This is a community driven project and everyone is welcome to contribute.
The project is hosted at the PyGMT GitHub repository.
The goal is to maintain a diverse community that’s pleasant for everyone. Please be considerate and respectful of others. Everyone must abide by our Code of Conduct and we encourage all to read it carefully.
Ways to Contribute¶
Tackle any issue that you wish! Some issues are labeled as “good first issues” to indicate that they are beginner friendly, meaning that they don’t require extensive knowledge of the project.
Make a tutorial or gallery example of how to do something.
Provide feedback about how we can improve the project or about your particular use case.
Contribute code you already have. It doesn’t need to be perfect! We will help you clean things up, test it, etc.
Discussion often happens on GitHub issues and pull requests. In addition, there is a Discourse forum for the project where you can ask questions.
Reporting a Bug¶
Find the Issues tab on the top of the GitHub repository and click New Issue. You’ll be prompted to choose between different types of issue, like bug reports and feature requests. Choose the one that best matches your need. The Issue will be populated with one of our templates. Please try to fillout the template with as much detail as you can. Remember: the more information we have, the easier it will be for us to solve your problem.
Editing the Documentation¶
If you’re browsing the documentation and notice a typo or something that could be improved, please consider letting us know by creating an issue or submitting a fix.
You can submit fixes to the documentation pages completely online without having to download and install anything:
On each documentation page, there should be an “Improve This Page” link at the very top.
Click on that link to open the respective source file (usually an
.rstfile in the
doc/folder or a
.pyfile in the
examples/folder) on GitHub for editing online (you’ll need a GitHub account).
Make your desired changes.
When you’re done, scroll to the bottom of the page.
Fill out the two fields under “Commit changes”: the first is a short title describing your fixes; the second is a more detailed description of the changes. Try to be as detailed as possible and describe why you changed something.
Choose “Create a new branch for this commit and start a pull request” and click on the “Propose changes” button to open a pull request.
The pull request will run the GMT automated tests and make a preview deployment. You can see how your change looks in the PyGMT documentation by clicking the “View deployment” button after the Vercel bot has finished (usually 5-10 minutes after the pull request was created).
We’ll review your pull request, recommend changes if necessary, and then merge them in if everything is OK.
Alternatively, you can make the changes offline to the files in the
doc folder or the
example scripts. See Contributing Code for instructions.
The gallery and tutorials are managed by
The source files for the example gallery are
.py scripts in
generate one or more figures. They are executed automatically by sphinx-gallery when
the documentation is built. The output is gathered and assembled
into the gallery.
You can add a new plot by placing a new
.py file in one of the folders inside the
examples/gallery folder of the repository. See the other examples to get an idea for the
General guidelines for making a good gallery plot:
Examples should highlight a single feature/command. Good: how to add a label to a colorbar. Bad: how to add a label to the colorbar and use two different CPTs and use subplots.
Try to make the example as simple as possible. Good: use only commands that are required to show the feature you want to highlight. Bad: use advanced/complex Python features to make the code smaller.
Use a sample dataset from
pygmt.datasetsif you need to plot data. If a suitable dataset isn’t available, open an issue requesting one and we’ll work together to add it.
Add comments to explain things are aren’t obvious from reading the code. Good: Use a Mercator projection and make the plot 15 centimeters wide. Bad: Draw coastlines and plot the data.
Describe the feature that you’re showcasing and link to other relevant parts of the documentation.
SI units should be used in the example code for gallery plots.
The tutorials (the User Guide in the docs) are also built by sphinx-gallery from the
.py files in the
examples/tutorials folder of the repository. To add a new tutorial:
.pyfile in the
examples/tutorialsfolder on the base of the repository.
Write the tutorial in “notebook” style with code mixed with paragraphs explaining what is being done. See the other tutorials for the format.
Include the tutorial in the table of contents of the documentation (side bar). Do this by adding a line to the User Guide
doc/index.rst. Notice that the file included is the
.rstgenerated by sphinx-gallery.
Choose the most representative figure as the thumbnail figure by adding a comment line
# sphinx_gallery_thumbnail_number = <fig_number>to any place (usually at the top) in the tutorial. The fig_number starts from 1.
