def is_similar(self, other_person_address):
if not self.is_populated or not other_person_address.is_populated:
return False
if self.id == other_person_address.id:
return True
return (jellyfish.damerau_levenshtein_distance(self.concat, other_person_address.concat) < 3 and
len(self.concat) > 20)
def names_conflict(name1, name2):
"""
For comparing name parts. We're just checking for conflicts. The following examples should help:
None does not conflict with a
a does not conflict with a
ambrose does not conflict with ambrose
a does not conflict with ambrose
t does conflict with ambrose
am does conflict with ambrose
ambrost does not conflict with ambrose # our fuzzy matching
Again, we're just trying to ensure there are no conflicts between the names. A single letter is considered an
initial which is why a single letter does not conflict with a full name, but two letters can.
We've introduced a wee little bit of fuzziness to the matching by allowing there to be one Damerau-Levenshtein
edit distance between names, provided the names have a little meat to them (obviously, "A" is one edit distance
from "Z", so we need to ensure the names have a little bit of length to them).
"""
return not (name1 is None or
name2 is None or
name1 == name2 or
((len(name1) == 1 or len(name2) == 1) and name1[0] == name2[0]) or
(len(name1) > 3 and len(name2) > 3 and jellyfish.damerau_levenshtein_distance(name1, name2) < 2))
def is_similar(self, other_person_name):
if self.family is None or other_person_name.family is None:
return False
if self.id == other_person_name.id:
return True
# This seems redundant-ish, but Levenshtein is fairly inefficient. If we can reduce runs through the
# here, it should help speed things up a bit.
if self.family != other_person_name.family and \
jellyfish.damerau_levenshtein_distance(self.family, other_person_name.family) > 1:
return False
if (self.given is None and self.other is None) or \
(other_person_name.given is None and other_person_name.other is None):
return True
if ((self.given is not None or other_person_name.other is not None) and
(self.other is not None or other_person_name.given is not None) and
not common.names_conflict(self.given, other_person_name.given) and
not common.names_conflict(self.other, other_person_name.other)):
return True
if ((self.given is not None or other_person_name.given is not None) and
(self.other is not None or other_person_name.other is not None) and
not common.names_conflict(self.given, other_person_name.other) and
not common.names_conflict(self.other, other_person_name.given)):
return True
return False
