From eaf3a4e857928eb9aaf018465f17c83845606735 Mon Sep 17 00:00:00 2001
From: Mathieu Loiseau <mathieu.loiseau@liris.cnrs.fr>
Date: Thu, 11 Jan 2024 11:45:31 +0100
Subject: [PATCH] {{audio|en|en-au-platypus.ogg}} sans 3e param
---
parsers/en_en.py | 20 +++++++++++++++++++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/parsers/en_en.py b/parsers/en_en.py
index 960944a..febb56a 100644
--- a/parsers/en_en.py
+++ b/parsers/en_en.py
@@ -2,6 +2,7 @@
from wikstraktor import Wikstraktor, Pronunciation, Sense, SubSense, Definition
from parsers.en_constants import string_values
+import re
debugEty = 0
@@ -13,6 +14,19 @@ class En_en_straktor(Wikstraktor):
self.constants = string_values
self.site = self.pwb.Site(f'wiktionary:en')
+ def process_audio_accent(self, audio, file_name = False):
+ if file_name: #on traite un nom de fichier
+ match = re.search(r'(e|E)(n|N)-(\w\w)-(.*)\.ogg', audio)
+ if match:
+ res = match.group(3).upper()
+ else:
+ res = None
+ else:
+ res = re.sub(r'((a|A)udio ?\(?)|\)?','',audio)
+ if res == "":
+ res = None
+ return res
+
def process_pronunciation(self, proContent):
# TODO: ne marche que pour les listes à 2 niveaux, voir water pour 3 niveaux
l = proContent.get_lists()[0]
@@ -29,7 +43,11 @@ class En_en_straktor(Wikstraktor):
elif t.normal_name() == self.constants['t_ipa']:
p.set_transcription(t.arguments[1].value)
elif t.normal_name() == self.constants['t_snd']:
- p.add_sound(self.get_file_url(t.arguments[1].value), t.arguments[2].value)
+ if len(t.arguments) > 2:
+ acc = self.process_audio_accent(t.arguments[2].value)
+ if acc == None:
+ acc = self.process_audio_accent(t.arguments[1].value, True)
+ p.add_sound(self.get_file_url(t.arguments[1].value), acc)
if p.ipa != None or p.has_accents() or p.has_sounds():
pronunciations.append(p)
else:
--
GitLab