Create a command line interface with minimal setup.
Clima handles loading and parsing command line arguments with some off-the-shelf features, including:
Creating a command-line interface for your program:
Example: to set up a configuration and a command-line interface ready to go.
from clima import c
@c
class Cli:
def say_hi(self):
print('oh hi - whatever this is..')
The command line usage form could be as simple as:
my_tool say_hi
from clima import c
# Defining the settings (configuration object)
class S(Schema):
place = 'world'
@c
class Cli:
def say_hi(self):
# using configuration object 'c'
print(f'oh hi - {c.place}')
Again, the command line usage form could be as simple as:
my_tool say_hi
my_tool say_hi --place 'other world'
See the examples folder and other sections for more examples. For example, the folder includes something that resembles
the example above.
pip install --user clima
See the example file in examples/script_example.py. Here's a rundown of the individual
parts in such a script (adapted from another example at module example).
First, import the required components:
from clima import c, Schema
In your code, define the Schema subclass:
class Configuration(Schema):
a: str = 'A' # a description
x: int = 1 # x description
Here "Configuration" is an arbitrary name, no magic there. The inherited Schema class
defines the attributes (i.e. a and x in this example).
Note the specific formatting of the Schema subclass:
# attribute[: type] = default value [# Description for the --help]
a: str = 'A' # a description
a is the attribute that can be called in the code later with c.a. In this example, it has a type of 'str' and a default
value of 'A'. The values in square brackets [] are
optional.
Clima parses the comment after the definition for the command-line help printout. In other words, Clima parses all of these parts to be displayed when the program is called using the argument '--help'. For example like this:
./script.py foo -h
Will now produce:
Usage: script.py foo [ARGS]
Description: Args:
--a (str): a description (Default is 'A')
--x (int): x description (Default is 1)
The example shown in this readme is at examples/readme_example.py.
Clima parses the methods as subcommands with their respective doc-strings, when the class is wrapped with the decorator @c, and will show these in the subcommand's help printout.
The subcommands should be defined somewhat as follows:
@c
class Cli:
def subcommand_foo(self):
"""This will be shown in --help for subcommand-foo"""
print('foo')
print(c.a)
print(c.x)
def subcommand_bar(self):
"""This will be shown in --help for subcommand-bar"""
print('bar')
Note the double usage of the c - first as a decorator and later as the parsed configuration
inside the method:
...
...
print(c.a)
print(c.x)
As a decorator, @c defines the class to be parsed as the subcommands. As an object c it is used to access all the arguments.
Tried and used on Linux, macOS, and windows. However, packaging and dependency management in python is sometimes hairy and your mileage may vary.
More examples in the examples directory with printouts of the defined subcommands and helps.
The examples can be tried out by cloning the repo and running from repo directory root (on linux and the like):
git clone https://github.com/d3rp/clima.git
cd clima
PYTHONPATH=$PWD python ./examples/readme_example.py foo -h
Running the examples that wrap a module:
PYTHONPATH=$PWD python ./examples/module_example/__main__.py -h
PYTHONPATH=$PWD python ./examples/module_example/__main__.py subcommand-foo -h
PYTHONPATH=$PWD python ./examples/module_example/__main__.py subcommand-bar
...
The output should resemble this (fire v0.1.3 prints out Args, fire v0.2.1 does not (though it looks much nicer))
$ tester subcommand-foo -- -h
Type: method
String form: <bound method Cli.subcommand_foo of <__main__.Cli object at 0x000002995AD74BE0>>
File: C:\Users\foobar\code\py\clima\tester\__main__.py
Line: 18
Docstring: This will be shown in --help for subcommand-foo
Args:
--a (str): a description (Default is 'A')
--x (int): x description (Default is 1)
Usage: __main__.py subcommand-foo [--X ...]
All of the example scripts can be run by installing poetry and running the run_examples.bash
script:
pip install --user poetry
./run_examples.bash
Version printing works via the version subcommand and provides a version check functionality for (poetry) packaged scripts. Thus with bumping the version with poetry, Clima will handle parsing the current version of the packaged tool thus providing a subcommand such as:
my_tool version
Clima parses the version attribute into the c object, so if you want control over it, you can overwrite it with the post_init or by handling the c.version otherwise.
