Package argparse on DUB

To use this package, run the following command in your project's root directory:

Parser for command-line arguments

argparse is a self-contained flexible utility to parse command-line arguments.

Features

Positional arguments:
- Automatic type conversion of the value.
- Required by default, can be marked as optional.
Named arguments:
- Multiple names are supported, including short (-v) and long (--verbose) ones.
- Case-sensitive/-insensitive parsing.
- Bundling of short names (-vvv is same as -v -v -v).
- Equals sign is accepted (-v=debug, --verbose=debug).
- Automatic type conversion of the value.
- Optional by default, can be marked as required.
Support different types of destination data member:
- Scalar (e.g., int, float, bool).
- String arguments.
- Enum arguments.
- Array arguments.
- Hash (associative array) arguments.
- Callbacks.
Different workflows are supported:
- Mixin to inject standard main function.
- Parsing of known arguments only (returning not recognized ones).
- Enforcing that there are no unknown arguments provided.
Shell completion.
Options terminator (e.g., parsing up to -- leaving any argument specified after it).
Arguments groups.
Subcommands.
Fully customizable parsing:
- Raw (string) data validation (i.e., before parsing).
- Custom conversion of argument value (string -> any destination type).
- Validation of parsed data (i.e., after conversion to destination type).
- Custom action on parsed data (doing something different from storing the parsed value in a member of destination object).
ANSI colors and styles.
Built-in reporting of error happened during argument parsing.
Built-in help generation.

Getting started

Here is the simple example showing the usage of argparse utility. It uses the basic approach when all members are considered arguments with the same name as the name of member:

import argparse;

struct Basic
{
    // Basic data types are supported:
        // '--name' argument
        string name;

        // '--number' argument
        int number;

        // '--boolean' argument
        bool boolean;

    // Argument can have default value if it's not specified in command line
        // '--unused' argument
        string unused = "some default value";


    // Enums are also supported
        enum Enum { unset, foo, boo }
        // '--choice' argument
        Enum choice;

    // Use array to store multiple values
        // '--array' argument
        int[] array;

    // Callback with no args (flag)
        // '--callback' argument
        void callback() {}

    // Callback with single value
        // '--callback1' argument
        void callback1(string value) { assert(value == "cb-value"); }

    // Callback with zero or more values
        // '--callback2' argument
        void callback2(string[] value) { assert(value == ["cb-v1","cb-v2"]); }
}

// This mixin defines standard main function that parses command line and calls the provided function:
mixin CLI!Basic.main!((args)
{
    // 'args' has 'Basic' type
    static assert(is(typeof(args) == Basic));

    // do whatever you need
    import std.stdio: writeln;
    args.writeln;
    return 0;
});

If you run the program above with -h argument, then you’ll see the following output:

Usage: hello_world [--name NAME] [--number NUMBER] [--boolean] [--unused UNUSED] [--choice {unset,foo,boo}] [--array ARRAY ...] [--callback] [--callback1 CALLBACK1] [--callback2 [CALLBACK2 ...]] [-h]

Optional arguments:
  --name NAME
  --number NUMBER
  --boolean
  --unused UNUSED
  --choice {unset,foo,boo}
  --array ARRAY ...
  --callback
  --callback1 CALLBACK1
  --callback2 [CALLBACK2 ...]
  -h, --help           Show this help message and exit

For more sophisticated CLI usage, argparse provides few UDAs:

struct Advanced
{
    // Positional arguments are required by default
    @PositionalArgument(0)
    string name;

    // Named arguments can be attributed in bulk (parentheses can be omitted)
    @NamedArgument
    {
        string unused = "some default value";
        int number;
        bool boolean;
    }

    // Named argument can have custom or multiple names
        @NamedArgument("apple","appl")
        int apple;

        @NamedArgument(["b","banana","ban"])
        int banana;
}

// This mixin defines standard main function that parses command line and calls the provided function:
mixin CLI!Advanced.main!((args, unparsed)
{
    // 'args' has 'Advanced' type
    static assert(is(typeof(args) == Advanced));

    // unparsed arguments has 'string[]' type
    static assert(is(typeof(unparsed) == string[]));

    // do whatever you need
    import std.stdio: writeln;
    args.writeln;
    writeln("Unparsed args: ", unparsed);
    return 0;
});

If you run it with -h argument, then you’ll see the following:

Usage: hello_world [--unused UNUSED] [--number NUMBER] [--boolean] [--apple APPLE] [-b BANANA] [-h] name

Required arguments:
  name

Optional arguments:
  --unused UNUSED
  --number NUMBER
  --boolean
  --apple APPLE, --appl APPLE
  -b BANANA, --banana BANANA, --ban BANANA
  -h, --help         Show this help message and exit

Calling the parser

argparse provides CLI template to call the parser covering different use cases. It has the following signatures:

template CLI(Config config, COMMAND) – this is main template that provides multiple API (see below) for all supported use cases.
template CLI(Config config, COMMANDS...) – convenience wrapper of the previous template that provides main template mixin only for the simplest use case with subcommands. See corresponding section for details about subcommands.
alias CLI(COMMANDS...) = CLI!(Config.init, COMMANDS) – alias provided for convenience that allows using default Config, i.e., config = Config.init.

