Coro/myhttpd/httpd.pl

use Coro;
use Coro::Semaphore;
use Coro::Event;
use Coro::Socket;
use Coro::Signal;

use HTTP::Date;
use POSIX ();

use Compress::Zlib ();

no utf8;
use bytes;

# at least on my machine, this thingy serves files
# quite a bit faster than apache, ;)
# and quite a bit slower than thttpd :(

$SIG{PIPE} = 'IGNORE';

our $accesslog;
our $errorlog;

our $NOW;
our $HTTP_NOW;

Event->timer(interval => 1, hard => 1, cb => sub {
   $NOW = time;
   $HTTP_NOW = time2str $NOW;
})->now;

if ($ERROR_LOG) {
   use IO::Handle;
   open $errorlog, ">>$ERROR_LOG"
     or die "$ERROR_LOG: $!";
   $errorlog->autoflush(1);
}

if ($ACCESS_LOG) {
   use IO::Handle;
   open $accesslog, ">>$ACCESS_LOG"
     or die "$ACCESS_LOG: $!";
   $accesslog->autoflush(1);
}

sub slog {
   my $level = shift;
   my $format = shift;
   my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
   printf "$NOW: $format\n", @_;
   printf $errorlog "$NOW: $format\n", @_ if $errorlog;
}

our $connections = new Coro::Semaphore $MAX_CONNECTS || 250;
our $httpevent   = new Coro::Signal;

our $queue_file  = new transferqueue $MAX_TRANSFERS;
our $queue_index = new transferqueue 10;

our $tbf_top     = new tbf rate => $TBF_RATE || 100000;

my $unused_bytes = 0;
my $unused_last  = time;

sub unused_bandwidth {
   $unused_bytes += $_[0];
   if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
      $unused_last = $NOW;
      $unused_bytes = 0;
      $queue_file->force_wake_next;
      slog 1, "forced filetransfer due to unused bandwidth";
   }
}

my @newcons;
my @pool;

# one "execution thread"
sub handler {
   while () {
      if (@newcons) {
         eval {
            conn->new(@{pop @newcons})->handle;
         };
         slog 1, "$@" if $@ && !ref $@;

         $httpevent->broadcast; # only for testing, but doesn't matter much

         $connections->up;
      } else {
         last if @pool >= $MAX_POOL;
         push @pool, $Coro::current;
         schedule;
      }
   }
}

sub listen_on {
   my $listen = $_[0];

   push @listen_sockets, $listen;

   # the "main thread"
   async {
      slog 1, "accepting connections";
      while () {
         $connections->down;
         push @newcons, [$listen->accept];
         #slog 3, "accepted @$connections ".scalar(@pool);
         if (@pool) {
            (pop @pool)->ready;
         } else {
            async \&handler;
         }
      }
   };
}

my $http_port = new Coro::Socket
        LocalAddr => $SERVER_HOST,
        LocalPort => $SERVER_PORT,
        ReuseAddr => 1,
        Listen => 50,
   or die "unable to start server";

listen_on $http_port;

if ($SERVER_PORT2) {
   my $http_port = new Coro::Socket
           LocalAddr => $SERVER_HOST,
           LocalPort => $SERVER_PORT2,
           ReuseAddr => 1,
           Listen => 50,
      or die "unable to start server";

   listen_on $http_port;
}

package conn;

use Socket;
use HTTP::Date;
use Convert::Scalar 'weaken';
use IO::AIO;

IO::AIO::min_parallel $::AIO_PARALLEL;

Event->io(fd => IO::AIO::poll_fileno,
          poll => 'r', async => 1,
          cb => \&IO::AIO::poll_cb);

our %conn; # $conn{ip}{self} => connobj
our %uri;  # $uri{ip}{uri}{self}
our %blocked;
our %mimetype;

sub read_mimetypes {
   local *M;
   if (open M, "<mime_types") {
      while (<M>) {
         if (/^([^#]\S+)\t+(\S+)$/) {
            $mimetype{lc $1} = $2;
         }
      }
   } else {
      print "cannot open mime_types\n";
   }
}

read_mimetypes;

sub new {
   my $class = shift;
   my $fh = shift;
   my $peername = shift;
   my $self = bless { fh => $fh }, $class;
   my (undef, $iaddr) = unpack_sockaddr_in $peername
      or $self->err(500, "unable to decode peername");

   $self->{remote_addr} =
      $self->{remote_id} = inet_ntoa $iaddr;

   $self->{time} = $::NOW;

   weaken ($Coro::current->{conn} = $self);

   $::conns++;
   $::maxconns = $::conns if $::conns > $::maxconns;

   $self;
}

sub DESTROY {
   #my $self = shift;
   $::conns--;
}

sub prune_cache {
   my $hash = $_[0];

   for (keys %$hash) {
      if (ref $hash->{$_} eq HASH::) {
         prune_cache($hash->{$_});
         unless (scalar keys %{$hash->{$_}}) {
            delete $hash->{$_};
            $d2++;
         }
      }
   }
}

sub prune_caches {
   prune_cache \%conn;
   prune_cache \%uri;

   for (keys %blocked) {
      delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
   }
}

