From 7a42265190acc197cb76a86ac3d7cd234cfaba4f Mon Sep 17 00:00:00 2001
From: pierre <ratinaud@univ-tlse2.fr>
Date: Sat, 9 Nov 2024 11:18:35 +0100
Subject: [PATCH] europresse parser

---
 parse_europress.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/parse_europress.py b/parse_europress.py
index a4c7e23..5159ef2 100755
--- a/parse_europress.py
+++ b/parse_europress.py
@@ -16,6 +16,7 @@ import os
 # import des fichiers du projet
 #------------------------------------
 from html.parser import HTMLParser
+from html import unescape
 
 
 mois = {'janvier' : '01',
@@ -169,7 +170,7 @@ def ParseEuropress(txtdir, fileout, encodage_in, encodage_out) :
             parser.doinit(outf)
             with codecs.open(f, 'r', encodage_in) as infile :
                 content = infile.read()
-                content = HTMLParser().unescape(content)
+                content = unescape(content)
             parser.feed(content)
             tot += parser.nb
     return tot
-- 
2.7.4