"""Functions which return a likeness score.
See Also:
The :class:`~.HeuristicScore` class.
"""
from collections.abc import Iterable as _Iterable
from . import exceptions
from .types import CandidateType, ContextType, ValueType
[docs]
def modified_hamming(
name: str,
candidates: _Iterable[str],
context: ContextType | None, # noqa: ARG001
*,
add_length_ratio_term: bool = True,
positional_penalty: float = 0.001,
) -> _Iterable[float]:
"""Compute hamming distance modified by length ratio, from the back. Score range is ``[0, 1]``.
Args:
name: A name that should be mapped one of the sources in `candidates`.
candidates: Candidate sources.
context: Should be ``None``. Always ignored, exists for compatibility.
add_length_ratio_term: If ``True``, score is divided by ``abs(len(name) - len(candidate))``.
positional_penalty: A penalty applied to prefer earlier `candidates`, according to the formulare
``penalty = index(candidate) * positional_penalty)``.
Examples:
>>> from id_translation.mapping.score_functions import modified_hamming
>>> list(modified_hamming("aa", ["aa", "a", "ab", "aa"], context=None))
[1.0, 0.499, 0.498, 0.997]
>>> list(
... modified_hamming(
... "aa", ["aa", "a", "ab", "aa"], context=None, positional_penalty=0
... )
... )
[1.0, 0.5, 0.5, 1.0]
>>> list(modified_hamming("face", ["face", "FAce", "race", "place"], context=None))
[1.0, 0.499, 0.748, 0.372]
"""
def _apply(candidate: str) -> float:
sz = min(len(candidate), len(name))
same = sum([name[i] == candidate[i] for i in range(-sz, 0)])
ratio = (1 / (1 + abs(len(candidate) - len(name)))) if add_length_ratio_term else 1
normalized_hamming = same / sz
return ratio * normalized_hamming
yield from (s - i * positional_penalty for i, s in enumerate(map(_apply, candidates)))
[docs]
def equality(
value: ValueType,
candidates: _Iterable[CandidateType],
context: ContextType | None, # noqa: ARG001
) -> _Iterable[float]:
"""Return 1.0 if ``k == c_i``, 0.0 otherwise.
Examples:
>>> from id_translation.mapping.score_functions import equality
>>> list(equality("a", "aAb", context=None))
[1.0, 0.0, 0.0]
"""
yield from map(float, (value == c for c in candidates))
[docs]
def disabled(
value: ValueType,
candidates: _Iterable[CandidateType],
context: ContextType | None,
strict: bool = True,
) -> _Iterable[float]:
"""Special value to indicate that scoring logic has been disabled.
This is a workaround to allow users to indicate that the scoring logic is disabled, and that overrides should be
used instead. The ``disabled``-function has no special meaning to the mapper, and will be called as any other
scoring function.
Returns:
If `strict` is ``False``, negative infinity for all `candidates`, serving as a catch-all removal filter.
Raises:
ScoringDisabledError: If `strict` is ``True``.
See Also:
The :ref:`override-only-mapping` documentation.
"""
if strict:
raise exceptions.ScoringDisabledError(value, candidates, context)
return [float("-inf")] * sum(1 for _ in candidates)
