开源软件名称：

开源软件地址：

开源编程语言：

开源软件介绍：

RethinkDB-go - RethinkDB Driver for Go

Go driver for RethinkDB

Current version: v6.2.1 (RethinkDB v2.4)

Please note that this version of the driver only supports versions of RethinkDB using the v0.4 protocol (any versions of the driver older than RethinkDB 2.0 will not work).

If you need any help you can find me on the RethinkDB slack in the #gorethink channel.

Installation

go get gopkg.in/rethinkdb/rethinkdb-go.v6

Replace v6 with v5 or v4 to use previous versions.

Example

package rethinkdb_test

import (
	"fmt"
	"log"

	r "gopkg.in/rethinkdb/rethinkdb-go.v6"
)

func Example() {
	session, err := r.Connect(r.ConnectOpts{
		Address: url, // endpoint without http
	})
	if err != nil {
		log.Fatalln(err)
	}

	res, err := r.Expr("Hello World").Run(session)
	if err != nil {
		log.Fatalln(err)
	}

	var response string
	err = res.One(&response)
	if err != nil {
		log.Fatalln(err)
	}

	fmt.Println(response)

	// Output:
	// Hello World
}

Connection

Basic Connection

Setting up a basic connection with RethinkDB is simple:

func ExampleConnect() {
	var err error

	session, err = r.Connect(r.ConnectOpts{
		Address: url,
	})
	if err != nil {
		log.Fatalln(err.Error())
	}
}

See the documentation for a list of supported arguments to Connect().

Connection Pool

The driver uses a connection pool at all times, by default it creates and frees connections automatically. It's safe for concurrent use by multiple goroutines.

To configure the connection pool InitialCap, MaxOpen and Timeout can be specified during connection. If you wish to change the value of InitialCap or MaxOpen during runtime then the functions SetInitialPoolCap and SetMaxOpenConns can be used.

func ExampleConnect_connectionPool() {
	var err error

	session, err = r.Connect(r.ConnectOpts{
		Address:    url,
		InitialCap: 10,
		MaxOpen:    10,
	})
	if err != nil {
		log.Fatalln(err.Error())
	}
}

Connect to a cluster

To connect to a RethinkDB cluster which has multiple nodes you can use the following syntax. When connecting to a cluster with multiple nodes queries will be distributed between these nodes.

func ExampleConnect_cluster() {
	var err error

	session, err = r.Connect(r.ConnectOpts{
		Addresses: []string{url},
		//  Addresses: []string{url1, url2, url3, ...},
	})
	if err != nil {
		log.Fatalln(err.Error())
	}
}

When DiscoverHosts is true any nodes are added to the cluster after the initial connection then the new node will be added to the pool of available nodes used by RethinkDB-go. Unfortunately the canonical address of each server in the cluster MUST be set as otherwise clients will try to connect to the database nodes locally. For more information about how to set a RethinkDB servers canonical address set this page http://www.rethinkdb.com/docs/config-file/.

User Authentication

To login with a username and password you should first create a user, this can be done by writing to the users system table and then grant that user access to any tables or databases they need access to. This queries can also be executed in the RethinkDB admin console.

err := r.DB("rethinkdb").Table("users").Insert(map[string]string{
    "id": "john",
    "password": "p455w0rd",
}).Exec(session)
...
err = r.DB("blog").Table("posts").Grant("john", map[string]bool{
    "read": true,
    "write": true,
}).Exec(session)
...

Finally the username and password should be passed to Connect when creating your session, for example:

session, err := r.Connect(r.ConnectOpts{
    Address: "localhost:28015",
    Database: "blog",
    Username: "john",
    Password: "p455w0rd",
})

Please note that DiscoverHosts will not work with user authentication at this time due to the fact that RethinkDB restricts access to the required system tables.

Query Functions

This library is based on the official drivers so the code on the API page should require very few changes to work.

