Docs Technical Guide

This document explains the tools and workflows used when contributing to ODK Documentation.


Developer and authoring tools have lots of options and alternatives. Local tools and workflows presented in this guide are what the authors feel would be easiest for a non-coding contributor to set up and use.

You should feel free to use your preferred tools.

Docs as Code

ODK Documentation follows (as much as possible) the Docs as Code philosophy.

This means:

  • Documentation source files are written in a plain text format. (We use reStructuredText.)
  • Documentation source files are kept under version control. (We use git and Github.)
  • Documentation is built from source to a published output using a static site generator. (We use Sphinx.)
  • Documentation builds are run, tested, and deployed automatically using continuous integration tools. (We use CircleCI.)

The 'Docs as Code' approach has many advantages, but we are aware that this approach can feel difficult for writers who aren't used to dealing with the command line. It can also be difficult for coders who are used to this approach, but who typically use simpler authoring tools like Jekyll and Markdown.

This section of the Contributor Guide walks through our authoring and publishing workflow and toolchain, to make it as easy as possible for you to contribute.

Overview of Workflow

When you first get started you'll need to:

  1. Fork the ODK Docs repo your own Github Account
  2. Clone it down to your local machine
  3. Install dependencies

And then each time you work you will:

  1. Make a branch for a specific task
  2. Make commits as you go
  3. Build and view the docs locally, running the style guide tests.
  • Correct any errors and commit those changes
  1. Push your branch to your Github fork
  2. Issue a pull request against the current working branch of the main repo (usually master)
  3. Pull the latest changes to master back to your local machine from the main repo.
  4. Repeat

Setting up your Environment

Terminal (Command Line)


This contributor guide is written from a Linux/Unix (Bash terminal) perspective, which is relevant to all flavors of Linux and macOS. We consider the Bash terminal commands to be the canonical way to build and work with the docs.

We have also provided explanations for how to adapt these commands to the Windows Command Prompt. This is different than Windows Powershell, and not all the commands will work in Powershell. For more details on the Windows Command Prompt, see this article.

If you are on a Windows machine, you may prefer to use the adapted Windows instructions here. Alternatively, you can follow the Bash commands:

The lead maintainers of this docs repo are not Windows users, so we rely on our contributor community to keep Windows-specific information complete and accurate. Contributions to this guide with explanations and help for Windows users are greatly appreciated.

Contributing to the docs requires interacting with git, GitHub, Python, and Sphinx, which requires the use of the Terminal. This is common among Linux users. Mac users unfamiliar with the Terminal can learn more from this tutorial.


Python 3

If you don't know, check to see if you have Python 3 installed:

$ python3

On windows:

> python

If you get an error, you probably don't have it and will need to install Python 3.

On Windows make sure to select the option "Add python to the Path" while installing, otherwise you'll need to add it manually.

See this video for more details.

If the Python command-line interpreter starts up, type quit() to exit.

Virtual Environment

A virtual environment is a Python tool for sandboxing dependencies. It lets you install whatever you need for a particular project, without messing up any other Python environments you might need.

Check to see if you have virtualenv installed:

$ virtualenv

If you get a help message with information about commands, you have it. If you don't have it, you'll get a command not found message.

If you don't have it, install it using pip by running:

$ pip install virtualenv

Then, create a directory called odk. This will contain your virtualenv and the docs repo as subdirectories.

$ mkdir odk
$ cd odk

Now, inside that directory, create a python3 virtualenv.

$ virtualenv -p python3 odkenv

On Windows use:

> path\to\python\python -m venv odkenv

The last part, odkenv, is the name of virtual environment. It can be whatever name you'd like to call it, but we'll use odkenv throughout these docs.

Activate your virtual environment with:

$ source odkenv/bin/activate

On Windows use:

> odkenv\Scripts\activate

When you are done working, deactivate it with:

$ deactivate

Github and git

Git is a distributed version control system. It makes it possible to track changes in files over time, merge changes made by different contributors, compare different versions of the same file, and revert a file to an earlier point.

Git is complicated, but you do not need to understand its advanced features or inner workings to use it.

GitHub is an online service for hosting git repositories. It also provides additional collaboration tools like issue trackers and project boards. Open Data Kit uses GitHub for its public code and documentation projects.

You will need to:

  1. Install git
  2. Make sure that git is installed properly by typing (git) in the terminal or command prompt
    • On Windows: If you get any errors, check if your environment variables are set up correctly. (See this StackOverflow answer for details.)
  3. Get a GitHub account.


We use Git Large File Storage (GLFS) to handle large binary files like images and videos. Once installed, you normally won't need to do anything else. GLFS is largely transparent when using git.

  1. Install GLFS.


On Windows

Make sure git-lfs.exe and git.exe are under the same main directory on Windows. (See this page for reference.

GLFS tracks binary files as defined in the .gitattributes file in the repo. Most common binary file formats are already listed, but there might be others we haven't thought of yet.

Adding new file formats to GLFS tracking

If you are adding binary files to the repo, and they are in formats not already tracked, it is your responsibility to make sure they are tracked by GLFS.

To make sure they are properly tracked, add them to the .gitattributes file.

