HiPoweredDev
/
xorm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
xorm/docs/QuickStart.md

20 KiB
Raw Blame History

Quick Start

1.Create ORM Engine

When using xorm, you can create multiple orm engines, an engine means a databse. So you can：

import (
	_ "github.com/go-sql-driver/mysql"
	"github.com/go-xorm/xorm"
)
engine, err := xorm.NewEngine("mysql", "root:123@/test?charset=utf8")
defer engine.Close()

or

import (
	_ "github.com/mattn/go-sqlite3"
	"github.com/go-xorm/xorm"
	)
engine, err := xorm.NewEngine("sqlite3", "./test.db")
defer engine.Close()

You can create many engines for different databases.Generally, you just need create only one engine. Engine supports run on go routines.

xorm supports four drivers now:

NewEngine's parameters are the same as sql.Open. So you should read the drivers' document for parameters' usage.

After engine created, you can do some settings.

1.Logs

  • engine.ShowSQL = true, Show SQL statement on standard output;
  • engine.ShowDebug = true, Show debug infomation on standard output;
  • engine.ShowError = true, Show error infomation on standard output;
  • engine.ShowWarn = true, Show warnning information on standard output;

2.If want to record infomation with another method: use engine.Logger as io.Writer:

f, err := os.Create("sql.log")
	if err != nil {
		println(err.Error())
		return
	}
engine.Logger = xorm.NewSimpleLogger(f)

3.Engine provide DB connection pool settings.

  • Use engine.SetMaxIdleConns() to set idle connections.
  • Use engine.SetMaxOpenConns() to set Max connections. This methods support only Go 1.2+.

2.Define struct

xorm maps a struct to a database table, the rule is below.

2.1.name mapping rule

use xorm.IMapper interface to implement. There are two IMapper implemented: SnakeMapper and SameMapper. SnakeMapper means struct name is word by word and table name or column name as 下划线. SameMapper means same name between struct and table.

SnakeMapper is the default.

engine.SetMapper(SameMapper{})

And you should notice:

  • If you want to use other mapping rule, implement IMapper
  • Tables's mapping rule could be different from Columns':
engine.SetTableMapper(SameMapper{})
engine.SetColumnMapper(SnakeMapper{})

2.2.Prefix mapping, Suffix Mapping and Cache Mapping

  • engine.NewPrefixMapper(SnakeMapper{}, "prefix") can add prefix string when naming based on SnakeMapper or SameMapper, or you custom Mapper.
  • engine.NewPrefixMapper(SnakeMapper{}, "suffix") can add suffix string when naming based on SnakeMapper or SameMapper, or you custom Mapper.
  • engine.NewCacheMapper(SnakeMapper{}) add naming Mapper for memory cache.

Of course, you can implement IMapper to make custom naming strategy.

2.3.Tag mapping

It's idealized of using IMapper for all naming. But if table or column is not in rule, we need new method to archive.

  • If struct or pointer of struct has TableName() string method, the return value will be the struct's table name.

  • engine.Table() can change the database table name for struct. The struct tag xorm:"'table_name'" can set column name for struct field. Use a pair of single quotes to prevent confusion for column's definition in struct tag. If not in confusion, ignore single quotes.

2.4.Column definition

Struct tag defines something for column as basic sql concepts, such as :

type User struct {
	Id 	 int64
	Name string  `xorm:"varchar(25) not null unique 'usr_name'"`
}

Data types are different in different DBMS. So xorm makes own data types definition to keep compatible. Details is in document Column Types.

