Coro/myhttpd/httpd.pl

use AnyEvent ();

use Coro;
use Coro::Semaphore;
use Coro::SemaphoreSet;
use Coro::EV;
use Coro::Socket;
use Coro::Signal;
use Coro::AIO ();

use Fcntl;
use HTTP::Date;
use POSIX ();

use Compress::Zlib ();

use common::sense;

# 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 @listen_sockets;

our $NOW;
our $HTTP_NOW;

our $ERROR_LOG;
our $ACCESS_LOG;
our $TRANSFER_LOCK = new Coro::SemaphoreSet; # lock to be acquired per ip

our $update_time = EV::periodic 0, 1, undef, sub {
   $NOW = time;
   $HTTP_NOW = time2str $NOW;
};
$update_time->invoke;

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;

   $format = sprintf $format, @_ if @_;

   my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
   print "$NOW: $format\n";
   print $errorlog "$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
         and slog 1, "forced filetransfer due to unused bandwidth";
   }
}

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

   push @listen_sockets, $listen;

   # the "main thread"
   async {
      slog 1, "accepting connections";
      while () {
         $connections->down;
         my @conn = $listen->accept;
         #slog 3, "accepted @$connections ".scalar(@pool);

         async_pool {
            eval {
               conn->new (@conn)->handle;
            };
            slog 1, "$@" if $@ && !ref $@;

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

            $connections->up;
         }
      }
   };
}

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 common::sense;

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

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

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

sub read_mimetypes {
   if (open my $fh, "<mime_types") {
      while (<$fh>) {
         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 $hash->{$_};
         }
      }
   }
}

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

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

our $PRUNE_WATCHER = AE::timer 60, 60, \&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};

   $host =~ /[\/\\]/
      and $self->err(400, "bad request");

   # 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 {

      Coro::AIO::aio_stat $path
         and $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 ($self->{h}{range} =~ /^bytes=(.*)$/i) {
      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 = $::NOW + 60;
         while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
            if ($timeout <= $::NOW) {
               $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 $fh = Coro::AIO::aio_open $self->{path}, Fcntl::O_RDONLY, 0
         or die "$self->{path}: late open failure ($!)";

      $h -= $l - 1;

      my $transfer = $queue->start_transfer ($h);
      my $locked;
      my $bufsize = $::WAIT_BUFSIZE; # initial buffer size

      while ($h > 0) {
         Coro::cede;
         my $transfer_lock = $TRANSFER_LOCK->guard ($self->{remote_id});

         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::;
         }

         Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), my $buf, 0
            or last;

