mirror of
https://github.com/explosion/spaCy.git
synced 2025-08-06 13:20:20 +03:00
exclude whitespace tokens
This commit is contained in:
parent
a1c984043a
commit
252e9ab3af
|
@ -849,7 +849,7 @@ class _FuzzyPredicate:
|
||||||
value = token.vocab.strings[get_token_attr_for_matcher(token.c, self.attr)]
|
value = token.vocab.strings[get_token_attr_for_matcher(token.c, self.attr)]
|
||||||
if self.value == value:
|
if self.value == value:
|
||||||
return True
|
return True
|
||||||
elif self.distance and token.is_oov:
|
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) <= self.distance)
|
||||||
return False
|
return False
|
||||||
|
|
||||||
|
@ -924,7 +924,7 @@ class _SetPredicate:
|
||||||
if self.predicate == "IN":
|
if self.predicate == "IN":
|
||||||
if value in self.value:
|
if value in self.value:
|
||||||
return True
|
return True
|
||||||
elif self.distance and token.is_oov:
|
elif self.distance and token.is_oov and not token.is_space:
|
||||||
for v in self.value:
|
for v in self.value:
|
||||||
if levenshtein(self.vocab.strings[value],
|
if levenshtein(self.vocab.strings[value],
|
||||||
self.vocab.strings[v]) <= self.distance:
|
self.vocab.strings[v]) <= self.distance:
|
||||||
|
@ -933,7 +933,7 @@ class _SetPredicate:
|
||||||
elif self.predicate == "NOT_IN":
|
elif self.predicate == "NOT_IN":
|
||||||
if value in self.value:
|
if value in self.value:
|
||||||
return False
|
return False
|
||||||
elif self.distance and token.is_oov:
|
elif self.distance and token.is_oov and not token.is_space:
|
||||||
for v in self.value:
|
for v in self.value:
|
||||||
if levenshtein(self.vocab.strings[value],
|
if levenshtein(self.vocab.strings[value],
|
||||||
self.vocab.strings[v]) <= self.distance:
|
self.vocab.strings[v]) <= self.distance:
|
||||||
|
|
Loading…
Reference in New Issue
Block a user