Guidelines for a good tutorial:
Each tutorial should focus on a particular set of tasks that a user might want to accomplish: plotting grids, interpolation, configuring the frame, projections, etc.
The tutorial code should be as simple as possible. Avoid using advanced/complex Python features or abbreviations.
Explain the options and features in as much detail as possible. The gallery has concise examples while the tutorials are detailed and full of text.
SI units should be used in the example code for tutorial plots.
Note that the
Figure.show() function needs to be called for a plot to be inserted into
Example code standards¶
When editing documentation, use the following standards to demonstrate the example code:
Python arguments, such as import statements, Boolean expressions, and function arguments should be wrapped as
codeby using `` on both sides of the code. Examples: ``import pygmt`` results in
import pygmt, ``True`` results in
True, ``style=”v”`` results in
Literal GMT arguments should be bold by wrapping the arguments with ** (two asterisks) on both sides. The argument description should be in italicized with * (single asterisk) on both sides. Examples:
**+l**\ *label*results in +llabel,
**05m**results in 05m.
Optional arguments are wrapped with [ ] (square brackets).
Arguments that are mutually exclusive are separated with a | (bar) to denote “or”.
Default arguments for parameters and configuration settings are wrapped with [ ] (square brackers) with the prefix “Default is”. Example: [Default is p].
Is this your first contribution? Please take a look at these resources to learn about git and pull requests (don’t hesitate to ask questions):
Aaron Meurer’s tutorial on the git workflow
We follow the git pull request workflow to make changes to our codebase. Every change made goes through a pull request, even our own, so that our continuous integration services have a chance to check that the code is up to standards and passes all our tests. This way, the master branch is always stable.
General guidelines for making a Pull Request (PR):
What should be included in a PR
Have a quick look at the titles of all the existing issues first. If there is already an issue that matches your PR, leave a comment there to let us know what you plan to do. Otherwise, open an issue describing what you want to do.
Each pull request should consist of a small and logical collection of changes; larger changes should be broken down into smaller parts and integrated separately.
Bug fixes should be submitted in separate PRs.
How to write and submit a PR
Use underscores for all Python (*.py) files as per PEP8, not hyphens. Directory names should also use underscores instead of hyphens.
Describe what your PR changes and why this is a good thing. Be as specific as you can. The PR description is how we keep track of the changes made to the project over time.
Do not commit changes to files that are irrelevant to your feature or bugfix (e.g.:
.gitignore, IDE project files, etc).
Write descriptive commit messages. Chris Beams has written a guide on how to write good commit messages.
Be willing to accept criticism and work on improving your code; we don’t want to break other users’ code, so care must be taken not to introduce bugs.
Be aware that the pull request review process is not immediate, and is generally proportional to the size of the pull request.
After you’ve submitted a pull request, you should expect to hear at least a comment within a couple of days. We may suggest some changes, improvements or alternative implementation details.
To increase the chances of getting your pull request accepted quickly, try to:
Submit a friendly PR
Write a good and detailed description of what the PR does.
Write some documentation for your code (docstrings) and leave comments explaining the reason behind non-obvious things.
Have a good coding style
Pull requests will automatically have tests run by GitHub Actions. This includes running both the unit tests as well as code linters. GitHub will show the status of these checks on the pull request. Try to get them all passing (green). If you have any trouble, leave a comment in the PR or get in touch.
Setting up your environment¶
We highly recommend using Anaconda and the
package manager to install and manage your Python packages.
It will make your life a lot easier!
The repository includes a conda environment file
environment.yml with the
specification for all development requirements to build and test the project.
Once you have forked and cloned the repository to your local machine, you can
use this file to create an isolated environment on which you can work.
Run the following on the base of the repository:
conda env create
Before building and testing the project, you have to activate the environment:
conda activate pygmt
You’ll need to do this every time you start a new terminal.
file for the list of dependencies and the environment name.
We have a
that provides commands for installing, running the tests and coverage analysis,
running linters, etc. If you don’t want to use
make, open the
copy the commands you want to run.
