#!/usr/bin/perl -T -w

# <@LICENSE>
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to you under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at:
# 
#     http://www.apache.org/licenses/LICENSE-2.0
# 
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# </@LICENSE>

use strict;
use warnings;
use re 'taint';

my $VERSION = 'svnunknown';
if ('$Id: sa-update.raw 1652181 2015-01-15 17:30:37Z mmartinec $' =~ ':') {
  # Subversion keyword "$Id: sa-update.raw 1652181 2015-01-15 17:30:37Z mmartinec $" has been successfully expanded.
  # Doesn't happen with automated launchpad builds:
  # https://bugs.launchpad.net/launchpad/+bug/780916
  $VERSION = 'svn' . (split(/\s+/, '$Id: sa-update.raw 1652181 2015-01-15 17:30:37Z mmartinec $'))[2];
}

my $PREFIX          = '/usr/local';             # substituted at 'make' time
my $DEF_RULES_DIR   = '/usr/local/share/spamassassin';      # substituted at 'make' time
my $LOCAL_RULES_DIR = '/etc/mail/spamassassin';    # substituted at 'make' time
my $LOCAL_STATE_DIR = '/var/lib/spamassassin';    # substituted at 'make' time
use lib '/usr/local/share/perl5';                   # substituted at 'make' time

# We want to do a small amount of macro processing during channel installs,
# based on the values as passed in via 'make'
my %MACRO_VALUES = (
	'VERSION'		=> '3.004001',
	'CONTACT_ADDRESS'	=> 'the administrator of that system',
	'PREFIX'		=> '/usr/local',
	'DEF_RULES_DIR'		=> '/usr/local/share/spamassassin',
	'LOCAL_RULES_DIR'	=> '/etc/mail/spamassassin',
	'LOCAL_STATE_DIR'	=> '/var/lib/spamassassin',
	'INSTALLSITELIB'	=> '/usr/local/share/perl5',
);

# Standard perl modules
use Errno qw(ENOENT EACCES);
use IO::File qw(O_RDONLY O_WRONLY O_RDWR O_CREAT O_EXCL);
use File::Spec;
use File::Path;
use Getopt::Long;
use Pod::Usage;
use Config;

BEGIN {                          # see comments in "spamassassin.raw" for doco
  my @bin = File::Spec->splitpath($0);
  my $bin = ($bin[0] ? File::Spec->catpath(@bin[0..1]) : $bin[1])
            || File::Spec->curdir;

  if (-e $bin.'/lib/Mail/SpamAssassin.pm'
        || !-e '/usr/local/share/perl5/Mail/SpamAssassin.pm' )
  {
    my $searchrelative;
    if ($searchrelative && $bin eq '../' && -e '../blib/lib/Mail/SpamAssassin.pm')
    {
      unshift ( @INC, '../blib/lib' );
    } else {
      foreach ( qw(lib ../lib/site_perl
                ../lib/spamassassin ../share/spamassassin/lib))
      {
        my $dir = File::Spec->catdir( $bin, split ( '/', $_ ) );
        if ( -f File::Spec->catfile( $dir, "Mail", "SpamAssassin.pm" ) )
        { unshift ( @INC, $dir ); last; }
      }
    }
  }
}

# These are the non-standard required modules
use Net::DNS;
use HTTP::Date qw(time2str);
use Archive::Tar 1.23;
use IO::Zlib 1.04;

use vars qw($have_lwp $io_socket_module_name $have_inet4 $have_inet6);

BEGIN {
  # Deal with optional modules

  eval { require Digest::SHA; import Digest::SHA qw(sha1_hex); 1 }
  or do { require Digest::SHA1; import Digest::SHA1 qw(sha1_hex) };

  $have_lwp = eval {
    require LWP::UserAgent;
  };

  if (eval { require IO::Socket::IP }) {  # handles IPv6 and IPv4
    $io_socket_module_name = 'IO::Socket::IP';
 
  } elsif (eval { require IO::Socket::INET6 }) {  # handles IPv6 and IPv4
    $io_socket_module_name = 'IO::Socket::INET6';

  } elsif (eval { require IO::Socket::INET }) {  # IPv4 only
    $io_socket_module_name = 'IO::Socket::INET';
  }

  $have_inet4 =  # can we create a PF_INET socket?
    defined $io_socket_module_name && eval {  
      my $sock =
        $io_socket_module_name->new(LocalAddr => '0.0.0.0', Proto => 'tcp');
      $sock->close or die "error closing socket: $!"  if $sock;
      $sock ? 1 : undef;
    };

  $have_inet6 =  # can we create a PF_INET6 socket?
    defined $io_socket_module_name &&
    $io_socket_module_name ne 'IO::Socket::INET' &&
    eval {
      my $sock =
        $io_socket_module_name->new(LocalAddr => '::', Proto => 'tcp');
      $sock->close or die "error closing socket: $!"  if $sock;
      $sock ? 1 : undef;
    };
}

# These should already be available
use Mail::SpamAssassin;
use Mail::SpamAssassin::Util qw(untaint_var untaint_file_path
                           proc_status_ok exit_status_str am_running_on_windows
                           secure_tmpfile secure_tmpdir);

# Make the main dbg() accessible in our package w/o an extra function
*dbg=\&Mail::SpamAssassin::dbg;
sub dbg;

$| = 1;  # autoflushing STDOUT makes verbose output consistent with warnings

# Clean up PATH appropriately
Mail::SpamAssassin::Util::clean_path_in_taint_mode();

##############################################################################

# Default list of GPG keys allowed to sign update releases
#
# pub   4096R/5244EC45 2005-12-20
#       Key fingerprint = 5E54 1DC9 59CB 8BAC 7C78  DFDC 4056 A61A 5244 EC45
# uid                  updates.spamassassin.org Signing Key <release@spamassassin.org>
# sub   4096R/24F434CE 2005-12-20
#
# note for gpg newbs: these are "long" gpg keyids.  It's common to also
# use the last 8 hex digits as a shorter keyid string.
#
my %valid_GPG = ( 
  '0C2B1D7175B852C64B3CDC716C55397824F434CE' => 1,
  '5E541DC959CB8BAC7C78DFDC4056A61A5244EC45' => 1,
);

# Default list of channels to update against
#
my @channels = ( 'updates.spamassassin.org' );

my $IGNORE_MIRBY_OLDER_THAN = (24 * 60 * 60 * 7);       # 1 week

##############################################################################

my %opt;
@{$opt{'gpgkey'}} = ();
@{$opt{'channel'}} = ();
my $GPG_ENABLED = 1;

$opt{'gpghomedir'} = File::Spec->catfile($LOCAL_RULES_DIR, 'sa-update-keys');

Getopt::Long::Configure(
  qw(bundling no_getopt_compat no_auto_abbrev no_ignore_case));
GetOptions(
  'debug|D:s'                           => \$opt{'debug'},
  'version|V'                           => \$opt{'version'},
  'help|h|?'                            => \$opt{'help'},
  'verbose|v+'                          => \$opt{'verbose'},
  'checkonly'                           => \$opt{'checkonly'},
  'allowplugins'                        => \$opt{'allowplugins'},
  'refreshmirrors'                      => \$opt{'refreshmirrors'},

  # allow multiple of these on the commandline
  'gpgkey=s'				=> $opt{'gpgkey'},
  'gpghomedir=s'			=> \$opt{'gpghomedir'},
  'channel=s'				=> $opt{'channel'},

  'install=s'                           => \$opt{'install'},
  'import=s'			        => \$opt{'import'},
  'gpgkeyfile=s'			=> \$opt{'gpgkeyfile'},
  'channelfile=s'			=> \$opt{'channelfile'},
  'updatedir=s'				=> \$opt{'updatedir'},
  'gpg!'				=> \$GPG_ENABLED,

  '4'                                   => sub { $opt{'force_pf'} = 'inet' },
  '6'                                   => sub { $opt{'force_pf'} = 'inet6' },

  # backward compatibility
  'usegpg'				=> \$GPG_ENABLED,

) or print_usage_and_exit();

if ( defined $opt{'help'} ) {               
  print_usage_and_exit("For more information read the sa-update man page.\n", 0);
} 
if ( defined $opt{'version'} ) {            
  print_version();
  exit(0);
}

if ( $opt{'force_pf'} && $opt{'force_pf'} eq 'inet' && !$have_inet4 ) {
  warn "Option -4 specified but support for the ".
       "INET protocol family is not available.\n";
}
if ( $opt{'force_pf'} && $opt{'force_pf'} eq 'inet6' && !$have_inet6 ) {
  warn "Option -6 specified but support for the ".
       "INET6 protocol family is not available.\n";
}

