Mailing List Archive

Author: felicity
Date: Fri Sep 1 19:37:06 2006
New Revision: 439524

URL: http://svn.apache.org/viewvc?rev=439524&view=rev
Log:
more cleanup of mass-check

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=439524&r1=439523&r2=439524&view=diff
==============================================================================
--- spamassassin/branches/tvd-multi-mass-check/masses/mass-check (original)
+++ spamassassin/branches/tvd-multi-mass-check/masses/mass-check Fri Sep 1 19:37:06 2006
@@ -1100,7 +1100,6 @@
# if we've been waiting for this result, process it, otherwise throw it on
# the ground. multiple clients could have been given the same messages to
# process, and we take whatever the first responder sends us.
-#print "received result for $k\n";
if (exists $msgsout->{$k}) {
my @d = Mail::SpamAssassin::ArchiveIterator::index_unpack($msgsout->{$k}->{'data'});

@@ -1115,7 +1114,6 @@

# if we got any results, clean out the results from the timestamp arrays
while ( my($k,$v) = each %timestamps ) {
-#print "dealing with timestamp $k\n";
my @temp = grep(!exists $v->{$_}, @{$timestamps->{$k}});

# if there are results left for a specific timestamp, update the array
@@ -1137,7 +1135,6 @@
my $total_count = 0;

while (($total_messages > $total_count) && ($message = read_line($tmpfd))) {
-#print ">>> $message\n";
my($class, undef, $date, undef, $result) = $iter->run_message($message);
if ($result) {
if ($opt_client) {
@@ -1146,7 +1143,6 @@
else {
result($class, $result, $date);
}
-#print ">>> $result\n";
}
$total_count++;
}
@@ -1279,7 +1275,7 @@
}
} while ($line !~ /^$/);

- return if ($headers{'finished'});
+ return 'finished' if ($headers{'finished'});

my $gzpath = '';
if ($headers{'content-length'}) {
@@ -1297,10 +1293,11 @@
return $gzpath;
}

-# Be conservative -- anything that's non-alphanumeric, encode!
+# Be conservative -- encode most things.
+# we could encode spaces to plusses, then decode that later, but...
sub post_encode {
my $string = shift;
- $string =~ s/(\W)/sprintf "%%%02x",unpack("C",$1)/egx;
+ $string =~ s/([^a-zA-Z0-9_,.\/\\-])/sprintf "%%%02x",unpack("C",$1)/egx;
return $string;
}

@@ -1351,13 +1348,13 @@

sub send_line {
my $fd = shift;
-
foreach ( @_ ) {
my $length = pack("V", length $_);
- $fd->print($length.$_);
+ $fd->print($length.$_);
}
}

+############################################################################

sub server_mode {
$opt_cs_max ||= 1000;
@@ -1404,15 +1401,13 @@
$messages = generate_messages($msgnum, $timestamps, $msgsout, $postdata->{'paths_only'});
}

-#print ">> sending messages\n";
-
if ($messages && open(MSG, $messages)) {
binmode(MSG);
local $/ = undef;

- # Content-Encoding: gzip
http_response($socket, "200 OK", {
- "Content-type" => "application/octet-stream",
+ 'Content-Type' => 'application/x-gzip',
+ 'Content-Encoding' => 'x-gzip',
"Content-Length" => (-s $messages),
},
scalar <MSG>);
@@ -1490,6 +1485,8 @@
# last if connection fails
last unless ($socket);

+ print "Requesting $msgnum messages from server\n";
+
# make request, include and drop results if there are any
my $result = http_make_request($socket, 'POST', $uri, {
'Host' => $http_host,
@@ -1507,16 +1504,19 @@
# we got an error?!? abort!
last;
}
- elsif (!$result) {
+ elsif ($result eq 'finished') {
+ # the server said that we're done
+ print "Server states that there is no more work, exiting.\n";
+ last;
+ }
+ elsif ($result eq '') {
# sleep for client_timeout seconds and try the request again
-print "Got no response, waiting $opt_cs_timeout seconds\n";
+ print "Received no messages from server, waiting $opt_cs_timeout seconds\n";
sleep $opt_cs_timeout;
}
else {
my $time_start = time;

-print "Got response: $result\n";
-
%postdata = ();
%real = ();
$total_count = $spam_count = $ham_count = 0;
@@ -1528,7 +1528,6 @@
# used for the temp queue file
my $tmppath;
($tmppath, $tmpfd) = Mail::SpamAssassin::Util::secure_tmpfile();
-print ">> $tmppath\n";
die "Can't make tempfile, exiting" unless $tmppath;

clean_dir($tmpdir) if ($tmpdir);
@@ -1537,21 +1536,20 @@
# 1- server message number in text format
# 2- server index string, binary packed format
# 3- message content
-print ">> writing out files\n";

# number of messages
$msgnum = $total_messages = read_line($gzfd);

-print ">> total of $total_messages messages\n";
+ print "Received $total_messages messages from the server\n";

for(my $i = 0 ; $i < $total_messages; $i++ ) {
my $num = read_line($gzfd);
last unless defined $num;
-#print "read in message $num\n";
+
my $index = read_line($gzfd);
-#print "read in index $index\n";
last unless defined $index;
-#print "output message $num\n";
+
+ # if we're doing paths_only, there'll be no message content
if (!$opt_cs_paths_only) {
my $msg = read_line($gzfd);
last unless defined $msg;
@@ -1575,15 +1573,12 @@
$real{$d[3]} = $num;
send_line($tmpfd, $index);
}
-#print "wrote mess $num\n";
}

-print "exited loop\n";
-
$gzfd->close;
unlink $result;

-print "beginning run\n";
+ print "Starting run\n";

# we're about to start running, so go back to the start of the file
seek $tmpfd, 0, 0;
@@ -1592,15 +1587,14 @@

unlink $tmppath;

-print "ended run\n";
# figure out new max messages, try keeping ~cs_timeout between runs
my $time_end = time;
if ($time_end == $time_start) {
$time_end++;
}
-print "ran $msgnum messages in ".($time_end-$time_start)." seconds\n";
+
+ print "Completed run in ".($time_end-$time_start)." seconds\n";
$msgnum = int($msgnum * $opt_cs_timeout / ($time_end-$time_start)) || 1;
-print "now requesting $msgnum messages\n";
}
}

@@ -1612,6 +1606,8 @@
}
}

+############################################################################
+
sub wanted_server {
my ($class, $id, $time, $dataref, $format) = @_;
return $dataref;
@@ -1637,6 +1633,6 @@
$postdata{$1} = $result;
}
else {
- warn ">> result is not in the correct format: $result\n";
+ warn ">> WTH!? result is not in the correct format: $result\n";
}
}