To install the current source code into your testing environment, run:
This installs your project in editable mode, meaning that changes made to the source code will be available when you import the package (even if you’re on a different directory).
We use some tools:
to format the code so we don’t have to think about it. Black and blackdoc loosely follows the PEP8 guide but with a few differences. Regardless, you won’t have to worry about formatting the code yourself. Before committing, run it to automatically format your code:
For consistency, we also use UNIX-style line endings (
\n) and file permission
-rw-r--r--) throughout the whole project.
Don’t worry if you forget to do it. Our continuous integration systems will
warn us and you can make a new commit with the formatted code.
Even better, you can just write
/format in the first line of any comment in a
Pull Request to lint the code automatically.
When wrapping a new alias, use an underscore to separate words bridged by vowels
(aeiou), such as
z_only. Do not use an underscore to separate
words bridged only by consonants, such as
convention is not applied by the code checking tools, but the PyGMT maintainers
will comment on any pull requests as needed.
make check # Runs black, blackdoc, docformatter, flake8 and isort (in check mode) make lint # Runs pylint, which is a bit slower
All docstrings should follow the numpy style guide. All functions/classes/methods should have docstrings with a full description of all arguments and return values.
While the maximum line length for code is automatically set by Black, docstrings must be formatted manually. To play nicely with Jupyter and IPython, keep docstrings limited to 79 characters per line.
Testing your code¶
Automated testing helps ensure that our code is as free of bugs as it can be. It also lets us know immediately if a change we make breaks any other part of the code.
All of our test code and data are stored in the
We use the pytest framework to run the test suite.
Please write tests for your code so that we can be sure that it won’t break any of the existing functionality. Tests also help us be confident that we won’t break your code in the future.
When writing tests, don’t test everything that the GMT function already tests, such as
the every unique combination arguments. An exception to this would be the most popular
modules, such as
basemap. The highest priority for tests should be the
Python-specific code, such as numpy, pandas, and xarray objects and the virtualfile
If you’re new to testing, see existing test files for examples of things to do. Don’t let the tests keep you from submitting your contribution! If you’re not sure how to do this or are having trouble, submit your pull request anyway. We will help you create the tests and sort out any kind of problem during code review.
Pull the baseline images, run the tests, and calculate test coverage using:
dvc status # should report any files 'not_in_cache' dvc pull # pull down files from DVC remote cache (fetch + checkout) make test
The coverage report will let you know which lines of code are touched by the tests.
If all the tests pass, you can view the coverage reports by opening
in your browser. Strive to get 100% coverage for the lines you changed.
It’s OK if you can’t or don’t know how to test something.
Leave a comment in the PR and we’ll help you out.
You can also run tests in just one test script using:
or run tests which contain names that match a specific keyword expression:
pytest -k KEYWORD pygmt/tests
Writing an image-based test is only slightly more difficult than a simple test.
The main consideration is that you must specify the “baseline” or reference
image, and compare it with a “generated” or test image. This is handled using
the decorator functions
whose usage are further described below.
This is the preferred way to test plots whenever possible.
This method uses the pytest-mpl
plug-in to test plot generating code.
Every time the tests are run,
pytest-mpl compares the generated plots with known
correct ones stored in
If your test created a
pygmt.Figure object, you can test it by adding a decorator and
@pytest.mark.mpl_image_compare def test_my_plotting_case(): "Test that my plotting function works" fig = Figure() fig.basemap(region=[0, 360, -90, 90], projection='W7i', frame=True) return fig
Your test function must return the
pygmt.Figure object and you can only
test one figure per function.
Before you can run your test, you’ll need to generate a baseline (a correct version) of your plot. Run the following from the repository root:
pytest --mpl-generate-path=baseline pygmt/tests/NAME_OF_TEST_FILE.py
This will create a
baseline folder with all the plots generated in your test
Visually inspect the one corresponding to your test function.
If it’s correct, copy it (and only it) to
When you run
make test the next time, your test should be executed and
Don’t forget to commit the baseline image as well!
The images should be pushed up into a remote repository using
dvc (instead of
git) as will be explained in the next section.