# Figure out what version of SpamAssassin we're using, and also figure out the
# reverse of it for the DNS query.  Handle x.yyyzzz as well as x.yz.
my $SAVersion = $Mail::SpamAssassin::VERSION;
if ($SAVersion =~ /^(\d+)\.(\d{3})(\d{3})$/) {
  $SAVersion = join(".", $1+0, $2+0, $3+0);
}
elsif ($SAVersion =~ /^(\d)\.(\d)(\d)$/) {
  $SAVersion = "$1.$2.$3";
}
else {
  die "fatal: SpamAssassin version number '$SAVersion' is in an unknown format!\n";
}
my $RevSAVersion = join(".", reverse split(/\./, $SAVersion));

  
# set debug areas, if any specified (only useful for command-line tools)
$SAVersion =~ /^(\d+\.\d+)/;
if ($1+0 > 3.0) {
  $opt{'debug'} ||= 'all' if (defined $opt{'debug'});
}
else {
  $opt{'debug'} = defined $opt{'debug'};
}


# Find the default site rule directory, also setup debugging and other M::SA bits
my $SA = Mail::SpamAssassin->new({
  debug => $opt{'debug'},
  local_tests_only => 1,
  dont_copy_prefs => 1,

  PREFIX          => $PREFIX,
  DEF_RULES_DIR   => $DEF_RULES_DIR,
  LOCAL_RULES_DIR => $LOCAL_RULES_DIR,
  LOCAL_STATE_DIR => $LOCAL_STATE_DIR,
});

if (defined $opt{'updatedir'}) {
  $opt{'updatedir'} = untaint_file_path($opt{'updatedir'});
}
else {
  $opt{'updatedir'} = $SA->sed_path('__local_state_dir__/__version__');
}


# check only disabled gpg
# https://issues.apache.org/SpamAssassin/show_bug.cgi?id=5854
if ( defined $opt{'checkonly'}) {
  $GPG_ENABLED=0;
  dbg("gpg: Disabling gpg requirement due to checkonly flag.");
}


dbg("generic: sa-update version $VERSION");
dbg("generic: using update directory: $opt{'updatedir'}");

# doesn't really display useful things for this script, but we do want
# a module/version listing, etc. sa-update may be used for older versions
# of SA that don't include this function, so eval around it.
eval { $SA->debug_diagnostics(); 1; };

$SA->finish();

# untaint the command-line args; since the root user supplied these, and
# we're not a setuid script, we trust them
foreach my $optkey (keys %opt) {
  next if ref $opt{$optkey};
  untaint_var(\$opt{$optkey});
}

##############################################################################

# Deal with gpg-related options

if (@{$opt{'gpgkey'}}) {
  $GPG_ENABLED = 1;
  foreach my $key (@{$opt{'gpgkey'}}) {
    unless (is_valid_gpg_key_id($key)) {
      dbg("gpg: invalid gpgkey parameter $key");
      next;
    }
    $key = uc $key;
    dbg("gpg: adding key id $key");
    $valid_GPG{$key} = 1;
  }
}

if (defined $opt{'gpgkeyfile'}) {
  $GPG_ENABLED = 1;
  open(GPG, $opt{'gpgkeyfile'})
    or die "cannot open $opt{'gpgkeyfile'} for reading: $!\n";

  dbg("gpg: reading in gpgfile ".$opt{'gpgkeyfile'});
  while(my $key = <GPG>) {
    chomp $key;

    $key =~ s/#.*$//;   # remove comments
    $key =~ s/^\s+//;   # remove leading whitespace
    $key =~ s/\s+$//;   # remove tailing whitespace
    next if $key eq ''; # skip empty lines

    unless (is_valid_gpg_key_id($key)) {
      dbg("gpg: invalid key id $key");
      next;
    }
    $key = uc $key;
    dbg("gpg: adding key id $key");
    $valid_GPG{$key} = 1;
  }
  close(GPG) or die "cannot close $opt{'gpgkeyfile'}: $!";
}

# At this point, we need to know where GPG is ...
my $GPGPath;
if ($GPG_ENABLED || $opt{'import'}) {
  # find GPG in the PATH
  # bug 4958: for *NIX it's "gpg", in Windows it's "gpg.exe"
  $GPGPath = 'gpg' . $Config{_exe};
  dbg("gpg: Searching for '$GPGPath'");

  if ($GPGPath = Mail::SpamAssassin::Util::find_executable_in_env_path($GPGPath)) {
    dbg("gpg: found $GPGPath");

    # bug 5030: if GPGPath has a space, put it in quotes
    if ($GPGPath =~ / /) {
      $GPGPath =~ s/"/\\"/g;
      $GPGPath = qq/"$GPGPath"/;
      dbg("gpg: path changed to $GPGPath");
    }
  }
  else {
    die "error: gpg required but not found!  It is not recommended, but you can use \"sa-update\" with the --no-gpg to skip the verification. \n";
  }

  # GPG was found, and we've been asked to import a key only
  if ( $opt{'import'} ) {
    my $ex = import_gpg_key($opt{'import'});
    exit $ex;
  }

  # does the sa-update keyring exist?  if not, import it
  if(!-f File::Spec->catfile($opt{'gpghomedir'}, "secring.gpg")) {
    import_default_keyring();
    # attempt to continue even if this fails, anyway
  }

  # specify which keys are trusted
  dbg("gpg: release trusted key id list: ".join(" ", keys %valid_GPG));

  # convert fingerprint gpg ids to keyids
  foreach (keys %valid_GPG) {
    my $id = substr $_, -8;
    $valid_GPG{$id} = 1;
  }
}

##############################################################################

# Deal with channel-related options
if (defined $opt{'channel'} && scalar @{$opt{'channel'}} > 0) {
  @channels = @{$opt{'channel'}};
}
if (defined $opt{'channelfile'}) {
  open(CHAN, $opt{'channelfile'})
    or die "cannot open $opt{'channelfile'} for reading: $!\n";

  dbg("channel: reading in channelfile ".$opt{'channelfile'});
  @channels = ();
  while(my $chan = <CHAN>) {
    chomp $chan;

    $chan =~ s/#.*$//;   # remove comments
    $chan =~ s/^\s+//;   # remove leading whitespace
    $chan =~ s/\s+$//;   # remove tailing whitespace
    next if $chan eq ''; # skip empty lines

    $chan = lc $chan;
    dbg("channel: adding $chan");
    push(@channels, $chan);
  }
  close(CHAN) or die "cannot close $opt{'channelfile'}: $!";
}

# untaint the channel listing
for(my $ind = 0; $ind < @channels; $ind++) {
  local($1); # bug 5061: prevent random taint flagging of $1
  if ($channels[$ind] =~ /^([a-zA-Z0-9._-]+)$/) {
    untaint_var(\$channels[$ind]);
  }
  else {
    dbg("channel: skipping invalid channel: $channels[$ind]");
    splice @channels, $ind, 1;
    $ind--; # the next element takes the place of the deleted one
  }
}

my ($res, $ua);

if ($opt{'install'}) {
  if (scalar @channels > 1) {
    die "fatal: --install cannot be used with multiple --channel switches.\n";
  }

} else {
  $res = Net::DNS::Resolver->new();
  $res->force_v4(1)  if $have_inet4 &&
                        $opt{'force_pf'} && $opt{'force_pf'} eq 'inet';
}

# Generate a temporary file to put channel content in for later use ...
my ($content_file, $tfh) = secure_tmpfile();
$tfh
  or die "fatal: could not create temporary channel content file: $!\n";
close $tfh
  or die "cannot close temporary channel content file $content_file: $!";
undef $tfh;

my $lint_failures = 0;
my $channel_failures = 0;
my $channel_successes = 0;

# Use a temporary directory for all update channels
my $UPDTmp;

# we only need to lint the site pre files once
my $site_pre_linted = 0;

# Go ahead and loop through all of the channels
foreach my $channel (@channels) {
  dbg("channel: attempting channel $channel");
  my %preserve_files;

  # Convert the channel to a nice-for-filesystem version
  my $nicechannel = $channel;
  $nicechannel =~ tr/A-Za-z0-9-/_/cs;

  my $UPDDir = File::Spec->catfile($opt{'updatedir'}, $nicechannel);
  my $CFFile = "$UPDDir.cf";
  my $PREFile = "$UPDDir.pre";

  if (-d $UPDDir) {
    dbg("channel: using existing directory $UPDDir");
  } else {
    # create the dir, if it doesn't exist
    dbg("channel: creating directory $UPDDir");
    mkpath([$UPDDir], 0, 0777)
      or die "channel: cannot create channel directory $UPDDir: $!\n";
  }
  dbg("channel: channel cf file $CFFile");
  dbg("channel: channel pre file $PREFile");

  my $instfile;
  if ($opt{'install'}) {
    $instfile = $opt{'install'};
    dbg("channel: installing from file $instfile");
  }

  my($mirby, $mirby_force_reload, $mirby_file_is_ok);
  my $mirby_path = File::Spec->catfile($UPDDir, "MIRRORED.BY");