Also, to enable autocompletion in IDEs, this hack suffices:
c: Configuration = c
Put it in the "global namespace" e.g. just after defining the template. See the examples/script_example.py for a specific example.
When all is complete, the imported c variable should have all the bits and pieces for the configuration. It can be
used inside the Cli class as well as imported around the codebase thus encapsulating all the configurations into one
container with quick access to the attributes (c.a, c.x, ...).
Running the script with -- --completion arguments should print an autocompletion declaration to include in a bash completions file:
my_tool -- --completion
TBD: zsh etc. completions
There are two ways to define a post_init hook depending on the Schema subclass or the Cli definition.
On some occasions, it is helpful to deduce specific defaults from the given arguments,
for example, in a cross-platform build allowing
only minimal CLI arguments. For those cases, there is a post_init hook
When defining the post_init() in the Cli class, for example:
@c
class Cli:
@staticmethod
def post_init(s):
if s.platform = 'win':
self.bin_path = 'c:/Users/foo/bar'
else:
s.bin_path = '/Users/mac/sth'
def subcommand(self):
pass
The method will have access to the cli args, but can not introduce new variables to the schema.
This is arguably the more useful of the two variations of post_inits.
Note: The signature of the post_init() differs in these two options. For the time being, it is a @staticmethod
This alternative provides post_init-like features that are overrideable with the CLI arguments.
class SoAdvanced(Schema):
platform: str = 'win' # a description
bin_path: pathlib.Path = '' # x description
def post_init(self, *args):
if self.platform = 'win':
self.win_specific_field = 'All your files are locked by us..'
Note: This post_init() does not have access to the CLI arguments, but the Schema's post_init can introduce new
attributes/properties/fields/arguments to the configuration, which the Cli-class post-init can't.
The Schema's post_init() hook is run after the schema object's initialisation, but BEFORE the command-line object's initialisation.
It is tedious to write a long list of parameters on the command line when most use cases follow a similar pattern. There are several options to choose from to facilitate the use of configurations.
The c decorator/configuration chains multiple configuration options together in order of priority (lower numbers refer to higher priority):
~/.password-store if gnugpg is installedSchemaThe configuration file should be named with either the postfix .cfg or .conf, for example foo.conf, and have an ini type formatting including
an explicit 'Clima' section:
# foo.conf
[Clima]
x = 2
The keys are the same as in the schema class' declaration. Default values can be defined for all, some or none of the attributes. The same applies to the env variables.
# linux example
X=2 tester subcommand-foo
A configuration file defined this way can be located in the current working directory or - if your Schema defines a
cwd field - there. Clima
will try to use the first configuration file it finds, so that might produce some caveats.
class Conf(Schema):
cwd = ''
# Running ./script.py --cwd <folder> would automatically load the first *.conf file in <folder>
The Schema definition can have type annotations, which are used to cast the given arguments. For example
class C(Schema):
p: Path = '' # Path to something
Results in c.p's type cast as Path.
You can also define the config file in the configuration class (one inheriting Schema) by defining the
magic field CFG.
For example, lets say the command my_tool (packaged etc) has a user configuration file at ~/.my_tool.conf. This
can now be handled with just adding CFG = Path.home() / '.my_tool.conf to the Schema:
from pathlib import Path
class S(Schema):
bing = 'bang'
CFG = Path.home() / '.my_tool.conf'
Then, for example, the configuration file would be written as:
#~/.my_tool.conf
[Clima]
bing = diudiu
Running the command my_tool would produce the value in the configuration file, though the arguments can still be overriden.
my_tool run
# diudiu
my_tool run --bing bam
# bam
This is handled by dotenv. In short, all the defaults defined in the
Schema subclass can be overridden either by:
<field> = <value>
or
export <field> = <value>
If gnugpg(2) is installed, clima can leverage it to decrypt secrets on-the-fly.
Note: Currently this works most conveniently with gpg-keys without a password. Gpg handles the password prompts which might fail on some platforms or configurations.
Note 2: Leading and trailing whitespace (including \n linefeeds) are stripped, when decrypted.
pass (not required) can be used to store passwords as gpg encrypted files under the home directory. Clima uses the default path of ~/.password-store and the files found within. It will then match the arguments with the stored passwords, for example:
tree -A ~/.password-store ✔ | 41s | anaconda3
/Users/me/.password-store
├── work
│ ├── ci
│ │ ├── sign_id.gpg
│ │ ├── sign_pw.gpg
... ... ...