         # readahead to work around rijk disk issues
         IO::AIO::aio_readahead $fh, $l + $bufsize, $bufsize;

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

      close $fh;
   }
}

1

Revision:	1.95
Committed:	Wed Apr 24 23:49:34 2013 UTC (11 years, 2 months ago) by root
Content type:	text/plain
Branch:	MAIN
CVS Tags:	rel-6_5, rel-6_512, rel-6_513, rel-6_511, rel-6_514, rel-6_46, rel-6_45, rel-6_51, rel-6_52, rel-6_53, rel-6_54, rel-6_55, rel-6_56, rel-6_43, rel-6_42, rel-6_41, rel-6_47, rel-6_44, rel-6_49, rel-6_48
Changes since 1.94:	+5 -3 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.94	use AnyEvent ();
2
3	root	1.1	use Coro;
4			use Coro::Semaphore;
5	root	1.90	use Coro::SemaphoreSet;
6	root	1.84	use Coro::EV;
7	root	1.1	use Coro::Socket;
8	root	1.44	use Coro::Signal;
9	root	1.79	use Coro::AIO ();
10	root	1.1
11	root	1.90	use Fcntl;
12	root	1.32	use HTTP::Date;
13	root	1.47	use POSIX ();
14	root	1.32
15	root	1.63	use Compress::Zlib ();
16
17	root	1.90	use common::sense;
18	root	1.1
19			# at least on my machine, this thingy serves files
20			# quite a bit faster than apache, ;)
21			# and quite a bit slower than thttpd :(
22
23			$SIG{PIPE} = 'IGNORE';
24	root	1.27
25			our $accesslog;
26	root	1.56	our $errorlog;
27	root	1.90	our @listen_sockets;
28	root	1.56
29			our $NOW;
30			our $HTTP_NOW;
31
32	root	1.80	our $ERROR_LOG;
33			our $ACCESS_LOG;
34	root	1.90	our $TRANSFER_LOCK = new Coro::SemaphoreSet; # lock to be acquired per ip
35	root	1.80
36	root	1.84	our $update_time = EV::periodic 0, 1, undef, sub {
37	root	1.56	$NOW = time;
38			$HTTP_NOW = time2str $NOW;
39	root	1.84	};
40			$update_time->invoke;
41	root	1.56
42			if ($ERROR_LOG) {
43			use IO::Handle;
44			open $errorlog, ">>$ERROR_LOG"
45			or die "$ERROR_LOG: $!";
46			$errorlog->autoflush(1);
47			}
48	root	1.27
49			if ($ACCESS_LOG) {
50			use IO::Handle;
51			open $accesslog, ">>$ACCESS_LOG"
52			or die "$ACCESS_LOG: $!";
53			$accesslog->autoflush(1);
54			}
55
56	root	1.1	sub slog {
57			my $level = shift;
58			my $format = shift;
59	root	1.92
60			$format = sprintf $format, @_ if @_;
61
62	root	1.56	my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
63	root	1.92	print "$NOW: $format\n";
64	root	1.93	print $errorlog "$NOW: $format\n", @_ if $errorlog;
65	root	1.1	}
66
67	root	1.90	our $connections = new Coro::Semaphore $::MAX_CONNECTS \|\| 250;
68	root	1.44	our $httpevent = new Coro::Signal;
69	root	1.34
70	root	1.90	our $queue_file = new transferqueue $::MAX_TRANSFERS;
71	root	1.63	our $queue_index = new transferqueue 10;
72	root	1.62
73	root	1.90	our $tbf_top = new tbf rate => $::TBF_RATE \|\| 100000;
74	root	1.67
75	root	1.68	my $unused_bytes = 0;
76			my $unused_last = time;
77
78			sub unused_bandwidth {
79			$unused_bytes += $_[0];
80			if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
81			$unused_last = $NOW;
82			$unused_bytes = 0;
83	root	1.91	$queue_file->force_wake_next
84			and slog 1, "forced filetransfer due to unused bandwidth";
85	root	1.68	}
86			}
87
88	root	1.40	sub listen_on {
89			my $listen = $_[0];
90
91			push @listen_sockets, $listen;
92
93			# the "main thread"
94			async {
95			slog 1, "accepting connections";
96			while () {
97			$connections->down;
98	root	1.88	my @conn = $listen->accept;
99	root	1.40	#slog 3, "accepted @$connections ".scalar(@pool);
100	root	1.88
101			async_pool {
102			eval {
103			conn->new (@conn)->handle;
104			};
105			slog 1, "$@" if $@ && !ref $@;
106
107			$httpevent->broadcast; # only for testing, but doesn't matter much
108
109			$connections->up;
110	root	1.40	}
111			}
112			};
113			}
114
115	root	1.4	my $http_port = new Coro::Socket
116	root	1.90	LocalAddr => $::SERVER_HOST,
117			LocalPort => $::SERVER_PORT,
118	root	1.4	ReuseAddr => 1,
119	root	1.81	Listen => 50,
120	root	1.4	or die "unable to start server";
121
122	root	1.40	listen_on $http_port;
123
124	root	1.90	if ($::SERVER_PORT2) {
125	root	1.41	my $http_port = new Coro::Socket
126	root	1.90	LocalAddr => $::SERVER_HOST,
127			LocalPort => $::SERVER_PORT2,
