Skip to content

Tool Shaping


Metadata can be defined using dunders (double underscore variables) at the very beginning of the script/tool and is used by Tinyscript to format the help message.

from tinyscript import *

__author__       = "John Doe"
__contributors__ = [
    {'author': "James McAdams", 'email': ""},
    {'author': "h4x0r1234", 'reason': "for his kind testing"},
__credits__      = "Thanks to Bob for his contribution"
__copyright__    = ("John Doe Inc.", 2020, 2023)
__description__  = "My Script"
__email__        = ""
__license__      = "agpl-3.0"
__version__      = "1.0"
__reference__    = "..."
__source__       = "..."
__training__     = "..."
__examples__     = ["..."]
__doc__          = "This tool ..."


This gives the following help message:

$ python -h
My Script v1.0
Author      : John Doe (
Contributors: James McAdams (
              h4x0r1234 - for his kind testing
Credits     : Thanks to Bob for his contribution
Copyright   : © 2020-2023 John Doe Inc.
License     : GNU Affero General Public License v3.0
Reference   : ...
Source      : ...
Training    : ...

usage: tool [-h] [-v]

This tool ...

extra arguments:
  -h             show usage message and exit
  --help         show this help message and exit
  -v, --verbose  verbose mode (default: False)

Usage examples:
  python ...

See example here

For more detailed information about metadata fields, see this section.

Package requirements checking

(only for Python3.8+)

Required module can be added via the __requires__ dunder in order to check for requirements before the tool starts.

__requires__  = {'tinyscript': "1.23"}

Help message styling

Help can be formatted using multiple markup languages for making the help text more user-friendly with colors and styling. This is configured by using the __docformat__ dunder (which can also be None, meaning no formatting). This feature is supported by mdv and the theme can be tuned by using the constant DOCFORMAT_THEME.

Currently, the following markup languages are supported: HTML, Markdown, RestructuredText and Textile.

__docformat__ = "html"  # None|"md"|"rst"|"textile"

Various formats support

The support for HTML, RestructuredText and Textile is based on document conversion with 'pypandoc' to get Markdown before using mdv. In some cases with complex format text, Pandoc can cause issues with indentation or break the layout. It is then advised to only use Markdown directly to avoid any conversion.

List of themes

Unfortunately, the documentation of mdv is a bit poor. However, one can find the names of the available themes in this JSON.

Script banner

Displaying a banner can be achieved in two ways:

  • by passing a keyword argument add_banner=True ; in this case, a random font is used
  • by defining the BANNER_FONT constant

The BANNER_STYLE constant also allows to style the banner with the following properties:

  • adjust: can be left | center | right ; by default, the banner is centered
  • bgcolor: determines the background color
  • fgcolor: determines the foreground color
BANNER_FONT = "roman"
BANNER_STYLE = {'adjust': "right", 'fgcolor': "blue"}

Multi-level debugging

This is achieved by passing a keyword argument multi_level_debug=[boolean] to initialize(...).


This modifies the classical -v/--verbose option to -v/-vv/-vvv.

Debug levels

  • '': logging.ERROR
  • -v: logging.WARNING
  • -vv: logging.INFO
  • -vvv: logging.DEBUG

See example here

Extended logging options

This is achieved by passing a keyword argument ext_logging=[boolean] to initialize(...).


This adds multiple options:

  • -f, --logfile: This sets the log filename for saving logging messages.
  • -r, --relative: This sets the log timestamps to the time relative to the start of the execution.
  • -s, --syslog: This allows to save the log messages to /var/log/syslog (Linux only).

Multi-level help

This is achieved by setting the __details__ metadata at the beginning of the script.

    __details__ = [
    """Extra documentation, displayed when using -hh. """,
    """Other extra documentation, displayed when using -hhh. """

This modifies the classical -h/--help option to -h/-hh/-hhh.

Help levels

  • -h: classical help message
  • -hh: classical help message + first string in details list
  • -hhh: classical help message + first and second strings in details list

Note: Strings beyond the two first elements of __details__ are not handled by purpose.

See example here

Short/Long help

This is enabled by the short_long_help parameter of initialize(...) and is set to True by default.


When set to True, -h only displays usage information and --help shows the full help message. If False, -h and --help both displays the full help message.

Argument groups ordering

Groups can be ordered by using the after and before keyword-arguments in add_argument_group.

parser.add_argument_group("custom arguments", before="extra arguments")

Subparser choices sorted per category

When defining subparsers under the main parser, it is possible to use the "category" keyword to set a category to get the subparser sorted in. This allows, when there are lots of choices, to sort them and enhance readability of the help message.

cmds = parser.add_subparsers(dest="command", metavar="CMD", title="positional argument", description="command to be executed")
cmd1 = cmds.add_parser("command1", category="category1", help="this does something")
cmd2 = cmds.add_parser("command2", category="category2", help="this does something")

This examples will yield a help message that proposes "CMD", the command to be executed, with category1 holding command1 and category2 holding command2.

Automatic Bash auto-completion with argcomplete

Using the related option, auto-completion can be enabled using argcomplete. This only applies argcomplete.autocomplete to the parser once initialized with the initialize(...) function, meaning that setting up the script and the Bash console accordingly is required as documented on the README of `argcomplete´.

In Bash:

$ activate-global-python-argcomplete --user

Your script, let us say

parser.add_argument("test", choices=["a", "b", "c"])

PATH needs to be in a folder registered in the PATH environment variable.

In Bash:

$ eval "$(register-python-argcomplete"