Event->timer(interval => 60, cb => \&prune_caches);

sub slog {
   my $self = shift;
   main::slog($_[0], "$self->{remote_id}> $_[1]");
}

sub response {
   my ($self, $code, $msg, $hdr, $content) = @_;
   my $res = "HTTP/1.1 $code $msg\015\012";
   my $GZ = "";

   if (exists $hdr->{Connection}) {
      if ($hdr->{Connection} =~ /close/) {
         $self->{h}{connection} = "close"
      }
   } else {
      if ($self->{version} < 1.1) {
         if ($self->{h}{connection} =~ /keep-alive/i) {
            $hdr->{Connection} = "Keep-Alive";
         } else {
            $self->{h}{connection} = "close"
         }
      }
   }

   if ($self->{method} ne "HEAD"
       && $self->{h}{"accept-encoding"} =~ /\bgzip\b/
       && 400 < length $content
       && $hdr->{"Content-Length"} == length $content
       && !exists $hdr->{"Content-Encoding"}
   ) {
      my $orig = length $content;
      $hdr->{"Content-Encoding"} = "gzip";
      $content = Compress::Zlib::memGzip(\$content);
      $hdr->{"Content-Length"} = length $content;
      $GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
   }

   $res .= "Date: $HTTP_NOW\015\012";
   $res .= "Server: $::NAME\015\012";

   while (my ($h, $v) = each %$hdr) {
      $res .= "$h: $v\015\012"
   }
   $res .= "\015\012";

   $res .= $content if defined $content and $self->{method} ne "HEAD";

   my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
             " $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
             " \"$self->{h}{referer}\"\n";

   print $::accesslog $log if $::accesslog;
   print STDERR $log;

   $tbf_top->request(length $res, 1e6);
   $self->{written} += print {$self->{fh}} $res;
}

sub err {
   my $self = shift;
   my ($code, $msg, $hdr, $content) = @_;

   unless (defined $content) {
      $content = "$code $msg\n";
      $hdr->{"Content-Type"} = "text/plain";
      $hdr->{"Content-Length"} = length $content;
   }
   $hdr->{"Connection"} = "close";

   $self->response($code, $msg, $hdr, $content);

   die bless {}, err::;
}

sub handle {
   my $self = shift;
   my $fh = $self->{fh};

   my $host;

   $fh->timeout($::REQ_TIMEOUT);
   while() {
      $self->{reqs}++;

      # read request and parse first line
      my $req = $fh->readline("\015\012\015\012");

      unless (defined $req) {
         if (exists $self->{version}) {
            last;
         } else {
            $self->err(408, "request timeout");
         }
      }

      $self->{h} = {};

      $fh->timeout($::RES_TIMEOUT);

      $req =~ /^(?:\015\012)?
                (GET|HEAD) \040+
                ([^\040]+) \040+
                HTTP\/([0-9]+\.[0-9]+)
                \015\012/gx
         or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });

      $self->{method} = $1;
      $self->{uri} = $2;
      $self->{version} = $3;

      $3 =~ /^1\./
         or $self->err(506, "http protocol version $3 not supported");

      # parse headers
      {
         my (%hdr, $h, $v);

         $hdr{lc $1} .= ",$2"
            while $req =~ /\G
                  ([^:\000-\040]+):
                  [\011\040]*
                  ((?: [^\015\012]+ | \015\012[\011\040] )*)
                  \015\012
               /gxc;

         $req =~ /\G\015\012$/
            or $self->err(400, "bad request");

         $self->{h}{$h} = substr $v, 1
            while ($h, $v) = each %hdr;
      }

      # remote id should be unique per user
      my $id = $self->{remote_addr};

      if (exists $self->{h}{"client-ip"}) {
         $id .= "[".$self->{h}{"client-ip"}."]";
      } elsif (exists $self->{h}{"x-forwarded-for"}) {
         $id .= "[".$self->{h}{"x-forwarded-for"}."]";
      }

      $self->{remote_id} = $id;

      weaken (local $conn{$id}{$self*1} = $self);

      if ($blocked{$id}) {
         $self->err_blocked
            if $blocked{$id}[0] > $::NOW;

         delete $blocked{$id};
      }

      # find out server name and port
      if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
         $host = $1;
      } else {
         $host = $self->{h}{host};
      }

      if (defined $host) {
         $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
      } else {
         ($self->{server_port}, $host)
            = unpack_sockaddr_in $self->{fh}->sockname
               or $self->err(500, "unable to get socket name");
         $host = inet_ntoa $host;
      }