And an according Schema definition:
class Conf(Schema):
sign_id: str = '' # signing id for the CI
sign_pw: str = '' # signing pw for the CI
Would accept those arguments as cli arguments, or if omitted, would traverse through the .password-store and decrypt the
found sign_id.gpg and sign_pw.gpg placing the values found in the configuration object c.
See the Python Fire's Flags documentation for nice additional features such as:
# e.g. tester.py is our cli program
tester.py subcommand-foo -- --trace
tester.py -- --interactive
tester.py -- --completion
This feature rose as an opinionated option, and I admit, it should be something the user could bypass. Even though I have used python for a few years professionally, I am still not satisfied with its error printing. When raising exceptions, Clima truncates the error lists and tries to provide a more readable version of the "first" point of failure. The whole traceback is written into a logfile exception_traceback.log to examine if the truncated output provides insufficient information.
Note: When running the examples, the exception_traceback.log file will be written inside the examples directory
Error (full trace in exception.log):
traceback_example.py:7 :: lumberjack() : self.bright_side_of_death() =>
traceback_example.py:12 :: bright_side_of_death() : return tuple()[0] => IndexError
IndexError: tuple index out of range
Here are some suggestions on how to wrap scripts using Clima.
Let's say those lines were written in a file named script.py. Command line usage in a terminal would then be e.g.:
python script.py foo
python script.py foo --a 42
Adding this line in script.py
#!/usr/bin/env python
and changing its execution permissions (mac, linux):
chmod +x script.py
Allows for a shorter means of execution:
./script.py foo
Now this could be linked as an adhoc command for example:
ln -s $PWD/script.py ~/.local/bin/my_command
For a pip-installable package, one could package this as a runnable command - publish in the public or one's private pypi etc - and then approach the convenience factor shown at first.
pip install my_tool
my_command foo -h
To publish with poetry is quite straight forward. First create an account in pypi.org and then:
cd <project directory>
poetry build
poetry publish
You can use version to bump up versions:
poetry version patch
This repo is based on poetry.
git clone https://github.com/d3rp/clima.git
cd clima
poetry install --no-dev
The --no-dev is for to install the running environment without development tooling.
You can write your subcommands as a class encapsulating the "business logic". Clima encapsulates the command-line arguments as a container, mapping its attributes as you declare them in a simple schema class.
In other words, you can use this to wrap your scripts as command line commands without resorting to Bash or maintaining argument parsing in python. It removes the need to duplicate comments to have --help remember what the arguments were and what they did. Sprinkling some decorator magic offers a typical user experience of a CLI program (e.g. argument parsing and validation, --help, subcommands, ...).
The premise behind Clima is that a simple script usually uses a script wide global configuration throughout the user code, or in other words, a context for the program accessed in different parts of the code. Clima populates that context, or configuration, with given arguments falling back on defaults in the code and some further complimentary options. Those are then made accessible via a global c variable, or container, that can be tossed around the code base with minimal additional effort.
With almost no additional effort, Clima can provide autocompletion in IDEs (as attributes). This feature helps when extending your code as the autocomplete kicks in after typing c. offering those fields in your "schema" as attributes.
Clima does not try to cater as a feature-complete CLI framework like the ones listed below. It is a package to help with boilerplate to get quick but reusable tools for your workflow.
Other options for a full-featured CLI experience:
gnugpg - this is pass through though. If it's not installed, the feature is not in use.
fire - python-fire from google does the cli wrapping / forked and included into the repo - I wanted to have the version 0.1.x formatting and help output with few hacks of my own
FAQs
Simple boilerplate for cli scripts
We found that clima demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Research
The Socket Research Team breaks down a malicious wrapper package that uses obfuscation to harvest credentials and exfiltrate sensitive data.
Research
Security News
Attackers used a malicious npm package typosquatting a popular ESLint plugin to steal sensitive data, execute commands, and exploit developer systems.
Security News
The Ultralytics' PyPI Package was compromised four times in one weekend through GitHub Actions cache poisoning and failure to rotate previously compromised API tokens.