gorm

paginate - Gorm Pagination

Simple way to paginate Gorm result. paginate is compatible with net/http and fasthttp. This library also supports many frameworks are based on net/http or fasthttp.

Table Of Contents

• Installation
• Configuration
• Paginate using http request
• Example usage
  • net/http
  • Fasthttp
  • Mux Router
  • Fiber
  • Echo
  • Gin
  • Martini
  • Beego
  • jQuery DataTable Integration
  • jQuery Select2 Integration
• Filter format
• Customize default configuration
• Override results
• Field Selector
• Dynamic Field Selector
• Speed up response with cache
  • In Memory Cache
  • Disk Cache
  • Redis Cache
  • Elasticsearch Cache
  • Custom cache
  • Clean up cache
• Limitations
• License

Installation

go get -u github.com/morkid/paginate

Configuration

var db *gorm.DB = ...
var req *http.Request = ...
// or
// var req *fasthttp.Request

model := db.Where("id > ?", 1).Model(&Article{})
pg := paginate.New()
page := pg.Response(model, req, &[]Article{})
// or 
page := pg.With(model).Request(req).Response(&[]Article{})

log.Println(page.Total)
log.Println(page.Items)
log.Println(page.First)
log.Println(page.Last)

you can customize config with paginate.Config struct.

pg := paginate.New(&paginate.Config{
    DefaultSize: 50,
})

see more about customize default configuration.

Note that Response was marked as a deprecated function. Please use With instead.
Old: pg.Response(model, req, &[]Article{}),
New: pg.With(model).Request(req).Response(&[]Article{})

Paginate using http request

example paging, sorting and filtering:

1. http://localhost:3000/?size=10&page=0&sort=-name
   produces:

   SELECT * FROM user ORDER BY name DESC LIMIT 10 OFFSET 0
   

   JSON response:

   {
       // result items
       "items": [
           {
               "id": 1,
               "name": "john",
               "age": 20
           }
       ],
       "page": 0, // current selected page
       "size": 10, // current limit or size per page
       "max_page": 0, // maximum page
       "total_pages": 1, // total pages
       "total": 1, // total matches including next page
       "visible": 1, // total visible on current page
       "last": true, // if response is first page
       "first": true // if response is last page
   }
   

2. http://localhost:3000/?size=10&page=1&sort=-name,id
   produces:

   SELECT * FROM user ORDER BY name DESC, id ASC LIMIT 10 OFFSET 10
   

3. http://localhost:3000/?filters=["name","john"]
   produces:

   SELECT * FROM user WHERE name = 'john' LIMIT 10 OFFSET 0
   

4. http://localhost:3000/?filters=["name","like","john"]
   produces:

   SELECT * FROM user WHERE name LIKE '%john%' LIMIT 10 OFFSET 0
   

5. http://localhost:3000/?filters=["age","between",[20, 25]]
   produces:

   SELECT * FROM user WHERE ( age BETWEEN 20 AND 25 ) LIMIT 10 OFFSET 0
   

6. http://localhost:3000/?filters=[["name","like","john%25"],["OR"],["age","between",[20, 25]]]
   produces:

   SELECT * FROM user WHERE (
      (name LIKE '%john\%%' ESCAPE '\') OR (age BETWEEN (20 AND 25))
   ) LIMIT 10 OFFSET 0
   

7. http://localhost:3000/?filters=[[["name","like","john"],["AND"],["name","not like","doe"]],["OR"],["age","between",[20, 25]]]
   produces:

   SELECT * FROM user WHERE (
      (
          (name LIKE '%john%')
                  AND
          (name NOT LIKE '%doe%')
      ) 
      OR 
      (age BETWEEN (20 AND 25))
   ) LIMIT 10 OFFSET 0
   

8. http://localhost:3000/?filters=["name","IS NOT",null]
   produces:

   SELECT * FROM user WHERE name IS NOT NULL LIMIT 10 OFFSET 0
   

9. Using POST method:

   curl -X POST \
   -H 'Content-type: application/json' \
   -d '{"page":"1","size":"20","sort":"-name","filters":["name","john"]}' \
   http://localhost:3000/
   