  # try to read metadata from channel.cf file
  my $currentV = -1;
  if (!open(CF, $CFFile)) {
    dbg("channel: error opening file %s: %s",
        $CFFile, $!)  unless $! == ENOENT;
  } else {
    while(<CF>) {
      local($1,$2);
      last unless /^# UPDATE\s+([A-Za-z]+)\s+(\S+)/;
      my($type, $value) = (lc $1,$2);

      dbg("channel: metadata $type = $value, from file $CFFile");

      if ($type eq 'version') {
        $value =~ /^(\d+)/;
        $currentV = $1;
      }
    }
    close(CF) or die "cannot close $CFFile: $!";
  }

  # obtain a version number which should be installed
  my $newV;
  if ($instfile) {
    # the /.*/ ensures we use the 3-digit string nearest to the end of string,
    # otherwise we might pick up something from the middle of the directory path
    local($1);
    if ($instfile !~ /(?:.*\D|^)(\d{3,})/) {        
      # this is a requirement
      die "channel: $channel: --install file $instfile does not contain a 3-digit version number!\n";
    }
    $newV = $1;

    if ( defined $opt{'checkonly'} ) {            
      dbg("channel: $channel: --install and --checkonly, claiming update available");
      $channel_successes++;
      next;
    }

  } else {  # not an install file, get the latest version number from network
    # Setup the channel version DNS query
    my $DNSQ = "$RevSAVersion.$channel";

    my $dnsV = join(' ', do_dns_query($DNSQ));
    local($1);
    if (defined $dnsV && $dnsV =~ /^(\d+)/) {
      $newV = $1 if (!defined $newV || $1 > $newV);
      dbg("dns: $DNSQ => $dnsV, parsed as $1");
    }

    # Not getting a response isn't a failure, there may just not be any updates
    # for this SA version yet.
    if (!defined $newV) {
      my @mirs = do_dns_query("mirrors.$channel");
      if (defined shift @mirs) {
        dbg("channel: no updates available, skipping channel");
      } else {
        channel_failed("channel: no 'mirrors.$channel' record found");
      }
      next;
    }

    # If this channel hasn't been installed before, or it's out of date,
    # keep going.  Otherwise, skip it.
    if ($currentV >= $newV) {
      dbg("channel: current version is $currentV, new version is $newV, ".
          "skipping channel");
      next;
    }

    print "Update available for channel $channel: $currentV -> $newV\n"  if $opt{'verbose'};

    # If we are only checking for update availability, exit now
    if ( defined $opt{'checkonly'} ) {            
      dbg("channel: $channel: update available, not downloading ".
          "in checkonly mode");
      $channel_successes++;
      next;
    }

  }

  # we need a directory we control that we can use to avoid loading any rules
  # when we lint the site pre files, we might as well use the channel temp dir
  dbg("channel: preparing temp directory for new channel");
  if (!$UPDTmp) {
    $UPDTmp = secure_tmpdir();
    dbg("channel: created tmp directory $UPDTmp");
  }
  else {
    dbg("channel: using existing tmp directory $UPDTmp");
    if (!clean_update_dir($UPDTmp)) {
      die "channel: attempt to clean update temp dir failed, aborting";
    }
  }

  # lint the site pre files (that will be used when lint checking the channel)
  # before downloading the channel update
  unless ($site_pre_linted) {
    dbg("generic: lint checking site pre files once before attempting channel updates");
    unless (lint_check_dir(File::Spec->catfile($UPDTmp, "doesnotexist"))) {
      dbg("generic: lint of site pre files failed, cannot continue"); 
      print "Lint of site pre files failed, cannot continue\n"  if $opt{'verbose'};
      $lint_failures++;
      last;
    }
    dbg("generic: lint check of site pre files succeeded, continuing with channel updates");
    $site_pre_linted = 1;
  }

  my $content;
  my $SHA1;
  my $GPG;

  if ($instfile) {
    dbg("channel: using --install files $instfile\{,.sha1,.asc\}");
    $content = read_install_file($instfile);
    $SHA1 = read_install_file($instfile.".sha1");
    $GPG = read_install_file($instfile.".asc") if $GPG_ENABLED;

  } else {  # not an install file, obtain fresh rules from network
    dbg("channel: protocol family available: %s%s",
        join(',', $have_inet4 ? 'inet'  : (),
                  $have_inet6 ? 'inet6' : ()),
        $opt{'force_pf'} ? '; force '.$opt{'force_pf'} : '' );

    # test if the MIRRORED.BY file for this channel exists,
    # is nonempty, and is reasonably fresh

    my(@mirr_stat_list) = stat($mirby_path);
    if (!@mirr_stat_list) {
      if ($! == ENOENT) {
        dbg("channel: no mirror file %s, will fetch it", $mirby_path);
      } else {
        # protection error, misconfiguration, file system error, ...
        warn "error: error accessing mirrors file $mirby_path: $!\n";
        channel_failed("channel: error accessing mirrors file $mirby_path: $!");
        next;
      }
    } elsif (-z _) {
      dbg("channel: file %s is empty, refreshing mirrors file", $mirby_path);
      $mirby_force_reload = 1;
    } elsif ($opt{'refreshmirrors'}) {
      dbg("channel: --refreshmirrors used, forcing mirrors file refresh ".
          "on channel $channel");
      $mirby_force_reload = 1;
    } elsif (time - $mirr_stat_list[9] > $IGNORE_MIRBY_OLDER_THAN) {
      dbg("channel: file %s is too old, refreshing mirrors file", $mirby_path);
      $mirby_file_is_ok = 1;  # mirrors file seems fine, but is old
      $mirby_force_reload = 1;
    } else {
      # mirror file $mirby_path exists, is nonempty, and is reasonably fresh
      $mirby_file_is_ok = 1;
    }

    if (!$mirby_file_is_ok || $mirby_force_reload) {
      # fetch a fresh list of mirrors
      dbg("channel: DNS lookup on mirrors.$channel");
      my @mirrors = do_dns_query("mirrors.$channel");
      unless (@mirrors) {
        warn "error: no mirror data available for channel $channel\n";
        channel_failed("channel: MIRRORED.BY file URL was not in DNS");
        next;
      }
      foreach my $mirror (@mirrors) {
        my $result_fname;
        ($mirby, $result_fname) =
          http_get($mirror, $UPDDir, $mirby_path, $mirby_force_reload);

        if ($mirby) {
          dbg("channel: MIRRORED.BY file for channel %s retrieved", $channel);
          $mirby_file_is_ok = 1;
          $mirby_force_reload = 0;
          $preserve_files{$mirby_path} = 1;

          # set file creation time to now, otherwise we'll keep refreshing
          # (N.B.: curl preserves time of a downloaded file)
          my $now = time;
          utime($now, $now, $mirby_path)
            or warn "error: error setting creation time of $mirby_path: $!\n";

          last;
        }
        dbg("channel: no mirror data available for channel %s from %s",
            $channel, $mirror);
      }
      if ($mirby_force_reload) {  # not refreshed?
        warn "error: unable to refresh mirrors file for channel $channel, ".
             "using old file\n";
      }
    }

    if (!$mirby_file_is_ok) {
      warn "error: no mirror data available for channel $channel\n";
      channel_failed("channel: MIRRORED.BY file contents were missing");
      next;
    } elsif ($mirby) {
      # file contents already in memory, no need to read it from a file
    } elsif (!open(MIRBY, $mirby_path)) {
      warn "error: error opening mirrors file $mirby_path: $!\n";
      channel_failed("channel: error opening mirrors file $mirby_path: $!");
      next;
    } else {
      dbg("channel: reading MIRRORED.BY file %s", $mirby_path);
      { local $/ = undef; $mirby = <MIRBY> }
      close(MIRBY) or die "cannot close $mirby_path: $!";
      $preserve_files{$mirby_path} = 1;
    }

    # Parse the list of mirrors
    dbg("channel: parsing MIRRORED.BY file for channel %s", $channel);
    my %mirrors;
    my @mirrors = split(/^/, $mirby);
    while(my $mirror = shift @mirrors) {
      chomp $mirror;

      $mirror =~ s/#.*$//;   # remove comments
      $mirror =~ s/^\s+//;   # remove leading whitespace
      $mirror =~ s/\s+$//;   # remove tailing whitespace
      next if $mirror eq ''; # skip empty lines

