From f2fe60bacfbe4faa2536ccca6eed3b0cdc9ce357 Mon Sep 17 00:00:00 2001 From: Adam Bittlingmayer Date: Tue, 29 Dec 2020 16:05:11 +0400 Subject: [PATCH 1/2] Update tokenizer_exceptions.py See https://github.com/explosion/spaCy/pull/6643 --- spacy/lang/es/tokenizer_exceptions.py | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/spacy/lang/es/tokenizer_exceptions.py b/spacy/lang/es/tokenizer_exceptions.py index fbfe75545..0dae457a2 100644 --- a/spacy/lang/es/tokenizer_exceptions.py +++ b/spacy/lang/es/tokenizer_exceptions.py @@ -51,6 +51,9 @@ for orth in [ "Dr.", "Dra.", "EE.UU.", + "Ee.Uu.", + "UU.", # For "EE. UU." + "Uu.", # For "Ee. Uu." "etc.", "fig.", "Gob.", @@ -65,9 +68,11 @@ for orth in [ "Prof.", "Profa.", "q.e.p.d.", - "Q.E.P.D." "S.A.", + "Q.E.P.D.", + "S.A.", "S.L.", - "S.R.L." "s.s.s.", + "S.R.L.", + "s.s.s.", "Sr.", "Sra.", "Srta.", From bced6309e5c9e8d4f0bc006f2a20b0230e2f289f Mon Sep 17 00:00:00 2001 From: Adriane Boyd Date: Fri, 29 Jan 2021 14:27:22 +0100 Subject: [PATCH 2/2] Add full exceptions with spaces --- spacy/lang/es/tokenizer_exceptions.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/spacy/lang/es/tokenizer_exceptions.py b/spacy/lang/es/tokenizer_exceptions.py index 0dae457a2..74cdc143d 100644 --- a/spacy/lang/es/tokenizer_exceptions.py +++ b/spacy/lang/es/tokenizer_exceptions.py @@ -52,8 +52,8 @@ for orth in [ "Dra.", "EE.UU.", "Ee.Uu.", - "UU.", # For "EE. UU." - "Uu.", # For "Ee. Uu." + "EE. UU.", + "Ee. Uu.", "etc.", "fig.", "Gob.",