Merge branch 'smhn'

smhnRSS.pl

@@ -1,5 +1,5 @@
 #!/usr/bin/perl
-use cPanelUserConfig; #for cpanel servers
+# use cPanelUserConfig; #for cpanel servers
 
 use 5.010;
 use strict;
@@ -63,6 +63,7 @@ my %list;
 # Make the list of URLS while parsing DATA
 my $listHTML = "<div class=\'listHTML\'><ul>\n";
 
+# Go through each URL in the DATA section and make the new list
 while ( my $url = <DATA>) {
     chomp $url;
     
@@ -96,14 +97,29 @@ while ( my $url = <DATA>) {
 			$testItem->{'feedURL'} = $_;
 			$testItem->{'feedName'} = $rss1->{'channel'}{'title'};
 			
+			# Find Author tags 
+			if ($testItem->{'dc'}{'creator'}) {
+				$testItem->{'itemAuthor'} = $testItem->{'dc'}{'creator'};
+			} elsif ($testItem->{'author'}) {
+				$testItem->{'itemAuthor'} = $testItem->{'author'}
+			} else {
+				$testItem->{'itemAuthor'} = $rss1->{'channel'}{'title'};
+			} # if author  
+						
 			# Clean up some of the artifacts in the RSS feed 'description' section
 			$testItem->{'description'} =~ s/\n\s*/\n/g; # get rid of excess white space
 			$testItem->{'description'} =~ s/(<a(.+?)<\/a>)$//s; # link at end of description
 			$testItem->{'description'} =~ s/<(\w) class=(.*?)>/<$1>/s; # remove class statements from text
 			$testItem->{'description'} =~ s/<figure ((.|\n)+?)figure>//sg; #remove "figure" can use pipe to add more
 			$testItem->{'description'} =~ s/<img(.+?)>//sg; # remove IMG tags
+			$testItem->{'description'} =~ s/<span ((.|\n)+?)>//sg; #remove "span" tags (mostly blogger)
+			$testItem->{'description'} =~ s/<\/span>//sg; #remove "span" endtags 
+			$testItem->{'description'} =~ s/<div class="separator" style(.+?)<\/div>//sg; # remove blogger DIV tags
+			$testItem->{'description'} =~ s/<br(.+?)>/<br>/sg; # remove blogger BR tags
+			$testItem->{'description'} =~ s/(<div><br><\/div>)+/<br>/sg; # remove blogger BR + DIV tags
 			$testItem->{'description'} =~ s/.{1200}\K.*//s; # limit length 
-			$testItem->{'description'} =~ s/(<[^<]+)$//s; # link at end of description
+			$testItem->{'description'} =~ s/<\/?div.*?>//sg; # remove div tags
+			$testItem->{'description'} =~ s/(<a[^<]+)$//s; # link at end of description
 			
 			#add continue reading to end. 
 			$testItem->{'description'} .= " <a href=\'" . $testItem->{'link'} . "\' target=\'_blank\'>Continue Reading</a>"; 
@@ -120,7 +136,7 @@ while ( my $url = <DATA>) {
 } # while DATA
 
 #close out the list URL html
-$listHTML .= "</div></ul>\n";
+$listHTML .= "</ul></div>\n";
 
 ###################################
 #  Make an RSS Feed! 
@@ -151,7 +167,7 @@ foreach my $name (reverse sort keys %list) {
         	link  => $list{$name}->{'link'},
         	description => $list{$name}->{'description'},
         	pubDate => $list{$name}->{'pubDate'},
-        	author => $list{$name}->{'dc'}{'creator'},
+        	author => $list{$name}->{'itemAuthor'},
         	source => $list{$name}->{'feedName'},
         	sourceUrl => $list{$name}->{'feedURL'},
        ); # $rss->channel
@@ -204,10 +220,9 @@ my $html = <<"HTML_END";
 		body{
 			margin:40px auto;
 			max-width:650px;
-			line-height:1.6;
+			font: 18px/1.6 Baskerville, TimesNewRoman, Times New Roman, Times, Georgia, serif;
-			font-size:18px;
+			padding:0 10px:
-			font-family: Baskerville, TimesNewRoman, Times New Roman, Times, Georgia, serif;
+			word-break: break-word;
-			padding:0 10px
 		}
 		h1, h2, h3 {font-weight: 200;}
 		hr{width:50%;}
@@ -238,7 +253,7 @@ my $html = <<"HTML_END";
 
 	<h1>$title</h1>
 	<p>$description</p>
-	<p>This site is updated every Sunday morning</p>
+	<p>This site is in BETA and will update daily as we test things out and add URLs. Let us know if there are sites that should be included!</p>
 	<p>updated: $printDate</p>
 
 </header>
@@ -304,12 +319,13 @@ close(FH);
 ###################################
 
 sub formatDate {
+	# takes a date, and converts it to a formatted string
 	my $testDate = $_[0];
 	$testDate =~ s/((\+|\-)(\d\d\d\d))//; 
 	# Convert to a TIME object
 	my $t = Time::Piece->strptime($testDate);
 	return($t->strftime('%a, %d %b %Y %H:%M:%S'));
-}
+} # sub formatDate
 
 # FIN
 
@@ -325,10 +341,10 @@ https://kelliowen.com/feed/
 https://mercedesmyardley.com/feed/
 https://mehitobel.com/feed/
 http://antoncancre.blogspot.com/feeds/posts/default?alt=rss
+https://marysangi.wordpress.com/feed/
 https://www.toddkeisling.com/news?format=rss
 https://www.jfgonzalez.org/news?format=rss
 https://www.briankeene.com/news?format=rss
-https://ghoulish.rip/feed/
 https://www.cemeterydance.com/extras/feed/
 https://www.gwendolynkiste.com/Blog/feed/
 https://katherinesilvaauthor.substack.com/feed
@@ -338,3 +354,14 @@ http://jacobhaddon.com/feed/
 http://apokrupha.com/feed/
 https://ellendatlow.com/feed/
 https://paulaguran.com/
+https://amandaheadlee.com/feed/
+https://theimbloglio.wordpress.com/feed/
+https://kennethwcain.com/feed/
+https://wellwortharead.blogspot.com/feeds/posts/default?alt=rss
+http://robertfordauthor.com/feed/
+https://tinyfrights.com/feed/
+https://weightlessbooks.com/feed/
+https://www.crystallakepub.com/feed/
+https://lynnehansen.zenfolio.com/blog.rss
+https://www.bevvincent.com/feed/
+http://liviallewellyn.com/feed/