Quickly create python projects from templates.
|Author:||Jim Carroll <email@example.com>|
|Description:||Quickly create python projects from templates|
Pyjig quickly creates new python projects using pre-created templates. Projects can be simple scripts, distutils packages or full blown applications. Pyjig can even add python source modules and c extensions to existing projects.
Pyjig is a wrapper around Cookiecutter, which is a command-line utility that creates projects from cookiecutters (project templates). Templates are downloaded from public repos (such as github.com and bitbucket.org). Templates are written in Jinja.
Pyjig uses four public repos hosted with github:
|https://github.com/jamercee/cookiecutter-pyapp||Python application type projects|
|https://github.com/jamercee/cookiecutter-pypkg||Python package type projects|
|https://github.com/jamercee/cookiecutter-pysource||Create python source|
|https://github.com/jamercee/cookiecutter-pyext||Create python extension|
Using python requires developers to employ a full ecosystem of tools. At Carroll-Net, all projects require the following tools; pylint and flake8 for static code analysis, sphinx for project documents and git for revision control. And this requires setting up directories, config files, a unittest infrastructure and a comprehensive Makefile for automating the daily build, test and install tasks.
With all these steps, it’s easy to miss one, or to make a typo when copying from another project which then causes developers to spend time debugging. What was needed was a way to ensure uniform deployment and configuration of our python architecture and toolchain.
Pyjig borrows it’s name from the concept of a jig which is a tool used in metal and woodworking. A jig is a template that allows one to make duplicates of pieces. The simplest example is a key duplication machine, which uses one key as the guide to make copies.
Created Project Layout
Each new project will create the following directories and files:
myproj <-- Project root | .gitignore | id.txt | Makefile | pylint.rc | setup.cfg | setup.py | +---.git <-- Git repository | ... | +---docs <-- Sphinx documentation | conf.py | index.rst | make.bat | Makefile | +---src <-- Project source code | __init__.py | \---tests <-- Unittest infrastructure __init__.py
Each project root folder includes a copy of id.txt. This file is a copy of the cookiecutter settings that were in effect when the project was created. It acts as sentinel for project root identification and should not be removed or renamed.
Each project will have a cutomized Makefile installed in the project’s root directory. It’s syntax is written to support GNU Make. It comes with the following pre-built recipes
|comp||Perform static analysis (default target)|
|tests||Run unittests after in-lace build|
|docs||Generate html documentation|
|dist||Build python software distributions|
|build||Build everything needed to install|
|install||Perform static analysis, run unittests and install to site-packages|
|viewdocs||Rebuild html docs & launch browser|
|clean||Meta-recipe to invoke clean-build, clean-pyc, clean-docs|
|clean-build||Remove all built outputs|
|clean-pyc||Remove python built elements (*.pyc, *.pyo, etc…)|
|debug||Generate Makefile diagnostic output|
|help||Display Makefile help|
Python’s flexible syntax means that coding errors are difficult to detect until runtime. Static analysis tries to solve this by scanning code for coding errors, bugs and bad style. It is an invaluable technique that has saved us untold hours in debugging.
We first started using pylint. Then later on we added a second static analysis tool flake8. Each tool has it’s strengths and we’ve found the combination of both has provided material reduction in time spent debugging.
Each of these two tools requires some tweaking before they will generate useful advice. Pyjig will handle configuring sane defaults for new projects to get them up to speed quickly.
To perform static analysis of code, from within the project’s root folder run make comp.
The Makefile recipe detects changes in *.py with reference to *.pyc. If the *.pyc is missing or older than it’s *.py, a static analysis pass will be done, and if the pass does not generate errors or warnings, the *.pyc will be re-built.
Carroll-Net has adopted Sphinx as our documentation generator for python projects. Sphinx converts reStructuredText into HTML websites. Sphinx can extract documentation from source modules and automatically generate browesable websites.
There are two Makefile recipes related to documentation; make docs which will rebuild documentation and make viewdocs which will rebuild docs and launch a webbrowser to read the rebuilt docs.
Two good references for authoring reST documents are
Carroll-Net has adopted Git as our version control system for software. Git is a fast, reliable distributed revision control system. Originally developed for Linux kernel development it is now the most widely used source code management tool.
Pyjig will initialize a git repository for each new project it creates using your local sytem defaults (see git config ….). And each time you use Pyjig to add to an existing project, pyjig will add the source to the repo.