      $self->{server_name} = $host;

      weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);

      eval {
         $self->map_uri;
         $self->respond;
      };

      die if $@ && !ref $@;

      last if $self->{h}{connection} =~ /close/i;

      $httpevent->broadcast;

      $fh->timeout($::PER_TIMEOUT);
   }
}

sub block {
   my $self = shift;

   $blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
   $self->slog(2, "blocked ip $self->{remote_id}");
   $self->err_blocked;
}

# uri => path mapping
sub map_uri {
   my $self = shift;
   my $host = $self->{server_name};
   my $uri = $self->{uri};

   # some massaging, also makes it more secure
   $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
   $uri =~ s%//+%/%g;
   $uri =~ s%/\.(?=/|$)%%g;
   1 while $uri =~ s%/[^/]+/\.\.(?=/|$)%%;

   $uri =~ m%^/?\.\.(?=/|$)%
      and $self->err(400, "bad request");

   $self->{name} = $uri;

   # now do the path mapping
   $self->{path} = "$::DOCROOT/$host$uri";

   $self->access_check;
}

sub _cgi {
   my $self = shift;
   my $path = shift;
   my $fh;

   # no two-way xxx supported
   if (0 == fork) {
      open STDOUT, ">&".fileno($self->{fh});
      if (chdir $::DOCROOT) {
         $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
         $ENV{HTTP_HOST}       = $self->{server_name};
         $ENV{HTTP_PORT}       = $self->{server_port};
         $ENV{SCRIPT_NAME}     = $self->{name};
         exec $path;
      }
      Coro::State::_exit(0);
   } else {
      die;
   }
}

sub server_hostport {
   $_[0]{server_port} == 80
      ? $_[0]{server_name}
      : "$_[0]{server_name}:$_[0]{server_port}";
}

sub respond {
   my $self = shift;
   my $path = $self->{path};

   if ($self->{name} =~ s%^/internal/([^/]+)%%) {
      if ($::internal{$1}) {
         $::internal{$1}->($self);
      } else {
         $self->err(404, "not found");
      }
   } else {

      stat $path
         or $self->err(404, "not found");

      $self->{stat} = [stat _];

      # idiotic netscape sends idiotic headers AGAIN
      my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
                ? str2time $1 : 0;

      if (-d _ && -r _) {
         # directory
         if ($path !~ /\/$/) {
            # create a redirect to get the trailing "/"
            # we don't try to avoid the :80
            $self->err(301, "moved permanently", { Location =>  "http://".$self->server_hostport."$self->{uri}/" });
         } else {
            $ims < $self->{stat}[9]
               or $self->err(304, "not modified");

            if (-r "$path/index.html") {
               # replace directory "size" by index.html filesize
               $self->{stat} = [stat ($self->{path} .= "/index.html")];
               $self->handle_file($queue_index, $tbf_top);
            } else {
               $self->handle_dir;
            }
         }
      } elsif (-f _ && -r _) {
         -x _ and $self->err(403, "forbidden");

         if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
            my $timeout = $::NOW + 10;
            while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
               if ($timeout < $::NOW) {
                  $self->block($::BLOCKTIME, "too many connections");
               } else {
                  $httpevent->wait;
               }
            }
         }

         $self->handle_file($queue_file, $tbf_top);
      } else {
         $self->err(404, "not found");
      }
   }
}

sub handle_dir {
   my $self = shift;
   my $idx = $self->diridx;

   $self->response(200, "ok",
         {
            "Content-Type"   => "text/html; charset=utf-8",
            "Content-Length" => length $idx,
            "Last-Modified"  => time2str ($self->{stat}[9]),
         },
         $idx);
}

sub handle_file {
   my ($self, $queue, $tbf) = @_;
   my $length = $self->{stat}[7];
   my $hdr = {
      "Last-Modified" => time2str ((stat _)[9]),
      "Accept-Ranges" => "bytes",
   };

   my @code = (200, "ok");
   my ($l, $h);

   if ($self->{h}{range} =~ /^bytes=(.*)$/) {
      for (split /,/, $1) {
         if (/^-(\d+)$/) {
            ($l, $h) = ($length - $1, $length - 1);
         } elsif (/^(\d+)-(\d*)$/) {
            ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1);
         } else {
            ($l, $h) = (0, $length - 1);
            goto ignore;
         }
         goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
      }
      $hdr->{"Content-Range"} = "bytes */$length";
      $hdr->{"Content-Length"} = $length;
      $self->err(416, "not satisfiable", $hdr, "");

satisfiable:
      # check for segmented downloads
      if ($l && $::NO_SEGMENTED) {
         my $timeout = $::NOW + 15;
         while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
            if ($timeout <= $::NOW) {
               $self->block($::BLOCKTIME, "segmented downloads are forbidden");
               #$self->err_segmented_download;
            } else {
               $httpevent->wait;
            }
         }
      }

