Measurement contexts

Added in version 5.0. measures whether code was run, but it can also record the context in which it was run. This can provide more information to help you understand the behavior of your tests.

There are two kinds of context: static and dynamic. Static contexts are fixed for an entire run, and are set explicitly with an option. Dynamic contexts change over the course of a single run.

Static contexts

A static context is set by an option when you run The value is fixed for the duration of a run. They can be any text you like, for example, “python3” or “with_numpy”. The context is recorded with the data.

When you combine multiple data files together, they can have differing contexts. All of the information is retained, so that the different contexts are correctly recorded in the combined file.

A static context is specified with the --context=CONTEXT option to the coverage run command, or the [run] context setting in the configuration file.

Dynamic contexts

Dynamic contexts are found during execution. They are most commonly used to answer the question “what test ran this line?,” but have been generalized to allow any kind of context tracking. As execution proceeds, the dynamic context changes to record the context of execution. Separate data is recorded for each context, so that it can be analyzed later.

There are three ways to enable dynamic contexts:

  • you can set the [run] dynamic_context option in your .coveragerc file, or

  • you can enable a dynamic context switcher plugin, or

  • another tool (such as a test runner) can call the Coverage.switch_context() method to set the context explicitly. The pytest plugin pytest-cov has a --cov-context option that uses this to set the dynamic context for each test.

The [run] dynamic_context setting has only one option now. Set it to test_function to start a new dynamic context for every test function:

  • .coveragerc
  • pyproject.toml
  • setup.cfg, tox.ini
dynamic_context = test_function
dynamic_context = "test_function"
dynamic_context = test_function

Each test function you run will be considered a separate dynamic context, and coverage data will be segregated for each. A test function is any function whose name starts with “test”.

If you have both a static context and a dynamic context, they are joined with a pipe symbol to be recorded as a single string.

Initially, when your program starts running, the dynamic context is an empty string. Any code measured before a dynamic context is set will be recorded in this empty context. For example, if you are recording test names as contexts, then the code run by the test runner before (and between) tests will be in the empty context.

Dynamic contexts can be explicitly disabled by setting dynamic_context to none.

Context reporting

The coverage report and coverage html commands both accept --contexts option, a comma-separated list of regular expressions. The report will be limited to the contexts that match one of those patterns.

The coverage html command also has --show-contexts. If set, the HTML report will include an annotation on each covered line indicating the number of contexts that executed the line. Clicking the annotation displays a list of the contexts.

Raw data

For more advanced reporting or analysis, the .coverage data file is a SQLite database. See database schema for details.