHERE id=111;u.WithContext(ctx).Where(u.Activate.Is(true)).UpdateSimple(u.Age.Value(17), u.Number.Zero(), u.Birthday.Null())// UPDATE users SET age=17, number=0, birthday=NULL, updated_at='2013-11-17 21:34:10' WHERE active=true;

NOTE When update with struct, GEN will only update non-zero fields, you might want to use map to update attributes or use Select to specify fields to update

Update selected fields

If you want to update selected fields or ignore some fields when updating, you can use Select, Omit

u := query.Use(db).User// Select with Map// User's ID is `111`:u.WithContext(ctx).Select(u.Name).Where(u.ID.Eq(111)).Updates(mapinterface{}{"name": "hello", "age": 18, "active": false})// UPDATE users SET name='hello' WHERE id=111;u.WithContext(ctx).Omit(u.Name).Where(u.ID.Eq(111)).Updates(mapinterface{}{"name": "hello", "age": 18, "active": false})// UPDATE users SET age=18, active=false, updated_at='2013-11-17 21:34:10' WHERE id=111;result, err := u.WithContext(ctx).Where(u.ID.Eq(111)).Updates(mapinterface{}{"name": "hello", "age": 18, "active": false})result.RowsAffected // affect rows numbererr                 // error

Delete

Delete record

e := query.Use(db).Email// Email's ID is `10`e.WithContext(ctx).Where(e.ID.Eq(10)).Delete()// DELETE from emails where id = 10;// Delete with additional conditionse.WithContext(ctx).Where(e.ID.Eq(10), e.Name.Eq("modi")).Delete()// DELETE from emails where id = 10 AND name = "modi";result, err := e.WithContext(ctx).Where(e.ID.Eq(10), e.Name.Eq("modi")).Delete()result.RowsAffected // affect rows numbererr                 // error

Delete with primary key

GEN allows to delete objects using primary key(s) with inline condition, it works with numbers.

u.WithContext(ctx).Where(u.ID.In(1,2,3)).Delete()// DELETE FROM users WHERE id IN (1,2,3);

Batch Delete

The specified value has no primary value, GEN will perform a batch delete, it will delete all matched records

e := query.Use(db).Emaile.WithContext(ctx).Where(e.Name.Like("%modi%")).Delete()// DELETE from emails where email LIKE "%modi%";

Soft Delete

If your model includes a gorm.DeletedAt field (which is included in gorm.Model), it will get soft delete ability automatically!

When calling Delete, the record WON’T be removed from the database, but GORM will set the DeletedAt‘s value to the current time, and the data is not findable with normal Query methods anymore.

// Batch Deleteu.WithContext(ctx).Where(u.Age.Eq(20)).Delete()// UPDATE users SET deleted_at="2013-10-29 10:23" WHERE age = 20;// Soft deleted records will be ignored when queryingusers, err := u.WithContext(ctx).Where(u.Age.Eq(20)).Find()// SELECT * FROM users WHERE age = 20 AND deleted_at IS NULL;

If you don’t want to include gorm.Model, you can enable the soft delete feature like:

type User struct {    ID      int    Deleted gorm.DeletedAt    Name    string}

Find soft deleted records

You can find soft deleted records with Unscoped

users, err := db.WithContext(ctx).Unscoped().Where(u.Age.Eq(20)).Find()// SELECT * FROM users WHERE age = 20;

Delete permanently

You can delete matched records permanently with Unscoped

o.WithContext(ctx).Unscoped().Where(o.ID.Eq(10)).Delete()// DELETE FROM orders WHERE id=10;

DIY method

Method interface

Method interface is an abstraction of query methods, all functions it contains are query methods and above comments describe the specific query conditions or logic. SQL supports simple where query or execute raw SQL. Simple query conditions wrapped by where(), and raw SQL wrapped by sql()（not required）

Method interface supports descriptive comment that describes how the method works.It starts with method name and followed descriptive message (not required). It is distinguished from query comment by blank line (with descriptive message) or space (without descriptive message).

type Method interface {    // where("name=@name and age=@age")    SimpleFindByNameAndAge(name string, age int) (gen.T, error)    // FindUserToMap query by id and return id->instance    //     // sql(select * from users where id=@id)    FindUserToMap(id int) (gen.M, error)        // InsertValue insert into users (name,age) values (@name,@age)    InsertValue(age int, name string) error}

Return values must contains less than 1 gen.T/gen.M/gen.RowsAffected and less than 1 error. You can also use bulitin type (like string/ int) as the return parameter，gen.T represents return a single result struct's pointer, <>gen.T represents return an array of result structs' pointer，

