From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sun, 4 Jun 2023 18:08:49 +0000 (+0200)
Subject: html2txt fix delete entities
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=699461386cdfcbddfa37967a85bebf98c1e26abb;p=nb.git

html2txt fix delete entities
---

diff --git a/lib/perl/NB/Functions.pm b/lib/perl/NB/Functions.pm
index 9e94d0da..ff275047 100644
--- a/lib/perl/NB/Functions.pm
+++ b/lib/perl/NB/Functions.pm
@@ -546,8 +546,12 @@ sub html2txt {
 
   # Delete
 	$v =~ s,
-		( (<|&lt;)/?\w+(\s+\w+=['"]+[^'"]+['"]+)*/?(>|&gt;) )
-		|( (<|&lt;)?\w+(\s+\w+=\S+)*/?(>|&gt;) )
+	# NB 04.06.23: NO we want entities !!!  
+		# NB 04.06.23 ( (<|&lt;)/?\w+(\s+\w+=['"]+[^'"]+['"]+)*/?(>|&gt;) )
+		# NB 04.06.23 |( (<|&lt;)?\w+(\s+\w+=\S+)*/?(>|&gt;) )
+
+		( (<)/?\w+(\s+\w+=['"]+[^'"]+['"]+)*/?(>) ) # start tag
+		|( (<)?\w+(\s+\w+=\S+)*/?(>) ) # end tag
 
 		|( (<|&lt;)!\[CDATA\[ ) # cdata begin
 		| ( \]\](>|&gt;) ) # cdata end