Fork me on GitHub

Guide to using begins

Overview

Command line programs for lazy humans.

  • Decorate a function to be your programs starting point.
  • Generate command line parser based on function signature.
  • Search system environment for option default values.

Why begins?

I write a lot of small programs in Python. These programs often accept a small number of simple command line arguments. Having to write command line parsing code in each of these small programs both breaks my train of thought and greatly increases the volume of code I am writting.

Begins was implemented to remove the boilerplate code from these Python programs. It’s not intended to replace the rich command line processing needed for larger applications.

Requirements

For Python versions earlier than Python 3.3, the funcsigs package from the Python Package Index is required.

For Python version 2.6, the argparse package from the Python Package Index is also required.

Both of these dependencies are listed in the package configuration. If using Pip to install begins then the required dependencies will be automatically installed.

Installation

begins is available for download from the Python Package Index. To install using Pip

$ pip install begins

Alternatively, the latest development version can be installed directly from Github.

$ pip install git+https://github.com/aliles/begins.git

Please note that begins is still in an alpha state and therefore the API or behaviour could change.

Setting a programs starting point

The begin.start() function can be used as a function call or a decorator. If called as a function it returns True when called from the __main__ module. To do this it inspects the stack frame of the caller, checking the __name__ global.

This allows the following Python pattern:

>>> if __name__ == '__main__':
...     pass

To be replace with:

>>> import begin
>>> if begin.start():
...    pass

If used as a decorator to annotate a function the function will be called if defined in the __main__ module as determined by inspecting the current stack frame. Any definitions that follow the decorated function wont be created until after the function call is complete.

Usage of begin.start() as a decorator looks like:

>>> import begin
>>> @begin.start
... def run():
...     pass

By deferring the execution of the function until after the remainder of the module has loaded ensures the main function doesn’t fail if depending on something defined in later code.

Parsing command line options

If begin.start() decorates a function accepts parameters begin.start() will process the command for options to pass as those parameters:

>>> import begin
>>> @begin.start
... def run(name='Arther', quest='Holy Grail', colour='blue', *knights):
...     "tis but a scratch!"

The decorated function above will generate the following command line help:

usage: example.py [-h] [-n NAME] [-q QUEST] [-c COLOUR]
                  [knights [knights ...]]

tis but a scratch!

positional arguments:
  knights

optional arguments:
  -h, --help            show this help message and exit
  -n NAME, --name NAME  (default: Arther)
  -q QUEST, --quest QUEST
                        (default: Holy Grail)
  -c COLOUR, --colour COLOUR
                        (default: blue)

In Python3, any function annotations for a parameter become the command line option help. For example:

>>> import begin
>>> @begin.start                                         
... def run(name: 'What, is your name?',
...         quest: 'What, is your quest?',
...         colour: 'What, is your favourite colour?'):
...     pass

Will generate command help like:

usage: holygrail_py3.py [-h] -n NAME -q QUEST -c COLOUR

optional arguments:
  -h, --help            show this help message and exit
  -n NAME, --name NAME  What, is your name?
  -q QUEST, --quest QUEST
                        What, is your quest?
  -c COLOUR, --colour COLOUR
                        What, is your favourite colour?

Command line parsing supports:

  • positional arguments
  • keyword arguments
  • default values
  • variable length arguments
  • annotations

Command line parsing does not support variable length keyword arguments, commonly written as **kwargs. If variable length keyword arguments are used by the decorated function an exception will be raised.

If a parameter does not have a default, failing to pass a value on the command line will cause running the program to print an error and exit.

For programs that have a large number of options it may be preferable to only use long options. To suppress short options, pass False as the short_args keyword argument to the begin.start decorator:

>>> import begin
>>> @begin.start(short_args=False)
... def run(name='Arther', quest='Holy Grail', colour='blue', *knights):
...     "tis but a scratch!"

This program will not accept -n, -q or -c as option names.

Similarity, a large number of command line options may be better displayed in alphabetical order. This can be achieved by passing lexical_order as True:

>>> import begin
>>> @begin.start(lexical_order=True)
... def main(charlie=3, alpha=1, beta=2):
...     pass

This program will list the command line options as alpha, beta, charlie instead of the order in which the function accepts them.

Boolean options

If a command line option has a default value that is a bool object. (True or False) The command line option will be flags rather than an option that accepts a value. Two flags are generated, one to set a True value and one to set a False value. The two commands will be of the form --flag and --no-flag. For example:

>>> import begin
>>> @begin.start
... def main(enable=False, disable=True):
...     pass

Using --enable or --no-disable when invoking this program will invert the associated option. The options --no-enable and --disable have not effect.