Example usage

NetHTTP Example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()

    http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
        model := db.Joins("User").Model(&Article{})
        paginated := pg.Response(model, r, &[]Article{})
        j, _ := json.Marshal(paginated)
        w.Header().Set("Content-type", "application/json")
        w.Write(j)
    })

    log.Fatal(http.ListenAndServe(":3000", nil))
}

Fasthttp Example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()

    fasthttp.ListenAndServe(":3000", func(ctx *fasthttp.RequestCtx) {
        model := db.Joins("User").Model(&Article{})
        paginated := pg.Response(model, &ctx.Request, &[]Article{})
        j, _ := json.Marshal(paginated)
        ctx.SetContentType("application/json")
        ctx.SetBody(j)
    })
}

Mux Router Example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()
    app := mux.NewRouter()
    app.HandleFunc("/", func(w http.ResponseWriter, req *http.Request) {
        model := db.Joins("User").Model(&Article{})
        paginated := pg.Response(model, req, &[]Article{})
        j, _ := json.Marshal(paginated)
        w.Header().Set("Content-type", "application/json")
        w.Write(j)
    }).Methods("GET")
    http.Handle("/", app)
    http.ListenAndServe(":3000", nil)
}

Fiber example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()
    app := fiber.New()
    app.Get("/", func(c *fiber.Ctx) error {
        model := db.Joins("User").Model(&Article{})
        return c.JSON(pg.Response(model, c.Request(), &[]Article{}))
    })

    app.Listen(":3000")
}

Echo example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()
    app := echo.New()
    app.GET("/", func(c echo.Context) error {
        model := db.Joins("User").Model(&Article{})
        return c.JSON(200, pg.Response(model, c.Request(), &[]Article{}))
    })

    app.Logger.Fatal(app.Start(":3000"))
}

Gin Example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()
    app := gin.Default()
    app.GET("/", func(c *gin.Context) {
        model := db.Joins("User").Model(&Article{})
        c.JSON(200, pg.Response(model, c.Request, &[]Article{}))
    })
    app.Run(":3000")
}

Martini Example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()
    app := martini.Classic()
    app.Use(render.Renderer())
    app.Get("/", func(req *http.Request, r render.Render) {
        model := db.Joins("User").Model(&Article{})
        r.JSON(200, pg.Response(model, req, &[]Article{}))
    })
    app.Run()
}

Beego Example

package main

import (
    "github.com/morkid/paginate"
    ...
)

func main() {
    // var db *gorm.DB
    pg := paginate.New()
    web.Get("/", func(c *context.Context) {
        model := db.Joins("User").Model(&Article{})
        c.Output.JSON(
            pg.Response(model, c.Request, &[]Article{}), false, false)
    })
    web.Run(":3000")
}

jQuery DataTable Integration

var logicalOperator = "OR"

$('#myTable').DataTable({

    columns: [
        {
            title: "Author",
            data: "user.name"
        }, {
            title: "Title",
            data: "title"
        }
    ],

    processing: true,
    
    serverSide: true,

    ajax: {
        cache: true,
        url: "http://localhost:3000/articles",
        dataSrc: function(json) {
            json.recordsTotal = json.visible
            json.recordsFiltered = json.total
            return json.items
        },
        data: function(params) {
            var custom = {
                page: !params.start ? 0 : Math.round(params.start / params.length),
                size: params.length
            }

            if (params.order.length > 0) {
                var sorts = []
                for (var o in params.order) {
                    var order = params.order[o]
                    if (params.columns[order.column].orderable != false) {
                        var sort = order.dir != 'desc' ? '' : '-'
                        sort += params.columns[order.column].data
                        sorts.push(sort)
                    }
                }
                custom.sort = sorts.join()
            }

            if (params.search.value) {
                var columns = []
                for (var c in params.columns) {
                    var col = params.columns[c]
                    if (col.searchable == false) {
                        continue
                    }
                    columns.push(JSON.stringify([col.data, "like", encodeURIComponent(params.search.value.toLowerCase())]))
                }
                custom.filters = '[' + columns.join(',["' + logicalOperator + '"],') + ']'
            }

            return custom
        }
    },
})

jQuery Select2 Integration

$('#mySelect').select2({
    ajax: {
        url: "http://localhost:3000/users",
        processResults: function(json) {
            json.items.forEach(function(item) {
                item.text = item.name
            })
            // optional
            if (json.first) json.items.unshift({id: 0, text: 'All'})

            return {
                results: json.items,
                pagination: {
                    more: json.last == false
                }
            }
        },
        data: function(params) {
            var filters = [
                ["name", "like", params.term]
            ]

            return {
                filters: params.term ? JSON.stringify(filters) : "",
                sort: "name",
                page: params.page && params.page - 1 ? params.page - 1 : 0
            }
        },
    }
})

Filter format

The format of filter param is a json encoded of multidimensional array.
Maximum array members is three, first index is column_name, second index is operator and third index is values, you can also pass array to values.

