Return to blacklist CVS log

Up to [Coherent Logic Development] / pandia

Annotation of pandia/blacklist, revision 1.3

1.1       snw         1: #!/usr/bin/env perl
                      2: 
                      3: # 
1.3     ! snw         4: # $Id: blacklist,v 1.2 2025/06/28 00:33:32 snw Exp $
1.1       snw         5: #  Copyright (C) 2025 Coherent Logic Development LLC
                      6: #
                      7: # Author: Serena Willis <snw@coherent-logic.com>
                      8: #
                      9: # Licensed AGPL-3.0
                     10: #
1.2       snw        11: # $Log: blacklist,v $
1.3     ! snw        12: # Revision 1.2  2025/06/28 00:33:32  snw
        !            13: # Update locking
        !            14: #
1.2       snw        15: # Revision 1.1  2025/06/27 16:20:30  snw
                     16: # Add blacklist
                     17: #
1.1       snw        18: #
                     19: 
                     20: use Getopt::Long;
1.2       snw        21: use DBI;
                     22: 
                     23: my $dbh = "";
                     24: my $dsn = "";
                     25: 
                     26: GetOptions("dbhost=s" => \$dbhost,
                     27:            "dbname=s" => \$dbname,
                     28:            "dbusername=s" => \$dbusername,
                     29:            "dbpw=s" => \$dbpw,
                     30:           "domain=s" => \$domain)
                     31:     or die("error in command line arguments");
                     32: 
                     33: $| = 1;
                     34: print "pandia blacklist v0.0.1\n";
                     35: print " Copyright (C) 2025 Coherent Logic Development LLC\n\n";
                     36: 
                     37: print "pandia:  connecting to $dbname database at $dbhost...";
                     38: $dsn = "DBI:mysql:database=$dbname;host=$dbhost;port=3306;mysql_connect_timeout=5;";
                     39: $dbh = DBI->connect($dsn, $dbusername, $dbpw, {RaiseError => 0, PrintError => 1});
                     40: die "pandia:  failed to connect to MySQL database: DBI->errstr()" unless $dbh;
                     41: print "[OK]\n";
                     42: 
                     43: print "pandia:  blacklisting domain $domain...";
                     44: my $sth = $dbh->prepare("INSERT INTO blacklist (url_domain) VALUES (?)");
                     45: $sth->execute($domain);
                     46: print "[OK]\n";
                     47: 
                     48: print "pandia:  removing blacklisted items from crawl queue...";
                     49: my $sth = $dbh->prepare("DELETE crawl_queue FROM crawl_queue JOIN blacklist ON crawl_queue.url_domain=blacklist.url_domain");
                     50: $sth->execute();
                     51: print "[OK]\n";
                     52: 
1.3     ! snw        53: print "pandia:  removing blacklisted items from index...";
        !            54: $sth = $dbh->prepare("DELETE url_fulltext FROM url_fulltext JOIN blacklist ON url_fulltext.url_domain=blacklist.url_domain");
        !            55: $sth->execute();
        !            56: print "[OK]\n";
        !            57: 
        !            58: