Mailing List Archive

Author: felicity
Date: Sat Aug 26 20:23:43 2006
New Revision: 437290

URL: http://svn.apache.org/viewvc?rev=437290&view=rev
Log:
server mode will actually send out mails now

Modified:
spamassassin/branches/tvd-multi-mass-check/masses/mass-check

Modified: spamassassin/branches/tvd-multi-mass-check/masses/mass-check
URL: http://svn.apache.org/viewvc/spamassassin/branches/tvd-multi-mass-check/masses/mass-check?rev=437290&r1=437289&r2=437290&view=diff
==============================================================================
--- spamassassin/branches/tvd-multi-mass-check/masses/mass-check (original)
+++ spamassassin/branches/tvd-multi-mass-check/masses/mass-check Sat Aug 26 20:23:43 2006
@@ -1,4 +1,7 @@
#!/usr/bin/perl -w
+use strict;
+eval { use IO::Zlib 1.04; };
+
#
# <@LICENSE>
# Licensed to the Apache Software Foundation (ASF) under one or more
@@ -20,7 +23,7 @@
sub usage {
my $status = shift;

- my $out = $status ? STDERR : STDOUT;
+ my $out = $status ? \*STDERR : \*STDOUT;
print $out <<EOF;
usage: mass-check [options] target ...

@@ -96,6 +99,7 @@
$opt_logmem $opt_after $opt_before $opt_rewrite $opt_deencap
$opt_learn $opt_reuse $opt_lint $opt_cache $opt_noisy
$total_messages $statusevery $opt_cachedir $opt_server $opt_client
+ $opt_server_max $opt_server_timeout
$tmpfd %reuse %orig_conf %reuse_conf $reuse_rules_loaded_p);

use FindBin;
@@ -123,6 +127,8 @@
$opt_spamlog = "spam.log";
$opt_learn = 0;
$reuse_rules_loaded_p = 0;
+$opt_server_max = 1000;
+$opt_server_timeout = 300;

my @ORIG_ARGV = @ARGV;
GetOptions("c=s", "p=s", "f=s", "j=i", "n", "o", "all", "bayes", "debug:s",
@@ -130,7 +136,9 @@
"progress", "rewrite:s", "showdots", "spamlog=s", "tail=i",
"rules=s", "restart=i", "after=s", "before=s", "loguris",
"deencap=s", "logmem", "learn=i", "reuse", "lint", "cache",
- "cachedir=s", "noisy", "server", "client=s",
+ "cachedir=s", "noisy",
+ "server", "server_max=i", "server_timeout=i",
+ "client=s",
"dir" => sub { $opt_format = "dir"; },
"file" => sub { $opt_format = "file"; },
"mbox" => sub { $opt_format = "mbox"; },
@@ -195,7 +203,7 @@
usage(1) if !@targets;
}

-$spamtest = new Mail::SpamAssassin ({
+my $spamtest = new Mail::SpamAssassin ({
'debug' => $opt_debug,
'rules_filename' => $opt_c,
'userprefs_filename' => $user_prefs,
@@ -356,8 +364,17 @@
status('server ready for connections');
}

+ my $timestamps = {};
+ my $msgsout = { 'curnum' => 0 };
+
my $select = IO::Select->new( $serv_socket );

+ my($respath, $resfd) = Mail::SpamAssassin::Util::secure_tmpfile();
+ die 'archive-iterator: failed to create temp file' unless $respath;
+
+ # TVD: remove later
+ close($resfd); unlink $respath;
+
my $sent_messages = 1;
while ($select->count()) {
foreach my $socket ($select->can_read()) {
@@ -365,57 +382,58 @@
$select->add($serv_socket->accept);
}
else {
- my %headers = ();
- my %postdata = ();
- my $postdata = '';
-
- # read in the request
- # read in headers, "key: value"
- my $line = $socket->getline();
- $line =~ s/\r\n$//;
- my ($type) = $line =~ /^([a-zA-Z]+)/;
-
- # we don't really care about the request right now
- do {
- $line = $socket->getline();
- last unless defined $line;
- $line =~ s/\r\n$//;
-
- my ($k,$v) = split(/:\s*/, $line, 2);
- $headers{lc $k} = $v;
- } while ($line !~ /^$/);
-
- if ($headers{'content-length'}) {
- $socket->read($postdata, $headers{'content-length'});
- %postdata = map {
- my($k,$v) = split(/=/, $_, 2);
- $k =~ s/\%([0-9a-fA-F]{2})/sprintf "%c", hex($1)/e;
- $v =~ s/\%([0-9a-fA-F]{2})/sprintf "%c", hex($1)/e;
- $k => $v;
- } split(/\&/, $postdata);
- }
+ my($type, $URI, $headers, $postdata) = handle_http_request($socket);

- if (1|| uc $type eq 'POST') {
- my @results = $iter->run_message($iter->read_line($tmpfd));
- #return($class, $format, $date, $where, &{$self->{wanted_sub}}($class, $where, $date, \@msg, $format));
- # for error, "501 Not Implemented"
- # Content-Encoding: gzip
+ if ($type eq 'GET') {
print $socket join("\r\n",
"HTTP/1.0 200 OK",
"Content-type: text/plain",
"Pragma: no-cache",
"Server: mass-check/0.0",
"\r\n"),join("\n",
- "Your IP Address: ".$socket->peerhost,
- "There are $total_messages messages, and ".($total_messages-$sent_messages++)." left",
- $results[0],
- $results[1],
- $results[2],
- $results[3],
- ""),
- @{$results[4]};
- }
+ "Your GET request came from IP Address: ".$socket->peerhost,
+ "");
+ }
+ elsif ($type eq 'POST') {
+ handle_post_results($postdata, $resfd);
+
+ my $messages = '';
+ if ($postdata->{'max_messages'}) {
+ my $msgnum = $postdata->{'max_messages'};
+ $msgnum = $opt_server_max if ($msgnum > $opt_server_max);
+ $messages = generate_messages($msgnum, $timestamps, $msgsout);
+ }
+
+ if ($messages && open(MSG, $messages)) {
+ binmode(MSG);
+ local $/ = undef;
+
+ # Content-Encoding: gzip
+ print $socket join("\r\n",
+ "HTTP/1.0 200 OK",
+ "Content-type: application/octet-stream",
+ "Pragma: no-cache",
+ "Server: mass-check/0.0",
+ "Content-Length: ".(-s $messages),
+ "\r\n"),
+ <MSG>;
+
+ close(MSG);
+ unlink $messages;
+ }
+ else {
+ print $socket join("\r\n",
+ "HTTP/1.0 200 OK",
+ "Content-type: text/plain",
+ "Pragma: no-cache",
+ "Server: mass-check/0.0",
+ "\r\n"),join("\n",
+ "Your POST request (sans max_messages) came from IP Address: ".$socket->peerhost,
+ "");
+ }
+ }
else {
+ # for error, "501 Not Implemented"
print $socket join("\r\n",
"HTTP/1.0 501 Not Implemented",
"");
@@ -423,11 +441,18 @@

$select->remove($socket);
$socket->close;
- $select->remove($serv_socket) if ($line =~ /^quit/i);
}
}
+
+ deal_with_results($timestamps, $msgsout, $resfd);
+
+ # drop the listener when ready
+ # $select->remove($serv_socket) if ($line =~ /^quit/i);
}

+ close($resfd);
+ unlink $respath;
+
exit;
}

@@ -717,7 +742,7 @@
if (defined $spam) {
my $result = ($spam ? "spam" : "ham");
my $status = $spamtest->learn($ma, undef, $spam, 0);
- $learned = $status->did_learn();
+ my $learned = $status->did_learn();
$result = "undef" if !defined $learned;
push(@extra, "learn=".$result);
}
@@ -1074,8 +1099,9 @@
my $iter = shift;

my $tmpf;
- ($tmpf, $tmpfd) = Mail::SpamAssassin::Util::secure_tmpfile()
- or die 'archive-iterator: failed to create temp file';
+ ($tmpf, $tmpfd) = Mail::SpamAssassin::Util::secure_tmpfile();
+ die 'archive-iterator: failed to create temp file' unless $tmpf;
+
unlink $tmpf or die "archive-iterator: unlink '$tmpf': $!";
undef $tmpf;

@@ -1103,4 +1129,133 @@
if (!$total_messages) {
die "archive-iterator: no messages to process\n";
}
+}
+
+sub handle_http_request {
+ my $socket = shift;
+
+ my $headers = {};
+ my $postdata = {};
+
+ # read in the request
+ # read in headers, "key: value"
+ my $line = $socket->getline();
+ $line =~ s/\r\n$//;
+ my ($type, $URI, $VERS) = $line =~ /^([a-zA-Z]+)\s+(\S+)(?:\s*(\S+))/;
+ unless ($type && $URI && $VERS) {
+ $type ||= '';
+ $URI ||= '';
+
+ return ($type, $URI, $headers, $postdata);
+ }
+
+ $type = uc $type;
+
+ # we don't really care about the request right now
+ do {
+ $line = $socket->getline();
+ last unless defined $line;
+ $line =~ s/\r\n$//;
+
+ if ($line) {
+ my ($k,$v) = split(/:\s*/, $line, 2);
+ $headers->{lc $k} = $v;
+ }
+ } while ($line !~ /^$/);
+
+ if ($type eq 'POST' && $headers->{'content-length'}) {
+ my $pd;
+ $socket->read($pd, $headers->{'content-length'});
+ $pd =~ s/[\r\n]+$//;
+
+ %{$postdata} = map {
+ my($k,$v) = split(/=/, $_, 2);
+ $k =~ s/\%([0-9a-fA-F]{2})/sprintf "%c", hex($1)/e;
+ $v =~ s/\%([0-9a-fA-F]{2})/sprintf "%c", hex($1)/e;
+ $k => $v;
+ } split(/\&/, $pd);
+ }
+
+ return($type, $URI, $headers, $postdata);
+}
+
+sub generate_messages {
+ my($msgs, $timestamps, $msgsout) = @_;
+
+ my @tosend = ();
+
+ # Find out if any of the messages we sent out before need to be sent out
+ # again because we haven't seen a response yet.
+ my $tooold = time - $opt_server_timeout;
+ foreach (keys %{$timestamps}) {
+ next if ($_ > $tooold);
+ my $wanted = $msgs - @tosend;
+
+ if (@{$timestamps->{$_}} > $wanted) {
+ # there are more entries in the timestamp than we want
+ push(@tosend, splice @{$timestamps->{$_}}, 0, $wanted);
+ }
+ else {
+ # we're going to take all of this timestamp's entries
+ push(@tosend, @{$timestamps->{$_}});
+ delete $timestamps->{$_};
+ }
+
+ last if (@tosend == $msgs);
+ }
+
+ if ($tmpfd) {
+ while (@tosend < $msgs) {
+ my $msg = $iter->read_line($tmpfd);
+
+ # no more messages from the temp file, close it out
+ unless ($msg) {
+ delete $msgsout->{'curnum'};
+ close $tmpfd;
+ undef $tmpfd;
+ last;
+ }
+
+ my $num = $msgsout->{'curnum'}++;
+ $msgsout->{$num}->{'data'} = $msg;
+ push(@tosend, $num);
+ }
+ }
+
+ # ok, at this point, @tosend ought to have a list of numbers, pointers into
+ # %{$msgsout}. turn that into a tar file.
+ return '' unless @tosend;
+
+ my($gzpath, $gzfd) = Mail::SpamAssassin::Util::secure_tmpfile();
+ die "Can't make tempfile, exiting" unless $gzpath;
+ close($gzfd);
+
+ $gzfd = IO::Zlib->new($gzpath, 'wb') || die "Can't create temp gzip file: $!";
+
+ # Generate an archive in the temp file
+ foreach my $num (@tosend) {
+ my $msg = ($iter->run_message($msgsout->{$num}->{'data'}))[4];
+ $iter->send_line($gzfd, join('', @{$msg}));
+ }
+
+ $gzfd->close;
+
+ #return($class, $format, $date, $where, &{$self->{wanted_sub}}($class, $where, $date, \@msg, $format));
+
+ # update timestamp entries
+ my $ts = time;
+ foreach (@tosend) {
+ $msgsout->{$_}->{'timestamp'} = $ts;
+ }
+ $timestamps->{$ts} = \@tosend;
+
+ return $gzpath;
+}
+
+sub handle_post_results {
+ my($postdata, $resfd) = @_;
+}
+
+sub deal_with_results {
+ my($timestamps, $msgsout, $resfd) = @_;
}