Testing packages#
When you are developing a package, you should write tests for it. The tests are automatically executed as soon as the package build and all it's run dependencies are ready.
Writing tests#
You can add one or more tests to your package in the tests
section of the recipe (or output).
Each test is run independently, in a separate environment.
One notable difference are the package_contents
tests that are executed right after the package
is prepared and do not create a new environment (as we only analyze the contents of the package).
tests:
# commands to run to test the package. If any of the commands
# returns with an error code, the test is considered failed.
- script:
- echo "Hello world"
- exit 1 # this will fail the test
# run a script from the recipe folder
- script: sometest.py
# run a Python script with the Python interpreter
- script:
interpreter: python
content: |
import mypkg
assert mypkg.__version__ == "0.1.0"
# execute `pytest` with the tests from the `tests` folder
- script:
- pytest ./tests
# additional requirements at test time
requirements:
run:
- pytest
- python 3.9.* # require an older python version
# extra files to be copied to the test folder from the recipe or source directory
files:
recipe:
- tests/
# python specific tests
- python:
# this test section tries to import the python modules and errors if it can't
imports:
- mypkg
pip_check: true
python_version: [3.9.*, 3.10.*] # run against multiple older python versions
- r:
libraries:
- dplyr
- perl:
modules:
- JSON
# test the contents of the package. If any of the globs does not match _any_ file, the test fails.
- package_contents:
files:
- share/package/*.txt
Testing package contents#
The package_contents
test is a special test that is executed right after the
package is prepared. It does not create a new environment, but instead checks the paths that will be part of the final package.
It can be very useful as a "sanity check" to ensure that the package contains the expected files.
It has multiple sub-keys that help when building cross-platform packages:
files
: a list of globs that should match at least one file in the package. If any of the globs do not match, the test fails.lib
: matches libraries in the package (.so
,.dll
,.dylib
files). The test fails if any of the libraries are not found. It's enough to specify the library name without any extension (e.g.foo
will matchlibfoo.so
,libfoo.dylib
, andfoo.dll
).include
: matches files under theinclude
directory in the package. You can specify the file name likefoo.h
.bin
: matches files under thebin
directory in the package. You can specify executable names likefoo
which will matchfoo.exe
on Windows andfoo
on Linux and macOS.site_packages
: matches files under thesite-packages
directory in the package. You can specify the import path likefoobar.api
which will matchfoobar/api.py
andfoobar/api/__init__.py
.
Testing existing packages#
The tests from the test section are actually added into your package and can also be executed straight from the existing package.
The idea behind adding the tests into the package is that you can execute the
tests independently from building the package. That is also why we are shipping
a test
subcommand that takes as input an existing package and executes the
tests:
Running the above command will extract the package and create a clean environment where the package and dependencies are installed. Then the tests are executed in this newly-created environment.
If you inspect the package contents, you would find the test files under
info/test/*
.
How tests are translated#
The tests
section allows you to define test configurations for your package.
Tests are serialized to info/tests/tests.yaml
in the created package and read from there during test execution.
When adding extra files to your tests:
- During package creation
- Files are copied to
$PREFIX/etc/conda/test-files/{pkg_name}/{idx}
{idx}
is a sequential number assigned to each test- Files can come from both
source
(work directory) andrecipe
locations
- Files are copied to
- During test execution
- Files are copied from
$PREFIX/etc/conda/test-files/{pkg_name}/{idx}
to a temporary directory - Tests run within this temporary directory
- Use relative paths to access these files in your test commands
- Files are copied from
This approach ensures test files are properly packaged and available during test execution.
Legacy tests#
Legacy tests (from conda-build
) are still supported for execution. These tests
are stored as files under the info/test/
folder.
The files are:
run_test.sh
(Unix)run_test.bat
(Windows)run_test.py
(for the Python import tests)test_time_dependencies.json
(for additional dependencies at test time)
Additionally, the info/test/
folder contains all the files specified in the test
section as source_files
and files
. The tests are executed pointing to this
directory as the current working directory.