Sub-Commands

begins supports using functions as sub-commands with the begin.subcommand() decorator:

>>> import begin
>>> @begin.subcommand                                    
... def name(answer):
...     "What is your name?"
...
>>> @begin.subcommand                                    
... def quest(answer):
...     "What is your quest?"
...
>>> @begin.subcommand                                    
... def colour(answer):
...     "What is your favourite colour?"
...
>>> @begin.start
... def main():
...     pass

This example registers three sub-commands for the program:

usage: subcommands.py [-h] {colour,name,quest} ...

optional arguments:
  -h, --help           show this help message and exit

Available subcommands:
  {colour,name,quest}
    colour             What is your favourite colour?
    name               What is your name?
    quest              What is your quest?

The main function will always be called with the provided command line arguments. If a sub-command was chosen the associated function will also be called.

It is possible to create a sub-command with a different name from the decorated function’s name. To do this pass the desired sub-command name using the name keyword argument:

>>> import begin
>>> @begin.subcommand(name='colour')                     
... def question(answer):
...     "What is your favourite colour?"

Sub-commands can also be registered with a specific named group by passing a group argument to the begin.subcommand decorator. The begin.start() decorator can use sub-commands from a named group by passing it a sub_group argument.

Similarly, sub-commands can be load from entry points by passing the name of the entry point through the plugins argument to the begin.start() decorator:

>>> import begin
>>> @begin.start(plugins='begins.plugin.demo')
... def main():
...     pass

Any functions from installed packages that are registered with the begins.plugin.demo entry point will be loaded as sub-commands.

Multiple Sub-Commands

Some commands may benefit from being able to be called with multiple subcommands on the command line. The enable multiple sub-commands a command separator value needs to be passed to be passed to begin.start() as the cmd_delim parameter:

>>> import begin
>>> @begin.subcommand                                    
... def subcmd():
...     pass
...
>>> @begin.start(cmd_delim='--')
... def main():
...     pass

When this program is called from the command line multiple instances of the sub-command may be called if separated by the command delimiter --.

Sub-Command Context

There are use cases where it is desirable to pass state from the main function to a subsequent sub-command. To support this Begins provides the begin.context object. This object will have the following properties:

  • last_return, value returned by previous command function.
  • return_values, iterable of all return values from previous commands.
  • opts_previous, iterable of options object used by previous commands.
  • opts_current, options object for current command.
  • opts_next, iterable of options object for following commands.
  • (deprecated) return_value, replaced by last_return.

Any other properties set on the begin.context object will not be altered by begins.

The last_return property and return_values will always be populated, even in the value returned from the main function or a sub-command function is the None object. The length and order of the return_values will match those of opts_previous.

Environment Variables

Environment variables can be used to override the default values for command line options. To use environment variables pass a prefix string to the begin.start() decorator through the env_prefix parameter:

>>> import begin
>>> @begin.start(env_prefix='MP_')
... def run(name='Arther', quest='Holy Grail', colour='blue', *knights):
...     "tis but a scratch!"

In the example above, if an environment variable MP_NAME existed, it’s value would be used as the default for the name option. The options value can still be set by explicitly passing a new value as a command line option.

Configuration files

Configuration files can also be used to override the default values of command line options. To use configuration files pass a base file name to the begin.start() decorator through the config_file parameter:

>>> import begin
>>> @begin.start(config_file='.camelot.cfg')
... def run(name='Arther', quest='Holy Grail', colour='blue', *knights):
...     "tis but a scratch!"

This example will look for configuration files named .camelot.cfg in the current directory and/or the user’s home directory. A command line option’s default value can be changed by an option value in a configuration file. The configuration section used matches the decorated function’s name by default. This can be changed by passing a config_section parameter to begin.start():

>>> import begin
>>> @begin.start(config_file='.camelot.cfg', config_section='camelot')
... def run(name='Arther', quest='Holy Grail', colour='blue', *knights):
...     "tis but a scratch!"

In this second example the section camelot will be used instead of a section named run.

Argument type casting

Command line arguments are always passed as strings. Sometimes thought it is more convenient to receive arguments of different types. For example, this is a possible function for starting a web application:

>>> import begin
>>> @begin.start
... def main(host='127.0.0.1', port='8080', debug='False'):
...    port = int(port)
...    debug = begin.utils.tobool(debug)
...    "Run web application"

Having to convert the port argument to an integer and the debug argument to a boolean is additional boilerplate code. To avoid this begins provides the begin.convert() decorator. This decorator accepts functions as keyword arguments where the argument name matches that of the decorator function. These functions are used to convert the types of arguments.

