README
¶
⚒ dig 
A reflection based dependency injection toolkit for Go.
Good for:
- Powering an application framework, e.g. Fx.
- Resolving the object graph during process startup.
Bad for:
- Using in place of an application framework, e.g. Fx.
- Resolving dependencies after the process has already started.
- Exposing to user-land code as a Service Locator.
Installation
We recommend consuming SemVer major version 1 using
your dependency manager of choice.
$ glide get 'go.uber.org/dig#^1'
$ dep ensure -add "go.uber.org/dig@v1"
$ go get 'go.uber.org/dig@v1'
Stability
This library is v1 and follows SemVer strictly.
No breaking changes will be made to exported APIs before v2.0.0.
Documentation
¶
Overview ¶
Package dig provides an opinionated way of resolving object dependencies.
Status ¶
STABLE. No breaking changes will be made in this major version.
Container ¶
Dig exposes type Container as an object capable of resolving a directed acyclic dependency graph. Use the New function to create one.
c := dig.New()
Provide ¶
Constructors for different types are added to the container by using the Provide method. A constructor can declare a dependency on another type by simply adding it as a function parameter. Dependencies for a type can be added to the graph both, before and after the type was added.
err := c.Provide(func(conn *sql.DB) (*UserGateway, error) {
// ...
})
if err != nil {
// ...
}
if err := c.Provide(newDBConnection); err != nil {
// ...
}
Multiple constructors can rely on the same type. The container creates a singleton for each retained type, instantiating it at most once when requested directly or as a dependency of another type.
err := c.Provide(func(conn *sql.DB) *CommentGateway {
// ...
})
if err != nil {
// ...
}
Constructors can declare any number of dependencies as parameters and optionally, return errors.
err := c.Provide(func(u *UserGateway, c *CommentGateway) (*RequestHandler, error) {
// ...
})
if err != nil {
// ...
}
if err := c.Provide(newHTTPServer); err != nil {
// ...
}
Constructors can also return multiple results to add multiple types to the container.
err := c.Provide(func(conn *sql.DB) (*UserGateway, *CommentGateway, error) {
// ...
})
if err != nil {
// ...
}
Constructors that accept a variadic number of arguments are treated as if they don't have those arguments. That is,
func NewVoteGateway(db *sql.DB, options ...Option) *VoteGateway
Is treated the same as,
func NewVoteGateway(db *sql.DB) *VoteGateway
The constructor will be called with all other dependencies and no variadic arguments.
Invoke ¶
Types added to to the container may be consumed by using the Invoke method. Invoke accepts any function that accepts one or more parameters and optionally, returns an error. Dig calls the function with the requested type, instantiating only those types that were requested by the function. The call fails if any type or its dependencies (both direct and transitive) were not available in the container.
err := c.Invoke(func(l *log.Logger) {
// ...
})
if err != nil {
// ...
}
err := c.Invoke(func(server *http.Server) error {
// ...
})
if err != nil {
// ...
}
Any error returned by the invoked function is propagated back to the caller.
Parameter Objects ¶
Constructors declare their dependencies as function parameters. This can very quickly become unreadable if the constructor has a lot of dependencies.
func NewHandler(users *UserGateway, comments *CommentGateway, posts *PostGateway, votes *VoteGateway, authz *AuthZGateway) *Handler {
// ...
}
A pattern employed to improve readability in a situation like this is to create a struct that lists all the parameters of the function as fields and changing the function to accept that struct instead. This is referred to as a parameter object.
Dig has first class support for parameter objects: any struct embedding dig.In gets treated as a parameter object. The following is equivalent to the constructor above.
type HandlerParams struct {
dig.In
Users *UserGateway
Comments *CommentGateway
Posts *PostGateway
Votes *VoteGateway
AuthZ *AuthZGateway
}
func NewHandler(p HandlerParams) *Handler {
// ...
}
Handlers can receive any combination of parameter objects and parameters.
func NewHandler(p HandlerParams, l *log.Logger) *Handler {
// ...