      $hdr->{"Content-Range"} = "bytes $l-$h/$length";
      @code = (206, "partial content");
      $length = $h - $l + 1;

ignore:
   } else {
      ($l, $h) = (0, $length - 1);
   }

   $self->{path} =~ /\.([^.]+)$/;
   $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream";
   $hdr->{"Content-Length"} = $length;

   $self->response(@code, $hdr, "");

   if ($self->{method} eq "GET") {
      $self->{time} = $::NOW;
      $self->{written} = 0;

      my $current = $Coro::current;

      my ($fh, $buf, $r);

      open $fh, "<", $self->{path}
         or die "$self->{path}: late open failure ($!)";

      $h -= $l - 1;

      if (0) { # !AIO
         if ($l) {
            sysseek $fh, $l, 0;
         }
      }
      
      my $transfer = $queue->start_transfer($h);
      my $locked;
      my $bufsize = $::WAIT_BUFSIZE; # initial buffer size

      while ($h > 0) {
         unless ($locked) {
            if ($locked ||= $transfer->try($::WAIT_INTERVAL)) {
               $bufsize = $::BUFSIZE;
               $self->{time} = $::NOW;
               $self->{written} = 0;
            }
         }

         if ($blocked{$self->{remote_id}}) {
            $self->{h}{connection} = "close";
            die bless {}, err::;
         }

         if (0) { # !AIO
            sysread $fh, $buf, $h > $bufsize ? $bufsize : $h
               or last;
         } else {
            aio_read($fh, $l, ($h > $bufsize ? $bufsize : $h),
                     $buf, 0, sub {
                        $r = $_[0];
                        Coro::ready($current);
                     });
            &Coro::schedule;
            last unless $r;
         }

         $tbf->request(length $buf);
         my $w = syswrite $self->{fh}, $buf
            or last;
         $::written += $w;
         $self->{written} += $w;
         $l += $r;
      }