      # We only support HTTP (and HTTPS) right now
      if ($mirror !~ m{^https?://}i) {
        dbg("channel: skipping non-HTTP mirror: $mirror");
        next;
      }

      dbg("channel: found mirror $mirror");

      my @data;
      ($mirror,@data) = split(/\s+/, $mirror);
      $mirror =~ s{/+\z}{};  # http://example.com/updates/ -> .../updates
      $mirrors{$mirror}->{weight} = 1;
      foreach (@data) {
        my($k,$v) = split(/=/, $_, 2);
        $mirrors{$mirror}->{$k} = $v;
      }
    }

    unless (%mirrors) {
      warn "error: no mirrors available for channel $channel\n";
      channel_failed("channel: no mirrors available");
      next;
    }

    # Now that we've laid the foundation, go grab the appropriate files
    #

    # Loop through all available mirrors, choose from them randomly
    # if the archive get fails, choose another mirror,
    # if the get for the sha1 or gpg signature files, the channel fails
    while (my $mirror = choose_mirror(\%mirrors)) {
      my $result_fname;
      # Grab the data hash for this mirror, then remove it from the list
      my $mirror_info = $mirrors{$mirror};
      delete $mirrors{$mirror};

      dbg("channel: selected mirror $mirror");

      # Actual archive file
      ($content, $result_fname) =
        http_get("$mirror/$newV.tar.gz", $UPDDir);
      next unless $content;
      $preserve_files{$result_fname} = 1;

      # SHA1 of the archive file
      ($SHA1, $result_fname) =
        http_get("$mirror/$newV.tar.gz.sha1", $UPDDir);
      last unless $SHA1;
      $preserve_files{$result_fname} = 1;

      # if GPG is enabled, the GPG detached signature of the archive file
      if ($GPG_ENABLED) {
        ($GPG, $result_fname) =
          http_get("$mirror/$newV.tar.gz.asc", $UPDDir);
        last unless $GPG;
        $preserve_files{$result_fname} = 1;
      }

      last;
    }

  }

  unless ($content && $SHA1 && (!$GPG_ENABLED || $GPG)) {
    channel_failed("channel: could not find working mirror");
    next;
  }

  # Validate the SHA1 signature before going forward with more complicated
  # operations.
  # The SHA1 file may be "signature filename" ala sha1sum, just use the signature
  { local($1);
    $SHA1 =~ /^([a-fA-F0-9]{40})/;
    $SHA1 = $1 || 'INVALID';
  }
  my $digest = sha1_hex($content);
  dbg("sha1: verification wanted: $SHA1");
  dbg("sha1: verification result: $digest");
  unless ($digest eq $SHA1) {
    channel_failed("channel: SHA1 verification failed");
    next;
  }

  # Write the content out to a temp file for GPG/Archive::Tar interaction
  dbg("channel: populating temp content file %s", $content_file);
  open(TMP, ">$content_file")
    or die "fatal: cannot create content temp file $content_file: $!\n";
  binmode TMP
    or die "fatal: cannot set binmode on content temp file $content_file: $!\n";
  print TMP $content
    or die "fatal: cannot write to content temp file $content_file: $!\n";
  close TMP
    or die "fatal: cannot close content temp file $content_file: $!\n";

  # to sign  : gpg -bas file
  # to verify: gpg --verify --batch --no-tty --status-fd=1 -q --logger-fd=1 file.asc file
  # look for : [GNUPG:] GOODSIG 6C55397824F434CE updates.spamassassin.org [...]
  #            [GNUPG:] VALIDSIG 0C2B1D7175B852C64B3CDC716C55397824F434CE [...]
  #            [GNUPG:] NO_PUBKEY 6C55397824F434CE
  if ($GPG) {
    dbg("gpg: populating temp signature file");
    my $sig_file;
    ($sig_file, $tfh) = secure_tmpfile();
    $tfh
      or die "fatal: couldn't create temp file for GPG signature: $!\n";
    binmode $tfh
      or die "fatal: cannot set binmode on temp file for GPG signature: $!\n";
    print $tfh $GPG
      or die "fatal: cannot write temp file for GPG signature: $!\n";
    close $tfh
      or die "fatal: cannot close temp file for GPG signature: $!\n";
    undef $tfh;

    dbg("gpg: calling gpg");

    my $gpghome = interpolate_gpghomedir();

    # TODO: we could also use "--keyserver pgp.mit.edu" or similar,
    # to autodownload missing keys...
    my $CMD = "$GPGPath $gpghome --verify --batch ".
        "--no-tty --status-fd=1 -q --logger-fd=1";

    unless (open(CMD, "$CMD $sig_file $content_file|")) {
      unlink $sig_file or warn "error: cannot unlink $sig_file: $!\n";
      die "fatal: couldn't execute $GPGPath: $!\n";
    }

    # Determine the fate of the signature
    my $signer = '';
    my $missingkeys = '';
    while(my $GNUPG = <CMD>) {
      chop $GNUPG;
      dbg ("gpg: $GNUPG");

      if ($GNUPG =~ /^gpg: fatal:/) {
        warn $GNUPG."\n";        # report bad news
      }

      local($1);
      if ($GNUPG =~ /^\Q[GNUPG:]\E NO_PUBKEY \S+(\S{8})$/) {
        $missingkeys .= $1." ";
      }

      next unless ($GNUPG =~ /^\Q[GNUPG:]\E (?:VALID|GOOD)SIG (\S{8,40})/);
      my $key = $1;

      # we want either a keyid (8) or a fingerprint (40)
      if (length $key > 8 && length $key < 40) {
        substr($key, 8) = '';
      }

      # use the longest match we can find
      $signer = $key if (length $key > length $signer);
    }

    my $errno = 0;  close CMD or $errno = $!;
    proc_status_ok($?,$errno)
      or warn("gpg: process '$GPGPath' finished: ".
              exit_status_str($?,$errno)."\n");

    unlink $sig_file or warn "cannot unlink $sig_file: $!\n";

    if ($signer) {
      my $keyid = substr $signer, -8;
      dbg("gpg: found signature made by key $signer");
      if (exists $valid_GPG{$signer}) {
	dbg("gpg: key id $signer is release trusted");
      }
      elsif (exists $valid_GPG{$keyid}) {
	dbg("gpg: key id $keyid is release trusted");
      }
      else {
	dbg("gpg: key id $keyid is not release trusted");
	$signer = undef;
      }
    }

    unless ($signer) {
      warn "error: GPG validation failed!\n";

      if ($missingkeys) {

        warn <<ENDOFVALIDATIONERR;
The update downloaded successfully, but it was not signed with a trusted GPG
key.  Instead, it was signed with the following keys:

    $missingkeys

Perhaps you need to import the channel's GPG key?  For example:

    wget http://spamassassin.apache.org/updates/GPG.KEY
    sa-update --import GPG.KEY

ENDOFVALIDATIONERR

      } else {

        warn <<ENDOFVALIDATIONERR;
The update downloaded successfully, but the GPG signature verification
failed.
ENDOFVALIDATIONERR

      }

      channel_failed("channel: GPG validation failed");
      next;
    }
  }

  # OK, we're all validated at this point, install the new version
  dbg("channel: file verification passed, testing update");

  dbg("channel: extracting archive");
  if (!taint_safe_archive_extract($UPDTmp, $content_file)) {
    channel_failed("channel: archive extraction failed");
    next;
  }

  # check --lint

  if (!lint_check_dir($UPDTmp)) {
    channel_failed("channel: lint check of update failed");
    next;
  }

  dbg("channel: lint check succeeded, extracting archive to $UPDDir...");

  my @totry = (
    {
      'try' => sub {
        if (-d $UPDDir) {
          # ok that worked, too late to stop now!   At this stage, if there are
          # errors, we have to attempt to carry on regardless, since we've already
          # blown away the old ruleset.
          dbg("channel: point of no return for existing $UPDDir");

          # clean out the previous channel files, if they exist
          if (-f $PREFile && ! unlink $PREFile ) {
            warn("channel: attempt to rm channel pre file failed, attempting to continue anyway: $!");
          }
          if (-f $CFFile && ! unlink $CFFile ) {
            warn("channel: attempt to rm channel cf file failed, attempting to continue anyway: $!");
          }
          if (!clean_update_dir($UPDDir, \%preserve_files)) {
            warn("channel: attempt to rm channel directory failed, attempting to continue anyway");
          }
        }
        else {
          # create the dir, if it doesn't exist
          dbg("channel: creating $UPDDir");
          mkpath([$UPDDir], 0, 0777)
            or die "channel: cannot create channel directory $UPDDir: $!\n";

          # ok, that test worked.  it's now likely that the .cf's will
          # similarly be ok to rename, too.   Too late to stop from here on
          dbg("channel: point of no return for new $UPDDir");
        }

	return 1;
      },

      'rollback' => sub {
	dbg("channel: attempting to remove the channel and update directories");

        # bug 4941: try to get rid of the empty directories to avoid leaving SA
        # with no rules.
        rmdir $UPDDir
          or dbg("channel: error removing dir %s: %s", $UPDDir, $!);
        rmdir $opt{'updatedir'}
          or dbg("channel: error removing dir %s: %s", $opt{'updatedir'}, $!);
      },
    },