}
Result Objects ¶
Result objects are the flip side of parameter objects. These are structs that represent multiple outputs from a single function as fields in the struct. Structs embedding dig.Out get treated as result objects.
func SetupGateways(conn *sql.DB) (*UserGateway, *CommentGateway, *PostGateway, error) {
// ...
}
The above is equivalent to,
type Gateways struct {
dig.Out
Users *UserGateway
Comments *CommentGateway
Posts *PostGateway
}
func SetupGateways(conn *sql.DB) (Gateways, error) {
// ...
}
Optional Dependencies ¶
Constructors often don't have a hard dependency on some types and are able to operate in a degraded state when that dependency is missing. Dig supports declaring dependencies as optional by adding an `optional:"true"` tag to fields of a dig.In struct.
Fields in a dig.In structs that have the `optional:"true"` tag are treated as optional by Dig.
type UserGatewayParams struct {
dig.In
Conn *sql.DB
Cache *redis.Client `optional:"true"`
}
If an optional field is not available in the container, the constructor will receive a zero value for the field.
func NewUserGateway(p UserGatewayParams, log *log.Logger) (*UserGateway, error) {
if p.Cache == nil {
log.Print("Logging disabled")
}
// ...
}
Constructors that declare dependencies as optional MUST handle the case of those dependencies being absent.
The optional tag also allows adding new dependencies without breaking existing consumers of the constructor.
Named Values ¶
Some use cases call for multiple values of the same type. Dig allows adding multiple values of the same type to the container with the use of Named Values.
Named Values can be produced by passing the dig.Name option when a constructor is provided. All values produced by that constructor will have the given name.
Given the following constructors,
func NewReadOnlyConnection(...) (*sql.DB, error) func NewReadWriteConnection(...) (*sql.DB, error)
You can provide *sql.DB into a Container under different names by passing the dig.Name option.
c.Provide(NewReadOnlyConnection, dig.Name("ro"))
c.Provide(NewReadWriteConnection, dig.Name("rw"))
Alternatively, you can produce a dig.Out struct and tag its fields with `name:".."` to have the corresponding value added to the graph under the specified name.
type ConnectionResult struct {
dig.Out
ReadWrite *sql.DB `name:"rw"`
ReadOnly *sql.DB `name:"ro"`
}
func ConnectToDatabase(...) (ConnectionResult, error) {
// ...
return ConnectionResult{ReadWrite: rw, ReadOnly: ro}, nil
}
Regardless of how a Named Value was produced, it can be consumed by another constructor by accepting a dig.In struct which has exported fields with the same name AND type that you provided.
type GatewayParams struct {
dig.In
WriteToConn *sql.DB `name:"rw"`
ReadFromConn *sql.DB `name:"ro"`
}
The name tag may be combined with the optional tag to declare the dependency optional.
type GatewayParams struct {
dig.In
WriteToConn *sql.DB `name:"rw"`
ReadFromConn *sql.DB `name:"ro" optional:"true"`
}
func NewCommentGateway(p GatewayParams, log *log.Logger) (*CommentGateway, error) {
if p.ReadFromConn == nil {
log.Print("Warning: Using RW connection for reads")
p.ReadFromConn = p.WriteToConn
}
// ...
}
Value Groups ¶
Added in Dig 1.2.
Dig provides value groups to allow producing and consuming many values of the same type. Value groups allow constructors to send values to a named, unordered collection in the container. Other constructors can request all values in this collection as a slice.
Constructors can send values into value groups by returning a dig.Out struct tagged with `group:".."`.
type HandlerResult struct {
dig.Out
Handler Handler `group:"server"`
}
func NewHelloHandler() HandlerResult {
..
}
func NewEchoHandler() HandlerResult {
..
}
Any number of constructors may provide values to this named collection. Other constructors can request all values for this collection by requesting a slice tagged with `group:".."`. This will execute all constructors that provide a value to that group in an unspecified order.
type ServerParams struct {
dig.In
Handlers []Handler `group:"server"`
}
func NewServer(p ServerParams) *Server {
server := newServer()
for _, h := range p.Handlers {
server.Register(h)
}
return server
}
Note that values in a value group are unordered. Dig makes no guarantees about the order in which these values will be produced.