# file type section heading
*.{extension-to-track} filter=lfs diff=lfs merge=lfs -text

You can also use the command line.

$ glfs track *.{file-extension}

This will add a line to .gitattributes.


Please keep .gitattributes organized by placing the new file format declaration in the appropriate section, or creating a new section as needed.


Updates to .gitattributes must be done in a commit before the commit that adds the new binary files.

We will not accept Pull Requests that include binary files untracked by GLFS.

Android Tools

Some testing and documentation tasks (including making screenshots from ODK Collect) require the Android Debug Bridge command line tool. You can either install Android Studio or install ADB as standalone SDK tool.

Android Studio

ADB is part of Android Studio, and is typically installed by default when you install Android Studio.

This is the best way to get adb if you plan to do any other Android development. To use it from the command line, add the SDK Platform tools to your path.

On Mac, add the following to your .bash_profile

export PATH=$PATH:~/Library/Android/sdk/tools/


On Windows, you have to run Android Studio once to complete the installation of ADB. The tool can be found in C:/Users/user-name/AppData/Local/Android/sdk/platform-tools. To add it to the environment variable path, use the following command:

set PATH=%PATH%;C:\Users\your user name\AppData\Local\Android\sdk\platform-tools


The path specified above assumes a default installation of Android Studio. You may have put Android Studio in a different location.

Getting ready to work

Fork the docs

Go to the ODK Doc repo on GitHub and use the Fork button (top right) to create your own copy. After the process completes, you'll be looking at your own fork on GitHub.

Clone to local

From your own fork of the repo on GitHub, select the Clone or download button. Copy the URI from the text box that opens up. It will be something like:

Open your terminal, and cd to your preferred directory. Then git clone the repo:

$ git clone
$ cd docs

The rest of the documentation assumes you are in the directory for the repo (the directory containing and index.rst).


  • The clone command creates a new directory inside the current one. So you do not need to create a new odk-docs directory first.
  • As noted above, we recommend a master odk directory that holds your virtualenv directory and your git repo in two separate subdirectories. So you would be in that master odk directory when you clone down the repo.
  • Double check that the right folders are in the right places
- odk/
  - odkenv/
  - docs/

Set the upstream remote

When you clone down a repo, the local copy calls your GitHub copy origin. You should also set upstream as the name of the main ODK Docs GitHub repo.

$ git remote add upstream

Or in Windows:

> git remote add upstream

Run git remote -v to check the status, you should see something like this:

$ origin (fetch)
$ origin (push)
$ upstream (fetch)
$ upstream (push)

Install Dependencies

The first time you clone down the repo, you need to install the dependencies. Make sure you have your Python 3 virtual environment set up and activated, then:

$ pip install -r requirements.txt


If you are working on the design, testing, or deployment of the docs, you might find the need to install an additional PyPi package. If you do, please update the requirements.txt file with pip freeze > requirements.txt. Pull Requests which change requirements.txt should include a note about why the new packages are needed.


If you have problems when running the Sphinx commands (see below), you may have a dependency issue. Try running pip install -r requirements.txt again.

Workflow details

Pull in changes from upstream

As other people make changes to the docs, you need to keep your local copy up to date.

You probably won't need to do this the first time, but you should always pull in any changes from the main repository before working.

$ git pull upstream


If you get this message:

You asked to pull from the remote 'upstream', but did not specify a branch.
Because this is not the default configured remote for your current branch,
you must specify a branch on the command line.

Try running git pull upstream master instead.

Make a New Branch

Choose a specific, deliverable task to work on. This should be an active issue from our issue tracker on GitHub.

Create a new branch in which you will work on this specific issue. The branch name should briefly describe what you are doing. For example, the original author of this contributor guide worked in a branch called contributing.

Also, make sure that all the branches are derived from master, to avoid mixing up work from different issues commits.

$ git checkout -b branch-name


Branch names should be short, lowercase, and use hyphens for separators.

Good branch names:

  • getting-started-guide
  • contributing
  • fix-issue-13

Bad branch names:

  • getting started guide
  • Getting started guide
  • Getting_started_guide
  • writing-the-getting-started-guide-adammichaelwood-july-2017-draft

Work on the Docs

Write and edit files in your favorite editor.


To work on ODK Docs, you need to work in a code editor.

If you've never used a code editor before, you should know that they are a little different than other writing environments like MS Word or your email editor.

People have strong opinions about code editors, and nearly everyone who uses them regularly has a favorite.

If you're new to using an editor, you might want to try Atom or Sublime, which are both popular and easy to use, and they both have decent support for reStructuredText syntax.

Spell check your work

After making changes to the docs, you need to run the spell checker. To run the spell checker:

$ sphinx-build -b spelling src build/spelling

If there are any warnings, make sure that you fix them to avoid build failure.

The error messages will be displayed on terminal as well as stored in a file /build/spelling/output.txt.

If you find a word which is not misspelled and will have repeated use in docs, add it to the spelling list in the file /src/spelling_wordlist.txt.

If you find a word which is not misspelled and is only required in a particular file, use the spelling directive to list the words before the file content.