The following table is field mapping rules, the keyword is not case sensitive except column name：

name or 'name'Column Name, optional
pkIf column is Primary Key
support over 30 kinds of column types, details in [Column Types](https://github.com/go-xorm/xorm/blob/master/docs/COLUMNTYPE.md)column type
autoincrIf autoincrement column
[not ]null | notnullif column could be blank
unique/unique(uniquename)column is Unique index; if add (uniquename), the column is used for combined unique index with the field that defining same uniquename.
index/index(indexname)column is index. if add (indexname), the column is used for combined index with the field that defining same indexname.
extendsuse for anonymous field, map the struct in anonymous field to database
-This field will not be mapping
->only write into database
<-only read from database
createdThis field will be filled in current time on insert
updatedThis field will be filled in current time on insert or update
versionThis field will be filled 1 on insert and autoincrement on update
default 0 | default 'name'column default value

Some default mapping rules：

    1. If field is name of Id and type of int64, xorm makes it as auto increment primary key. If another field, use struct tag xorm:"pk".
    1. String is corresponding to varchar(255).
    1. Support custom type as type MyString string，slice, map as field type. They are saving as Text column type and json-encode string. Support Blob column type with field type []byte or []uint8.
    1. You can implement Conversion interface to define your custom mapping rule between field and database data.
type Conversion interface {
	FromDB([]byte) error
	ToDB() ([]byte, error)
}
    1. If one struct has a Conversion field, so we need set an implementation to the field before get data from database. We can implement BeforeSet(name string, cell xorm.Cell) on struct to do this. For example: testConversion

3. database meta information

xorm provides methods to getting and setting table schema. For less schema changing production, engine.Sync() is enough.

3.1 retrieve database meta info

  • DBMetas() engine.DBMetas() returns all tables schema information.

3.2.directly table operation

  • CreateTables() engine.CreateTables(struct) creates table with struct or struct pointer. engine.Charset() and engine.StoreEngine() can change charset or storage engine for mysql database.

  • IsTableEmpty() check table is empty or not.

  • IsTableExist() check table is existed or not.

  • DropTables() engine.DropTables(struct) drops table and indexes with struct or struct pointer. engine.DropTables(string) only drops table except indexes.

3.3.create indexes and uniques

  • CreateIndexes create indexes with struct.

  • CreateUniques create unique indexes with struct.

3.4.Synchronize database schema

xorm watches tables and indexes and sync schema:

  1. use table name to create or drop table
  2. use column name to alter column
  3. use the indexes definition in struct field tag to create or drop indexes.
err := engine.Sync(new(User))

4.Insert data

Inserting records use Insert method.

  • Insert one record
user := new(User)
user.Name = "myname"
affected, err := engine.Insert(user)

After inseted, user.Id will be filled with primary key column value.

fmt.Println(user.Id)
  • Insert multiple records by Slice on one table
users := make([]User, 0)
users[0].Name = "name0"
...
affected, err := engine.Insert(&users)
  • Insert multiple records by Slice of pointer on one table
users := make([]*User, 0)
users[0] = new(User)
users[0].Name = "name0"
...
affected, err := engine.Insert(&users)
  • Insert one record on two table.
user := new(User)
user.Name = "myname"
question := new(Question)
question.Content = "whywhywhwy?"
affected, err := engine.Insert(user, question)
  • Insert multiple records on multiple tables.
users := make([]User, 0)
users[0].Name = "name0"
...
questions := make([]Question, 0)
questions[0].Content = "whywhywhwy?"
affected, err := engine.Insert(&users, &questions)
  • Insert one or multple records on multiple tables.
user := new(User)
user.Name = "myname"
...
questions := make([]Question, 0)
questions[0].Content = "whywhywhwy?"
affected, err := engine.Insert(user, &questions)

Notice: If you want to use transaction on inserting, you should use session.Begin() before calling Insert.

5. Chainable APIs

5.1. Chainable APIs for Queries, Execusions and Aggregations

Queries and Aggregations is basically formed by using Get, Find, Count methods, with conjunction of following chainable APIs to form conditions, grouping and ordering:

  • Id([]interface{}) Primary Key lookup

  • Where(string, …interface{}) As SQL conditional WHERE clause

  • And(string, …interface{}) Conditional AND

  • Or(string, …interface{}) Conditional OR

  • Sql(string, …interface{}) Custom SQL query

  • Asc(…string) Ascending ordering on 1 or more fields

  • Desc(…string) Descending ordering on 1 or more fields

  • OrderBy(string) As SQL ORDER BY

  • In(string, …interface{}) As SQL Conditional IN

  • Cols(…string) Explicity specify query or update columns. e.g.,:

engine.Cols("age", "name").Find(&users)
// SELECT age, name FROM user
engine.Cols("age", "name").Update(&user)
// UPDATE user SET age=? AND name=?
  • Omit(...string) Inverse function to Cols, to exclude specify query or update columns. Warning: Don't use with Cols()
engine.Omit("age").Update(&user)
// UPDATE user SET name = ? AND department = ?
  • Distinct(…string) As SQL DISTINCT
engine.Distinct("age", "department").Find(&users)
// SELECT DISTINCT age, department FROM user

Caution: this method will not lookup from caching store

  • Table(nameOrStructPtr interface{}) Specify table name, or if struct pointer is passed into the name is extract from struct type name by IMapper conversion policy

  • Limit(int, …int) As SQL LIMIT with optional second param for OFFSET

  • Top(int) As SQL LIMIT

  • Join(type, tableName, criteria string) As SQL JOIN, support type: either of these values [INNER, LEFT OUTER, CROSS] are supported now tableName: joining table name criteria: join criteria

  • GroupBy(string) As SQL GROUP BY

  • Having(string) As SQL HAVING

5.2. Override default behavior APIs

  • NoAutoTime() No auto timestamp for Created and Updated fields for INSERT and UPDATE

  • NoCache() Disable cache lookup

  • UseBool(...string) xorm's default behavior is fields with 0, "", nil, false, will not be used during query or update, use this method to explicit specify bool type fields for query or update

  • Cascade(bool) Do cascade lookup for associations

5.3.Get one record

Fetch a single object by user

var user = User{Id:27}
has, err := engine.Get(&user)
// or has, err := engine.Id(27).Get(&user)

var user = User{Name:"xlw"}
has, err := engine.Get(&user)

5.4.Find

Fetch multipe objects into a slice or a map, use Find：

var everyone []Userinfo
err := engine.Find(&everyone)

users := make(map[int64]Userinfo)
err := engine.Find(&users)
  • also you can use Where, Limit
var allusers []Userinfo
err := engine.Where("id > ?", "3").Limit(10,20).Find(&allusers) //Get id>3 limit 10 offset 20
  • or you can use a struct query
var tenusers []Userinfo
err := engine.Limit(10).Find(&tenusers, &Userinfo{Name:"xlw"}) //Get All Name="xlw" limit 10 offset 0
  • or In function
var tenusers []Userinfo
err := engine.In("id", 1, 3, 5).Find(&tenusers) //Get All id in (1, 3, 5)
  • The default will query all columns of a table. Use Cols function if you want to select some columns
var tenusers []Userinfo
err := engine.Cols("id", "name").Find(&tenusers) //Find only id and name

5.5.Iterate records

Iterate, like find, but handle records one by one

err := engine.Where("age > ? or name=?)", 30, "xlw").Iterate(new(Userinfo), func(i int, bean interface{})error{
	user := bean.(*Userinfo)
	//do somthing use i and user
})

5.6.Count method usage

An ORM pointer struct is required for Count method in order to determine which table to retrieve from.

user := new(User)
total, err := engine.Where("id >?", 1).Count(user)

6.Update

更新数据使用Update方法，Update方法的第一个参数为需要更新的内容，可以为一个结构体指针或者一个Map[string]interface{}类型。当传入的为结构体指针时，只有非空和0的field才会被作为更新的字段。当传入的为Map类型时，key为数据库Column的名字，value为要更新的内容。

user := new(User)
user.Name = "myname"
affected, err := engine.Id(id).Update(user)

这里需要注意，Update会自动从user结构体中提取非0和非nil得值作为需要更新的内容，因此，如果需要更新一个值为0，则此种方法将无法实现，因此有两种选择：

  1. 通过添加Cols函数指定需要更新结构体中的哪些值，未指定的将不更新，指定了的即使为0也会更新。
affected, err := engine.Id(id).Cols("age").Update(&user)
  1. 通过传入map[string]interface{}来进行更新，但这时需要额外指定更新到哪个表，因为通过map是无法自动检测更新哪个表的。
affected, err := engine.Table(new(User)).Id(id).Update(map[string]interface{}{"age":0})

6.1.Optimistic Lock

To enable object optimistic lock, add 'version' tag value: type User struct { Id int64 Name string Version int xorm:"version" }

The version starts with 1 when inserted to DB. For updating make sure originated version value is used for optimistic lock check.

var user User
engine.Id(1).Get(&user)
// SELECT * FROM user WHERE id = ?
engine.Id(1).Update(&user)
// UPDATE user SET ..., version = version + 1 WHERE id = ? AND version = ?