    {
      'try' => sub {
        # extract the files again for the last time
        if (!taint_safe_archive_extract($UPDDir, $content_file)) {
          channel_failed("channel: archive extraction failed");
	  return 0;
        }

	return 1;
      },

      'rollback' => sub {
	dbg("channel: attempting to clean out the channel update directory");

        # bug 4941: try to get rid of the empty directories to avoid leaving SA
        # with no rules.
        if (!clean_update_dir($UPDDir, \%preserve_files)) {
          warn "channel: attempt to clean up failed extraction also failed!\n";
        }
      },
    },

    {
      'try' => sub {
        if ($instfile) {
          dbg("channel: not creating MIRRORED.BY file due to --install");
          return 1;
        }

      # The $mirby_path file should have already been created by http_get
      # and preserved past clean_update_dir()
      #
      # # Write out the mirby file, not fatal if it doesn't work
      # dbg("channel: creating MIRRORED.BY file");
      # open(MBY, ">$mirby_path")
      #   or die "cannot create a new MIRRORED.BY file: $!\n";
      # print MBY $mirby or die "error writing to $mirby_path: $!";
      # close(MBY) or die "cannot close $mirby_path: $!";

	return 1;
      },

      'rollback' => sub {
      },
    },

    {
      'try' => sub {
        # the last step is to create the .cf and .pre files to include the
        # channel files
        my @CF;
        my @PRE;

        dbg("channel: creating update cf/pre files");

        # Put in whatever metadata we need
        push(@CF, "# UPDATE version $newV\n");


        # Find all of the cf and pre files
        opendir(DIR, $UPDDir)
          or die "fatal: cannot access $UPDDir: $!\n";

        my @files;
        while(my $file = readdir(DIR)) {
	  next if $file eq '.' || $file eq '..';
          untaint_var(\$file);
          my $path = File::Spec->catfile($UPDDir, $file);
          next unless (-f $path);   # shouldn't ever happen
          push(@files, $file);
        }

        # bug 5371: ensure files are sorted
        foreach my $file ( sort @files ) {
          if ($file =~ /\.cf$/) {
            push(@CF, "include $nicechannel/$file\n");
          }
          elsif ($file =~ /\.pre$/) {
            push(@PRE, "include $nicechannel/$file\n");
          }
          else {
            next;
          }

          dbg("channel: adding $file");
        }
        closedir(DIR) or die "cannot close directory $UPDDir: $!";

        # Finally, write out the files to include the update files
        if (!write_channel_file($PREFile, \@PRE)) {
          channel_failed("channel: writing of $PREFile failed");
          return 0;
        }
        if (!write_channel_file($CFFile, \@CF)) {
          channel_failed("channel: writing of $CFFile failed");
          return 0;
        }

        # if all went fine, remove the .tar.gz, .sha1 and .asc files
        delete_files( grep(!m{/\QMIRRORED.BY\E\z}, keys %preserve_files) );

        $channel_successes++;
        dbg("channel: update complete");

	return 1;
      },

      'rollback' => sub {
      },
    },

  );

  my $error;
  my $eval_stat;
  for(my $elem = 0; $elem <= $#totry; $elem++) {
    my $success;
    eval {
      $success = &{$totry[$elem]->{'try'}}();  1;
    } or do {
      $eval_stat = $@ ne '' ? $@ : "errno=$!";  chomp $eval_stat;
    };
    if (!$success) {
      $error = $elem;
      $eval_stat = "step $elem unsuccessful"  if !defined $eval_stat;
      last;
    }
  }
  if (defined $error) {
    dbg("channel: channel failed, attempting rollback: %s", $eval_stat);
    for(my $elem = $error; $elem >= 0; $elem--) {
      &{$totry[$elem]->{'rollback'}}();
    }
  }
}

##############################################################################

# clean out the temp dir
if ($UPDTmp) {
  dbg("generic: cleaning up temporary directory/files");
  if (!clean_update_dir($UPDTmp)) {
    warn "error: unable to clean out the files in $UPDTmp\n";
  }
}

# clear out the temp files if they still exist
foreach ( $content_file, $UPDTmp ) {
  next unless defined $_;
  my $stat_errn = stat($_) ? 0 : 0+$!;
  next if $stat_errn == ENOENT;

  if ($stat_errn != 0) {
    warn "error: cannot access $_: $!\n";
  }
  elsif (-d _) {
    rmdir $_ or warn "error: cannot remove directory $_: $!\n";
  }
  elsif (-f _) {
    unlink $_ or warn "error: cannot remove file $_: $!\n";
  }
  else {
    warn "error: '$_' isn't a file nor a directory, skipping\n";
  }
}

my $exit;
if ($lint_failures) {
  # 2: lint of site pre files failed, cannot continue
  $exit = 2;
}
elsif (!$channel_failures) {
  # 0: updates found and successfully applied
  # 1: no updates were needed (success with nothing to do)
  $exit = $channel_successes ? 0 : 1;
}
else {  # at least some failures
  # 3: some failures, but at least one channel suceeded
  # 4 or higher means all channels failed
  $exit = $channel_successes ? 3 : 4;
}

dbg("diag: updates complete, exiting with code $exit");

if ($opt{'verbose'}) {
  if (!$exit) {
    if (defined $opt{'checkonly'}) {
      print "Update was available, but not installed in checkonly mode\n";
    } else {
      print "Update was available, and was downloaded and installed successfully\n";
    }
  } elsif ($exit == 1) {
    print "Update finished, no fresh updates were available\n";
  } elsif ($exit == 3) {
    print "Update of at least one channel finished, other updates failed\n";
  } else {
    print "Update failed, exiting with code $exit\n";
  }
}

exit $exit;

##############################################################################

sub read_install_file {
  my ($file) = @_;
  open (IN, "<$file") or die "cannot open $file\n";
  my $all;
  { local $/ = undef; $all = <IN> }
  close IN or die "cannot close $file: $!";
  return $all;
}

##############################################################################

sub write_channel_file {
  my ($filename, $contents) = @_;

  return 1 unless @{$contents};

  if (open(FILE, ">$filename")) {
    print FILE @{$contents} or die "cannot write to $filename: $!";
    close FILE or return 0;
    return 1;
  }

  return 0;
}

##############################################################################

sub channel_failed {
  my $reason = shift;
  warn("$reason, channel failed\n");
  $channel_failures++;
}

##############################################################################

sub taint_safe_archive_extract {
  my $todir = shift;
  my $input = shift;

  my $tfh = IO::Zlib->new($input, "rb");
  $tfh or die "fatal: couldn't read content tmpfile $content_file: $!\n";

  my $tar = Archive::Tar->new($tfh);
  $tar or die "fatal: couldn't open tar archive!\n";

  # stupid Archive::Tar is not natively taint-safe! duh.
  # return $tar->extract();
  # instead, get the file list, untaint, and extract one-by-one.

  my @files = $tar->list_files();
  foreach my $file (@files) {
    next if ($file =~ /^\/$/);  # ignore dirs

    local($1);
    $file =~ /^([-\.\,\/a-zA-Z0-9_]+)$/;
    my $outfname = $1;
    $outfname =~ s/\.\.\//__\//gs;      # avoid "../" dir traversal attacks
    $outfname = File::Spec->catfile($todir, $outfname);

    dbg "extracting: $outfname";
    if (!open OUT, ">".$outfname) {
      warn "error: failed to create $outfname: $!";
      goto failed;
    } else {
      my $content = $tar->get_content($file);

      if ($outfname =~ /\.(?:pre|cf)$/) {
        # replace macros in the update files if it's a .pre or .cf
        local($1);
        $content =~ s/\@\@([^\@]+)\@\@/$MACRO_VALUES{$1} || "\@\@$1\@\@"/ge;

        # also, if --allowplugins is not specified, comment out
        # all loadplugin or tryplugin lines
        if ( !$opt{'allowplugins'} ) {
          $content =~ s{^\s*((?:load|try)plugin|bayes_store_module|auto_whitelist_factory)\b}
            {#(commented by sa-update, no --allowplugins switch specified)# $1}gmx;
        }
      }

      print OUT $content
        or do { warn "error writing to $outfname: $!"; goto failed };
      close OUT
        or do { warn "error: write failed to $outfname: $!"; goto failed }
    }
  }

  return @files;

failed:
  return;       # undef = failure
}

##############################################################################

# Do a generic DNS query
sub do_dns_query {
  my($query, $rr_type) = @_;
  $rr_type = 'TXT'  if !defined $rr_type;

  my $RR = $res->query($query, $rr_type);
  my @result;