128	root	1.41	ReuseAddr => 1,
129	root	1.81	Listen => 50,
130	root	1.41	or die "unable to start server";
131	root	1.40
132	root	1.41	listen_on $http_port;
133			}
134	root	1.4
135	root	1.1	package conn;
136
137	root	1.90	use common::sense;
138	root	1.80
139	root	1.1	use Socket;
140			use HTTP::Date;
141	root	1.2	use Convert::Scalar 'weaken';
142	root	1.77	use IO::AIO;
143	root	1.94	use AnyEvent::AIO;
144	root	1.16
145	root	1.77	IO::AIO::min_parallel $::AIO_PARALLEL;
146	root	1.16
147	root	1.26	our %conn; # $conn{ip}{self} => connobj
148			our %uri; # $uri{ip}{uri}{self}
149	root	1.3	our %blocked;
150	root	1.9	our %mimetype;
151
152			sub read_mimetypes {
153	root	1.81	if (open my $fh, "<mime_types") {
154			while (<$fh>) {
155	root	1.9	if (/^([^#]\S+)\t+(\S+)$/) {
156			$mimetype{lc $1} = $2;
157			}
158			}
159			} else {
160	root	1.10	print "cannot open mime_types\n";
161	root	1.9	}
162			}
163	root	1.1
164	root	1.10	read_mimetypes;
165
166	root	1.1	sub new {
167			my $class = shift;
168	root	1.42	my $fh = shift;
169	root	1.6	my $peername = shift;
170	root	1.2	my $self = bless { fh => $fh }, $class;
171	root	1.6	my (undef, $iaddr) = unpack_sockaddr_in $peername
172	root	1.81	or $self->err (500, "unable to decode peername");
173	root	1.7
174	root	1.50	$self->{remote_addr} =
175			$self->{remote_id} = inet_ntoa $iaddr;
176	root	1.60
177	root	1.11	$self->{time} = $::NOW;
178	root	1.2
179	root	1.49	weaken ($Coro::current->{conn} = $self);
180
181	root	1.82	++$::conns;
182	root	1.53	$::maxconns = $::conns if $::conns > $::maxconns;
183	root	1.13
184	root	1.81	$self
185	root	1.2	}
186
187			sub DESTROY {
188	root	1.82	my $self = shift;
189
190	root	1.81	--$::conns;
191	root	1.1	}
192
193	root	1.73	sub prune_cache {
194			my $hash = $_[0];
195
196			for (keys %$hash) {
197			if (ref $hash->{$_} eq HASH::) {
198			prune_cache($hash->{$_});
199			unless (scalar keys %{$hash->{$_}}) {
200			delete $hash->{$_};
201			}
202			}
203			}
204			}
205
206			sub prune_caches {
207			prune_cache \%conn;
208			prune_cache \%uri;
209
210			for (keys %blocked) {
211	root	1.74	delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
212	root	1.73	}
213			}
214
215	root	1.94	our $PRUNE_WATCHER = AE::timer 60, 60, \&prune_caches;
216	root	1.73
217	root	1.1	sub slog {
218	root	1.4	my $self = shift;
219	root	1.50	main::slog($_[0], "$self->{remote_id}> $_[1]");
220	root	1.1	}
221
222	root	1.4	sub response {
223	root	1.1	my ($self, $code, $msg, $hdr, $content) = @_;
224	root	1.17	my $res = "HTTP/1.1 $code $msg\015\012";
225	root	1.63	my $GZ = "";
226	root	1.1
227	root	1.46	if (exists $hdr->{Connection}) {
228			if ($hdr->{Connection} =~ /close/) {
229			$self->{h}{connection} = "close"
230			}
231			} else {
232			if ($self->{version} < 1.1) {
233			if ($self->{h}{connection} =~ /keep-alive/i) {
234			$hdr->{Connection} = "Keep-Alive";
235			} else {
236			$self->{h}{connection} = "close"
237			}
238			}
239			}
240	root	1.28
241	root	1.63	if ($self->{method} ne "HEAD"
242			&& $self->{h}{"accept-encoding"} =~ /\bgzip\b/
243	root	1.64	&& 400 < length $content
244	root	1.63	&& $hdr->{"Content-Length"} == length $content
245			&& !exists $hdr->{"Content-Encoding"}
246			) {
247			my $orig = length $content;
248			$hdr->{"Content-Encoding"} = "gzip";
249			$content = Compress::Zlib::memGzip(\$content);
250			$hdr->{"Content-Length"} = length $content;
251			$GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
252			}
253
254	root	1.32	$res .= "Date: $HTTP_NOW\015\012";
255	root	1.71	$res .= "Server: $::NAME\015\012";
256	root	1.1
257			while (my ($h, $v) = each %$hdr) {
258			$res .= "$h: $v\015\012"
259			}
260	root	1.10	$res .= "\015\012";
261	root	1.4
262	root	1.13	$res .= $content if defined $content and $self->{method} ne "HEAD";
263	root	1.1
264	root	1.65	my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
265	root	1.63	" $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
266	root	1.50	" \"$self->{h}{referer}\"\n";
267	root	1.27
268	root	1.65	print $::accesslog $log if $::accesslog;
269	root	1.27	print STDERR $log;
