Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time
  - for easy Shell Scripting

Build Status

This framework is intended to help on easy bash script development. It is totally modularized. It helps you developing new Bash Script programs by forcing you to modularize and organize your code in functions, so that your program can be tested.


You can clone the bang repository in any path and run make install. For instance,

git clone git://
cd bangsh
sudo make prefix=/usr/local install

Hello World

Basically, allows you to run any bash file. A "hello world" program would be like this:

$ echo 'echo "Hello World"' >
$ bang run

Creating a new project

Bang projects are an organized way of placing your files. In order to create a new project, you can just run:

bang new my_project

This command will create a directory called my_project/. There will be some directories which are intended to place some specific files.

 - my_project
 - modules/
 - tasks/
 - tests/

My project is a executable file that can be run as ./my_project or with bang run my_project. modules/ and tasks/ are where you place your files. All files in the modules/ path is automatically loaded, so you can define any function in there, that it will be available in my_project file.

The "b" namespace

After sourcing the file, several functions namespaced by b will be available, for instance:

  • use b.module.require unittest to require modules from bangsh or from your modules folder
  • use b.get bang.working_dir to get information, like where the command was called from:

In the next sections there's more detailed information with some examples that will let you understand how the b namespace is used.


A module is a bunch of functions that have a certain domain. It works like a namespace for aggregating functions. The general idea is to have it isolated, so it could be copied and pasted into another project in a such way it would not rely on any other dependency but Bang, for example:

# modules/

function my_first_module_says () {
  echo "My first module says: $*"

Now, you can use the module in your executable file:

#!/usr/bin/env bang run

my_first_module_says "Hey!"

Remember that exactly like your own modules, those included with bang by default must be required before using them. e.g: b.module.require path


A task is like an action your executable will perform. It is how bang new and bang test work.

Every task file should include at least one function (the entrypoint) following the pattern:

# tasks/<name>.sh
function btask.<name>.run () {
  # code to run on 'yourprogram <name>'

Additionally, as a convention, the other functions defined in the task files should be named with a preceding underscore, e.g:

function _create_module_path () {
  # task helper functions start with _

This is useful to distinguish "where" a given function comes from and to avoid name collisions with executables in the path.

To see more about tasks, check bang's executable to get an idea of how to use the task module in the main executable of your program as a way to create subcommands.

Keep in mind that you can trigger tasks from withing other tasks, allowing you to create nested subcommands for very expressive CLI tools. Related to this, the opt module allows you to have a configuration layer for each task, allowing to have this king of calls: bang test --test-specific-option and bang new --new-task-option


The unittest module provides useful functions for testing. In order to peform assertions, you can use the following functions:

  • b.unittest.assert_error: Asserts a function exit code is 1
  • b.unittest.assert_equal: Asserts a function output is the same as required
  • b.unittest.assert_raise: Asserts a function will raise a given exception
  • b.unittest.assert_success: Asserts a function exit code is zero

There are a few extra functions for more complex scenarios, you can learn about them in the unittest module source

There's a great test task used by the bang binary itself. It prints colors and have some nice defaults that you can use in your project during development. Simply execute the bang test command inside your project's root path. If you put that path in your PATH env var, you can simply run bang test:

$ exprot PATH="$PATH:/path/to/"
$ cd my_project/
$ bang test

str module

Provides three helpers to work with strings: replace, part and trim. You can see details on how to use these in the

path module

Exposes functions to simplify common checks on paths, like: doest this file exists? The API docs can be found in

opt module

Helps you parse options. You can define flags, option aliases and define required arguments. The module can generate usage information if a required argument was missing. You can see the code in the parsing arguments example or check the source that is pretty expressive about the exposed API.

Misc. utils

By sourcing you get by default some useful utils, e.g:

  • Check if a value is inside an array: in_array? "one" (one two)
  • Check for a key in an associative hash: declare -A TEST_HASH=(["foo"]="bar"); key_exists? "foo" TEST_HASH
  • Check if something is a function: is_function? foo
  • Raise and error and exit with code 2: b.abort
  • Print to stderr print_e
  • Sanitize an argument with sanitize_arg this is useful to remove semicolons, pipes, ampersands, etc. that could lead to code injections. Examples can be found in the bang tests
  • Escape argument: escape_arg (turns -- into --)
  • Check whether the argument is a valid module: is_module?

All this helpers together with others (like the global scope variables), are defined in the source

Checking for dependencies

If your program uses external dependencies you can define them and execute a fallback function if the command's missing. For example, check if git is present and print the confirmation or run print_how_to_install_git otherwise: b.depends_on git
b.catch DependencyNotMetException print_how_to_install_git

echo 'You have git installed!'

Handling exceptions

You can handle custom exceptions inside your functions: b.raise FileNotFoundException. You can make WhateverException name you want!

To work with exceptions you define the function to run, the functions that catch specific exceptions and, optionally, a finally clause that gets executed even if no exception was raised. Here's an example: run "$1"
b.catch FileNotFoundException file_not_found
b.catch IDontKnowException i_dont_know_what_it_is
b.finally try_again

In the previos example i_dont_know_what_it_is and file_not_found are simply functions triggered on the applicable cases. try_again however will always run.

Global scope variables

To store information accessible from anywhere in the program during it's excution, you can use b.set using the format: b.set <registry>.<key> <value>.

As a convention, the bang registry is reserved for things like: b.set bang.working_dir $(pwd) (allows you to know exactly from what folder the program executiong began).

The bang registry also stores information about added tasks, e.g: b.get bang.tasks.taskname would return the description of taskname.

You can use other "registries" for your own programs, e.g: b.set "myprogram.version" "1.0"

Here's a list of useful methods related to globally scoped data storage:

  • b.set
  • b.get
  • b.is_set?
  • b.unset


A Bash4 Framework for easy app creation







No packages published