API Reference

Module-Level Instance

import pantry

Importing pantry returns a Pantry instance. All methods below are available directly on this object.

pantry.has("numpy")
pantry.has("numpy", "pandas")

Return True if all listed packages are installed. Uses distribution metadata only — does not import the module.

PIL = pantry["pillow"]
User = pantry["myapp.models.User"]   # if lazy_import'd

Return a module (or lazy-resolved object). Checks lazy imports first, then installed packages.

np = pantry.get("numpy")
redis = pantry.get("redis", "fallback")

Return the imported module, or default if unavailable. The module is imported lazily on first access and cached.

Parameters:: pkg (str) — pip package name; default (object, optional) — value to return when unavailable
Returns:: types.ModuleType | None

pantry.version("numpy")   # "1.26.4" or None

Return the installed version of pkg, or None.

@pantry("pillow", "numpy")
def process(path):
    ...

Decorator that guards a function. Raises RuntimeError at call-time if any listed package is missing.

print(pantry.report("numpy", "pandas", "pillow"))
print(pantry.report())  # all previously queried packages

Return a formatted availability table. With arguments, probes those packages. Without arguments, reports all packages queried in this session.

pantry.lazy_import("myapp.models.User")

Register dotted paths for deferred import. No import at registration time. Resolve on first pantry["path"] access. Cached after first resolution.

For your own modules — to break circular imports. Separate from the external dependency system.

with pantry.simulate_missing("numpy"):
    assert pantry.has("numpy") is False

Context manager that temporarily hides packages. All methods (has, get, [], decorator, report) see them as unavailable. Restored on exit.