Wrapper for main function

The recommended and most convenient way to use argparse is through CLI!(...).main(alias newMain) mixin template. It declares the standard main function that parses command-line arguments and calls provided newMain function with an object that contains parsed arguments.

newMain function must satisfy these requirements:

It must accept COMMAND type as a first parameter if CLI template is used with one COMMAND.
It must accept all COMMANDS types as a first parameter if CLI template is used with multiple COMMANDS.... argparse uses std.sumtype.match for matching. Possible implementation of such newMain function would be a function that is overridden for every command type from COMMANDS. Another example would be a lambda that does compile-time checking of the type of the first parameter (see examples below for details).
Optionally newMain function can take a string[] parameter as a second argument. Providing such a function will mean that argparse will parse known arguments only and all unknown ones will be passed as a second parameter to newMain function. If newMain function doesn’t have such parameter, then argparse will error out if there is an unknown argument provided in command line.
Optionally newMain can return a result that can be cast to int. In this case, this result will be returned from standard main function.

Usage examples:

struct T
{
    string a;
    string b;
}

mixin CLI!T.main!((args)
{
    // 'args' has 'T' type
    static assert(is(typeof(args) == T));

    // do whatever you need
    import std.stdio: writeln;
    args.writeln;
    return 0;
});

struct cmd1
{
    string a;
}

struct cmd2
{
    string b;
}

mixin CLI!(cmd1, cmd2).main!((args, unparsed)
{
    // 'args' has either 'cmd1' or 'cmd2' type
    static if(is(typeof(args) == cmd1))
        writeln("cmd1: ", args);
    else static if(is(typeof(args) == cmd2))
        writeln("cmd2: ", args);
    else
        static assert(false); // this would never happen

    // unparsed arguments has 'string[]' type
    static assert(is(typeof(unparsed) == string[]));

    return 0;
});

Providing a new `main` function without wrapping standard `main`

If wrapping of standard main function doesn’t fit your needs (e.g., you need to do some initialization before parsing the command line), then you can use CLI!(...).parseArgs function:

int parseArgs(alias newMain)(string[] args, COMMAND initialValue = COMMAND.init)

Parameters:

newMain – function that’s called with object of type COMMAND as a first parameter filled with the data parsed from command line; optionally it can take string[] as a second parameter which will contain unknown arguments (see Wrapper for main function section for details).
args – raw command-line arguments (excluding argv[0] – first command-line argument in main function).
initialValue – initial value for the object passed to newMain function.

Return value:

If there is an error happened during the parsing, then non-zero value is returned. In case of no error, if newMain function returns a value that can be cast to int, then this value is returned, or 0 otherwise.

Usage example:

struct COMMAND
{
    string a;
    string b;
}

int my_main(COMMAND command)
{
    // Do whatever is needed
    return 0;
}

int main(string[] args)
{
    // Do initialization here
    // If needed, termination code can be done as 'scope(exit) { ...code... }' here as well

    return CLI!COMMAND.parseArgs!my_main(args[1..$]);
}

Providing an object for values of command line arguments

For the cases when providing newMain function is not possible or feasible, parseArgs function can accept a reference to an object that receives the values of command line arguments:

Result parseArgs(ref COMMAND receiver, string[] args)

Parameters:

receiver – object that is populated with parsed values.
args – raw command-line arguments (excluding argv[0] – first command-line argument in main function).

Return value:

An object that can be cast to bool to check whether the parsing was successful or not. Note that this function will error out if command line contains unknown arguments.

Usage example:

struct COMMAND
{
    string a;
    string b;
}

int main(string[] argv)
{
    COMMAND cmd;

    if(!CLI!COMMAND.parseArgs(cmd, argv[1..$]))
      return 1; // parsing failure

    // Do whatever is needed

    return 0;
}

Partial argument parsing

Sometimes a program may only parse a few of the command-line arguments, processing the remaining arguments in different way. In these cases, CLI!(...).parseKnownArgs function can be used. It works much like CLI!(...).parseArgs except that it does not produce an error when unknown arguments are present. It has the following signatures:

Result parseKnownArgs(ref COMMAND receiver, string[] args, out string[] unrecognizedArgs)

Parameters:

receiver – the object that’s populated with parsed values.
args – raw command-line arguments (excluding argv[0] – first command-line argument in main function).
unrecognizedArgs – raw command-line arguments that were not parsed.

Return value:

An object that can be cast to bool to check whether the parsing was successful or not.

Result parseKnownArgs(ref COMMAND receiver, ref string[] args)

Parameters:

receiver – the object that’s populated with parsed values.
args – raw command-line arguments that are modified to have parsed arguments removed (excluding argv[0] – first command-line argument in main function).

Return value:

An object that can be cast to bool to check whether the parsing was successful or not.

Usage example:

struct T
{
    string a;
}

auto args = [ "-a", "A", "-c", "C" ];

T result;
assert(CLI!T.parseKnownArgs(result, args));
assert(result == T("A"));
assert(args == ["-c", "C"]);

Shell completion

argparse supports tab completion of last argument for certain shells (see below). However, this support is limited to the names of arguments and subcommands.

Wrappers for main function

If you are using CLI!(...).main(alias newMain) mixin template in your code then you can easily build a completer (program that provides completion) by defining argparse_completion version (-version=argparse_completion option of dmd). Don’t forget to use different file name for completer than your main program (-of option in dmd). No other changes are necessary to generate completer, but you should consider minimizing the set of imported modules when argparse_completion version is defined. For example, you can put all imports into your main function that is passed to CLI!(...).main(alias newMain) – newMain parameter is not used in completer.

If you prefer having separate main module for completer, then you can use CLI!(...).mainComplete mixin template:

mixin CLI!(...).mainComplete;

In case if you prefer to have your own main function and would like to call completer by yourself, you can use int CLI!(...).complete(string[] args) function. This function executes the completer by parsing provided args (note that you should remove the first argument from argv passed to main function). The returned value is meant to be returned from main function, having zero value in case of success.