As the baseline images are quite large blob files that can change often (e.g.
with new GMT versions), it is not ideal to store them in
git (which is meant
for tracking plain text files). Instead, we will use
which is like
git but for data. What
dvc does is to store the hash (md5sum)
of a file. For example, given an image file like
test_logo.png.dvc plain text file containing the hash of the
test_logo.png.dvc file can be stored as usual on GitHub, while
test_logo.png file can be stored separately on our
dvc remote at
To pull or sync files from the
dvc remote to your local repository, use
the commands below. Note how
dvc commands are very similar to
dvc status # should report any files 'not_in_cache' dvc pull # pull down files from DVC remote cache (fetch + checkout)
Once the sync/download is complete, you should notice two things. There will be
images stored in the
pygmt/tests/baseline folder (e.g.
these images are technically reflinks/symlinks/copies of the files under the
.dvc/cache folder. You can now run the image comparison test suite as per
pytest pygmt/tests/test_logo.py # run only one test make test # run the entire test suite
To push or sync changes from your local repository up to the
at DAGsHub, you will first need to set up authentication using the commands
below. This only needs to be done once, i.e. the first time you contribute a
test image to the PyGMT project.
dvc remote modify upstream --local auth basic dvc remote modify upstream --local user "$DAGSHUB_USER" dvc remote modify upstream --local password "$DAGSHUB_PASS"
The configuration will be stored inside your
.dvc/config.local file. Note
that the $DAGSHUB_PASS token can be generated at
https://dagshub.com/user/settings/tokens after creating a DAGsHub account
(can be linked to your GitHub account). Once you have an account set up, please
ask one of the PyGMT maintainers to add you as a collaborator at
proceeding with the next steps.
The entire workflow for generating or modifying baseline test images can be summarized as follows:
# Sync with both git and dvc remotes git pull dvc pull # Generate new baseline images pytest --mpl-generate-path=baseline pygmt/tests/test_logo.py mv baseline/*.png pygmt/tests/baseline/ # Generate hash for baseline image and stage the *.dvc file in git git rm -r --cached 'pygmt/tests/baseline/test_logo.png' # only run if migrating existing image from git to dvc dvc status # check which files need to be added to dvc dvc add pygmt/tests/baseline/test_logo.png git add pygmt/tests/baseline/test_logo.png.dvc # Commit changes and push to both the git and dvc remotes git commit -m "Add test_logo.png into DVC" git push dvc push
This approach draws the same figure using two different methods (the reference
method and the tested method), and checks that both of them are the same.
It takes two
pygmt.Figure objects (‘fig_ref’ and ‘fig_test’), generates a png
image, and checks for the Root Mean Square (RMS) error between the two.
Here’s an example:
@check_figures_equal() def test_my_plotting_case(): "Test that my plotting function works" fig_ref, fig_test = Figure(), Figure() fig_ref.grdimage("@earth_relief_01d_g", projection="W120/15c", cmap="geo") fig_test.grdimage(grid, projection="W120/15c", cmap="geo") return fig_ref, fig_test
Most documentation sources are in Python
*.py files under the
folder, and the code docstrings can be found e.g. under the
pygmt/datasets/ folders. The documentation are written in
built by Sphinx. Please refer to
if you are new to reStructuredText.
Building the documentation¶
To build the HTML files from sources:
cd doc make all
This will build the HTML files in
doc/_build/html/index.html in your browser to view the pages.
Cross-referencing with Sphinx¶
The API reference is manually assembled in
The autodoc sphinx extension will automatically create pages for each
function/class/module listed there.
You can reference functions, classes, methods, and modules from anywhere (including docstrings) using:
An example would be to use
:meth:`pygmt.Figure.grdview` to link
PyGMT documentation that is not a class, method,
or module can be linked with
:doc:`Any Link Text </path/to/the/file>`.
:doc:`Install instructions </install>` links
Linking to the GMT documentation and GMT configuration parameters can be done using:
An example would be using
:gmt-docs:`makecpt.html` to link to
For GMT configuration parameters, an example is
:gmt-term:`COLOR_FOREGROUND` to link to
Sphinx will create a link to the automatically generated page for that function/class/module.