// Format:
["column_name", "operator", "values"]

// Example:
["age", "=", 20]
// Shortcut:
["age", 20]

// Produces:
// WHERE age = 20

Single array member is known as Logical Operator.

// Example
[["age", "=", 20],["or"],["age", "=", 25]]

// Produces:
// WHERE age = 20 OR age = 25

You are allowed to send array inside a value.

["age", "between", [20, 30] ]
// Produces:
// WHERE age BETWEEN 20 AND 30

["age", "not in", [20, 21, 22, 23, 24, 25, 26, 26] ]
// Produces:
// WHERE age NOT IN(20, 21, 22, 23, 24, 25, 26, 26)

You can filter nested condition with deep array.

[
    [
        ["age", ">", 20],
        ["and"]
        ["age", "<", 30]
    ],
    ["and"],
    ["name", "like", "john"],
    ["and"],
    ["name", "like", "doe"]
]
// Produces:
// WHERE ( (age > 20 AND age < 20) and name like '%john%' and name like '%doe%' )

For null value, you can send string "null" or null value, (lower)

// Wrong request
[ "age", "is", NULL ]
[ "age", "is", Null ]
[ "age", "is not", NULL ]
[ "age", "is not", Null ]

// Right request
[ "age", "is", "NULL" ]
[ "age", "is", "Null" ]
[ "age", "is", "null" ]
[ "age", "is", null ]
[ "age", null ]
[ "age", "is not", "NULL" ]
[ "age", "is not", "Null" ]
[ "age", "is not", "null" ]
[ "age", "is not", null ]

Customize default configuration

You can customize the default configuration with paginate.Config struct.

pg := paginate.New(&paginate.Config{
    DefaultSize: 50,
})

Override results

You can override result with custom function.

// var db = *gorm.DB
// var httpRequest ... net/http or fasthttp instance
// Example override function
override := func(article *Article) {
    if article.UserID > 0 {
        article.Title = fmt.Sprintf(
            "%s written by %s", article.Title, article.User.Name)
    }
}

var articles []Article
model := db.Joins("User").Model(&Article{})

pg := paginate.New()
result := pg.Response(model, httpRequest, &articles)
for index := range articles {
    override(&articles[index])
}

log.Println(result.Items)

Field selector

To implement a custom field selector, struct properties must have a json tag with omitempty.

// real gorm model
type User {
    gorm.Model
    Name string `json:"name"`
    Age  int64  `json:"age"`
}

// fake gorm model
type UserNullable {
    ID        *string    `json:"id,omitempty"`
    CreatedAt *time.Time `json:"created_at,omitempty"`
    UpdatedAt *time.Time `json:"updated_at,omitempty"`
    Name      *string    `json:"name,omitempty"`
    Age       *int64     `json:"age,omitempty"`
}

// usage
nameAndIDOnly := []string{"name","id"}
model := db.Model(&User{})

page := pg.With(model).
   Request(req).
   Fields(nameAndIDOnly).
   Response([]&UserNullable{})

// response
{
    "items": [
        {
            "id": 1,
            "name": "John"
        }
    ],
    ...
}

Dynamic field selector

If the request contains query parameter fields (eg: ?fieilds=name,id), then the response will show only name and id. To activate this feature, please set FieldSelectorEnabled to true.

config := paginate.Config{
    FieldSelectorEnabled: true,
}

pg := paginate.New(config)

Speed up response with cache

You can speed up results without looking database directly with cache adapter. See more about cache adapter.