Low level completion

In case if none of the above methods is suitable, argparse provides string[] CLI!(...).completeArgs(string[] args) function. It takes arguments that should be completed and returns all possible completions.

completeArgs function expects to receive all command-line arguments (excluding argv[0] – first command-line argument in main function) in order to provide completions correctly (set of available arguments depends on subcommand). This function supports two workflows:

If the last argument in args is empty and it’s not supposed to be a value for a command-line argument, then all available arguments and subcommands (if any) are returned.
If the last argument in args is not empty and it’s not supposed to be a value for a command-line argument, then only those arguments and subcommands (if any) are returned that start with the same text as the last argument in args.

For example, if there are --foo, --bar and --baz arguments available, then:

Completion for args=[""] will be ["--foo", "--bar", "--baz"].
Completion for args=["--b"] will be ["--bar", "--baz"].

Using the completer

Completer that is provided by argparse supports the following shells:

bash
zsh
tcsh
fish

Its usage consists of two steps: completion setup and completing of the command line. Both are implemented as subcommands (init and complete accordingly).

Completion setup

Before using completion, completer should be added to the shell. This can be achieved by using init subcommand. It accepts the following arguments (you can get them by running <completer> init --help):

--bash: provide completion for bash.
--zsh: provide completion for zsh. Note: zsh completion is done through bash completion so you should execute bashcompinit first.
--tcsh: provide completion for tcsh.
--fish: provide completion for fish.
--completerPath <path>: path to completer. By default, the path to itself is used.
--commandName <name>: command name that should be completed. By default, the first name of your main command is used.

Either --bash, --zsh, --tcsh or --fish is expected.

As a result, completer prints the script to setup completion for requested shell into standard output (stdout) which should be executed. To make this more streamlined, you can execute the output inside the current shell or to do this during shell initialization (e.g., in .bashrc for bash). To help doing so, completer also prints sourcing recommendation to standard output as a comment.

Example of completer output for <completer> init --bash --commandName mytool --completerPath /path/to/completer arguments:

# Add this source command into .bashrc:
#       source <(/path/to/completer init --bash --commandName mytool)
complete -C 'eval /path/to/completer --bash -- $COMP_LINE ---' mytool

Recommended workflow is to install completer into a system according to your installation policy and update shell initialization/config file to source the output of init command.

Completing of the command line

Argument completion is done by complete subcommand (it’s default one). It accepts the following arguments (you can get them by running <completer> complete --help):

--bash: provide completion for bash.
--tcsh: provide completion for tcsh.
--fish: provide completion for fish.

As a result, completer prints all available completions, one per line, assuming that it’s called according to the output of init command.

Argument declaration

Positional arguments

Positional arguments are expected to be at a specific position within the command line. This argument can be declared using PositionalArgument UDA:

struct Params
{
    @PositionalArgument(0)
    string firstName;

    @PositionalArgument(1, "lastName")
    string arg;
}

Parameters of PositionalArgument UDA:

#	Name	Type	Optional/ Required	Description
1	`position`	`uint`	required	Zero-based unsigned position of the argument.
2	`name`	`string`	optional	Name of this argument that is shown in help text. If not provided, then the name of data member is used.

Named arguments

As an opposite to positional, there can be named arguments (they are also called as flags or options). They can be declared using NamedArgument UDA:

struct Params
{
    @NamedArgument
    string greeting;

    @NamedArgument(["name", "first-name", "n"])
    string name;

    @NamedArgument("family", "last-name")
    string family;
}

Parameters of NamedArgument UDA:

#	Name	Type	Optional/ Required	Description
1	`name`	`string` or `string[]`	optional	Name(s) of this argument that can show up in command line.

Named arguments might have multiple names, so they should be specified either as an array of strings or as a list of parameters in NamedArgument UDA. Argument names can be either single-letter (called as short options) or multi-letter (called as long options). Both cases are fully supported with one caveat: if a single-letter argument is used with a double dash (e.g., --n) in command line, then it behaves the same as a multi-letter option. When an argument is used with a single dash, then it is treated as a single-letter argument.

The following usages of the argument in the command line are equivalent: --name John, --name=John, --n John, --n=John, -nJohn, -n John, -n=John. Note that any other character can be used instead of = – see Parser customization for details.

Optional and required arguments

Arguments can be marked as required or optional by adding .Required() or .Optional() to UDA. If required argument is not present, parser will error out. Positional arguments are required by default.

struct T
{
    @(PositionalArgument(0, "a").Optional())
    string a = "not set";

    @(NamedArgument.Required())
    int b;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T("not set", 4)); })(["-b", "4"]) == 0);

Limit the allowed values

In some cases an argument can receive one of the limited set of values so AllowedValues can be used here:

struct T
{
    @(NamedArgument.AllowedValues!(["apple","pear","banana"]))
    string fruit;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T("apple")); })(["--fruit", "apple"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["--fruit", "kiwi"]) != 0);    // "kiwi" is not allowed

For the value that is not in the allowed list, this error will be printed:

Error: Invalid value 'kiwi' for argument '--fruit'.
Valid argument values are: apple,pear,banana

Note that if the type of destination variable is enum, then the allowed values are automatically limited to those listed in the enum.

