namer

Documentation

Overview

Package namer has support for making different type naming systems.

This is because sometimes you want to refer to the literal type, sometimes you want to make a name for the thing you're generating, and you want to make the name based on the type. For example, if you have `type foo string`, you want to be able to generate something like `func FooPrinter(f *foo) { Print(string(*f)) }`; that is, you want to refer to a public name, a literal name, and the underlying literal name.

This package supports the idea of a "Namer" and a set of "NameSystems" to support these use cases.

Additionally, a "RawNamer" can optionally keep track of what needs to be imported.

Index

• func IC(in string) string
• func IL(in string) string
• func Joiner(first, others func(string) string) func(pre string, in []string, post string) string
• func NewRawNamer(tracker ImportTracker) *rawNamer
• type ImportTracker
• type NameStrategy
  • func NewPrivateNamer(prependPackageNames int, ignoreWords ...string) *NameStrategy
  • func NewPublicNamer(prependPackageNames int, ignoreWords ...string) *NameStrategy
  • func (ns *NameStrategy) Name(t *types.Type) string
• type NameSystems
• type Namer
• type Names
• type Orderer
  • func (o *Orderer) Order(u types.Universe) []*types.Type

Functions

func IC

func IC(in string) string

IC ensures the first character is uppercase.

func IL

func IL(in string) string

IL ensures the first character is lowercase.

func Joiner

func Joiner(first, others func(string) string) func(pre string, in []string, post string) string

Joiner lets you specify functions that preprocess the various components of a name before joining them. You can construct e.g. camelCase or CamelCase or any other way of joining words. (See the IC and IL convenience functions.)

func NewRawNamer

func NewRawNamer(tracker ImportTracker) *rawNamer

NewRawNamer will return a Namer that makes a name by which you would directly refer to a type, optionally keeping track of the import paths necessary to reference the names it provides. Tracker may be nil.

For example, if the type is map[string]int, a raw namer will literally return "map[string]int".

Or if the type, in package foo, is "type Bar struct { ... }", then the raw namer will return "foo.Bar" as the name of the type, and if 'tracker' was not nil, will record that package foo needs to be imported.

Types

type ImportTracker

type ImportTracker interface {
	AddType(*types.Type)
	LocalNameOf(packagePath string) string
}

ImportTracker allows a raw namer to keep track of the packages needed for import. You can implement yourself or use the one in the generation package.

type NameStrategy

type NameStrategy struct {
	Prefix, Suffix string
	Join           func(pre string, parts []string, post string) string

	// Add non-meaningful package directory names here (e.g. "proto") and
	// they will be ignored.
	IgnoreWords map[string]bool

	// If > 0, prepend exactly that many package directory names (or as
	// many as there are).  Package names listed in "IgnoreWords" will be
	// ignored.
	//
	// For example, if Ignore words lists "proto" and type Foo is in
	// pkg/server/frobbing/proto, then a value of 1 will give a type name
	// of FrobbingFoo, 2 gives ServerFrobbingFoo, etc.
	PrependPackageNames int

	// A cache of names thus far assigned by this namer.
	Names
}

NameStrategy is a general Namer. The easiest way to use it is to copy the Public/PrivateNamer variables, and modify the members you wish to change.

The Name method produces a name for the given type, of the forms: Anonymous types: <Prefix><Type description><Suffix> Named types: <Prefix><Optional Prepended Package name(s)><Original name><Suffix>

In all cases, every part of the name is run through the capitalization functions.

The IgnoreWords map can be set if you have directory names that are semantically meaningless for naming purposes, e.g. "proto".

Prefix and Suffix can be used to disambiguate parallel systems of type names. For example, if you want to generate an interface and an implementation, you might want to suffix one with "Interface" and the other with "Implementation". Another common use-- if you want to generate private types, and one of your source types could be "string", you can't use the default lowercase private namer. You'll have to add a suffix or prefix.

func NewPrivateNamer

func NewPrivateNamer(prependPackageNames int, ignoreWords ...string) *NameStrategy

NewPrivateNamer is a helper function that returns a namer that makes camelCase names. See the NameStrategy struct for an explanation of the arguments to this constructor.

func NewPublicNamer

func NewPublicNamer(prependPackageNames int, ignoreWords ...string) *NameStrategy

NewPublicNamer is a helper function that returns a namer that makes CamelCase names. See the NameStrategy struct for an explanation of the arguments to this constructor.

func (*NameStrategy) Name

func (ns *NameStrategy) Name(t *types.Type) string

See the comment on NameStrategy.

type NameSystems

type NameSystems map[string]Namer

NameSystems is a map of a system name to a namer for that system.

type Namer

type Namer interface {
	Name(*types.Type) string
}

Namer takes a type, and assigns a name.

The purpose of this complexity is so that you can assign coherent side-by-side systems of names for the types. For example, you might want a public interface, a private implementation struct, and also to reference literally the type name.

Note that it is safe to call your own Name() function recursively to find the names of keys, elements, etc. This is because anonymous types can't have cycles in their names, and named types don't require the sort of recursion that would be problematic.

type Names

type Names map[*types.Type]string

Names is a map from Type to name, as defined by some Namer.

type Orderer

type Orderer struct {
	Namer
}

Orderer produces an ordering of types given a Namer.

func (*Orderer) Order

func (o *Orderer) Order(u types.Universe) []*types.Type

Order assigns a name to every type, and returns a list sorted by those names.