A modern, powerful commmand line argument parser.
Batteries included.
βοΈ Report a bug
Β·
π‘ Request feature
commandr handles all kinds of command-line arguments with a nice and clean interface.
Comes with help generation, shell auto-complete scripts and validation.
Add this entry to your dub.json
file:
"dependencies": {
...
"commandr": "~>0.1"
...
}
-
Does it use templates/compile-time magic?
No, at least currently not. Right now everything is done at runtime, so there's not much overhead on compilation time resources. In the future I'll probably look into generation of compile-time struct.
The reason is that I want it to be rather simple and easy to learn, and having a lot of generated code hurts e.g. generated documentation and some minor things such as IDE auto-complete (even right now mixin-s cause some problems).
-
Are the results typesafe? / Does it use UDA?
No, parsed arguments are returned in a
ProgramArgs
class instance that allow to fetch parsed data,However it should be possible to generate program definition from struct/class with UDA and then fill the parsed data into struct instance, but it is currently out of scope of this project (at least for now).
-
Flags (boolean values)
- Short and long forms are supported (
-v
,--verbose
) - Supports stacking of flags (
-vvv
is same as-v -v -v
)
- Short and long forms are supported (
-
Options (taking a string value)
- Short and long forms are supported (
-c test
,--config test
) - Equals sign accepted (
-c=1
,--config=test
) - Repeated options are supported (
-c 1 -c 2
) - Default values can be specified.
- Options be marked as required.
- Short and long forms are supported (
-
Arguments (positional)
- Required by default, can be marked as optional
- Default values can be specified.
- Repeated options are supported (only last argument)
-
Commands (git-style)
- Infinitely recursive subcommands (you can go as deep as needed)
- Contains own set of flags/options and arguments
- Dedicated help output
- Comfortable command handling with
ProgramArgs.on()
-
Provided help output
- Generated help output for your program and sub-commands
- Can be configured to suit your needs, such as disabling colored output.
- Provided usage, help and version information.
- Completly detached from core
Program
, giving you complete freedom in writing your own help output. - You can categorize commands for better help output
-
Consistency checking
- When you build your program model,
commandr
checks its consistency. - Detects name duplications as well as short/long options.
- Detects required parameters with default value.
- When you build your program model,
-
BASH auto-complete script
- You can generate completion script with single function call
- Completion script works on flags, options and sub-commands (at any depth)
- Acknowledges difference between flags and options
-
Validators
- Passed values can be checked for correctness
- Simple process of creating custom validating logic
- Provided validators for common cases:
EnumValidator
,FileSystemValidator
andDelegateValidator
-
Suggestions
- Suggestion with correct flag, option or sub-command name is provided when user passes invalid value
- Also supported for
EnumValidator
(acceptsValues
)
Simple example showing how to create a basic program and parse arguments:
import std.stdio;
import commandr;
void main(string[] args) {
auto a = new Program("test", "1.0")
.summary("Command line parser")
.author("John Doe <[email protected]>")
.add(new Flag("v", null, "turns on more verbose output")
.name("verbose")
.repeating)
.add(new Option(null, "test", "some teeeest"))
.add(new Argument("path", "Path to file to edit"))
.parse(args);
writeln("verbosity level", a.occurencesOf("verbose"));
writeln("arg: ", a.arg("path"));
}
You can create subcommands in your program or command using .add
. You can nest commands.
Adding subcommands adds a virtual required argument at the end to your program. This makes you unable to declare repeating or optional arguments (because you cannot have required argument past these).
Default command can be set with .defaultCommand(name)
call after defining all commands.
After parsing, every subcommand gets its own ProgramArgs
instance, forming a hierarchy. Nested args inherit arguments from parent, so that options defined higher
in hierarchy are copied.
ProgramArgs defines a helper method on
, that allows to dispatch method on specified command.
