Return to blacklist CVS log

Up to [Coherent Logic Development] / pandia

Annotation of pandia/blacklist, revision 1.2

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