      close $fh;
   }
}

1
Revision:	1.78
Committed:	Sun Dec 4 19:36:28 2005 UTC (20 years, 9 months ago) by root
Content type:	text/plain
Branch:	MAIN
CVS Tags:	rel-2_5, rel-2_0, rel-2_1, rel-1_9, rel-1_7, rel-1_6, stack_sharing
Changes since 1.77:	+1 -1 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	use Coro;
2			use Coro::Semaphore;
3			use Coro::Event;
4			use Coro::Socket;
5	root	1.44	use Coro::Signal;
6	root	1.1
7	root	1.32	use HTTP::Date;
8	root	1.47	use POSIX ();
9	root	1.32
10	root	1.63	use Compress::Zlib ();
11
12	root	1.1	no utf8;
13			use bytes;
14
15			# at least on my machine, this thingy serves files
16			# quite a bit faster than apache, ;)
17			# and quite a bit slower than thttpd :(
18
19			$SIG{PIPE} = 'IGNORE';
20	root	1.27
21			our $accesslog;
22	root	1.56	our $errorlog;
23
24			our $NOW;
25			our $HTTP_NOW;
26
27			Event->timer(interval => 1, hard => 1, cb => sub {
28			$NOW = time;
29			$HTTP_NOW = time2str $NOW;
30			})->now;
31
32			if ($ERROR_LOG) {
33			use IO::Handle;
34			open $errorlog, ">>$ERROR_LOG"
35			or die "$ERROR_LOG: $!";
36			$errorlog->autoflush(1);
37			}
38	root	1.27
39			if ($ACCESS_LOG) {
40			use IO::Handle;
41			open $accesslog, ">>$ACCESS_LOG"
42			or die "$ACCESS_LOG: $!";
43			$accesslog->autoflush(1);
44			}
45
46	root	1.1	sub slog {
47			my $level = shift;
48			my $format = shift;
49	root	1.56	my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
50			printf "$NOW: $format\n", @_;
51			printf $errorlog "$NOW: $format\n", @_ if $errorlog;
52	root	1.1	}
53
54	root	1.32	our $connections = new Coro::Semaphore $MAX_CONNECTS \|\| 250;
55	root	1.44	our $httpevent = new Coro::Signal;
56	root	1.34
57	root	1.63	our $queue_file = new transferqueue $MAX_TRANSFERS;
58			our $queue_index = new transferqueue 10;
59	root	1.62
60	root	1.72	our $tbf_top = new tbf rate => $TBF_RATE \|\| 100000;
61	root	1.67
62	root	1.68	my $unused_bytes = 0;
63			my $unused_last = time;
64
65			sub unused_bandwidth {
66			$unused_bytes += $_[0];
67			if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
68			$unused_last = $NOW;
69			$unused_bytes = 0;
70			$queue_file->force_wake_next;
71			slog 1, "forced filetransfer due to unused bandwidth";
72			}
73			}
74
75	root	1.6	my @newcons;
76	root	1.1	my @pool;
77
78	root	1.2	# one "execution thread"
79	root	1.1	sub handler {
80			while () {
81	root	1.38	if (@newcons) {
82	root	1.1	eval {
83	root	1.38	conn->new(@{pop @newcons})->handle;
84	root	1.1	};
85			slog 1, "$@" if $@ && !ref $@;
86	root	1.56
87			$httpevent->broadcast; # only for testing, but doesn't matter much
88
89	root	1.1	$connections->up;
90			} else {
91			last if @pool >= $MAX_POOL;
92			push @pool, $Coro::current;
93			schedule;
94			}
95			}
96			}
97
98	root	1.40	sub listen_on {
99			my $listen = $_[0];
100
101			push @listen_sockets, $listen;
102
103			# the "main thread"
104			async {
105			slog 1, "accepting connections";
106			while () {
107			$connections->down;
108			push @newcons, [$listen->accept];
109			#slog 3, "accepted @$connections ".scalar(@pool);
110			if (@pool) {
111			(pop @pool)->ready;
112			} else {
113			async \&handler;
114			}
115			}
116			};
117			}
118
119	root	1.4	my $http_port = new Coro::Socket
120			LocalAddr => $SERVER_HOST,
121			LocalPort => $SERVER_PORT,
122			ReuseAddr => 1,
123	root	1.13	Listen => 50,
124	root	1.4	or die "unable to start server";
125
126	root	1.40	listen_on $http_port;
127
128	root	1.41	if ($SERVER_PORT2) {
129			my $http_port = new Coro::Socket
130			LocalAddr => $SERVER_HOST,
131			LocalPort => $SERVER_PORT2,
132			ReuseAddr => 1,
133			Listen => 50,
134			or die "unable to start server";
135	root	1.40
136	root	1.41	listen_on $http_port;
137			}
138	root	1.4
139	root	1.1	package conn;
140
141			use Socket;
142			use HTTP::Date;
143	root	1.2	use Convert::Scalar 'weaken';
144	root	1.77	use IO::AIO;
145	root	1.16
146	root	1.77	IO::AIO::min_parallel $::AIO_PARALLEL;
147	root	1.16
148	root	1.77	Event->io(fd => IO::AIO::poll_fileno,
149	root	1.17	poll => 'r', async => 1,
150	root	1.77	cb => \&IO::AIO::poll_cb);
151	root	1.16
152	root	1.26	our %conn; # $conn{ip}{self} => connobj
153			our %uri; # $uri{ip}{uri}{self}
154	root	1.3	our %blocked;
155	root	1.9	our %mimetype;
156
157			sub read_mimetypes {
158			local *M;
159	root	1.10	if (open M, "<mime_types") {
160	root	1.9	while (<M>) {
161			if (/^([^#]\S+)\t+(\S+)$/) {
162			$mimetype{lc $1} = $2;
163			}
164			}
165			} else {
166	root	1.10	print "cannot open mime_types\n";
167	root	1.9	}
168			}
169	root	1.1
170	root	1.10	read_mimetypes;
171