Value groups can be used to provide multiple values for a group from a dig.Out using slices, however considering groups are retrieved by requesting a slice this implies that the values must be retrieved using a slice of slices. As of dig v1.9.0, if you want to provide individual elements to the group instead of the slice itself, you can add the `flatten` modifier to the group from a dig.Out.
type IntResult struct {
dig.Out
Handler []int `group:"server"` // [][]int from dig.In
Handler []int `group:"server,flatten"` // []int from dig.In
}
Example (Minimal) ¶
package main
import (
"encoding/json"
"log"
"os"
"go.uber.org/dig"
)
func main() {
type Config struct {
Prefix string
}
c := dig.New()
// Provide a Config object. This can fail to decode.
err := c.Provide(func() (*Config, error) {
// In a real program, the configuration will probably be read from a
// file.
var cfg Config
err := json.Unmarshal([]byte(`{"prefix": "[foo] "}`), &cfg)
return &cfg, err
})
if err != nil {
panic(err)
}
// Provide a way to build the logger based on the configuration.
err = c.Provide(func(cfg *Config) *log.Logger {
return log.New(os.Stdout, cfg.Prefix, 0)
})
if err != nil {
panic(err)
}
// Invoke a function that requires the logger, which in turn builds the
// Config first.
err = c.Invoke(func(l *log.Logger) {
l.Print("You've been invoked")
})
if err != nil {
panic(err)
}
}
Output: [foo] You've been invoked
Index ¶
- Constants
- func CanVisualizeError(err error) bool
- func IsCycleDetected(err error) bool
- func IsIn(o interface{}) bool
- func IsOut(o interface{}) bool
- func RootCause(err error) error
- func Visualize(c *Container, w io.Writer, opts ...VisualizeOption) error
- type Container
- type In
- type InvokeOption
- type Option
- type Out
- type ProvideOption
- type VisualizeOption
Examples ¶
Constants ¶
const Version = "1.11.0-dev"
Version of the library.
Variables ¶
This section is empty.
Functions ¶
func CanVisualizeError ¶ added in v1.4.0
CanVisualizeError returns true if the error is an errVisualizer.
func IsCycleDetected ¶ added in v1.5.0
IsCycleDetected returns a boolean as to whether the provided error indicates a cycle was detected in the container graph.
func IsIn ¶ added in v1.0.0
func IsIn(o interface{}) bool
IsIn checks whether the given struct is a dig.In struct. A struct qualifies as a dig.In struct if it embeds the dig.In type or if any struct that it embeds is a dig.In struct. The parameter may be the reflect.Type of the struct rather than the struct itself.
A struct MUST qualify as a dig.In struct for its fields to be treated specially by dig.
See the documentation for dig.In for a comprehensive list of supported tags.
func IsOut ¶ added in v1.0.0
func IsOut(o interface{}) bool
IsOut checks whether the given struct is a dig.Out struct. A struct qualifies as a dig.Out struct if it embeds the dig.Out type or if any struct that it embeds is a dig.Out struct. The parameter may be the reflect.Type of the struct rather than the struct itself.
A struct MUST qualify as a dig.Out struct for its fields to be treated specially by dig.
See the documentation for dig.Out for a comprehensive list of supported tags.
Types ¶
type Container ¶
type Container struct {
// contains filtered or unexported fields
}
Container is a directed acyclic graph of types and their dependencies.
func (*Container) Invoke ¶
func (c *Container) Invoke(function interface{}, opts ...InvokeOption) error
Invoke runs the given function after instantiating its dependencies.
Any arguments that the function has are treated as its dependencies. The dependencies are instantiated in an unspecified order along with any dependencies that they might have.
The function may return an error to indicate failure. The error will be returned to the caller as-is.
func (*Container) Provide ¶
func (c *Container) Provide(constructor interface{}, opts ...ProvideOption) error
Provide teaches the container how to build values of one or more types and expresses their dependencies.
