scripts: sitemap.pl

1 files changed, 81 insertions, 0 deletions

diff --git a/scripts/sitemap.pl b/scripts/sitemap.pl
new file mode 100644
index 00000000..0363fd44
--- /dev/null
+++ b/scripts/sitemap.pl
@@ -0,0 +1,81 @@
+#!/usr/bin/perl
+
+use strict;
+use DBI;
+use CGI;
+use utf8;
+
+my $dbh = DBI->connect("dbi:Pg:dbname=juick");
+
+my $sql =
+"SELECT users.nick,messages.message_id FROM messages INNER JOIN users ON messages.user_id=users.id WHERE users.banned=0 AND messages.privacy=1";
+my $sth = $dbh->prepare($sql) or die;
+$sth->execute or die;
+my $filescount = int( $sth->rows / 50000 ) + 1;
+for ( my $next = 0 ; $next < $filescount ; $next++ ) {
+    open( TEMP, ">_sitemap-messages" . $next . ".xml" );
+    print TEMP '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
+    print TEMP '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'
+      . "\n";
+    for ( my $i = 0 ; $i < 50000 ; $i++ ) {
+        my ( $nick, $mid ) = $sth->fetchrow() or goto NEXT;
+        print TEMP '<url><loc>https://juick.com/'
+          . $nick . '/'
+          . $mid
+          . '</loc></url>' . "\n";
+    }
+  NEXT: print TEMP '</urlset>' . "\n";
+}
+open( TEMP, ">_sitemap-users.xml" );
+print TEMP '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
+print TEMP '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'
+  . "\n";
+
+my $sql =
+'SELECT nick, to_char(lastmessage, \'YYYY-MM-DD"T"HH24:MI:SS"Z"\') FROM users INNER JOIN messages ON users.id=messages.user_id WHERE users.banned=0 GROUP BY users.id HAVING COUNT(message_id)>4';
+my $sth = $dbh->prepare($sql) or die;
+$sth->execute or die;
+while ( my ( $nick, $lastmod ) = $sth->fetchrow ) {
+    my $data =
+        '<url><loc>https://juick.com/'
+      . CGI::escape($nick)
+      . '/</loc><lastmod>'
+      . $lastmod
+      . '</lastmod></url>' . "\n";
+    print $data;
+    print TEMP $data;
+}
+
+print TEMP '</urlset>' . "\n";
+
+open( TEMP, ">_sitemap-tags.xml" );
+print TEMP '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
+print TEMP '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'
+  . "\n";
+
+my $sql =
+"SELECT tags.name FROM tags WHERE noindex=0 AND synonym_id IS NULL AND stat_messages>9 AND stat_users>2";
+my $sth = $dbh->prepare($sql) or die;
+$sth->execute or die;
+while ( my ($tag) = $sth->fetchrow ) {
+    print TEMP '<url><loc>https://juick.com/tag/'
+      . CGI::escape($tag)
+      . '</loc></url>' . "\n";
+}
+
+print TEMP '</urlset>' . "\n";
+
+open( TEMP, ">_sitemap-index.xml" );
+print TEMP '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
+print TEMP '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'
+  . "\n";
+print TEMP '<sitemap><loc>https://juick.com/_sitemap-users.xml</loc></sitemap>'
+  . "\n";
+print TEMP '<sitemap><loc>https://juick.com/_sitemap-tags.xml</loc></sitemap>'
+  . "\n";
+for ( my $next = 0 ; $next < $filescount ; $next++ ) {
+    print TEMP '<sitemap><loc>https://juick.com/_sitemap-messages'
+      . $next
+      . '.xml</loc></sitemap>' . "\n";
+}
+print TEMP '</sitemapindex>' . "\n";