Resolver

The library provides its resolution machinery via an object oriented interface, which allows users to explore the structure of data before or during resolution using DependencyGroupInclude and DependencyGroupResolver.

For example,

from dependency_groups import DependencyGroupResolver

groups = {
  "test": ["pytest", {"include-group": "runtime"}],
  "runtime": ["flask"],
}

resolver = DependencyGroupResolver(groups)

# you can lookup a group without resolving it
resolver.lookup("test")  # [Requirement('pytest'), DependencyGroupInclude('runtime')]

# and resolve() produces packaging Requirements
resolver.resolve("test")  # [Requirement('pytest'), Requirement('flask')]

Models

class dependency_groups.DependencyGroupInclude(include_group: 'str')

Resolver

class dependency_groups.DependencyGroupResolver(dependency_groups: Mapping[str, str | Mapping[str, str]])

A resolver for Dependency Group data.

This class handles caching, name normalization, cycle detection, and other parsing requirements. There are only two public methods for exploring the data: lookup() and resolve().

Parameters:

dependency_groups – A mapping, as provided via pyproject [dependency-groups].

lookup(group: str) → tuple[Requirement | DependencyGroupInclude, ...]

Lookup a group name, returning the parsed dependency data for that group. This will not resolve includes.

Parameters:

group – the name of the group to lookup

Raises:

• ValueError – if the data does not appear to be valid dependency group data

• TypeError – if the data is not a string

• LookupError – if group name is absent

• packaging.requirements.InvalidRequirement – if a specifier is not valid

resolve(group: str) → tuple[Requirement, ...]

Resolve a dependency group to a list of requirements.

Parameters:

group – the name of the group to resolve

Raises:

• TypeError – if the inputs appear to be the wrong types

• ValueError – if the data does not appear to be valid dependency group data

• LookupError – if group name is absent

• packaging.requirements.InvalidRequirement – if a specifier is not valid

resolve_all() → Mapping[str, tuple[Requirement, ...]]

Resolve all dependency groups, returning a mapping of normalized group names to resolved requirements.

This is more efficient than calling resolve() on each group individually because it avoids repeated work when groups share common includes.

Raises:

• TypeError – if the data appears to be the wrong types

• ValueError – if the data does not appear to be valid dependency group data

• packaging.requirements.InvalidRequirement – if a specifier is not valid

Errors

class dependency_groups.CyclicDependencyError(requested_group: str, group: str, include_group: str)

An error representing the detection of a cycle.