270	root	1.2
271	root	1.67	$tbf_top->request(length $res, 1e6);
272			$self->{written} += print {$self->{fh}} $res;
273	root	1.1	}
274
275			sub err {
276			my $self = shift;
277			my ($code, $msg, $hdr, $content) = @_;
278
279			unless (defined $content) {
280	root	1.35	$content = "$code $msg\n";
281	root	1.1	$hdr->{"Content-Type"} = "text/plain";
282			$hdr->{"Content-Length"} = length $content;
283			}
284	root	1.17	$hdr->{"Connection"} = "close";
285	root	1.1
286	root	1.81	$self->response ($code, $msg, $hdr, $content);
287	root	1.1
288	root	1.81	die bless {}, err::
289	root	1.1	}
290
291			sub handle {
292			my $self = shift;
293			my $fh = $self->{fh};
294
295	root	1.29	my $host;
296
297	root	1.17	$fh->timeout($::REQ_TIMEOUT);
298	root	1.81	while () {
299	root	1.17	$self->{reqs}++;
300	root	1.1
301			# read request and parse first line
302			my $req = $fh->readline("\015\012\015\012");
303
304	root	1.17	unless (defined $req) {
305			if (exists $self->{version}) {
306			last;
307			} else {
308			$self->err(408, "request timeout");
309			}
310			}
311
312			$self->{h} = {};
313	root	1.1
314	root	1.17	$fh->timeout($::RES_TIMEOUT);
315	root	1.3
316	root	1.1	$req =~ /^(?:\015\012)?
317			(GET\|HEAD) \040+
318			([^\040]+) \040+
319			HTTP\/([0-9]+\.[0-9]+)
320			\015\012/gx
321	root	1.14	or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
322	root	1.1
323			$self->{method} = $1;
324			$self->{uri} = $2;
325	root	1.17	$self->{version} = $3;
326
327	root	1.20	$3 =~ /^1\./
328	root	1.17	or $self->err(506, "http protocol version $3 not supported");
329	root	1.1
330			# parse headers
331			{
332			my (%hdr, $h, $v);
333
334			$hdr{lc $1} .= ",$2"
335			while $req =~ /\G
336			([^:\000-\040]+):
337	root	1.66	[\011\040]*
338			((?: [^\015\012]+ \| \015\012[\011\040] )*)
339	root	1.1	\015\012
340			/gxc;
341
342			$req =~ /\G\015\012$/
343			or $self->err(400, "bad request");
344
345			$self->{h}{$h} = substr $v, 1
346			while ($h, $v) = each %hdr;
347			}
348
349	root	1.36	# remote id should be unique per user
350			my $id = $self->{remote_addr};
351
352			if (exists $self->{h}{"client-ip"}) {
353			$id .= "[".$self->{h}{"client-ip"}."]";
354			} elsif (exists $self->{h}{"x-forwarded-for"}) {
355			$id .= "[".$self->{h}{"x-forwarded-for"}."]";
356			}
357
358			$self->{remote_id} = $id;
359
360	root	1.56	weaken (local $conn{$id}{$self*1} = $self);
361
362	root	1.36	if ($blocked{$id}) {
363	root	1.56	$self->err_blocked
364			if $blocked{$id}[0] > $::NOW;
365	root	1.36
366			delete $blocked{$id};
367			}
368
369	root	1.29	# find out server name and port
370			if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
371			$host = $1;
372			} else {
373			$host = $self->{h}{host};
374			}
375
376			if (defined $host) {
377			$self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
378			} else {
379			($self->{server_port}, $host)
380	root	1.43	= unpack_sockaddr_in $self->{fh}->sockname
381	root	1.29	or $self->err(500, "unable to get socket name");
382			$host = inet_ntoa $host;
383			}
384
385			$self->{server_name} = $host;
386
387	root	1.56	weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
388	root	1.1
389	root	1.24	eval {
390			$self->map_uri;
391			$self->respond;
392			};
393
394			die if $@ && !ref $@;
395	root	1.17
396	root	1.56	last if $self->{h}{connection} =~ /close/i;
397	root	1.17
398	root	1.44	$httpevent->broadcast;
399
400	root	1.17	$fh->timeout($::PER_TIMEOUT);
401			}
402	root	1.1	}
403
404	root	1.56	sub block {
405			my $self = shift;
406
407			$blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
408			$self->slog(2, "blocked ip $self->{remote_id}");
409			$self->err_blocked;
410			}
411
412	root	1.1	# uri => path mapping
413			sub map_uri {
414			my $self = shift;
415	root	1.29	my $host = $self->{server_name};
416	root	1.1	my $uri = $self->{uri};
417
418	root	1.85	$host =~ /[\/\\]/
419			and $self->err(400, "bad request");
420
421	root	1.1	# some massaging, also makes it more secure
422			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