  # NOTE:  $rr->rdatastr returns the result encoded in a DNS zone file
  # format, i.e. enclosed in double quotes if a result contains whitespace
  # (or other funny characters), and may use \DDD encoding or \X quoting as
  # per RFC 1035.  Using $rr->txtdata instead avoids this unnecessary encoding
  # step and a need for decoding by a caller, returning an unmodified string.
  # Caveat: in case of multiple RDATA <character-string> fields contained
  # in a resource record (TXT, SPF, HINFO), starting with Net::DNS 0.69
  # the $rr->txtdata in a list context returns these strings as a list.
  # The $rr->txtdata in a scalar context always returns a single string
  # with <character-string> fields joined by a single space character as
  # a separator.  The $rr->txtdata in Net::DNS 0.68 and older returned
  # such joined space-separated string even in a list context.
  #
  # From Net::DNS maintainers (Willem Toorop, NLnet Labs):
  #   I encourage you to use txtdata for getting the values of
  # <version>.updates.spamassassin.org and mirros.updates.spamassassin.org.
  # As those records have only a single rdata field, txtdata would return
  # the same value since Net::DNS 0.34.
  #
  if ($RR) {
    foreach my $rr ($RR->answer) {
      next if !$rr;  # no answer records, only rcode
      next if $rr->type ne $rr_type;
      # scalar context!
      my $text = $rr->UNIVERSAL::can('txtdata') ? $rr->txtdata : $rr->rdatastr;
      push(@result,$text)  if defined $text && $text ne '';
    }
    printf("DNS %s query: %s -> %s\n", $rr_type, $query, join(", ",@result))
      if $opt{'verbose'} && $opt{'verbose'} > 1;
  }
  else {
    dbg("dns: query failed: $query => " . $res->errorstring);
    printf("DNS %s query %s failed: %s\n", $rr_type, $query, $res->errorstring)
      if $opt{'verbose'} && $opt{'verbose'} > 1;
  }

  return @result;
}

##############################################################################

sub init_lwp {
  if ($have_inet6 &&
      (!$opt{'force_pf'} || $opt{'force_pf'} eq 'inet6') &&
      ($io_socket_module_name eq 'IO::Socket::IP' ||
       $io_socket_module_name eq 'IO::Socket::INET6') )
  {
    # LWP module has no support for IPv6.  Use hotpatching,
    # copying IO::Socket::IP or IO::Socket::INET6 to IO::Socket::INET.
    # 'Borrowed' from Net::INET6Glue::INET_is_INET6 :

    printf("http: (lwp) hotpatching IO::Socket::INET by module %s\n",
           $io_socket_module_name) if $opt{'verbose'};
    my $io_socket_module_hash_name = $io_socket_module_name . '::';
    my $io_socket_module_path = $io_socket_module_name . '.pm';
    $io_socket_module_path =~ s{::}{/}g;
    $INC{'IO/Socket/INET.pm'} = $INC{$io_socket_module_path};
    no strict 'refs';
    no warnings 'redefine';
    for ( keys %{$io_socket_module_hash_name} ) {
      ref(my $v = $io_socket_module_hash_name->{$_}) and next;
      *{ 'IO::Socket::INET::'.$_ } =
        \&{ $io_socket_module_hash_name . $_ } if *{$v}{CODE};
    }
  }
  my $ua = LWP::UserAgent->new();
  $ua->agent("sa-update/$VERSION/$SAVersion");
  $ua->timeout(60);      # a good long timeout; 10 is too short for Coral!
  $ua->env_proxy;

# if ($opt{'force_pf'}) {
#   # No longer needed and can be harmful as we don't know which address family
#   # will be picked by the IO::Socket::* module in case of multihomed servers.
#   # The IO::Socket::IP should choose the right protocol family automatically.
#   if ($have_inet4 && $opt{'force_pf'} eq 'inet') {
#     $ua->local_address('0.0.0.0');
#   } elsif ($have_inet6 && $opt{'force_pf'} eq 'inet6') {
#     $ua->local_address('::');
#   }
# }

  return $ua;
}

# Do a GET request via HTTP for a certain URL
# Use the optional time_t value to do an IMS GET
sub http_get_lwp {
  my($url, $ims, $dir) = @_;

  $have_lwp  or die "http_get_lwp: module LWP not available";
  $ua = init_lwp()  if !$ua;

  my $response;
  my $text;

  # retry 3 times; this works better with Coral
  foreach my $retries (1 .. 3) {
    my $request = HTTP::Request->new("GET");
    $request->url($url);

    if (defined $ims) {
      my $str = time2str($ims);
      $request->header('If-Modified-Since', $str);
      dbg("http: IMS GET request, $url, $str");
    }
    else {
      dbg("http: GET request, $url");
    }

    $response = $ua->request($request);

    printf("http: (lwp) %sGET %s, %s\n",
           defined $ims ? 'IMS ' : '',  $url,
           !$response ? '(no response)' : $response->status_line )
           if $opt{'verbose'};

    if ($response->is_success) {
      return $response->content;
    }

    # could be a "304 not modified" or similar.
    # TODO: should use a special return type for "not modified" here
    # instead of overloading the failure return type
    if ($ims && $response->status_line =~ /^3/) {
      return;
    }
    if ($response->status_line =~ /^[45]/) {
      # client error or server error, makes no sense retrying
      return;
    }

    # include the text in the debug output; it's useful in some cases,
    # e.g. proxies that require authentication, diagnosing fascist
    # filtering false positives, etc.
    $text = $response->content;
    $text ||= "(no body)";
    $text =~ s/\s+/ /gs;

    dbg ("http: GET $url request failed, retrying: " .
                    $response->status_line.": ".$text);
  }

  # this should be a user-readable warning without --debug
  warn "http: GET $url request failed: " .
                    $response->status_line.": ".$text."\n";
  return;
}

# Do a GET request via HTTP for a given URL using an external program,
# or fall back to LWP if no external downloading program is available.
# Use the optional time_t value to do an IMS GET
sub http_get {
  my($url, $dir, $suggested_out_fname, $force_reload) = @_;
  my $content;

  my $out_fname;
  if (defined $suggested_out_fname) {
    $out_fname = $suggested_out_fname;
  } else {
    local $1;
    $url =~ m{([^/]*)\z}s;
    my $url_fname = $1;
    $out_fname = File::Spec->catfile($dir, $url_fname);
  }
  $out_fname = untaint_var($out_fname);

  # construct a short filename, relative to a current working directory $dir
  my $out_fname_short = $out_fname;
  $out_fname_short =~ s{^\Q$dir\E/*}{};

  printf("fetching %s\n", $url)  if $opt{'verbose'} && $opt{'verbose'} > 1;
  dbg("http: url: %s", $url);

  my $out_fname_exists = -e $out_fname;
  dbg("http: downloading to: %s, %s", $out_fname,
      !$out_fname_exists ? 'new' : $force_reload ? 'replace' : 'update');

  my($ext_prog, $cmd, @args);
  foreach my $try_prog ('curl', 'wget', 'fetch') {
    $cmd = Mail::SpamAssassin::Util::find_executable_in_env_path($try_prog);
    if (defined $cmd && $cmd ne '') { $ext_prog = $try_prog; last }
  }

  if (defined $ext_prog && $ext_prog eq 'curl') {
    push(@args, qw(-s -L -O --remote-time -g --max-redirs 2
                   --connect-timeout 30 --max-time 300
                   --fail -o), $out_fname_short);
    push(@args, '-z', $out_fname_short)  if $out_fname_exists && !$force_reload;
  } elsif (defined $ext_prog && $ext_prog eq 'wget') {
    push(@args, qw(-q --max-redirect=2 --tries=3
                   --dns-timeout=20 --connect-timeout=30 --read-timeout=300));
    push(@args, defined $suggested_out_fname ? ('-O', $out_fname_short)
                : $force_reload ? () : ('-N') );
  } elsif (defined $ext_prog && $ext_prog eq 'fetch') {
    push(@args, qw(-q -n -a -w 20 -m -o), $out_fname_short);
    push(@args, '-m')  if $out_fname_exists && !$force_reload;
  } elsif ($have_lwp) {
    dbg("http: no external tool for download, fallback to using LWP");
    my $ims;
    if ($out_fname_exists && !$force_reload) {
      my @out_fname_stat = stat($out_fname);
      my $size = $out_fname_stat[7];
      $ims = $out_fname_stat[9]  if $size;  # only if nonempty
    }
    my $out_fh = IO::File->new;
    $out_fh->open($out_fname,'>',0640)
      or die "Cannot create a file $out_fname: $!";
    binmode($out_fh) or die "Can't set binmode on $out_fname: $!";
    $content = http_get_lwp($url, $ims, $dir);
    if (!defined $content) {
      dbg("http: (lwp) no content downloaded from %s", $url);
    } else {
      $out_fh->print($content) or die "Error writing to $out_fname: $!";
    }
    $out_fh->close or die "Error closing file $out_fname: $!";
    return ($content, $out_fname);
  } else {
    die "http: no downloading tool available";
  }