In memory cache

in memory cache is not recommended for production environment:

import (
    "github.com/morkid/gocache"
    ...
)

func main() {
    ...
    adapterConfig := gocache.InMemoryCacheConfig{
        ExpiresIn: 1 * time.Hour,
    }
    pg := paginate.New(&paginate.Config{
        CacheAdapter: gocache.NewInMemoryCache(adapterConfig),
    })

    page := pg.With(model).
               Request(req).
               Cache("article"). // set cache name
               Response(&[]Article{})
    ...
}

Disk cache

Disk cache will create a file for every single request. You can use disk cache if you don't care about inode.

import (
    "github.com/morkid/gocache"
    ...
)

func main() {
    adapterConfig := gocache.DiskCacheConfig{
        Directory: "/writable/path/to/my-cache-dir",
        ExpiresIn: 1 * time.Hour,
    }
    pg := paginate.New(&paginate.Config{
        CacheAdapter: gocache.NewDiskCache(adapterConfig),
    })

    page := pg.With(model).
               Request(req).
               Cache("article"). // set cache name
               Response(&[]Article{})
    ...
}

Redis cache

Redis cache require redis client for golang.

import (
    cache "github.com/morkid/gocache-redis/v8"
    "github.com/go-redis/redis/v8"
    ...
)

func main() {
    client := redis.NewClient(&redis.Options{
        Addr:     "localhost:6379",
        Password: "",
        DB:       0,
    })

    adapterConfig := cache.RedisCacheConfig{
        Client:    client,
        ExpiresIn: 1 * time.Hour,
    }
    pg := paginate.New(&paginate.Config{
        CacheAdapter: cache.NewRedisCache(adapterConfig),
    })

    page := pg.With(model).
               Request(req).
               Cache("article").
               Response(&[]Article{})
    ...
}

if your code already adopts another redis client, you can implement the redis adapter according to its version. See more about redis adapter.

Elasticsearch cache

Elasticsearch cache require official elasticsearch client for golang.

import (
    cache "github.com/morkid/gocache-elasticsearch/v7"
    "github.com/elastic/go-elasticsearch/v7"
    ...
)

func main() {
    config := elasticsearch.Config{
        Addresses: []string{
            "http://localhost:9200",
        },
    }
    es, err := elasticsearch.NewClient(config)
    if nil != err {
        panic(err)
    }

    adapterConfig := cache.ElasticCacheConfig{
        Client:    es,
        Index:     "exampleproject",
        ExpiresIn: 1 * time.Hour,
    }
    pg := paginate.New(&paginate.Config{
        CacheAdapter: cache.NewElasticCache(adapterConfig),
    })

    page := pg.With(model).
               Request(req).
               Cache("article").
               Response(&[]Article{})
    ...
}

if your code already adopts another elasticsearch client, you can implement the elasticsearch adapter according to its version. See more about elasticsearch adapter.

Custom cache

Create your own cache adapter by implementing gocache AdapterInterface. See more about cache adapter.

// AdapterInterface interface
type AdapterInterface interface {
    // Set cache with key
    Set(key string, value string) error
    // Get cache by key
    Get(key string) (string, error)
    // IsValid check if cache is valid
    IsValid(key string) bool
    // Clear clear cache by key
    Clear(key string) error
    // ClearPrefix clear cache by key prefix
    ClearPrefix(keyPrefix string) error
    // Clear all cache
    ClearAll() error
}

Clean up cache

Clear cache by cache name

pg.ClearCache("article")

Clear multiple cache

pg.ClearCache("cache1", "cache2", "cache3")

Clear all cache

pg.ClearAllCache()

Limitations

Paginate doesn't support has many relationship. You can make API with separated endpoints for parent and child:

GET /users

{
    "items": [
        {
            "id": 1,
            "name": "john",
            "age": 20,
            "addresses": [...] // doesn't support
        }
    ],
    ...
}

Best practice:

GET /users
{
    "items": [
        {
            "id": 1,
            "name": "john",
            "age": 20
        }
    ],
    ...
}

GET /users/1/addresses
{
    "items": [
        {
            "id": 1,
            "name": "home",
            "street": "home street"
            "user": {
                "id": 1,
                "name": "john",
                "age": 20
            }
        }
    ],
    ...
}

Paginate doesn't support for customized json or table field name.
Make sure your struct properties have same name with gorm column and json property before you expose them.

Example bad configuration:

type User struct {
    gorm.Model
    UserName       string `gorm:"column:nickname" json:"name"`
    UserAddress    string `gorm:"column:user_address" json:"address"`
}

// request: GET /path/to/endpoint?sort=-name,address
// response: "items": [] with sql error (column name not found)

Best practice:

type User struct {
    gorm.Model
    Name       string `gorm:"column:name" json:"name"`
    Address    string `gorm:"column:address" json:"address"`
}

License

Published under the MIT License.