py-qs-example 0.2.11

Python package PyPi Sphinx quickstart example app

pypi-sphinx-quickstart

Overview

This repo is a template to use for starting a new Python package which is hosted on PyPi and uses Sphinx for documentation hosted on Github pages. It has a built-in CI/CD system using Github Actions, all you need to do for that is hook up the secrets. The CI system has the following features:

• Runs any tests in tests with pytest
• Lints code using flake8
• Static code checks with mypy
• Deploys PyPI package
• Deploys Sphinx documentation on Github Pages
  • Autodoc/autosummary already set up
  • Automatic sitemap.xml generated
  • Just add Google Analytics ID to enable tracking
  • Read the Docs Theme with Custom CSS
  • Notebook-style examples with Sphinx Gallery complete with download and Binder links
    • Auto-converts Jupyter notebooks in nbexamples
• Auto-merges pull requests by maintainers
• Auto-drafts release notes based on merged pull requests
• Syncs new changes from the cookiecutter template using cruft on a cron workflow. Creates a PR with the changes to be merged manually. Creates an issue if it is not possible to commit the changes (when there are changes to workflow files).
• Collects TODO comments and converts them into issues (optional)
• Closes TODO issues once comments are removed (optional)

Getting Started

Required Steps

Follow the steps in one of the two ways to start, then the following steps.

Two Ways to Start

Cookiecutter

This template is available as a cookiecutter. Find the cookiecutter for this template here. This is the recommended way to get started if you are already familiar with how everything works. No example files will be included.

Repo Template

Click the "Use this template" button at the top of the repo page, then fill out the name and description your new repo. This is the recommended way to get started if you are figuring out how everything works as it gives you a fully working demo with example files.

Adding Secrets

Go into the repo settings, under Secrets, and add the following secrets:

• pypi_password: Personal token for PyPI
• gh_token: Github personal access token
• CODECOV_TOKEN (optional): codecov.io token for this project
• TODO_ACTIONS_MONGO_URL (optional): MongoDB connection url, complete with username and password. See Setup MongoDB.

conf.py

Edit conf.py in the main repo directory. This contains the main settings for the PyPi package. If you used the repo template, you will need to change all the settings. Otherwise, the cookiecutter will have filled out most of the settings, but specific package details such as dependencies still need to be added.

Adding Project Source

Delete the folder py_qs_example, and add your own package with the name you set in conf.PACKAGE_NAME.

Adding Global Requirements to Build

If you do not already have pipenv installed, you will need to run:

pip install pipenv

Then regardless of whether you already had pipenv installed, you will need to navigate to the repo folder and run:

pipenv update

Setting up Documentation

If you used the repo template, edit docsrc/Makefile to change SPHINXPROJ to set it to the name you set in conf.PACKAGE_NAME (with cookiecutter this step is not necessary.)

Edit docsrc/source/index.rst to remove the example included files. Replace with your own if you wish or entirely delete the My Module and My Package sections if don't wish to use the autosummary directive.

Edit docsrc/source/tutorial.rst to put your own tutorial, or remove it and remove it from the toctree directive in docsrc/source/index.rst.

You may further modify Sphinx configuration in docsrc/source/conf.py if you wish.

Add Sphinx Gallery examples in the examples folder. You can also add Jupyter notebook examples in the nbexamples folder, and they will automatically be converted to Sphinx Gallery-style examples and included with examples in the build of the documentation.

Adding Labels

The following labels are used in the CI/CD. They should be added in Labels in the repo settings:

• no auto merge: added to prevent automatic merging of pull requests by maintainers
• maintenance: one of the output categories for release notes
• automated pr: Used by automated template update cron workflow which uses cruft to check for changes in the template and opens a PR automatically if so.
• automated issue: Due to limitations in Github Actions, the template update cron workflow is not able to commit to the repo if the changes include changes to workflow files. It instead raises an issue to update the template in this case. This label is applied to these issues as well as the maintenance label.

Commit and Push

After the preceding steps, now commit your changes and push to master if not done already. After a few minutes, Github Actions should create a gh-pages branch which has the documentation HTML in it.

Github Pages Setup

Note: This should happen automatically after Github Actions creates the gh-pages branch. But follow these steps if your docs still do not work.

Go to repo settings, Github Pages section. For the Source dropdown, select "gh-pages branch". The settings page should reload, and in the Github Pages section it should show the URL of your documentation. You should be able to see the documentation at the URL after a few seconds, but it will still be the example documentation.

If "gh-pages branch" is not shown in the dropdown, you need to make one release commit and push it, so that the gh-pages branch will be added to your repo. After doing that, you can go into the repo settings and select "gh-pages branch" as described.

Optional Steps

Set Master to Protected Branch

It is recommended to make master a protected branch so that nobody can delete it.

Setup Codecov

Go to codecov.io, log in via Github, click Repositories then "Add new repository" and select this repository from the list. Copy the token for Codecov to use in the next step.

Setup MongoDB (optional, for TODO integration)

For the TODO integration to work, you need a MongoDB instance. You can get one for free at mlab.com. After creating the database, create a database user. The MLab interface will show you the format of the connection url string, which you will fill in the database user's username and password and use that as the TODO_ACTIONS_MONGO_URL secret, as the Adding Secrets section shows.

Built-in CI/CD

On Every Push

Github Actions are used to automatically run the following steps on every push:

• Check Python syntax with flake8
• Run pytest
• Static typing checks with mypy

When Branch is master

If the branch is the master branch, then it will also:

• Upload pytest results to codecov

If there is a change in docsrc

If the branch is the master branch, and there was a change in docsrc, it will do all the steps in On Every Push and When Branch is master, then it will:

• Build documentation HTML using Sphinx
• Create gh-pages branch and copy HTML there
• Push to gh-pages branch, which will update the hosted documentation

If there is a change in the package version

If the branch is the master branch, and there was a change in the package version in conf.py, it will do all the steps in On Every Push and When Branch is master, then it will:

• Build documentation HTML using Sphinx
• Create gh-pages branch and copy HTML there
• Push to gh-pages branch, which will update the hosted documentation
• Build Python package
• Upload Python package to PyPI

If a Pull Request is Opened

The CI/CD system will check whether the pull request was opened by a maintainer (configured in conf.py). If so, it will auto-merge the pull request after it has passed CI checks. It will then run the deployment pipeline. To avoid this auto-merge behavior, add the label "no auto merge" to the pull request.

Regular Usage

Once everything is set up, just commit your changes. The built-in CI/CD will take care of testing, build, and deployment of PyPI package and documentation. If you use pull requests on Github then it will show you whether it passes the CI tests.

Releases

If you want to create Github releases notes for your package, they will already be drafted, just edit them as desired before posting.

Local Usage

Building the documentation locally makes sense if you are updating it but don't want to make it live yet. You can view the HTML files in the docs folder via a browser after building them.

Building Documentation

Navigate into the docsrc folder and run:

pipenv run make github

This should generate documentation HTML in the docs folder.

Uploading to PyPi

Navigate to the repo base folder and run:

pipenv run python upload.py

Updating Build Requirements

The Github Actions CI/CD uses Pipfile.lock to install its requirements. Run pipenv update locally to update the Pipfile.lock with the newest dependencies and push into the master branch to get the dependencies updated on the CI/CD system.

Syncing with the cookiecutter template

There is a built-in workflow which runs daily to check for updates in the cookiecutter template. If it finds an update, it will use cruft to apply the update and raise a PR with the changes. Manually review the changes, adjusting if needed, then merge the PR to keep updated with the template.

Links

See the example generated documentation here.