Argument dependencies

Mutually exclusive arguments

Mutually exclusive arguments (i.e., those that can’t be used together) can be declared using MutuallyExclusive() UDA:

struct T
{
    @MutuallyExclusive()
    {
        string a;
        string b;
    }
}

// Either or no argument is allowed
assert(CLI!T.parseArgs!((T t) {})(["-a","a"]) == 0);
assert(CLI!T.parseArgs!((T t) {})(["-b","b"]) == 0);
assert(CLI!T.parseArgs!((T t) {})([]) == 0);

// Both arguments are not allowed
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-a","a","-b","b"]) != 0);

Note that parentheses are required in this UDA to work correctly.

Set of mutually exclusive arguments can be marked as required in order to require exactly one of the arguments:

struct T
{
    @(MutuallyExclusive().Required())
    {
        string a;
        string b;
    }
}

// Either argument is allowed
assert(CLI!T.parseArgs!((T t) {})(["-a","a"]) == 0);
assert(CLI!T.parseArgs!((T t) {})(["-b","b"]) == 0);

// Both arguments or no argument is not allowed
assert(CLI!T.parseArgs!((T t) { assert(false); })([]) != 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-a","a","-b","b"]) != 0);

Mutually required arguments

Mutually required arguments (i.e., those that require other arguments) can be declared using RequiredTogether() UDA:

struct T
{
    @RequiredTogether()
    {
        string a;
        string b;
    }
}

// Both or no argument is allowed
assert(CLI!T.parseArgs!((T t) {})(["-a","a","-b","b"]) == 0);
assert(CLI!T.parseArgs!((T t) {})([]) == 0);

// Only one argument is not allowed
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-a","a"]) != 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-b","b"]) != 0);

Note that parentheses are required in this UDA to work correctly.

Set of mutually required arguments can be marked as required in order to require all arguments:

struct T
{
    @(RequiredTogether().Required())
    {
        string a;
        string b;
    }
}

// Both arguments are allowed
assert(CLI!T.parseArgs!((T t) {})(["-a","a","-b","b"]) == 0);

// Single argument or no argument is not allowed
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-a","a"]) != 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-b","b"]) != 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })([]) != 0);

Subcommands

Sophisticated command-line tools, like git, have many subcommands (e.g., commit, push, etc.), each with its own set of arguments. There are few ways how to use subcommands with argparse.

Enumerating subcommands in CLI mixin

All commands can be listed as template parameters to Main.CLI and provided main function must be able to handle all command types:

struct sum
{
  int[] numbers;  // --numbers argument
}

struct min
{
  int[] numbers;  // --numbers argument
}

struct max
{
  int[] numbers;  // --numbers argument
}

int main_(max cmd)
{
  import std.algorithm: maxElement;

  writeln("max = ", cmd.numbers.maxElement);

  return 0;
}

int main_(min cmd)
{
  import std.algorithm: minElement;

  writeln("min = ", cmd.numbers.minElement);

  return 0;
}

int main_(sum cmd)
{
  import std.algorithm: sum;

  writeln("sum = ", cmd.numbers.sum);

  return 0;
}

// This mixin defines standard main function that parses command line and calls the provided function:
mixin CLI!(sum, min, max).main!main_;

Subcommands with shared common arguments

In some cases command-line tool has arguments that are common across all subcommands. They can be specified as regular arguments in a struct that represents the whole program. In this case subcommands must be listed as regular data member having SubCommand type that is passed with a list of types of all subcommands. The main function should accept a parameter for the program, not for each subcommand:

struct sum {}
struct min {}
struct max {}

struct Program
{
  int[] numbers;  // --numbers argument

  // name of the command is the same as a name of the type
  SubCommand!(sum, min, max) cmd;
}

// This mixin defines standard main function that parses command line and calls the provided function:
mixin CLI!Program.main!((prog)
{
  static assert(is(typeof(prog) == Program));

  int result = prog.cmd.match!(
    (.max)
    {
      import std.algorithm: maxElement;
      return prog.numbers.maxElement;
    },
    (.min)
    {
      import std.algorithm: minElement;
      return prog.numbers.minElement;
    },
    (.sum)
    {
      import std.algorithm: sum;
      return prog.numbers.sum;
    }
  );

  writeln("result = ", result);

  return 0;
});

Subcommand name and aliases

To define a command name that is not the same as the type that represents this command, one should use Command UDA – it accepts a name and list of name aliases. All these names are recognized by the parser and are displayed in the help text. For example:

@(Command("maximum", "max")
.ShortDescription("Print the maximum")
)
struct MaxCmd
{
    int[] numbers;
}

Would result in this help fragment:

  maximum,max    Print the maximum

If Command has no names listed, then the name of the type is used as a command name:

  MaxCmd         Print the maximum

Default subcommand

The default command is a command that is ran when user does not specify any command in the command line. To mark a command as default, use Default template:

SubCommand!(sum, min, Default!max) cmd;

Help generation

Command

Command UDA provides few customizations that affect help text. It can be used for top-level command and subcommands.

Program name (i.e., the name of top-level command) and subcommand name can be provided to Command UDA as a parameter. If program name is not provided, then Runtime.args[0] (a.k.a. argv[0] from main function) is used. If subcommand name is not provided (e.g., @(Command.Description(...))), then the name of the type that represents the command is used.
Usage – allows custom usage text. By default, the parser calculates the usage message from the arguments it contains but this can be overridden with Usage call. If the custom text contains %(PROG) then it will be replaced by the command/program name.
Description – used to provide a description of what the command/program does and how it works. In help messages, the description is displayed between the usage string and the list of the command arguments.
ShortDescription – used to provide a brief description of what the subcommand does. It is applicable to subcommands only and is displayed in Available commands section on help screen of the parent command.
Epilog – custom text that is printed after the list of the arguments.

