From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sat, 8 Jul 2023 22:12:34 +0000 (+0200)
Subject: html2txt links
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=0db0888fd7a671cb230738499ddadd0d47bed470;p=nb.git

html2txt links
---

diff --git a/lib/perl/NB/Functions.pm b/lib/perl/NB/Functions.pm
index 7146358b..8e7c6b64 100644
--- a/lib/perl/NB/Functions.pm
+++ b/lib/perl/NB/Functions.pm
@@ -557,10 +557,16 @@ sub html2txt {
 		( \\n )
 	,\n,gix;
 
+  # Links
+	$v =~ s,
+  	<a[^>]+href="([^"]+)"[^>]*>([^<]+)</a>
+  ,$2 ( $1 ),gix;
+
   # Delete
 	$v =~ s,(^_DUMMY_$) # never happend only for easy regexp change order
 
 		|( <!--\[if [^\]]+\]>.*?<!\[endif\]--> ) # condition
+		|( <!--.*?--> ) # comment
 
 		|( <style[^>]*>.*?</style> ) # style
 		|( <noscript[^>]*>.*?</noscript> ) # noscript
@@ -581,6 +587,7 @@ sub html2txt {
 	,,sgix;
 
 	# NB 03.07.23: Multiple empty nosy lines 
+	#$v =~ s/([ \t\r]*\n){2}([ \t\r]*\n)*/$1/sg;
 	$v =~ s/\n(\s*\n)+/\n/sg;
 
 	# Mutiple tabs