Pyjig will not create the repo if invoked with --excludegit or if the dirctory is a subdirectory of an existing git repository. It detects repository membership by invoking git status.
Pyjig is hosted on git hub at https://github.com/jamercee/pyjig
Installation using git:
git clone https://github.com/jamercee/pyjig cd pyjig python setup.py install
Pyjig can also be installed with pip:
pip install pyjig
Command line options
usage: pyjig [-?] [-d] [--pkg PKG] [--app APP] [--ext EXT [EXT ...]] [-x] [source [source ..]]
Add one or more source file(s) to project. If the current directry is not part of an existing project, the source file will be created, but no project related activities will be taken (no unittest generation, no sphix-docs generation, not added to git…)
-?, -h, –help Display help and exit.
|-d, --debug||Generate diagnotic output.|
|--pkg PKG||Create a distutils package project.|
|--app APP||Create an application type project.|
–ext EXT [EXT …] Add an extension module(s) to the existing project.
|Do not initialize git repo and do not add new source to git repo.|
In the examples that follow, the --quiet flag is used to accept the default cookiecutter answers (and to keep our example brief). Some of the default answers may not be appropriate for your project until you configure cookiecutter. An example of how todo this is also provided.
New package project with python source files
Typically, the workflow is to create a new project and then to add source files to it. For example, to create a new package called mypkg and to the then add three source files, you would do the following:
$ pyjig --quiet --pkg mypkg $ cd mypkg $ pyjig --quiet s1 s2 s3 $ git status --short A docs/s1.rst A docs/s2.rst A docs/s3.rst A src/s1.py A src/s2.py A src/s3.py A tests/tests1.py A tests/tests2.py A tests/tests3.py
New application project with python source
Application projects are similar to package projects with the main difference being how the setup.py is created. Application projects use the setuptools entry_points attribute to cause the install to create a python command script:
$ pyjig --quiet --app myapp $ cd myapp $ pyjig --quiet mycmd $ git status --short A docs/mycmd.rst A src/mycmd.py A tests/testmycmd.py
Create a single python source file
If you only need to create a python source file without the application or package ecosystem, you can just use the source command. A simple example:
$ pyjig -q source
This will create a single source.py in your current directory.
New project with C++ extension
C++ Extensions enable developers to extend the Python interpreter with new modules. Pyjig comes with support for creating new projects with these extensions. To create a new C++ Extension you would typically do:
$ pyjig --quiet --pkg mymod $ cd mymod $ pyjig --quiet --ext e1 $ git status --short A docs/e1.rst A src/e1_module.cpp A tests/teste1.py
Pyjig will not add the new module to the setup.py file. This is an important step that needs to be done by the developer to cause the module to be rebuilt. The instructions for how todo this are included as a comment at the top of the newly created module file:
$ head -n 15 src/e1_module.cpp /* This module needs to be manually added to your setup.py. Consider adding the following lines: from setuptools import Extension module = Extension('mymod.e1', sources = ['src/e1_module.cpp'], ) setup(... ext_modules = [ module ], ) */
Defining New Types with C++ Extension
Extensions can be used to create new types that can be manipulated from Python code, much like strings and lists in core Python. The pyjig C++ Extension system has specialized sections that can be included to create these types. You cannot create these modules using --quiet as it requies the developer to specify the name of the new type during the cookicutter build step. Here’s an example:
$ pyjig --quiet --pkg mytype $ cd mytype $ pyjig --ext mytype >>> Option: Add extension module 'mytype'. You've cloned C:\cygwin64\home\jimc\.cookiecutters\cookiecutter-pyext before. Is it okay to delete and re-clone it? [yes]: Cloning into 'cookiecutter-pyext'... remote: Counting objects: 29, done. remote: Compressing objects: 100% (21/21), done. remote: Total 29 (delta 13), reused 24 (delta 8), pack-reused 0 Unpacking objects: 100% (29/29), done. Checking connectivity... done. module [mytype]: module_short_description [short module description]: project [mytype]: new_type : Mytype <--- Give your new type a name here version [1.0]: release [1.0.1]: python [python2.7]: author [Jim Carroll]: email [firstname.lastname@example.org]: year : copyright [Copyright(c) 2015, Carroll-Net, Inc., All Rights Reserved]:
In the example above, the fifth question allows the developer to give their new type a name. Any non-default answer will cause additional code to be included in the project to create new custom types.
Remember to add the new module to setup.py