Usage, Description, ShortDescription and Epilog modifiers take either string or string delegate() value – the latter can be used to return a value that is not known at compile time.

Argument

There are some customizations supported on argument level for both PositionalArgument and NamedArgument UDAs:

Description – provides brief description of the argument. This text is printed next to the argument in the argument-list section of a help message. Description takes either string or string delegate() value – the latter can be used to return a value that is not known at compile time.
HideFromHelp – can be used to indicate that the argument shouldn’t be printed in help message.
Placeholder – provides custom text that is used to indicate the value of the argument in help message.

Help text styling

argparse uses Config.styling to determine what style should be applied to different parts of the help text. Please refer to Styling scheme for available settings.

Example

Here is an example of how this customization can be used:

@(Command("MYPROG")
 .Description("custom description")
 .Epilog(() => "custom epilog")
)
struct T
{
  @NamedArgument  string s;
  @(NamedArgument.Placeholder("VALUE"))  string p;

  @(NamedArgument.HideFromHelp())  string hidden;

  enum Fruit { apple, pear };
  @(NamedArgument("f","fruit").Required().Description("This is a help text for fruit. Very very very very very very very very very very very very very very very very very very very long text")) Fruit f;

  @(NamedArgument.AllowedValues!([1,4,16,8])) int i;

  @(PositionalArgument(0).Description(() => "This is a help text for param0. Very very very very very very very very very very very very very very very very very very very long text")) string param0;
  @(PositionalArgument(1).AllowedValues!(["q","a"])) string param1;
}

CLI!T.parseArgs!((T t) {})(["-h"]);

This example will print the following help message:

Usage: MYPROG [-s S] [-p VALUE] -f {apple,pear} [-i {1,4,16,8}] [-h] param0 {q,a}

custom description

Required arguments:
  -f {apple,pear}, --fruit {apple,pear}
                   This is a help text for fruit. Very very very very very very
                   very very very very very very very very very very very very
                   very long text
  param0           This is a help text for param0. Very very very very very very
                   very very very very very very very very very very very very
                   very long text
  {q,a}

Optional arguments:
  -s S
  -p VALUE
  -i {1,4,16,8}
  -h, --help       Show this help message and exit

custom epilog

Argument groups

By default, parser groups command-line arguments into “required arguments” and “optional arguments” when displaying help message. When there is a better conceptual grouping of arguments than this default one, appropriate groups can be created using ArgumentGroup UDA.

This UDA has some customization for displaying text:

Description – provides brief description of the group. This text is printed right after group name. It takes either string or string delegate() value – the latter can be used to return a value that is not known at compile time.

Example:

struct T
{
    @(ArgumentGroup("group1").Description("group1 description"))
    {
        @NamedArgument
        {
            string a;
            string b;
        }
        @PositionalArgument(0) string p;
    }

    @(ArgumentGroup("group2").Description("group2 description"))
    @NamedArgument
    {
        string c;
        string d;
    }
    @PositionalArgument(1) string q;
}

When an argument is attributed with a group, the parser treats it just like a normal argument, but displays the argument in a separate group for help messages:

Usage: MYPROG [-a A] [-b B] [-c C] [-d D] [-h] p q

group1:
  group1 description

  -a A
  -b B
  p

group2:
  group2 description

  -c C
  -d D

Required arguments:
  q

Optional arguments:
  -h, --help    Show this help message and exit

ANSI colors and styles

Using colors in your command’s output does not just look good: contrasting important elements like argument names from the rest of the text reduces the cognitive load on the user. argparse uses ANSI escape sequences to add coloring and styling to help text. In addition, argparse offers public API to apply colors and styles to any text printed to the console (see below).

Default styling

Styles and colors

The argparse.ansi submodule provides supported styles and colors. You can use any combinations of them:

Font styles:

bold
italic
underline

Foreground colors:

black
red
green
yellow
blue
magenta
cyan
lightGray
darkGray
lightRed
lightGreen
lightYellow
lightBlue
lightMagenta
lightCyan
white

Background colors:

onBlack
onRed
onGreen
onYellow
onBlue
onMagenta
onCyan
onLightGray
onDarkGray
onLightRed
onLightGreen
onLightYellow
onLightBlue
onLightMagenta
onLightCyan
onWhite

There is also a “virtual” style noStyle that means no styling is applied. It’s useful in ternary operations as a fallback for the case when styling is disabled. See below example for details.

All styles above can be combined using . and even be used in regular output:

// `enableStyle` is a flag indicating that styling should be enabled
void printText(bool enableStyle)
{
  // style is enabled at runtime when `enableStyle` is true
  auto myStyle = enableStyle ? bold.italic.cyan.onRed : noStyle;

  // "Hello" is always printed in green;
  // "world!" is printed in bold, italic, cyan and on red when `enableStyle` is true, or "as is" otherwise
  writeln(green("Hello "), myStyle("world!"));
}

This example shows how styling can be used in custom help text (Usage, Description, ShortDescription, Epilog API):

    @(NamedArgument.Description(bold.underline("Colorize the output:")~" make everything "~red("red")))
    bool red;

Styling mode

By default argparse will try to detect whether ANSI styling is supported, and if so, it will apply styling to the help text. In some cases this behavior should be adjusted or overridden. To do so, you can use Config.stylingMode. Argparse provides the following setting to control the styling:

If it’s set to Config.StylingMode.on, then styling is always enabled.
If it’s set to Config.StylingMode.off, then styling is always disabled.
If it’s set to Config.StylingMode.autodetect, then heuristics are used to determine whether styling will be applied.