Syntax of template

placeholder

• gen.T represents specified struct or table
• gen.M represents mapinterface
• gen.RowsAffected represents SQL executed rowsAffected (type:int64)
• @@table represents table's name (if method's parameter doesn't contains variable table, GEN will generate table from model struct)
• @@<columnName> represents column's name or table's name
• @<name> represents normal query variable

template

Logical operations must be wrapped in {{}},and end must used {{end}}, All templates support nesting

• if/else if/else the condition accept a bool parameter or operation expression which conforms to Golang syntax.
• where The where clause will be inserted only if the child elements return something. The key word and or or in front of clause will be removed. And and will be added automatically when there is no junction keyword between query condition clause.
• Set The set clause will be inserted only if the child elements return something. The , in front of columns array will be removed.And , will be added automatically when there is no junction keyword between query coulmns.
• ... Coming soon

Ifclause

{{if cond1}}    // do something here{{else if cond2}}    // do something here{{else}}    // do something here{{end}}

Use case in raw SQL:

// select * from users where {{if name !=""}} name=@name{{end}}methond(name string) (gen.T,error) 

Use case in raw SQL template:

select * from @@table where{{if age>60}}    status="older"{{else if age>30}}    status="middle-ager"{{else if age>18}}    status="younger"{{else}}    {{if sex=="male"}}        status="boys"    {{else}}        status="girls"    {{end}}{{end}}

Whereclause

{{where}}    // do something here{{end}}

Use case in raw SQL

// select * from {{where}}id=@id{{end}}methond(id int) error

Use case in raw SQL template

select * from @@table {{where}}    {{if cond}}id=@id {{end}}    {{if name != ""}}@@key=@value{{end}}{{end}}

Setclause

{{set}}    // sepecify update expression here{{end}}

Use case in raw SQL

// update users {{set}}name=@name{{end}}methond() error

Use case in raw SQL template

update @@table {{set}}    {{if name!=""}} name=@name {{end}}    {{if age>0}} age=@age {{end}}{{end}}where id=@id

Method interface example

type Method interface {    // Where("name=@name and age=@age")    SimpleFindByNameAndAge(name string, age int) (gen.T, error)        // select * from users where id=@id    FindUserToMap(id int) (gen.M, error)        // sql(insert into @@table (name,age) values (@name,@age) )    InsertValue(age int, name string) error        // select name from @@table where id=@id    FindNameByID(id int) string        // select * from @@table    //  {{where}}    //      id>0    //      {{if cond}}id=@id {{end}}    //      {{if key!="" && value != ""}} or @@key=@value{{end}}    //  {{end}}    FindByIDOrCustom(cond bool, id int, key, value string) (<>gen.T, error)        // update @@table    //  {{set}}    //      update_time=now()    //      {{if name != ""}}    //          name=@name    //      {{end}}    //  {{end}}    //  {{where}}    //      id=@id    //  {{end}}    UpdateName(name string, id int) (gen.RowsAffected,error)}

Smart select fields

GEN allows select specific fields with Select, if you often use this in your application, maybe you want to define a smaller struct for API usage which can select specific fields automatically, for example:

type User struct {  ID     uint  Name   string  Age    int  Gender string  // hundreds of fields}type APIUser struct {  ID   uint  Name string}type Method interface{    // select * from user    FindSome() (<>APIUser, error)}apiusers, err := u.WithContext(ctx).Limit(10).FindSome()// SELECT `id`, `name` FROM `users` LIMIT 10

Advanced Topics

Hints

Optimizer hints allow to control the query optimizer to choose a certain query execution plan, GORM supports it with gorm.io/hints, e.g:

import "gorm.io/hints"u := query.Use(db).Userusers, err := u.WithContext(ctx).Clauses(hints.New("MAX_EXECUTION_TIME(10000)")).Find()// SELECT * /*+ MAX_EXECUTION_TIME(10000) */ FROM `users`

Index hints allow passing index hints to the database in case the query planner gets confused.

import "gorm.io/hints"u := query.Use(db).Userusers, err := u.WithContext(ctx).Clauses(hints.UseIndex("idx_user_name")).Find()// SELECT * FROM `users` USE INDEX (`idx_user_name`)users, err := u.WithContext(ctx).Clauses(hints.ForceIndex("idx_user_name", "idx_user_id").ForJoin()).Find()// SELECT * FROM `users` FORCE INDEX FOR JOIN (`idx_user_name`,`idx_user_id`)"