  # only reached if invoking an external program is needed (not lwp)
  if ($opt{'force_pf'}) {
    if    ($opt{'force_pf'} eq 'inet')  { push(@args, '-4') }
    elsif ($opt{'force_pf'} eq 'inet6') { push(@args, '-6') }
  }
  push(@args, '--', untaint_var($url));
  dbg("http: %s", join(' ',$cmd,@args));

  # avoid a system() call, use fork/exec to make sure we avoid invoking a shell
  my $pid;
  eval {
    # use eval, the fork() sometimes signals an error
    # instead of returning a failure status
    $pid = fork(); 1;
  } or do { $@ = "errno=$!" if $@ eq ''; chomp $@; die "http fork: $@" };
  defined $pid or die "spawning $cmd failed: $!";
  if (!$pid) {  # child
    chdir($dir) or die "Can't chdir to $dir: $!";
    $cmd = untaint_file_path($cmd);
    exec {$cmd} ($cmd,@args);
    die "failed to exec $cmd: $!";
  }

  # parent
  waitpid($pid,0);
  my $child_stat = $?;
  dbg("http: process [%s], exit status: %s",
      $pid, exit_status_str($child_stat,0));

  if (!$opt{'verbose'}) {
    # silent
  } elsif ($child_stat == 0) {
    printf("http: (%s) GET %s, success\n", $ext_prog, $url);
  } else {
    printf("http: (%s) GET %s, FAILED, status: %s\n",
           $ext_prog, $url, exit_status_str($child_stat,0));
  }

  if ($child_stat == 0) {
    my $inp = IO::File->new;
    $inp->open($out_fname,'<') or die "Cannot open file $out_fname: $!";
    my($nbytes,$buff); $content = '';
    while (($nbytes=$inp->read($buff,16384)) > 0) { $content .= $buff }
    defined $nbytes or die "Error reading from file $out_fname: $!";
    $inp->close or die "Error closing $out_fname: $!";
  }

  return ($content, $out_fname);
}

##############################################################################

# choose a random integer between 0 and the total weight of all mirrors
# loop through the mirrors from largest to smallest weight
# if random number is < largest weight, use it
# otherwise, random number -= largest, remove mirror from list, try again
# eventually, there'll just be 1 mirror left in $mirrors[0] and it'll be used
# 
sub choose_mirror {
  my($mirror_list) = @_;

  # Sort the mirror list by reverse weight (largest first)
  my @mirrors = sort { $mirror_list->{$b}->{weight} <=> $mirror_list->{$a}->{weight} } keys %{$mirror_list};

  return unless @mirrors;

  if (keys %{$mirror_list} > 1) {
    # Figure out the total weight
    my $weight_total = 0;
    foreach (@mirrors) {
      $weight_total += $mirror_list->{$_}->{weight};
    }

    # Pick a random int
    my $value = int(rand($weight_total));

    # loop until we find the right mirror, or there's only 1 left
    while (@mirrors > 1) {
      if ($value < $mirror_list->{$mirrors[0]}->{weight}) {
        last;
      }
      $value -= $mirror_list->{$mirrors[0]}->{weight};
      shift @mirrors;
    }
  }

  return $mirrors[0];
}

##############################################################################

sub print_version {
  printf("sa-update version %s\n  running on Perl version %s\n", $VERSION, 
         join(".", map( 0+($_||0), ( $] =~ /(\d)\.(\d{3})(\d{3})?/ ))));
}

##############################################################################

sub print_usage_and_exit {
  my ( $message, $exitval ) = @_;
  $exitval ||= 64;
  
  if ($exitval == 0) {
    print_version();
    print("\n");
  }
  pod2usage(
    -verbose => 0,
    -message => $message,
    -exitval => $exitval,
  );
}

##############################################################################

sub usage {
  my ( $verbose, $message ) = @_;
  print "sa-update version $VERSION\n";
  pod2usage( -verbose => $verbose, -message => $message, -exitval => 64 );
}

##############################################################################

sub interpolate_gpghomedir {
  my $gpghome = '';
  if ($opt{'gpghomedir'}) {
    $gpghome = $opt{'gpghomedir'};
    if (am_running_on_windows()) {
      # windows is single-quote-phobic; bug 4958 cmt 7
      $gpghome =~ s/\"/\\\"/gs;
      $gpghome = "--homedir=\"$gpghome\"";
    } else {
      $gpghome =~ s/\'/\\\'/gs;
      $gpghome = "--homedir='$gpghome'";
    }
  }
  return $gpghome;
}

##############################################################################

sub import_gpg_key {
  my $keyfile = shift;

  my $gpghome = interpolate_gpghomedir();

  my $CMD = "$GPGPath $gpghome --batch ".
      "--no-tty --status-fd=1 -q --logger-fd=1 --import";

  unless (open(CMD, "$CMD $keyfile|")) {
    die "fatal: couldn't execute $GPGPath: $!\n";
  }

  # Determine the fate of the signature
  while(my $GNUPG = <CMD>) {
    chop $GNUPG;
    dbg ("gpg: $GNUPG");

    if ($GNUPG =~ /^gpg: /) {
      warn $GNUPG."\n";        # report bad news
    }

    if ($GNUPG =~ /^IMPORTED /) {
      dbg("gpg: gpg key imported successfully");
    }
  }

  my $errno = 0;  close CMD or $errno = $!;
  proc_status_ok($?,$errno)
    or warn("gpg: process '$CMD' finished: ".exit_status_str($?,$errno)."\n");

  return ($? >> 8);
}

##############################################################################

sub import_default_keyring {
  my $defkey = File::Spec->catfile ($DEF_RULES_DIR, "sa-update-pubkey.txt");
  unless (-f $defkey) {
    dbg("gpg: import of default keyring failed, couldn't find sa-update-pubkey.txt");
    return;
  }

  dbg("gpg: importing default keyring to ".$opt{gpghomedir});
  unless (-d $opt{gpghomedir}) {
    # use 0700 to avoid "unsafe permissions" warning
    mkpath([$opt{'gpghomedir'}], 0, 0700)
      or die "cannot mkpath $opt{gpghomedir}: $!";
  } 
  import_gpg_key($defkey);
}

##############################################################################

sub is_valid_gpg_key_id {
  # either a keyid (8 bytes) or a fingerprint (40 bytes)
  return ($_[0] =~ /^[a-fA-F0-9]+$/ && (length $_[0] == 8 || length $_[0] == 40));
}

##############################################################################

sub clean_update_dir {
  my($dir, $preserve_files_ref) = @_;

  dbg("generic: cleaning directory %s", $dir);
  unless (opendir(DIR, $dir)) {
    warn "error: cannot opendir $dir: $!\n";
    dbg("generic: attempt to opendir ($dir) failed");
    return;
  }
  while(my $file = readdir(DIR)) {
    next if $file eq '.' || $file eq '..';

    my $path = File::Spec->catfile($dir, $file);
    if ($preserve_files_ref && $preserve_files_ref->{$path}) {
      dbg("generic: preserving $file");
      next;
    }
    untaint_var(\$path);
    next unless -f $path;

    dbg("generic: unlinking $file");
    if (!unlink $path) {
      warn "error: cannot remove file $path: $!\n";
      closedir(DIR) or die "cannot close directory $dir: $!";
      return;
    }
  }
  closedir(DIR) or die "cannot close directory $dir: $!";
  return 1;
}

sub delete_files {
  my(@filenames) = @_;
  foreach my $path (@filenames) {
    dbg("generic: unlinking $path");
    unlink $path  or warn "error: cannot unlink file $path: $!\n";
  }
  return 1;
}

##############################################################################

sub lint_check_dir {
  my $dir = shift;

  # due to the Logger module's globalness (all M::SA objects share the same
  # Logger setup), we can't change the debug level here to only include
  # "config" or otherwise be more terse. :(
  my $spamtest = new Mail::SpamAssassin( {
    rules_filename       => $dir,
    site_rules_filename  => $LOCAL_RULES_DIR,
    ignore_site_cf_files => 1,
    userprefs_filename   => File::Spec->catfile($dir, "doesnotexist"),

    local_tests_only     => 1,
    dont_copy_prefs      => 1,

    PREFIX               => $PREFIX,
    DEF_RULES_DIR        => $DEF_RULES_DIR,
    LOCAL_RULES_DIR      => $LOCAL_RULES_DIR,
    LOCAL_STATE_DIR      => $LOCAL_STATE_DIR,
  });

