prompter

Helper for go-prompt. Under heavy development.

Thanks to @parsiya for taking my pile of code and turning it into something useful. Check out his borrowedtime app that uses prompter.

Example

Main example is at example/main.go.

Usage

The package was not designed to work in a vacuum. It's meant to be used in conjunction with go-prompt.

Creating Commands

Commands are created with Command struct.

type Command struct {
	Name        string // Name of command (to be typed on cli)
	Description string // Description shown for command by the completer
	Executor    ExecutorFunc // Function that executes when command is run
	Hide        func() bool  // function to optionally hide a command from the completer
	SubCommands []Command    // List of completable subcommands
	Arguments   []Argument   // List of completable arguments
	Completer   *Completer
}

Commands must have a Name and should have a useful Description. Commands that implement the Executor field will be executable. Command can be optionally hidden by defining a Hide function.

helloCmd := prompter.Command{
	Name: "hello",
	Description: "print hello world", 
	Executor: func(_ prompter.CmdArgs) error {
		fmt.Println("Hello World")
		return nil
  },
}

Exit

One useful pre-packaged command is "exit" for quitting the cli application (via os.Exit(0)):

exitCmd := prompter.ExitCommand("exit", "exit the application")

Starting the prompt loop

To connect commands to go-prompt, create a prompter.Completer object, call RegisterCommands(...Command), and create the go-prompt object with the completer.Execute and completer.Complete properties,

// Create the prompter completer
completer := prompter.NewCompleter()
completer.RegisterCommands(helloCmd, exitCmd)

// Start go-prompt
p := prompt.New(completer.Execute, completer.Complete,
  prompt.OptionPrefix(">>> "),
  prompt.OptionTitle("pewpew"),
  prompt.OptionPrefixTextColor(prompt.White),
)
p.Run()

SubCommands

Subcommands are normal commands and added to main commands. Main commands usually do not do anything by themselves. One or more subcommands can be added to another command using the AddSubCommands(...Command) call.

sayCmd := prompter.Command{
  Name: "say",
  Description: "say some words",
}

fooSubCmd := prompter.Command{
  Name: "foo",
  Description: "say foo",
  Executor: func(_ prompter.CmdArgs) error {
    fmt.Println("Prompter says \"foo\"")
    return nil
  },
}

sayCmd.AddSubCommands(fooSubCmd)

Arguments

Commands can have zero or more arguments using the Argument struct.

type Argument struct {
	Name            string // Name of argument (to be typed on cli)
	Description     string // Description shown for argument by the completer
	OptionCompleter CompleterFunc // Function used to complete argument fields
	Repeatable      bool // Boolean to allow argument to be provided multiple times with different values
}

The following code can be called on the command line like: greet --name Bob

nameArg := prompter.Argument{
	Name: "--name",
	Description: "your name",
}

greetCmd := prompter.Command{
  Name: "greet",
  Description: "say a greeting",
  Executor: greetFunction,
}

greetCmd.AddArguments(nameArg)

Executor Functions and CmdArgs

The Executor function of the Command struct has the following type

// CmdArgs contains the arguments passed to the command.
type CmdArgs map[string][]string

type ExecutorFunc func(CmdArgs) error

CmdArgs will contain the passed arguments and their value(s). This is where the subcommand is processed and action upon. From the above greet command example, the following defines the greetFunction which executes the command given the --name argument value.

func greetFunction(args prompter.CmdArgs) error {
	if ! args.Contains("--name") {
		return errors.New("must provide a name")
	}
	name, err := args.GetFirstValue("--name")
	if err != nil {
		return err
	}
	fmt.Printf("Hello to %s\n", name)
	return nil
}

The CmdArgs.GetFirstValue(name string) function is a shortcut for CmdArgs.GetValue(name string, n int) which gets the nth occurance of the named argument when repeatable arguments are in use.

Completers

Arguments can also have completers of type:

type CompleterFunc func(string, []string) []prompt.Suggest

CompleterFunc is used to display suggestions for each argument. These suggestions can be hardcoded or dynamic (or a mix of both). The current argument is passed in optName. This allows multiple arguments to have the same completer. While there's only one option in this example, a switch is used to act as a blueprint for multiple arguments.

endpointCompleter := func (optName string, _ []string) []prompt.Suggest {
	// Create an empty list of suggestions.
	sugs := []prompt.Suggest{}
	switch optName {
  case "-endpoint":
    // Hardcoded suggestion.
    sugs = append(sugs, prompt.Suggest{Text: "aws", Description: "Amazon endpoint"})
    
    // Do something to get some dynamic suggestions.
    // Assuming newEndpoints is a []string of endpoints.
    newEndpoints := GetEndpointsFromAPI()
    
    for _, ed := range newEndpoints {
        sugs = append(sugs,
      prompt.Suggest{Text: ed, Description: ed})
    }
  }
	return sugs
}

endpointArg := prompter.Argument{
	Name: "-endpoint",
	ArgumentCompleter: endpointCompleter,
}

Licence

Opensourced under the Apache License v 2.0 license. See LICENSE for details.