To view full documentation for the query functions check the API reference or GoDoc

Slice Expr Example

r.Expr([]interface{}{1, 2, 3, 4, 5}).Run(session)

Map Expr Example

r.Expr(map[string]interface{}{"a": 1, "b": 2, "c": 3}).Run(session)

Get Example

r.DB("database").Table("table").Get("GUID").Run(session)

Map Example (Func)

r.Expr([]interface{}{1, 2, 3, 4, 5}).Map(func (row Term) interface{} {
    return row.Add(1)
}).Run(session)

Map Example (Implicit)

r.Expr([]interface{}{1, 2, 3, 4, 5}).Map(r.Row.Add(1)).Run(session)

Between (Optional Args) Example

r.DB("database").Table("table").Between(1, 10, r.BetweenOpts{
    Index: "num",
    RightBound: "closed",
}).Run(session)

For any queries which use callbacks the function signature is important as your function needs to be a valid RethinkDB-go callback, you can see an example of this in the map example above. The simplified explanation is that all arguments must be of type r.Term, this is because of how the query is sent to the database (your callback is not actually executed in your Go application but encoded as JSON and executed by RethinkDB). The return argument can be anything you want it to be (as long as it is a valid return value for the current query) so it usually makes sense to return interface{}. Here is an example of a callback for the conflict callback of an insert operation:

r.Table("test").Insert(doc, r.InsertOpts{
    Conflict: func(id, oldDoc, newDoc r.Term) interface{} {
        return newDoc.Merge(map[string]interface{}{
            "count": oldDoc.Add(newDoc.Field("count")),
        })
    },
})

Optional Arguments

As shown above in the Between example optional arguments are passed to the function as a struct. Each function that has optional arguments as a related struct. This structs are named in the format FunctionNameOpts, for example BetweenOpts is the related struct for Between.

Cancelling queries

For query cancellation use Context argument at RunOpts. If Context is nil and ReadTimeout or WriteTimeout is not 0 from ConnectionOpts, Context will be formed by summation of these timeouts.

For unlimited timeouts for Changes() pass context.Background().

Results

Different result types are returned depending on what function is used to execute the query.

• Run returns a cursor which can be used to view all rows returned.
• RunWrite returns a WriteResponse and should be used for queries such as Insert, Update, etc...
• Exec sends a query to the server and closes the connection immediately after reading the response from the database. If you do not wish to wait for the response then you can set the NoReply flag.

Example:

res, err := r.DB("database").Table("tablename").Get(key).Run(session)
if err != nil {
    // error
}
defer res.Close() // Always ensure you close the cursor to ensure connections are not leaked

Cursors have a number of methods available for accessing the query results

• Next retrieves the next document from the result set, blocking if necessary.
• All retrieves all documents from the result set into the provided slice.
• One retrieves the first document from the result set.

Examples:

var row interface{}
for res.Next(&row) {
    // Do something with row
}
if res.Err() != nil {
    // error
}

var rows []interface{}
err := res.All(&rows)
if err != nil {
    // error
}

var row interface{}
err := res.One(&row)
if err == r.ErrEmptyResult {
    // row not found
}
if err != nil {
    // error
}

Encoding/Decoding

When passing structs to Expr(And functions that use Expr such as Insert, Update) the structs are encoded into a map before being sent to the server. Each exported field is added to the map unless

• the field's tag is "-", or
• the field is empty and its tag specifies the "omitempty" option.

Each fields default name in the map is the field name but can be specified in the struct field's tag value. The "rethinkdb" key in the struct field's tag value is the key name, followed by an optional comma and options. Examples:

// Field is ignored by this package.
Field int `rethinkdb:"-"`
// Field appears as key "myName".
Field int `rethinkdb:"myName"`
// Field appears as key "myName" and
// the field is omitted from the object if its value is empty,
// as defined above.
Field int `rethinkdb:"myName,omitempty"`
// Field appears as key "Field" (the default), but
// the field is skipped if empty.
// Note the leading comma.
Field int `rethinkdb:",omitempty"`
// When the tag name includes an index expression
// a compound field is created
Field1 int `rethinkdb:"myName[0]"`
Field2 int `rethinkdb:"myName[1]"`

NOTE: It is strongly recommended that struct tags are used to explicitly define the mapping between your Go type and how the data is stored by RethinkDB. This is especially important when using an Id field as by default RethinkDB will create a field named id as the primary key (note that the RethinkDB field is lowercase but the Go version starts with a capital letter).