  # need to kluge disabling bayes since it may try to expire the DB, and
  # without the proper config it's not going to be good.
  $spamtest->{conf}->{use_bayes} = 0;

  my $res = $spamtest->lint_rules();
  $spamtest->finish();

  return $res == 0;
}

##############################################################################

=head1 NAME

sa-update - automate SpamAssassin rule updates

=head1 SYNOPSIS

B<sa-update> [options]

Options:

  --channel channel       Retrieve updates from this channel
                          Use multiple times for multiple channels
  --channelfile file      Retrieve updates from the channels in the file
  --checkonly             Check for update availability, do not install
  --install filename      Install updates directly from this file. Signature
                          verification will use "file.asc" and "file.sha1"
  --allowplugins          Allow updates to load plugin code
  --gpgkey key            Trust the key id to sign releases
                          Use multiple times for multiple keys
  --gpgkeyfile file       Trust the key ids in the file to sign releases
  --gpghomedir path       Store the GPG keyring in this directory
  --gpg and --nogpg       Use (or do not use) GPG to verify updates
                          (--gpg is assumed by use of the above
                          --gpgkey and --gpgkeyfile options)
  --import file           Import GPG key(s) from file into sa-update's
                          keyring. Use multiple times for multiple files
  --updatedir path        Directory to place updates, defaults to the
                          SpamAssassin site rules directory
                          (default: /var/lib/spamassassin/3.004001)
  --refreshmirrors        Force the MIRRORED.BY file to be updated
  -D, --debug [area=n,...]  Print debugging messages
  -v, --verbose           Be verbose, like print updated channel names;
                          For more verbosity specify multiple times
  -V, --version           Print version
  -h, --help              Print usage message
  -4                      Force using the inet protocol (IPv4), not inet6
  -6                      Force using the inet6 protocol (IPv6), not inet

=head1 DESCRIPTION

sa-update automates the process of downloading and installing new rules and
configuration, based on channels.  The default channel is
I<updates.spamassassin.org>, which has updated rules since the previous
release.

Update archives are verified using SHA1 hashes and GPG signatures, by default.

Note that C<sa-update> will not restart C<spamd> or otherwise cause
a scanner to reload the now-updated ruleset automatically.  Instead,
C<sa-update> is typically used in something like the following manner:

        sa-update && /etc/init.d/spamassassin reload

This works because C<sa-update> only returns an exit status of C<0> if
it has successfully downloaded and installed an updated ruleset.

The program sa-update uses the underlying operating system umask for the
updated rule files it installs.  You may wish to run sa-update from a script
that sets the umask prior to calling sa-update.  For example:

	#!/bin/sh
	umask 022
	sa-update

=head1 OPTIONS

=over 4

=item B<--channel>

sa-update can update multiple channels at the same time.  By default, it will
only access "updates.spamassassin.org", but more channels can be specified via
this option.  If there are multiple additional channels, use the option
multiple times, once per channel.  i.e.:

	sa-update --channel foo.example.com --channel bar.example.com

=item B<--channelfile>

Similar to the B<--channel> option, except specify the additional channels in a
file instead of on the commandline.  This is useful when there are a
lot of additional channels.

=item B<--checkonly>

Only check if an update is available, don't actually download and install it.
The exit code will be C<0> or C<1> as described below.

=item B<--install>

Install updates "offline", from the named tar.gz file, instead of performing
DNS lookups and HTTP invocations.

Files named B<file>.sha1 and B<file>.asc will be used for the SHA-1 and GPG
signature, respectively.  The filename provided must contain a version number
of at least 3 digits, which will be used as the channel's update version
number.

Multiple B<--channel> switches cannot be used with B<--install>.  To install
multiple channels from tarballs, run C<sa-update> multiple times with different
B<--channel> and B<--install> switches, e.g.:

        sa-update --channel foo.example.com --install foo-34958.tgz
        sa-update --channel bar.example.com --install bar-938455.tgz

=item B<--allowplugins>

Allow downloaded updates to activate plugins.  The default is not to
activate plugins; any C<loadplugin> or C<tryplugin> lines will be commented
in the downloaded update rules files.

=item B<--gpg>, B<--nogpg>

sa-update by default will verify update archives by use of a SHA1 checksum
and GPG signature.  SHA1 hashes can verify whether or not the downloaded
archive has been corrupted, but it does not offer any form of security
regarding whether or not the downloaded archive is legitimate (aka:
non-modifed by evildoers).  GPG verification of the archive is used to
solve that problem.

If you wish to skip GPG verification, you can use the B<--nogpg> option
to disable its use.  Use of the following gpgkey-related options will
override B<--nogpg> and keep GPG verification enabled.

Note: Currently, only GPG itself is supported (ie: not PGP).  v1.2 has been
tested, although later versions ought to work as well.

=item B<--gpgkey>

sa-update has the concept of "release trusted" GPG keys.  When an archive is
downloaded and the signature verified, sa-update requires that the signature
be from one of these "release trusted" keys or else verification fails.  This
prevents third parties from manipulating the files on a mirror, for instance,
and signing with their own key.

By default, sa-update trusts key ids C<24F434CE> and C<5244EC45>, which are
the standard SpamAssassin release key and its sub-key.  Use this option to
trust additional keys.  See the B<--import> option for how to add keys to
sa-update's keyring.  For sa-update to use a key it must be in sa-update's
keyring and trusted.

For multiple keys, use the option multiple times.  i.e.:

	sa-update --gpgkey E580B363 --gpgkey 298BC7D0

Note: use of this option automatically enables GPG verification.

=item B<--gpgkeyfile>

Similar to the B<--gpgkey> option, except specify the additional keys in a file
instead of on the commandline.  This is extremely useful when there are a lot
of additional keys that you wish to trust.

=item B<--gpghomedir>

Specify a directory path to use as a storage area for the C<sa-update> GPG
keyring.  By default, this is

        /etc/mail/spamassassin/sa-update-keys

=item B<--import>

Use to import GPG key(s) from a file into the sa-update keyring which is
located in the directory specified by B<--gpghomedir>.  Before using channels
from third party sources, you should use this option to import the GPG key(s)
used by those channels.  You must still use the B<--gpgkey> or B<--gpgkeyfile>
options above to get sa-update to trust imported keys.

To import multiple keys, use the option multiple times.  i.e.:

	sa-update --import channel1-GPG.KEY --import channel2-GPG.KEY

Note: use of this option automatically enables GPG verification.

=item B<--refreshmirrors>

Force the list of sa-update mirrors for each channel, stored in the MIRRORED.BY
file, to be updated.  By default, the MIRRORED.BY file will be cached for up to
7 days after each time it is downloaded.

=item B<--updatedir>

By default, C<sa-update> will use the system-wide rules update directory:

        /var/lib/spamassassin/3.004001

If the updates should be stored in another location, specify it here.

Note that use of this option is not recommended; if you're just using sa-update
to download updated rulesets for a scanner, and sa-update is placing updates in
the wrong directory, you probably need to rebuild SpamAssassin with different
C<Makefile.PL> arguments, instead of overriding sa-update's runtime behaviour.

=item B<-D> [I<area,...>], B<--debug> [I<area,...>]

Produce debugging output.  If no areas are listed, all debugging information is
printed.  Diagnostic output can also be enabled for each area individually;
I<area> is the area of the code to instrument. For example, to produce
diagnostic output on channel, gpg, and http, use:

        sa-update -D channel,gpg,http

For more information about which areas (also known as channels) are
available, please see the documentation at
L<http://wiki.apache.org/spamassassin/DebugChannels>.

=item B<-h>, B<--help>

Print help message and exit.

=item B<-V>, B<--version>

Print sa-update version and exit.

=back

=head1 EXIT CODES

In absence of a --checkonly option, an exit code of C<0> means:
an update was available, and was downloaded and installed successfully.
If --checkonly was specified, an exit code of C<0> means: an update was
available.

An exit code of C<1> means no fresh updates were available.

An exit code of C<2> means that at least one update is available but that a
lint check of the site pre files failed.  The site pre files must pass a lint
check before any updates are attempted.

An exit code of C<3> means that at least one update succeeded while 
other channels failed.  If using sa-compile, you should proceed with it.

An exit code of C<4> or higher, indicates that errors occurred while
attempting to download and extract updates, and no channels were updated.

=head1 SEE ALSO

Mail::SpamAssassin(3)
Mail::SpamAssassin::Conf(3)
spamassassin(1)
spamd(1)
<http://wiki.apache.org/spamassassin/RuleUpdates>

=head1 PREREQUESITES

C<Mail::SpamAssassin>

=head1 BUGS

See <http://issues.apache.org/SpamAssassin/>

=head1 AUTHORS

The Apache SpamAssassin(tm) Project <http://spamassassin.apache.org/>

=head1 COPYRIGHT

SpamAssassin is distributed under the Apache License, Version 2.0, as
described in the file C<LICENSE> included with the distribution.

=cut