In some cases styling control should be exposed to a user as a command-line argument (similar to --color argument in ls and grep commands). Argparse supports this use case – just add an argument to your command (you can customize it with @NamedArgument UDA):

static auto color = ansiStylingArgument;

This will add the following argument:

  --color [{always,auto,never}]
                          Colorize the output. If value is omitted then 'always'
                          is used.

If you want to determine whether --color argument was specified in command line, you can simply check the value of that data member:

struct Arguments
{
    static auto color = ansiStylingArgument;
}

mixin CLI!Arguments.main!((args)
{
    // 'autodetect' is converted to either 'on' or 'off'
    if(args.color)
      writeln("Colors are enabled");
    else
      writeln("Colors are disabled");
});

Heuristics for enabling styling

Below is the exact sequence of steps argparse uses to determine whether or not to emit ANSI escape codes (see detectSupport() function here for details):

If environment variable NO_COLOR != "", then styling is disabled. See here for details.
If environment variable CLICOLOR_FORCE != "0", then styling is enabled. See here for details.
If environment variable CLICOLOR == "0", then styling is disabled. See here for details.
If environment variable ConEmuANSI == "OFF", then styling is disabled. See here for details.
If environment variable ConEmuANSI == "ON", then styling is enabled. See here for details.
If environment variable ANSICON is defined (regardless of its value), then styling is enabled. See here for details.
Windows only (version(Windows)):
If environment variable TERM contains "cygwin" or starts with "xterm", then styling is enabled.
If GetConsoleMode call for STD_OUTPUT_HANDLE returns a mode that has ENABLE_VIRTUAL_TERMINAL_PROCESSING set, then styling is enabled.
If SetConsoleMode call for STD_OUTPUT_HANDLE with ENABLE_VIRTUAL_TERMINAL_PROCESSING mode was successful, then styling is enabled.
Posix only (version(Posix)):
If STDOUT is not redirected, then styling is enabled.
If none of the above applies, then styling is disabled.

Supported types

Boolean

Boolean types usually represent command-line flags. argparse supports multiple ways of providing flag value including negation (i.e., --no-flag):

struct T
{
    @NamedArgument("b","boo")
    bool b;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(true)); })(["-b"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(true)); })(["-b=true"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(false)); })(["-b=false"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(false)); })(["--no-b"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-b","true"]) == 1);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-b","false"]) == 1);

assert(CLI!T.parseArgs!((T t) { assert(t == T(true)); })(["--boo"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(true)); })(["--boo=true"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(false)); })(["--boo=false"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(false)); })(["--no-boo"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["--boo","true"]) == 1);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["--boo","false"]) == 1);

Numeric

Numeric arguments are converted using std.conv.to:

struct T
{
    int i;
    uint u;
    double d;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(-5,8,12.345)); })(["-i","-5","-u","8","-d","12.345"]) == 0);

String

argparse supports string arguments as pass through:

struct T
{
    string a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T("foo")); })(["-a","foo"]) == 0);

Enum

If an argument is bound to an enum, an enum symbol as a string is expected as a value, or right within the argument separated with an “=” sign:

struct T
{
    enum Fruit { apple, pear };

    Fruit a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(T.Fruit.apple)); })(["-a","apple"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(T.Fruit.pear)); })(["-a=pear"]) == 0);

In some cases the value for command-line argument might have characters that are not allowed in enum identifiers. There is ArgumentValue UDA that can be used to adjust allowed values:

struct T
{
    enum Fruit {
        apple,
        @ArgumentValue("no-apple","noapple")
        noapple
    };

    Fruit a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(T.Fruit.apple)); })(["-a","apple"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(T.Fruit.noapple)); })(["-a=no-apple"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T(T.Fruit.noapple)); })(["-a","noapple"]) == 0);

Counter

Counter argument is the parameter that tracks the number of times the argument occurred on the command line:

struct T
{
    @(NamedArgument.Counter()) int a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(3)); })(["-a","-a","-a"]) == 0);

Array

If an argument is bound to 1D array, a new element is appended to this array each time the argument is provided in command line. In case if an argument is bound to 2D array then new elements are grouped in a way as they appear in command line and then each group is appended to this array:

struct T
{
    int[]   a;
    int[][] b;
}

assert(CLI!T.parseArgs!((T t) { assert(t.a == [1,2,3,4,5]); })(["-a","1","2","3","-a","4","5"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t.b == [[1,2,3],[4,5]]); })(["-b","1","2","3","-b","4","5"]) == 0);

Alternatively you can set Config.arraySep to allow multiple elements in one parameter:

struct T
{
    int[] a;
}

enum cfg = {
    Config cfg;
    cfg.arraySep = ',';
    return cfg;
}();

assert(CLI!(cfg, T).parseArgs!((T t) { assert(t == T([1,2,3,4,5])); })(["-a","1,2,3","-a","4","5"]) == 0);

Specifying number of values

In case the argument is bound to static array then the maximum number of values is set to the size of the array. For dynamic array, the number of values is not limited. The minimum number of values is 1 in all cases. This behavior can be customized by calling the following functions:

NumberOfValues(ulong min, ulong max) – sets both minimum and maximum number of values.
NumberOfValues(ulong num) – sets both minimum and maximum number of values to the same value.
MinNumberOfValues(ulong min) – sets minimum number of values.
MaxNumberOfValues(ulong max) – sets maximum number of values.

