Clamp
"Clamp" is a minimal framework for command-line utilities.
It handles boring stuff like parsing the command-line, and generating help, so you can get on with making your command actually do stuff.
Not another one!
Yeah, sorry. There are a bunch of existing command-line parsing libraries out there, and Clamp draws inspiration from a variety of sources, including Thor, optparse, and Clip. In the end, though, I wanted a slightly rounder wheel.
Quick Start
Clamp models a command as a Ruby class; a subclass of Clamp::Command
. They look something like this:
class SpeakCommand < Clamp::Command
option "--loud", :flag, "say it loud"
option ["-n", "--iterations"], "N", "say it N times", :default => 1 do |s|
Integer(s)
end
parameter "WORDS ...", "the thing to say"
def execute
signal_usage_error "I have nothing to say" if arguments.empty?
the_truth = arguments.join(" ")
the_truth.upcase! if loud?
iterations.times do
puts the_truth
end
end
end
Class-level methods (like option
and parameter
) are available to declare command-line options, and document usage.
The command can be invoked by instantiating the class, and asking it to run:
SpeakCommand.new("speak").run(["--loud", "a", "b", "c"])
but it's more typical to use the class-level "run
" method:
SpeakCommand.run
which takes arguments from ARGV
, and includes some handy error-handling.
Declaring options
Options are declared using the option
method. The three required arguments are:
- the option switch (or switches),
- a short description of the option argument type, and
- a description of the option itself
For example:
option "--flavour", "FLAVOUR", "ice-cream flavour"
It works a little like attr_accessor
, defining reader and writer methods on the command class. The attribute name is derived from the switch (in this case, "flavour
"). When you pass options to your command, Clamp will populate the attributes, which are then available for use in your #execute
method.
def execute
puts "You chose #{flavour}. Excellent choice!"
end
If you don't like the inferred attribute name, you can override it:
option "--type", "TYPE", "type of widget", :attribute_name => :widget_type
# to avoid clobbering Object#type
Short/long option switches
The first argument to option
can be an array, rather than a single string, in which case all the switches are treated as aliases:
option ["-s", "--subject"], "SUBJECT", "email subject line"
Flag options
Some options are just boolean flags. Pass ":flag
" as the second parameter to tell Clamp not to expect an option argument:
option "--verbose", :flag, "be chatty"
For flag options, Clamp appends "?
" to the generated reader method; ie. you get a method called "verbose?
", rather than just "verbose
".
Negatable flags are easy to generate, too:
option "--[no-]force", :flag, "be forceful (or not)"
Clamp will handle both "--force
" and "--no-force
" options, setting the value of "#force?
" appropriately.
Validation and conversion of option arguments
If a block is passed to option
, it will be called with the raw string option argument, and is expected to coerce that String to the correct type, e.g.
option "--port", "PORT", "port to listen on" do |s|
Integer(s)
end
If the block raises an ArgumentError, Clamp will catch it, and report that the option value was bad:
ERROR: option '--port': invalid value for Integer: "blah"
Declaring parameters
The parameter
method is used to declare positional command parameters:
parameter "FILE ...", "source files"
parameter "DIR", "target directory"
Use of parameter
is entirely for documentation purposes. Whether or not you declare and describe your expected arguments, the actual arguments that remain after option parsing will be available as arguments
when your #execute
method is called.
Sub-commands
The subcommand
method declares sub-commands:
class MainCommand < Clamp::Command
subcommand "init", "Initialize the repository" do
def execute
# ...
end
end
end
Clamp generates an anonymous sub-class of the current class, to represent the sub-command. Additional options may be declared within subcommand blocks, but all options declared on the parent class are also accepted.
Alternatively, you can provide an explicit sub-command class, rather than a block:
class MainCommand < Clamp::Command
subcommand "init", "Initialize the repository", InitCommand
end
class InitCommand < Clamp::Command
def execute
# ...
end
end
When a command has sub-commands, Clamp will attempt to delegate based on the first command-line argument, before options are parsed. Remaining arguments will be passed on to the sub-command.
Getting help
All Clamp commands support a "--help
" option, which outputs brief usage documentation, based on those seemingly useless extra parameters that you had to pass to option
and parameter
.
$ speak --help
Usage:
speak [OPTIONS] WORDS ...
Arguments:
WORDS ... the thing to say
Options:
--loud say it loud
-n, --iterations N say it N times
--help print help
Contributing to Clamp
Source-code for Clamp is on Github.