When encoding maps with non-string keys the key values are automatically converted to strings where possible, however it is recommended that you use strings where possible (for example map[string]T).

If you wish to use the json tags for RethinkDB-go then you can call SetTags("rethinkdb", "json") when starting your program, this will cause RethinkDB-go to check for json tags after checking for rethinkdb tags. By default this feature is disabled. This function will also let you support any other tags, the driver will check for tags in the same order as the parameters.

NOTE: Old-style gorethink struct tags are supported but deprecated.

Pseudo-types

RethinkDB contains some special types which can be used to store special value types, currently supports are binary values, times and geometry data types. RethinkDB-go supports these data types natively however there are some gotchas:

• Time types: To store times in RethinkDB with RethinkDB-go you must pass a time.Time value to your query, due to the way Go works type aliasing or embedding is not support here
• Binary types: To store binary data pass a byte slice ([]byte) to your query
• Geometry types: As Go does not include any built-in data structures for storing geometry data RethinkDB-go includes its own in the github.com/rethinkdb/rethinkdb-go/types package, Any of the types (Geometry, Point, Line and Lines) can be passed to a query to create a RethinkDB geometry type.

Compound Keys

RethinkDB unfortunately does not support compound primary keys using multiple fields however it does support compound keys using an array of values. For example if you wanted to create a compound key for a book where the key contained the author ID and book name then the ID might look like this ["author_id", "book name"]. Luckily RethinkDB-go allows you to easily manage these keys while keeping the fields separate in your structs. For example:

type Book struct {
  AuthorID string `rethinkdb:"id[0]"`
  Name     string `rethinkdb:"id[1]"`
}
// Creates the following document in RethinkDB
{"id": [AUTHORID, NAME]}

References

Sometimes you may want to use a Go struct that references a document in another table, instead of creating a new struct which is just used when writing to RethinkDB you can annotate your struct with the reference tag option. This will tell RethinkDB-go that when encoding your data it should "pluck" the ID field from the nested document and use that instead.

This is all quite complicated so hopefully this example should help. First lets assume you have two types Author and Book and you want to insert a new book into your database however you dont want to include the entire author struct in the books table. As you can see the Author field in the Book struct has some extra tags, firstly we have added the reference tag option which tells RethinkDB-go to pluck a field from the Author struct instead of inserting the whole author document. We also have the rethinkdb_ref tag which tells RethinkDB-go to look for the id field in the Author document, without this tag RethinkDB-go would instead look for the author_id field.

type Author struct {
    ID      string  `rethinkdb:"id,omitempty"`
    Name    string  `rethinkdb:"name"`
}

type Book struct {
    ID      string  `rethinkdb:"id,omitempty"`
    Title   string  `rethinkdb:"title"`
    Author  Author `rethinkdb:"author_id,reference" rethinkdb_ref:"id"`
}

The resulting data in RethinkDB should look something like this:

{
    "author_id": "author_1",
    "id":  "book_1",
    "title":  "The Hobbit"
}

If you wanted to read back the book with the author included then you could run the following RethinkDB-go query:

r.Table("books").Get("1").Merge(func(p r.Term) interface{} {
    return map[string]interface{}{
        "author_id": r.Table("authors").Get(p.Field("author_id")),
    }
}).Run(session)

You are also able to reference an array of documents, for example if each book stored multiple authors you could do the following:

type Book struct {
    ID       string  `rethinkdb:"id,omitempty"`
    Title    string  `rethinkdb:"title"`
    Authors  []Author `rethinkdb:"author_ids,reference" rethinkdb_ref:"id"`
}

{
    "author_ids": ["author_1", "author_2"],
    "id":  "book_1",
    "title":  "The Hobbit"
}

The query for reading the data back is slightly more complicated but is very similar:

r.Table("books").Get("book_1").Merge(func(p r.Term) interface{} {
    return map[string]interface{}{
        "author_ids": r.Table("authors").GetAll(r.Args(p.Field("author_ids"))).CoerceTo("array"),
    }
})

Custom Marshalers/Unmarshalers

Sometimes the default behaviour for converting Go types to and from ReQL is not desired, for these situations the driver allows you to implement both the Marshaler and Unmarshaler interfaces. These interfaces might look familiar if you are using to using the encoding/json package however instead of dealing with []byte the interfaces deal with interface{} values (which are later encoded by the encoding/json package when communicating with the database).