.. spelling::


Upgrading Aggregate

Style-test your work

After making changes to the docs, you need to run the style-guide checks. To run the style guide checks on all the docs:

$ python

To run the style guide checks on specified files:

$ python filename1.rst filename2.rst ...

If you want to run the style guide checks on the files you modified, use the option -d or --diff.

$ python -d


Using this option, only checks the files that have been edited since the last git commit. So, if you modify the files, make sure to check them before making a commit.

To run the tests on modified files and some other specified files:

$ python -d filename1.rst filename2.rst ...

The output will consist of a list of warnings and errors. Make sure to go through the warnings and eliminate the ones which violate the style guide rules.


It is not necessary to fix all the warnings but you should go through each warning and decide if a change makes sense.

If there are any errors, you need to fix them to avoid build failure. You can manually fix the errors or use the option -f or --fix.

$ python -f

If you want to fix the errors in some specified files:

$ python -f filename1.rst filename2.rst ...


After automatic fixing of errors, make sure to go through the changes made.

To ignore any part of a file from being checked, you can enclose it in comments:

.. startignore

Some text which is to be ignored while testing.

.. endignore

You can also generate .csv format output using the -o or --out_path with the output filename:

$ python -o output.csv

To generate output for a some specified files:

$ python -o output.csv filename1.rst filename2.rst ...


Make sure to specify the output file before the input files when you use the option -o.

When you run the style guide checks, style testing scripts are generated before the test starts and are removed automatically after the testing is complete. If you want to keep these scripts for debugging, use the option -s or --store:

$ python -s

Use the option -h or --help to get any help about style guide testing:

$ python -h

Build, View, and Debug

To build the documentation into a viewable website:

$ sphinx-build -b dirhtml src build

This calls the sphinx-build utility. The -b switch specifies the builder, which in this case is html. src refers to the src directory which contains all .rst files (the build source) and build refers to the target of the build (the built files will be put into a directory labeled build).

When you run the build, you may see error or warning messages. These indicate potential problems with the documentation, like:

  • syntax errors
  • broken links
  • terms not included in the glossary

Error and warning messages include a file name and line number for tracking them down. Try to resolve all your errors and warnings before issuing a pull request. If this is not possible, please add a note in your pull request so that we can help you debug the problem.

We will not merge Pull Requests that have warnings or errors in them.


Because of a bug in Sphinx the line numbers in error and warning messages will be off by the length of rst_prolog in

To view the documentation in your web browser, you can use Python's built-in web server.

$ cd build
$ python -m http.server 8000

Then open your browser and go to http://localhost:8000

Read through your doc edits in the browser and correct any issues in your source files. You'll need to shut down the web server (CTRL C) before rebuilding, then return to the main directory of the repo ( cd .. ).

It's a good idea to delete the build directory before each rebuild.

$ rm -rf build
$ sphinx-build -b dirhtml . build


The script automatically runs all the build commands. It saves typing. In the future, it will also become the canonical build script for ODK Docs, including additional tests and other build tasks.

Push Your Branch

Once your work on the issue is completed, add the files you've changed or created, and write a relevant commit message describing the changes.

$ git add my_changed_files
$ git commit -m "A small but relevant commit message"

Then, push the changes. The first time you do this on any branch, you'll need to specify the branch name:

$ git push origin branch-name

After that, you only need to use the push command:

$ git push


origin is the local label for your GitHub fork.

Issue a Pull Request

A pull request (or PR) is a request from you to the ODK Docs maintainers, for us to pull in your changes to the main repo.

Go the main docs repo on GitHub. You'll see a message there referencing your recently pushed branches. Select Compare & pull request to start a pull request.

Follow GitHub's instructions. The Base fork should be the main repo, and base should be master. Your repo and working fork should be listed beside them. (This should all populate by default, but be sure to double check.) If there is a green Able to be merged message, you can proceed.

You must include a PR comment. Things to include:

  • A summary of what you did.
  • A note about anything that probably should have been done, but you didn't do.
  • A note about any new work this PR will create.
  • The issue number you are working on. If the PR completes the issue, include the text Closes # and the issue number.
  • A note about any errors or warnings, and why you did not or could not resolve them.
  • A note justifying any changes to requirements.txt.
  • A note about any difficulties, questions, or concerns that came up while working on this issue.

Complete the pull request. The maintainers will review it as quickly as possible. If there are any problems the maintainers can't deal with, they will reach out to you.


If you happen to rename any document file (*.rst), then be sure that you add the redirect in your PR.

To add the redirect go to s3_website.yml, and add a mapping from the old file name to the new file name below the redirects: line, one mapping per line.

If you have renamed old-name.rst to new-name.rst:

 old-name/index.html: new-name

Notice the inclusion of /index.html on the left side.

Keep Going

Once the PR is merged, you'll need to pull in the changes from the main repo ( upstream ) into your local copy.

$ git checkout master
$ git pull upstream master

Then you should push those change to your copy on GitHub ( origin ).

$ git push

If you want to delete your branch from before, you can do that:

$ git branch -d branch-name

Now you can find a new issue to work on, create a new branch, and get to work again.