The first argument of Provide is a function that accepts zero or more parameters and returns one or more results. The function may optionally return an error to indicate that it failed to build the value. This function will be treated as the constructor for all the types it returns. This function will be called AT MOST ONCE when a type produced by it, or a type that consumes this function's output, is requested via Invoke. If the same types are requested multiple times, the previously produced value will be reused.
In addition to accepting constructors that accept dependencies as separate arguments and produce results as separate return values, Provide also accepts constructors that specify dependencies as dig.In structs and/or specify results as dig.Out structs.
type In ¶
type In struct {
// contains filtered or unexported fields
}
In may be embedded into structs to request dig to treat them as special parameter structs. When a constructor accepts such a struct, instead of the struct becoming a dependency for that constructor, all its fields become dependencies instead. See the section on Parameter Objects in the package-level documentation for more information.
Fields of the struct may optionally be tagged to customize the behavior of dig. The following tags are supported,
name Requests a value with the same name and type from the
container. See Named Values for more information.
optional If set to true, indicates that the dependency is optional and
the constructor gracefully handles its absence.
group Name of the Value Group from which this field will be filled.
The field must be a slice type. See Value Groups in the
package documentation for more information.
type InvokeOption ¶ added in v1.0.0
type InvokeOption interface {
// contains filtered or unexported methods
}
An InvokeOption modifies the default behavior of Invoke. It's included for future functionality; currently, there are no concrete implementations.
type Option ¶ added in v1.0.0
type Option interface {
// contains filtered or unexported methods
}
Option configures a Container. It's included for future functionality; currently, there are no concrete implementations.
func DeferAcyclicVerification ¶ added in v1.5.0
func DeferAcyclicVerification() Option
DeferAcyclicVerification is an Option to override the default behavior of container.Provide, deferring the dependency graph validation to no longer run after each call to container.Provide. The container will instead verify the graph on first `Invoke`.
Applications adding providers to a container in a tight loop may experience performance improvements by initializing the container with this option.
type Out ¶
type Out struct {
// contains filtered or unexported fields
}
Out may be embedded into structs to request dig to treat them as special result structs. When a constructor returns such a struct, instead of the struct becoming a result of the constructor, all its fields become results of the constructor. See the section on Result Objects in the package-level documentation for more information.
Fields of the struct may optionally be tagged to customize the behavior of dig. The following tags are supported,
name Specifies the name of the value. Only a field on a dig.In
struct with the same 'name' annotation can receive this
value. See Named Values for more information.
group Name of the Value Group to which this field's value is being
sent. See Value Groups in the package documentation for more
information.
type ProvideOption ¶ added in v1.0.0
type ProvideOption interface {
// contains filtered or unexported methods
}
A ProvideOption modifies the default behavior of Provide.
func Group ¶ added in v1.7.0
func Group(group string) ProvideOption
Group is a ProvideOption that specifies that all values produced by a constructor should be added to the specified group. See also the package documentation about Value Groups.
This option cannot be provided for constructors which produce result objects.
func Name ¶ added in v1.4.0
func Name(name string) ProvideOption
Name is a ProvideOption that specifies that all values produced by a constructor should have the given name. See also the package documentation about Named Values.
Given,
func NewReadOnlyConnection(...) (*Connection, error) func NewReadWriteConnection(...) (*Connection, error)
The following will provide two connections to the container: one under the name "ro" and the other under the name "rw".
c.Provide(NewReadOnlyConnection, dig.Name("ro"))
c.Provide(NewReadWriteConnection, dig.Name("rw"))
This option cannot be provided for constructors which produce result objects.
type VisualizeOption ¶ added in v1.4.0
type VisualizeOption interface {
// contains filtered or unexported methods
}
A VisualizeOption modifies the default behavior of Visualize.
func VisualizeError ¶ added in v1.4.0
func VisualizeError(err error) VisualizeOption
VisualizeError includes a visualization of the given error in the output of Visualize if an error was returned by Invoke or Provide.
if err := c.Provide(...); err != nil {
dig.Visualize(c, w, dig.VisualizeError(err))
}
This option has no effect if the error was nil or if it didn't contain any information to visualize.
Source Files
Directories