Rewriting the example above using the begin.convert() decorator:

>>> import begin
>>> @begin.start
... @begin.convert(port=int, debug=begin.utils.tobool)
... def main(host='127.0.0.1', port=8080, debug=False):
...    "Run web application"

The module begin.utils contains useful functions for converting argument types.

Automatic casting

For simple, built-in types begins can automatically type cast arguments. This is achieved by passing the parameter _automatic to begin.convert():

>>> import begin
>>> @begin.start
... @begin.convert(_automatic=True)
... def main(host='127.0.0.1', port=8080, debug=False):
...     "Run web application"

This example is functionally equivalent to the example above.

Automatic type casting works for the following built-in types.

  • int or long
  • float
  • boolean
  • tuple or list

Additional casting functions can be provided with the same call to the begin.convert() decorator.

Alternatively, use of begin.convert() can be dispensed by passing True to begin.start() via the auto_convert parameter:

>>> import begin
>>> @begin.start(auto_convert=True)
... def main(host='127.0.0.1', port=8080, debug=False):
...     "Run web application"

Again, this example is functionally equivalent to the example above.

The limitation of using auto_convert is that it is not longer possible to provide additional casting functions.

Command Line Extensions

There are behaviours that are common to many command line applications, such as configuring the logging and cgitb modules. begins provides function decorators that extend a program’s command line arguments to configure these modules.

  • begin.tracebacks()
  • begin.logging()

To use these decorators they need to decorate the main function before begin.start() is applied.

Tracebacks

The begin.tracebacks() decorator adds command line options for extended traceback reports to be generated for unhandled exceptions:

>>> import begin
>>> @begin.start
... @begin.tracebacks
... def main(*message):
...     pass

The example above will now have the following additional argument group:

tracebacks:
  Extended traceback reports on failure

  --tracebacks   Enable extended traceback reports
  --tbdir TBDIR  Write tracebacks to directory

Passing --tracebacks will cause extended traceback reports to be generated for unhandled exceptions.

Traceback options may also be set using configuration files, if Configuration files are supported. The follow options are used.

  • enabled: use any of true, t, yes, y, on or 1 to enable tracebacks.
  • directory: write tracebacks to this directory.

Options are expected to be in a tracebacks section.

Logging

The begin.logging() decorator adds command line options for configuring the logging module:

>>> import logging
>>> import begin
>>> @begin.start
... @begin.logging
... def main(*message):
...     for msg in message:
...         logging.info(msg)

The example above will now have two additional optional arguments as well as an additional argument group:

optional arguments:
  -h, --help            show this help message and exit
  -v, --verbose         Increse logging output
  -q, --quiet           Decrease logging output

logging:
  Detailed control of logging output

  --loglvl {DEBUG,INFO,WARNING,ERROR,CRITICAL}
                        Set explicit log level
  --logfile LOGFILE     Ouput log messages to file
  --logfmt LOGFMT       Log message format

The logging level defaults to INFO. It can be adjusted by passing --quiet, --verbose or explicitly using --loglvl.

The default log format depends on whether log output is being directed to standard out or file. The raw log text is written to standard out. The log message written to file output includes:

  • Time
  • Log level
  • Filename and line number
  • Message

The message format can be overridden using the --logfmt option.

Logging options may also be set using configuration files, if Configuration files are supported. The follow options are used.

  • level: log level, must be one of DEBUG, INFO, WARNING, ERROR or CRITICAL.
  • file: output log messages to this file.
  • format: log message format.

Options are expected to be in a logging section.

Command Line Formatting

The default argparse help formatter may not always meet your needs. An alternate formatter can be provided using the formatter_class argument to begin.start():

>>> import begin, argparse
>>> @begin.start(formatter_class=argparse.RawTextHelpFormatter)
... def main():
...     pass

Any of the formatter classes provided by the argparse module can be used.

Alternatively, begin.formatters provides a mechanism to compose new formatter class according to your requirements.:

>>> from begin import formatters
>>> formatter_class = formatters.compose(formatters.RawDescription, formatters.RawArguments)

The following mixin classes are provided for use with begin.formatters.compose()

  • RawDescription
  • RawArguments
  • ArgumentDefaults
  • RemoveSubcommandsLine

One or more of these may be passed to begin.formatters.compose() to create a new formatter class.

Entry Points

The setuptools package supports automatic script creation to automatically create command line scripts. These command line scripts use the entry points system from setuptools.

To support the use of entry points, functions decorated by begin.start() have an instance method called start() that must be used to configure the entry point:

setup(
    # ...
    entry_points = {
        'console_scripts': [
            'program = package.module:main.start'
        ]
    }

Use of the start() method is required because the main function is not called from the __main__ module by the entryp points system.

Issues

Any bug reports or feature requests can be made using Github’s issues system.