don't allow more edits than characters

This commit is contained in:
Kevin Humphreys 2022-09-15 16:14:24 -07:00
parent 252e9ab3af
commit 4a677acf5d

View File

@ -850,7 +850,7 @@ class _FuzzyPredicate:
if self.value == value:
return True
elif self.distance and token.is_oov and not token.is_space:
return bool(levenshtein(self.value, value) <= self.distance)
return bool(levenshtein(self.value, value) <= min(self.distance, len(token.text)-1))
return False
@ -927,7 +927,7 @@ class _SetPredicate:
elif self.distance and token.is_oov and not token.is_space:
for v in self.value:
if levenshtein(self.vocab.strings[value],
self.vocab.strings[v]) <= self.distance:
self.vocab.strings[v]) <= min(self.distance, len(token.text)-1):
return True
return False
elif self.predicate == "NOT_IN":
@ -936,7 +936,7 @@ class _SetPredicate:
elif self.distance and token.is_oov and not token.is_space:
for v in self.value:
if levenshtein(self.vocab.strings[value],
self.vocab.strings[v]) <= self.distance:
self.vocab.strings[v]) <= min(self.distance, len(token.text)-1):
return False
return True
elif self.predicate == "IS_SUBSET":