172	root	1.1	sub new {
173			my $class = shift;
174	root	1.42	my $fh = shift;
175	root	1.6	my $peername = shift;
176	root	1.2	my $self = bless { fh => $fh }, $class;
177	root	1.6	my (undef, $iaddr) = unpack_sockaddr_in $peername
178			or $self->err(500, "unable to decode peername");
179	root	1.7
180	root	1.50	$self->{remote_addr} =
181			$self->{remote_id} = inet_ntoa $iaddr;
182	root	1.60
183	root	1.11	$self->{time} = $::NOW;
184	root	1.2
185	root	1.49	weaken ($Coro::current->{conn} = $self);
186
187	root	1.13	$::conns++;
188	root	1.53	$::maxconns = $::conns if $::conns > $::maxconns;
189	root	1.13
190	root	1.2	$self;
191			}
192
193			sub DESTROY {
194	root	1.56	#my $self = shift;
195	root	1.13	$::conns--;
196	root	1.1	}
197
198	root	1.73	sub prune_cache {
199			my $hash = $_[0];
200
201			for (keys %$hash) {
202			if (ref $hash->{$_} eq HASH::) {
203			prune_cache($hash->{$_});
204			unless (scalar keys %{$hash->{$_}}) {
205			delete $hash->{$_};
206			$d2++;
207			}
208			}
209			}
210			}
211
212			sub prune_caches {
213			prune_cache \%conn;
214			prune_cache \%uri;
215
216			for (keys %blocked) {
217	root	1.74	delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
218	root	1.73	}
219			}
220
221			Event->timer(interval => 60, cb => \&prune_caches);
222
223	root	1.1	sub slog {
224	root	1.4	my $self = shift;
225	root	1.50	main::slog($_[0], "$self->{remote_id}> $_[1]");
226	root	1.1	}
227
228	root	1.4	sub response {
229	root	1.1	my ($self, $code, $msg, $hdr, $content) = @_;
230	root	1.17	my $res = "HTTP/1.1 $code $msg\015\012";
231	root	1.63	my $GZ = "";
232	root	1.1
233	root	1.46	if (exists $hdr->{Connection}) {
234			if ($hdr->{Connection} =~ /close/) {
235			$self->{h}{connection} = "close"
236			}
237			} else {
238			if ($self->{version} < 1.1) {
239			if ($self->{h}{connection} =~ /keep-alive/i) {
240			$hdr->{Connection} = "Keep-Alive";
241			} else {
242			$self->{h}{connection} = "close"
243			}
244			}
245			}
246	root	1.28
247	root	1.63	if ($self->{method} ne "HEAD"
248			&& $self->{h}{"accept-encoding"} =~ /\bgzip\b/
249	root	1.64	&& 400 < length $content
250	root	1.63	&& $hdr->{"Content-Length"} == length $content
251			&& !exists $hdr->{"Content-Encoding"}
252			) {
253			my $orig = length $content;
254			$hdr->{"Content-Encoding"} = "gzip";
255			$content = Compress::Zlib::memGzip(\$content);
256			$hdr->{"Content-Length"} = length $content;
257			$GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
258			}
259
260	root	1.32	$res .= "Date: $HTTP_NOW\015\012";
261	root	1.71	$res .= "Server: $::NAME\015\012";
262	root	1.1
263			while (my ($h, $v) = each %$hdr) {
264			$res .= "$h: $v\015\012"
265			}
266	root	1.10	$res .= "\015\012";
267	root	1.4
268	root	1.13	$res .= $content if defined $content and $self->{method} ne "HEAD";
269	root	1.1
270	root	1.65	my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
271	root	1.63	" $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
272	root	1.50	" \"$self->{h}{referer}\"\n";
273	root	1.27
274	root	1.65	print $::accesslog $log if $::accesslog;
275	root	1.27	print STDERR $log;
276	root	1.2
277	root	1.67	$tbf_top->request(length $res, 1e6);
278			$self->{written} += print {$self->{fh}} $res;
279	root	1.1	}
280
281			sub err {
282			my $self = shift;
283			my ($code, $msg, $hdr, $content) = @_;
284
285			unless (defined $content) {
286	root	1.35	$content = "$code $msg\n";
287	root	1.1	$hdr->{"Content-Type"} = "text/plain";
288			$hdr->{"Content-Length"} = length $content;
289			}
290	root	1.17	$hdr->{"Connection"} = "close";
291	root	1.1
292	root	1.4	$self->response($code, $msg, $hdr, $content);
293	root	1.1
294			die bless {}, err::;
295			}
296
297			sub handle {
298			my $self = shift;
299			my $fh = $self->{fh};
300
301	root	1.29	my $host;
302
303	root	1.17	$fh->timeout($::REQ_TIMEOUT);
304			while() {
305			$self->{reqs}++;
306	root	1.1
307			# read request and parse first line
308			my $req = $fh->readline("\015\012\015\012");
309
310	root	1.17	unless (defined $req) {
311			if (exists $self->{version}) {
312			last;
313			} else {
314			$self->err(408, "request timeout");
315			}
316			}
317
318			$self->{h} = {};
319	root	1.1
320	root	1.17	$fh->timeout($::RES_TIMEOUT);
321	root	1.3
322	root	1.1	$req =~ /^(?:\015\012)?
323			(GET\|HEAD) \040+
324			([^\040]+) \040+
325			HTTP\/([0-9]+\.[0-9]+)
326			\015\012/gx
327	root	1.14	or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
328	root	1.1
329			$self->{method} = $1;
330			$self->{uri} = $2;
331	root	1.17	$self->{version} = $3;
332
333	root	1.20	$3 =~ /^1\./
