Configuration

Principles

For this project's multi-stage workflow, some variables remain constant across multiple commands. For example, the paths necessary to differentiate between standard project types, standard library types and types from third-party dependencies in the virtualenv are needed for both the tracing and unification processes.

All such variables are stored in a pytypes.toml file in the root of the project to trace, which will be read in by the program without the user having to respecify them.

Example

The following configuration is used for a project called PyTypes. The project is located in /home/name/repos/pytypes, the used Python binary's standard library is located at /usr/lib/python3.10, and the virtual environment is located at /home/name/.cache/pypoetry/venv/pytypes-xvtnrWJT. To read up on why these paths are necessary, read up on the Resolver class and how types are stored into our trace data.

Furthermore, customisable unifiers that are used during the annotation generation process are stored by a name and identified by the kind attribute. To read up on how these unifiers come into play, read up on the annotation generation process.

[pytypes]
project = "PyTypes"

proj_path = "/home/name/repos/pytypes"
stdlib_path = "/usr/lib/python3.10"
venv_path = "/home/name/.cache/pypoetry/venv/pytypes-xvtnrWJT"

[[unifier]]
name = "remove_dups"
kind = "dedup"

[[unifier]]
name = "ignore_test"
kind = "drop_test"
test_name_pat = "test_"

[[unifier]]
name = "drop_implicit_2"
kind = "drop_mult_var"

[[unifier]]
name = "drop_explicit_5"
kind = "drop_mult_var"
min_amount_types_to_drop = 5

[[unifier]]
name = "unify_subtypes_relaxed"
kind = "unify_subty"

[[unifier]]
name = "min_threshold"
kind = "drop_min_threshold"
min_threshold = 0.3

[[unifier]]
name = "unify"
kind = "union"