Pytemplator aims to streamline the creation of dynamic templates. It supports the format from CookieCutter package but also offers the option to generate the context using Python, which in practice provides a better user experience and more flexibility.
- Free software: Apache Software License 2.0
- Documentation: https://arnaudblois.github.io/pytemplator/.
Install the package pytemplator using pip or poetry.
In a shell:
$ pytemplate <target>
Where <target> can be either a local path to the directory of a Pytemplator template or the url to a git repo.
There are options to specify which branch should be used for templating, the output directory and the config directory. More details can be obtained with:
$ pytemplate --help
See this project example to get an idea of an actual pytemplator template.
A typical Pytemplator template project can live either as a local directory or as a Git repo. It relies on three elements: - a templates folder where all folders and files to be templated should be placed. Under the hood, pytemplator relies on jinja2. - an initialize.py at the root level with a function "generate_context". More details below. - a finalize.py whose finalize function is run after the templating.
The generate_context function should return a dictionary mapping the variables in the template to their values. The idea is to use the extra flexibility to offer sensible default values to make the user experience smoother.
generate_context must accept no_input as argument. This tells what should happen in purely programmatic environment. It is up to you how you'd like to address this, you can provide default values if this makes sense and choose not to handle it, in which case a NoInputOptionNotHandledByTemplateError will be raised.
There are several utility classes to help, Context and Question.
The following code illustrates how they can be used:
import datetime
from pytemplator.utils import Question as Q, Context
def generate_context(no_input, *args, **kwargs):
"""Generate context."""
context = Context()
context.questions = [
Q("pypi_name", ask="Name of the package on Pypi"),
Q("module_name", ask=False, default=lambda: context["pypi_name"].replace("-","_").lower()),
Q("year", default=date.today().year, ask=False),
]
context.resolve(no_input)
return context.as_dict()
A Question takes several arguments: - the key that will be put in the context, required. - ask is the prompt displayed to the user, a default inferred from the key is displayed if this is left to None. Set this to False to take the default without asking the user. - default is the value by default. This can be either a value or a callable. The latter allows for lazy evaluation, especially useful to look into the context to use answers from previous questions. - no_input_default is the value used when no_input is True. If None, default is used.
All help is much appreciated and credit is always given. Please consult CONTRIBUTING.rst for details on how to assist me.
This package is inspired from the excellent CookieCutter package and audreyr/cookiecutter-pypackage project template.