334	root	1.17	or $self->err(506, "http protocol version $3 not supported");
335	root	1.1
336			# parse headers
337			{
338			my (%hdr, $h, $v);
339
340			$hdr{lc $1} .= ",$2"
341			while $req =~ /\G
342			([^:\000-\040]+):
343	root	1.66	[\011\040]*
344			((?: [^\015\012]+ \| \015\012[\011\040] )*)
345	root	1.1	\015\012
346			/gxc;
347
348			$req =~ /\G\015\012$/
349			or $self->err(400, "bad request");
350
351			$self->{h}{$h} = substr $v, 1
352			while ($h, $v) = each %hdr;
353			}
354
355	root	1.36	# remote id should be unique per user
356			my $id = $self->{remote_addr};
357
358			if (exists $self->{h}{"client-ip"}) {
359			$id .= "[".$self->{h}{"client-ip"}."]";
360			} elsif (exists $self->{h}{"x-forwarded-for"}) {
361			$id .= "[".$self->{h}{"x-forwarded-for"}."]";
362			}
363
364			$self->{remote_id} = $id;
365
366	root	1.56	weaken (local $conn{$id}{$self*1} = $self);
367
368	root	1.36	if ($blocked{$id}) {
369	root	1.56	$self->err_blocked
370			if $blocked{$id}[0] > $::NOW;
371	root	1.36
372			delete $blocked{$id};
373			}
374
375	root	1.29	# find out server name and port
376			if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
377			$host = $1;
378			} else {
379			$host = $self->{h}{host};
380			}
381
382			if (defined $host) {
383			$self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
384			} else {
385			($self->{server_port}, $host)
386	root	1.43	= unpack_sockaddr_in $self->{fh}->sockname
387	root	1.29	or $self->err(500, "unable to get socket name");
388			$host = inet_ntoa $host;
389			}
390
391			$self->{server_name} = $host;
392
393	root	1.56	weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
394	root	1.1
395	root	1.24	eval {
396			$self->map_uri;
397			$self->respond;
398			};
399
400			die if $@ && !ref $@;
401	root	1.17
402	root	1.56	last if $self->{h}{connection} =~ /close/i;
403	root	1.17
404	root	1.44	$httpevent->broadcast;
405
406	root	1.17	$fh->timeout($::PER_TIMEOUT);
407			}
408	root	1.1	}
409
410	root	1.56	sub block {
411			my $self = shift;
412
413			$blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
414			$self->slog(2, "blocked ip $self->{remote_id}");
415			$self->err_blocked;
416			}
417
418	root	1.1	# uri => path mapping
419			sub map_uri {
420			my $self = shift;
421	root	1.29	my $host = $self->{server_name};
422	root	1.1	my $uri = $self->{uri};
423
424			# some massaging, also makes it more secure
425			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
426			$uri =~ s%//+%/%g;
427			$uri =~ s%/\.(?=/\|$)%%g;
428			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
429
430			$uri =~ m%^/?\.\.(?=/\|$)%
431			and $self->err(400, "bad request");
432
433			$self->{name} = $uri;
434
435			# now do the path mapping
436			$self->{path} = "$::DOCROOT/$host$uri";
437	root	1.7
438			$self->access_check;
439	root	1.1	}
440
441			sub _cgi {
442			my $self = shift;
443			my $path = shift;
444			my $fh;
445
446			# no two-way xxx supported
447			if (0 == fork) {
448			open STDOUT, ">&".fileno($self->{fh});
449			if (chdir $::DOCROOT) {
450			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
451	root	1.29	$ENV{HTTP_HOST} = $self->{server_name};
452			$ENV{HTTP_PORT} = $self->{server_port};
453	root	1.1	$ENV{SCRIPT_NAME} = $self->{name};
454	root	1.10	exec $path;
455	root	1.1	}
456			Coro::State::_exit(0);
457			} else {
458	root	1.29	die;
459	root	1.1	}
460			}
461
462	root	1.29	sub server_hostport {
463			$_[0]{server_port} == 80
464			? $_[0]{server_name}
465			: "$_[0]{server_name}:$_[0]{server_port}";
466			}
467
468	root	1.1	sub respond {
469			my $self = shift;
470			my $path = $self->{path};
471
472	root	1.49	if ($self->{name} =~ s%^/internal/([^/]+)%%) {
473			if ($::internal{$1}) {
474			$::internal{$1}->($self);
475			} else {
476			$self->err(404, "not found");
477			}
478			} else {
479	root	1.1
480	root	1.49	stat $path
481			or $self->err(404, "not found");
482	root	1.10
483	root	1.49	$self->{stat} = [stat _];
484	root	1.1
485	root	1.49	# idiotic netscape sends idiotic headers AGAIN
486			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
487			? str2time $1 : 0;
488
489			if (-d _ && -r _) {
490			# directory
491			if ($path !~ /\/$/) {
492			# create a redirect to get the trailing "/"
493			# we don't try to avoid the :80
494			$self->err(301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
495	root	1.25	} else {
496	root	1.49	$ims < $self->{stat}[9]
497			or $self->err(304, "not modified");
498
499			if (-r "$path/index.html") {
500			# replace directory "size" by index.html filesize
501	root	1.55	$self->{stat} = [stat ($self->{path} .= "/index.html")];