auto args = new Program("test", "1.0")
.add(new Flag("v", null, "turns on more verbose output")
.name("verbose")
.repeating)
.add(new Command("greet")
.add(new Argument("name", "name of person to greet")))
.add(new Command("farewell")
.add(new Argument("name", "name of person to say farewell")))
.parse(args);
args
.on("greet", (args) {
// args.flag("verbose") works
writefln("Hello %s!", args.arg("name"));
})
.on("farewell", (args) {
writefln("Bye %s!", args.arg("name"));
});
Delegate passed to on
function receives ProgramArgs
instance for that subcommand. Because it is also ProgramArgs
, on
chain can be nested, as in:
// assuming program has nested subcommands
a.on("branch", (args) {
args
.on("add", (args) {
writefln("adding branch %s", args.arg("name"));
})
.on("rm", (args) {
writefln("removing branch %s", args.arg("name"));
});
});
You can attach one or more validators to options and arguments with validate
method. Every validator has its own helper function that simplifies adding it do option (usually starting with accepts
):
new Program("test")
// adding validator manually
.add(new Option("s", "scope", "")
.validate(new EnumValidator(["local", "global", "system"]))
)
// helper functionnew Program("test")
.add(new Option("s", "scope", "")
.acceptsValues(["local", "global", "system"]));
-
EnumValidator - Allows to pass values from white-list.
Helpers:
.acceptsValues(values)
-
FileSystemValidator - Verifies whenever passed values are files/directories or just exist (depending on configuration).
Helpers:
.acceptsFiles()
,.acceptsDirectories()
,.acceptsPaths(bool existing)
-
DelegateValidator - Verifies whenever passed values with user-defined delegate.
Helpers:
.validateWith(delegate)
,validateEachWith(delegate)
You can create custom validators either by implementing IValidator
interface, or by using DelegateValidator
:
new Program("test")
// adding validator manually
.add(new Option("s", "scope", "")
.validateEachWith(opt => opt.isDirectory, "must be a valid directory")
);
You can print help for program or any subcommand with printHelp()
function:
program.printHelp(); // prints program help
program.commands["test"].printHelp();
To customise help output, pass HelpOutput
struct instance:
HelpOutput helpOptions;
helpOptions.colors = false;
helpOptions.optionsLimit = 2;
program.printHelp(helpOptions);
Commandr can generate BASH autocompletion script. During installation of your program you can save the generated script to /etc/bash_completion.d/<programname>.bash
(or any other path depending on distro).
import commandr;
import commandr.completion.bash;
string script = program.createBashCompletionScript();
// save script to file
TODO
Overview of available entries that can be added to program or command with .add
method:
What | Type | Example | Definition |
---|---|---|---|
Flag | bool | --verbose |
new Flag(abbrev?, full?, summary?) |
Option | string[] | --db=test |
new Option(abbrev?, full?, summary?) |
Argument | string[] | 123 |
new Argument(name, summary?) |
Shows how to access values after parsing args.
Examples assume args
variable contains result of parse()
or parseArgs()
function calls (an instance of ProgramArgs
)
ProgramArgs args = program.parse(args);
What | Type | Fetch |
---|---|---|
Flag | bool | args.flag(name) |
Flag | int | args.occurencesOf(name) |
Option | string | args.option(name) |
Option | string[] | args.options(name) |
Argument | string | args.arg(name) |
Argument | string[] | args.args(name) |
Table below shows which fields exist and which don't (or should not be used).
Column name
contains name of the method to set the value. All methods return
this
to allow chaining.
Name | Program | Command | Flag | Option | Argument |
---|---|---|---|---|---|
.name |
βοΈ | βοΈ | βοΈ | βοΈ | βοΈ |
.version_ |
βοΈ | βοΈ | β | οΈβ | β |
.summary |
βοΈοΈ | οΈ:heavy_check_mark: | β | οΈβ | β |
.description |
β | οΈβ | βοΈ | οΈ:heavy_check_mark: | βοΈ |
.abbrev |
β | β | βοΈ | βοΈ | β |
.full |
β | β | βοΈοΈ | οΈ:heavy_check_mark: | β |
.tag |
β | β | β | οΈ:heavy_check_mark: | βοΈοΈ |
.defaultValue |
β | β | β | οΈ:heavy_check_mark: | βοΈοΈ |
.required |
β | β | β | οΈ:heavy_check_mark: | βοΈοΈ |
.optional |
β | β | β | οΈ:heavy_check_mark: | βοΈοΈ |
.repeating |
β | β | βοΈ | οΈ:heavy_check_mark: | βοΈοΈ |
.topic |
β | βοΈ | β | οΈβ | β |
.topicGroup |
βοΈ | βοΈ | β | οΈβ | β |
.authors |
βοΈ | β | β | οΈβ | β |
.binaryName |
βοΈ | β | β | οΈβ | β |
Current major missing features are:
- Command/Option aliases
- Combined short flags/options (e.g.
-qLob
) - EnumValidator/FileSystemValidator auto-completion hinting
See the open issues for a list of proposed features (and known issues).