7.Delete one or more records

Delete one or more records

  • delete by id
err := engine.Id(1).Delete(&User{})
  • delete by other conditions
err := engine.Delete(&User{Name:"xlw"})

8.Execute SQL query

Of course, SQL execution is also provided.

If select then use Query

sql := "select * from userinfo"
results, err := engine.Query(sql)

9.Execute SQL command

If insert, update or delete then use Exec

sql = "update userinfo set username=? where id=?"
res, err := engine.Exec(sql, "xiaolun", 1)

10.Transaction

session := engine.NewSession()
defer session.Close()

// add Begin() before any action
err := session.Begin()	
user1 := Userinfo{Username: "xiaoxiao", Departname: "dev", Alias: "lunny", Created: time.Now()}
_, err = session.Insert(&user1)
if err != nil {
	session.Rollback()
	return
}
user2 := Userinfo{Username: "yyy"}
_, err = session.Where("id = ?", 2).Update(&user2)
if err != nil {
	session.Rollback()
	return
}

_, err = session.Exec("delete from userinfo where username = ?", user2.Username)
if err != nil {
	session.Rollback()
	return
}

// add Commit() after all actions
err = session.Commit()
if err != nil {
	return
}

11.Built-in LRU memory cache provider

  1. Global Cache Xorm implements cache support. Defaultly, it's disabled. If enable it, use below code.
cacher := xorm.NewLRUCacher(xorm.NewMemoryStore(), 1000)
engine.SetDefaultCacher(cacher)

If disable some tables' cache, then:

engine.MapCacher(&user, nil)
  1. Table's Cache If only some tables need cache, then:
cacher := xorm.NewLRUCacher(xorm.NewMemoryStore(), 1000)
engine.MapCacher(&user, cacher)

Caution:

  1. When use Cols methods on cache enabled, the system still return all the columns.

  2. When using Exec method, you should clear cache：

engine.Exec("update user set name = ? where id = ?", "xlw", 1)
engine.ClearCache(new(User))

Cache implement theory below:

cache design

12.xorm tool

xorm commandl line tool

12.1.Reverse command

Please visit xorm tool

13.Examples

Please visit https://github.com/go-xorm/xorm/tree/master/examples

14.Cases

15.FAQ

  • How the xorm tag use both with json?

    Use space.

type User struct {
    Name string `json:"name" xorm:"name"`
}

  • Does xorm support composite primary key?

    Yes. You can use pk tag. All fields have tag will as one primary key by fields order on struct. When use, you can use xorm.PK{1, 2}. For example: Id(xorm.PK{1, 2}).

  • How to use join？

    We can use Join() and extends tag to do join operation. For example:

    type Userinfo struct { Id int64 Name string DetailId int64 }

    type Userdetail struct { Id int64 Gender int }

    type User struct { Userinfo xorm:"extends" Userdetail xorm:"extends" }

    var users = make([]User, 0) err := engine.Table(&Userinfo{}).Join("LEFT", "userdetail", "userinfo.detail_id = userdetail.id").Find(&users)

    //assert(User.Userinfo.Id != 0 && User.Userdetail.Id != 0)

Please notice that Userinfo field on User should be before Userdetail because of the order on join SQL stsatement. If the order is wrong, the same name field may be set a wrong value.

Of course, If join statment is very long, you could directly use Sql():

err := engine.Sql("select * from userinfo, userdetail where userinfo.detail_id = userdetail.id").Find(&users)

//assert(User.Userinfo.Id != 0 && User.Userdetail.Id != 0)