struct T
{
  @(NamedArgument.NumberOfValues(1,3))
  int[] a;
  @(NamedArgument.NumberOfValues(2))
  int[] b;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T([1,2,3],[4,5])); })(["-a","1","2","3","-b","4","5"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t == T([1],[4,5])); })(["-a","1","-b","4","5"]) == 0);

Associative array

If an argument is bound to an associative array, a string of the form “name=value” is expected as the next entry in command line, or right within the option separated with an “=” sign:

struct T
{
    int[string] a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(["foo":3,"boo":7])); })(["-a=foo=3","-a","boo=7"]) == 0);

Alternatively you can set Config.arraySep to allow multiple elements in one parameter:

struct T
{
    int[string] a;
}

enum cfg = {
    Config cfg;
    cfg.arraySep = ',';
    return cfg;
}();

assert(CLI!(cfg, T).parseArgs!((T t) { assert(t == T(["foo":3,"boo":7])); })(["-a=foo=3,boo=7"]) == 0);
assert(CLI!(cfg, T).parseArgs!((T t) { assert(t == T(["foo":3,"boo":7])); })(["-a","foo=3,boo=7"]) == 0);

In general, the keys and values can be of any parsable types.

Callback

An argument can be bound to a function with one of the following signatures (return value, if any, is ignored):

... function()

In this case, the argument is treated as a flag and the function is called every time when the argument is seen in command line.

... function(string)

In this case, the argument has exactly one value and the function is called every time when the argument is seen in command line and the value specified in command line is provided into string parameter.

... function(string[])

In this case, the argument has zero or more values and the function is called every time when the argument is seen in command line and the set of values specified in command line is provided into string[] parameter.

... function(RawParam)

In this case, the argument has one or more values and the function is called every time when the argument is seen in command line and the set of values specified in command line is provided into parameter.

struct T
{
    int a;

    @(NamedArgument("a")) void foo() { a++; }
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(4)); })(["-a","-a","-a","-a"]) == 0);

Custom types

Any arbitrary type can be used to receive command-line-argument values. argparse supports this use case – you just need to provide parsing function:

struct Value
{
    string a;
}
struct T
{
    @(NamedArgument.Parse!((string s) { return Value(s); }))
    Value s;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(Value("foo"))); return 12345; })(["-s","foo"]) == 12345);

Argument parsing customization

Sometime the functionality that is provided out of the box is not enough and needs to be tuned.

Parsing of command-line string values into some typed receiver member consists of following steps:

Pre-validation – argument values are validated as raw strings.
Parsing – raw argument values are converted to a different type (usually the type of the receiver).
Validation – converted value is validated.
Action – depending on a type of the receiver, it might be either assignment of converted value to a receiver, appending value if receiver is an array or other operation.

In case if argument does not have any value to parse, then the only one step is involved:

Action if no value – similar to Action step above but without converted value.

If any of the steps fails, then the command-line parsing fails as well.

Each of the steps above can be customized with UDA modifiers below. These modifiers take a function that might accept either argument value(s) or Param struct that has the fields below:

config- Config object that is passed to parsing function.
name – Argument name that is specified in command line.
value – Array of argument values that are provided in command line.

There is also an alias, RawParam, where the type of the value field is string[]. This alias represents "raw" values from command line.

Pre-validation

PreValidation modifier can be used to customize the validation of raw string values. It accepts a function with one of the following signatures:

bool validate(string value)
Result validate(string value)
bool validate(string[] value)
Result validate(string[] value)
bool validate(RawParam param)
Result validate(RawParam param)

Note that the first function will be called once for every value specified in command line for an argument.

Parameters:

value/param values to be parsed.

Return value:

true/Result.Success if validation passed or
false/Result.Error otherwise.

Parsing

Parse modifier allows providing custom conversion from raw string to a typed value. It accepts a function with one of the following signatures:

ParseType parse(string value)
ParseType parse(string[] value)
ParseType parse(RawParam param)
bool parse(ref ParseType receiver, RawParam param)
void parse(ref ParseType receiver, RawParam param)
Result parse(ref ParseType receiver, RawParam param)

Note that ParseType is a type that a string value is supposed to be parsed to and it is not required be the same as a target type.

Parameters:

value/param values to be parsed.
receiver is an output variable for parsed value.

Return value:

true/Result.Success if parsing was successful or
false/Result.Error otherwise.

Validation

Validation modifier can be used to validate parsed value. It accepts a function with one of the following signatures:

bool validate(ParseType value)
Result validate(ParseType value)
bool validate(ParseType[] value)
Result validate(ParseType[] value)
bool validate(Param!ParseType param)
Result validate(Param!ParseType param)

Parameters:

value/param contains a value returned from Parse step.

Return value:

true/Result.Success if validation passed or
false/Result.Error otherwise.

Action

Action modifier allows providing a custom logic of how receiver should be changed when argument has a value in command line. It accepts a function with one of the following signatures:

bool action(ref T receiver, ParseType value)
void action(ref T receiver, ParseType value)
Result action(ref T receiver, ParseType value)
bool action(ref T receiver, Param!ParseType param)
void action(ref T receiver, Param!ParseType param)
Result action(ref T receiver, Param!ParseType param)

Parameters:

receiver is a receiver (destination field) which is supposed to be changed based on a value/param.
value/param has a value returned from Parse step.

Return value:

true/Result.Success if operation was successful or
false/Result.Error otherwise.

Arguments with no values

Sometimes arguments are allowed to have no values in command line. Here are two cases that arise in this situation:

Argument should get specific default value if there is no value provided in command line. AllowNoValue modifier should be used in this case.
Argument must not have any values in command line. In this case RequireNoValue modifier should be used.