502	root	1.67	$self->handle_file($queue_index, $tbf_top);
503	root	1.49	} else {
504			$self->handle_dir;
505			}
506	root	1.1	}
507	root	1.49	} elsif (-f _ && -r _) {
508			-x _ and $self->err(403, "forbidden");
509	root	1.56
510	root	1.58	if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
511	root	1.56	my $timeout = $::NOW + 10;
512	root	1.58	while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
513	root	1.56	if ($timeout < $::NOW) {
514			$self->block($::BLOCKTIME, "too many connections");
515			} else {
516			$httpevent->wait;
517			}
518			}
519			}
520
521	root	1.67	$self->handle_file($queue_file, $tbf_top);
522	root	1.49	} else {
523			$self->err(404, "not found");
524	root	1.1	}
525			}
526			}
527
528			sub handle_dir {
529			my $self = shift;
530	root	1.10	my $idx = $self->diridx;
531
532			$self->response(200, "ok",
533			{
534	root	1.76	"Content-Type" => "text/html; charset=utf-8",
535	root	1.10	"Content-Length" => length $idx,
536	root	1.55	"Last-Modified" => time2str ($self->{stat}[9]),
537	root	1.10	},
538			$idx);
539	root	1.1	}
540
541			sub handle_file {
542	root	1.67	my ($self, $queue, $tbf) = @_;
543	root	1.34	my $length = $self->{stat}[7];
544	root	1.1	my $hdr = {
545	root	1.75	"Last-Modified" => time2str ((stat _)[9]),
546			"Accept-Ranges" => "bytes",
547	root	1.1	};
548
549			my @code = (200, "ok");
550			my ($l, $h);
551
552			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
553			for (split /,/, $1) {
554			if (/^-(\d+)$/) {
555			($l, $h) = ($length - $1, $length - 1);
556			} elsif (/^(\d+)-(\d*)$/) {
557			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
558			} else {
559			($l, $h) = (0, $length - 1);
560			goto ignore;
561			}
562	root	1.26	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
563	root	1.1	}
564			$hdr->{"Content-Range"} = "bytes */$length";
565	root	1.24	$hdr->{"Content-Length"} = $length;
566			$self->err(416, "not satisfiable", $hdr, "");
567	root	1.1
568			satisfiable:
569	root	1.4	# check for segmented downloads
570	root	1.10	if ($l && $::NO_SEGMENTED) {
571	root	1.56	my $timeout = $::NOW + 15;
572	root	1.58	while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
573	root	1.56	if ($timeout <= $::NOW) {
574			$self->block($::BLOCKTIME, "segmented downloads are forbidden");
575			#$self->err_segmented_download;
576	root	1.29	} else {
577	root	1.46	$httpevent->wait;
578	root	1.29	}
579	root	1.4	}
580			}
581
582	root	1.1	$hdr->{"Content-Range"} = "bytes $l-$h/$length";
583			@code = (206, "partial content");
584			$length = $h - $l + 1;
585
586			ignore:
587			} else {
588			($l, $h) = (0, $length - 1);
589			}
590
591	root	1.9	$self->{path} =~ /\.([^.]+)$/;
592			$hdr->{"Content-Type"} = $mimetype{lc $1} \|\| "application/octet-stream";
593	root	1.1	$hdr->{"Content-Length"} = $length;
594
595	root	1.4	$self->response(@code, $hdr, "");
596	root	1.1
597			if ($self->{method} eq "GET") {
598	root	1.32	$self->{time} = $::NOW;
599	root	1.71	$self->{written} = 0;
600	root	1.32
601	root	1.49	my $current = $Coro::current;
602	root	1.35
603	root	1.49	my ($fh, $buf, $r);
604	root	1.32
605	root	1.1	open $fh, "<", $self->{path}
606			or die "$self->{path}: late open failure ($!)";
607
608			$h -= $l - 1;
609
610	root	1.49	if (0) { # !AIO
611	root	1.19	if ($l) {
612			sysseek $fh, $l, 0;
613			}
614			}
615	root	1.48
616	root	1.56	my $transfer = $queue->start_transfer($h);
617	root	1.51	my $locked;
618	root	1.48	my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
619	root	1.19
620	root	1.1	while ($h > 0) {
621	root	1.51	unless ($locked) {
622			if ($locked \|\|= $transfer->try($::WAIT_INTERVAL)) {
623	root	1.48	$bufsize = $::BUFSIZE;
624			$self->{time} = $::NOW;
625	root	1.71	$self->{written} = 0;
626	root	1.48	}
627	root	1.56	}
628
629			if ($blocked{$self->{remote_id}}) {
630			$self->{h}{connection} = "close";
631	root	1.59	die bless {}, err::;
632	root	1.48	}
633
634	root	1.49	if (0) { # !AIO
635	root	1.48	sysread $fh, $buf, $h > $bufsize ? $bufsize : $h
636	root	1.19	or last;
637			} else {
638	root	1.48	aio_read($fh, $l, ($h > $bufsize ? $bufsize : $h),
639	root	1.19	$buf, 0, sub {
640			$r = $_[0];
641	root	1.37	Coro::ready($current);
642	root	1.19	});
643			&Coro::schedule;
644			last unless $r;
645			}
646	root	1.67
647			$tbf->request(length $buf);
648	root	1.37	my $w = syswrite $self->{fh}, $buf
649	root	1.1	or last;
650	root	1.11	$::written += $w;
651			$self->{written} += $w;
652	root	1.16	$l += $r;
653	root	1.1	}
654	root	1.32
655			close $fh;
656	root	1.1	}
657	root	1.7	}
658
659	root	1.78	1