From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sun, 4 Jun 2023 19:04:08 +0000 (+0200)
Subject: fix html entities
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=47d34a14d63396951c176a0ffb8406002621cf56;p=nb.git

fix html entities
---

diff --git a/lib/perl/NB/Functions.pm b/lib/perl/NB/Functions.pm
index 75d14cde..aa43fd9b 100644
--- a/lib/perl/NB/Functions.pm
+++ b/lib/perl/NB/Functions.pm
@@ -554,11 +554,18 @@ sub html2txt {
 		( (<)/?\w+(\s+\w+=['"]+[^'"]+['"]+)*/?(>) ) # start tag
 		|( (<)?\w+(\s+\w+=\S+)*/?(>) ) # end tag
 
-		|( (<|&lt;)!\[CDATA\[ ) # cdata begin
-		| ( \]\](>|&gt;) ) # cdata end
+	# NB 04.06.23: NO we want entities !!!  
+		# NB 04.06.23 |( (<|&lt;)!\[CDATA\[ ) # cdata begin
+		# NB 04.06.23 | ( \]\](>|&gt;) ) # cdata end
+		|( <!\[CDATA\[ ) # cdata begin
+		|( \]\]> ) # cdata end
+
+	# NB 04.06.23: NO we want entities !!!  
+		# NB 04.06.23 |( (<|&lt;)/\w+\s*$ ) # incomplet html
+		# NB 04.06.23 |( ^(<|&lt;)\w+ .* ) # incomplet html
+		|( </\w+\s*$ ) # incomplet html
+		|( ^<\w+ .* ) # incomplet html
 
-		|( (<|&lt;)/\w+\s*$ ) # incomplet html
-		|( ^(<|&lt;)\w+ .* ) # incomplet html
     |(<\!DOCTYPE[^>]+>)
     |([\w_-]+="[^"]+") # attrs
 		|( <\w[^>]+> ) # tags