An good example of how to use these interfaces is in the types package, in this package the Point type is encoded as the GEOMETRY pseudo-type instead of a normal JSON object.

On the other side, you can implement external encode/decode functions with SetTypeEncoding function.

Logging

By default the driver logs are disabled however when enabled the driver will log errors when it fails to connect to the database. If you would like more verbose error logging you can call r.SetVerbose(true).

Alternatively if you wish to modify the logging behaviour you can modify the logger provided by github.com/sirupsen/logrus. For example the following code completely disable the logger:

// Enabled
r.Log.Out = os.Stderr
// Disabled
r.Log.Out = ioutil.Discard

Tracing

The driver supports opentracing-go. You can enable this feature by setting UseOpentracing to true in the ConnectOpts. Then driver will expect opentracing.Span in the RunOpts.Context and will start new child spans for queries. Also you need to configure tracer in your program by yourself.

The driver starts span for the whole query, from the first byte is sent to the cursor closed, and second-level span for each query for fetching data.

So you can trace how much time you program spends for RethinkDB queries.

Mocking

The driver includes the ability to mock queries meaning that you can test your code without needing to talk to a real RethinkDB cluster, this is perfect for ensuring that your application has high unit test coverage.

To write tests with mocking you should create an instance of Mock and then setup expectations using On and Return. Expectations allow you to define what results should be returned when a known query is executed, they are configured by passing the query term you want to mock to On and then the response and error to Return, if a non-nil error is passed to Return then any time that query is executed the error will be returned, if no error is passed then a cursor will be built using the value passed to Return. Once all your expectations have been created you should then execute you queries using the Mock instead of a Session.

Here is an example that shows how to mock a query that returns multiple rows and the resulting cursor can be used as normal.

func TestSomething(t *testing.T) {
	mock := r.NewMock()
	mock.On(r.Table("people")).Return([]interface{}{
		map[string]interface{}{"id": 1, "name": "John Smith"},
		map[string]interface{}{"id": 2, "name": "Jane Smith"},
	}, nil)

	cursor, err := r.Table("people").Run(mock)
	if err != nil {
		t.Errorf("err is: %v", err)
	}

	var rows []interface{}
	err = cursor.All(&rows)
	if err != nil {
		t.Errorf("err is: %v", err)
	}

	// Test result of rows

	mock.AssertExpectations(t)
}

If you want the cursor to block on some of the response values, you can pass in a value of type chan interface{} and the cursor will block until a value is available to read on the channel. Or you can pass in a function with signature func() interface{}: the cursor will call the function (which may block). Here is the example above adapted to use a channel.

func TestSomething(t *testing.T) {
	mock := r.NewMock()
	ch := make(chan []interface{})
	mock.On(r.Table("people")).Return(ch, nil)
	go func() {
		ch <- []interface{}{
			map[string]interface{}{"id": 1, "name": "John Smith"},
			map[string]interface{}{"id": 2, "name": "Jane Smith"},
		}
		ch <- []interface{}{map[string]interface{}{"id": 3, "name": "Jack Smith"}}
		close(ch)
	}()
	cursor, err := r.Table("people").Run(mock)
	if err != nil {
		t.Errorf("err is: %v", err)
	}

	var rows []interface{}
	err = cursor.All(&rows)
	if err != nil {
		t.Errorf("err is: %v", err)
	}

	// Test result of rows

	mock.AssertExpectations(t)
}

The mocking implementation is based on amazing https://github.com/stretchr/t