Both AllowNoValue and RequireNoValue modifiers accept a value that should be used when no value is provided in command line. The difference between them can be seen in this example:

struct T
{
    @(NamedArgument.AllowNoValue  !10) int a;
    @(NamedArgument.RequireNoValue!20) int b;
}

assert(CLI!T.parseArgs!((T t) { assert(t.a == 10); })(["-a"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t.b == 20); })(["-b"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(t.a == 30); })(["-a","30"]) == 0);
assert(CLI!T.parseArgs!((T t) { assert(false); })(["-b","30"]) != 0);

Usage example

All the above modifiers can be combined in any way:

struct T
{
    @(NamedArgument
     .PreValidation!((string s) { return s.length > 1 && s[0] == '!'; })
     .Parse        !((string s) { return s[1]; })
     .Validation   !((char v) { return v >= '0' && v <= '9'; })
     .Action       !((ref int a, char v) { a = v - '0'; })
    )
    int a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(4)); })(["-a","!4"]) == 0);

Parser customization

argparse provides decent amount of settings to customize the parser. All customizations can be done by creating Config object with required settings (see below).

Assign character

Config.assignChar – the assignment character used in arguments with value: -a=5, -b=foo.

Default is equal sign =.

Array separator

Config.arraySep – when set to char.init, values to array and associative-array receivers are treated as an individual value. That is, only one argument is appended/inserted per appearance of the argument. If arraySep is set to something else, then each value is first split by the separator, and the individual pieces are treated as values to the same argument.

Default is char.init.

struct T
{
    string[] a;
}

assert(CLI!T.parseArgs!((T t) { assert(t == T(["1,2,3","4","5"])); })(["-a","1,2,3","-a","4","5"]) == 0);

enum cfg = {
    Config cfg;
    cfg.arraySep = ',';
    return cfg;
}();

assert(CLI!(cfg, T).parseArgs!((T t) { assert(t == T(["1","2","3","4","5"])); })(["-a","1,2,3","-a","4","5"]) == 0);

Named argument prefix character

Config.namedArgPrefix – the character that named arguments begin with.

Default is dash (-).

End of named arguments

Config.endOfNamedArgs – the string that conventionally marks the end of all named arguments. All arguments that are specified after this one are treated as positional regardless to the value which can start with namedArgPrefix (dash - by default).

Default is double dash (--).

Case sensitivity

Config.caseSensitive – by default argument names are case-sensitive. You can change that behavior by setting this member to false.

Default is true.

Bundling of single-letter arguments

Config.bundling – when it is set to true, single-letter arguments can be bundled together, i.e., -abc is the same as -a -b -c.

Default is false.

Adding help generation

Config.addHelp – when it is set to true, then -h and --help arguments are added to the parser. In case if the command line has one of these arguments, then the corresponding help text is printed and the parsing will be stopped. If CLI!(...).parseArgs(alias newMain) or CLI!(...).main(alias newMain) is used, then provided newMain function will not be called.

Default is true.

Styling mode

Config.stylingMode – styling mode for the text output (error messages and help screen). It has the following type: enum StylingMode { autodetect, on, off }. If it's set to Config.StylingMode.on then styling is always enabled. If it's set to Config.StylingMode.off then styling is always disabled.

Default value is Config.StylingMode.autodetect, which means that styling will be enabled when possible.

See ANSI coloring and styling for details.

Styling scheme

Config.styling – contains style for the text output (error messages and help screen). It has the following members:

programName: style for the program name. Default is bold.
subcommandName: style for the subcommand name. Default is bold.
argumentGroupTitle: style for the title of argument group. Default is bold.underline.
argumentName: style for the argument name. Default is lightYellow.
namedArgumentValue: style for the value of named argument. Default is italic.
positionalArgumentValue: style for the value of positional argument. Default is lightYellow.
errorMessagePrefix: style for Error: prefix in error messages. Default is red.

See ANSI coloring and styling for details.

Error handling

Config.errorHandler – this is a handler function for all errors occurred during parsing the command line. It a function that receives string parameter which would be an error message.

The default behavior is to print error message to stderr.

1.4.0	2024-Oct-31
1.3.0	2023-May-10
1.2.0	2022-Sep-16
1.1.1	2022-Aug-29
1.1.0	2022-Aug-24

argparse ~ldc-check

Parser for command-line arguments

Features

Getting started

Calling the parser

Wrapper for main function

Providing a new main function without wrapping standard main

Providing an object for values of command line arguments

Partial argument parsing

Shell completion

Wrappers for main function

Low level completion

Using the completer

Completion setup

Completing of the command line

Argument declaration

Positional arguments

Named arguments

Optional and required arguments

Limit the allowed values

Argument dependencies

Mutually exclusive arguments

Mutually required arguments

Subcommands

Enumerating subcommands in CLI mixin

Subcommands with shared common arguments

Subcommand name and aliases

Default subcommand

Help generation

Command

Argument

Help text styling

Example

Argument groups

ANSI colors and styles

Styles and colors

Styling mode

Heuristics for enabling styling

Supported types

Boolean

Numeric

String

Enum

Counter

Array

Specifying number of values

Associative array

Callback

Custom types

Argument parsing customization

Pre-validation

Parsing

Validation

Action

Arguments with no values

Usage example

Parser customization

Assign character

Array separator

Named argument prefix character

End of named arguments

Case sensitivity

Bundling of single-letter arguments

Adding help generation

Styling mode

Styling scheme

Error handling

Providing a new `main` function without wrapping standard `main`