423			$uri =~ s%//+%/%g;
424			$uri =~ s%/\.(?=/\|$)%%g;
425			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
426
427			$uri =~ m%^/?\.\.(?=/\|$)%
428			and $self->err(400, "bad request");
429
430			$self->{name} = $uri;
431
432			# now do the path mapping
433			$self->{path} = "$::DOCROOT/$host$uri";
434	root	1.7
435			$self->access_check;
436	root	1.1	}
437
438			sub _cgi {
439			my $self = shift;
440			my $path = shift;
441			my $fh;
442
443			# no two-way xxx supported
444			if (0 == fork) {
445			open STDOUT, ">&".fileno($self->{fh});
446			if (chdir $::DOCROOT) {
447			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
448	root	1.29	$ENV{HTTP_HOST} = $self->{server_name};
449			$ENV{HTTP_PORT} = $self->{server_port};
450	root	1.1	$ENV{SCRIPT_NAME} = $self->{name};
451	root	1.10	exec $path;
452	root	1.1	}
453			Coro::State::_exit(0);
454			} else {
455	root	1.29	die;
456	root	1.1	}
457			}
458
459	root	1.29	sub server_hostport {
460			$_[0]{server_port} == 80
461			? $_[0]{server_name}
462			: "$_[0]{server_name}:$_[0]{server_port}";
463			}
464
465	root	1.1	sub respond {
466			my $self = shift;
467			my $path = $self->{path};
468
469	root	1.49	if ($self->{name} =~ s%^/internal/([^/]+)%%) {
470			if ($::internal{$1}) {
471			$::internal{$1}->($self);
472			} else {
473	root	1.81	$self->err (404, "not found");
474	root	1.49	}
475			} else {
476	root	1.1
477	root	1.89	Coro::AIO::aio_stat $path
478			and $self->err (404, "not found");
479	root	1.10
480	root	1.49	$self->{stat} = [stat _];
481	root	1.1
482	root	1.49	# idiotic netscape sends idiotic headers AGAIN
483			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
484			? str2time $1 : 0;
485
486			if (-d _ && -r _) {
487			# directory
488			if ($path !~ /\/$/) {
489			# create a redirect to get the trailing "/"
490			# we don't try to avoid the :80
491	root	1.81	$self->err (301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
492	root	1.25	} else {
493	root	1.49	$ims < $self->{stat}[9]
494	root	1.81	or $self->err (304, "not modified");
495	root	1.49
496			if (-r "$path/index.html") {
497			# replace directory "size" by index.html filesize
498	root	1.55	$self->{stat} = [stat ($self->{path} .= "/index.html")];
499	root	1.81	$self->handle_file ($queue_index, $tbf_top);
500	root	1.49	} else {
501			$self->handle_dir;
502			}
503	root	1.1	}
504	root	1.49	} elsif (-f _ && -r _) {
505	root	1.81	-x _ and $self->err (403, "forbidden");
506	root	1.56
507	root	1.58	if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
508	root	1.56	my $timeout = $::NOW + 10;
509	root	1.58	while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
510	root	1.56	if ($timeout < $::NOW) {
511			$self->block($::BLOCKTIME, "too many connections");
512			} else {
513			$httpevent->wait;
514			}
515			}
516			}
517
518	root	1.81	$self->handle_file ($queue_file, $tbf_top);
519	root	1.49	} else {
520	root	1.81	$self->err (404, "not found");
521	root	1.1	}
522			}
523			}
524
525			sub handle_dir {
526			my $self = shift;
527	root	1.10	my $idx = $self->diridx;
528
529	root	1.81	$self->response (200, "ok",
530	root	1.10	{
531	root	1.76	"Content-Type" => "text/html; charset=utf-8",
532	root	1.10	"Content-Length" => length $idx,
533	root	1.55	"Last-Modified" => time2str ($self->{stat}[9]),
534	root	1.10	},
535			$idx);
536	root	1.1	}
537
538			sub handle_file {
539	root	1.67	my ($self, $queue, $tbf) = @_;
540	root	1.34	my $length = $self->{stat}[7];
541	root	1.1	my $hdr = {
542	root	1.75	"Last-Modified" => time2str ((stat _)[9]),
543			"Accept-Ranges" => "bytes",
544	root	1.1	};
545
546			my @code = (200, "ok");
547			my ($l, $h);
548
549	root	1.86	if ($self->{h}{range} =~ /^bytes=(.*)$/i) {
550	root	1.1	for (split /,/, $1) {
551			if (/^-(\d+)$/) {
552			($l, $h) = ($length - $1, $length - 1);
553			} elsif (/^(\d+)-(\d*)$/) {
554			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
555			} else {
556			($l, $h) = (0, $length - 1);
557			goto ignore;
558			}
559	root	1.26	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
560	root	1.1	}
561			$hdr->{"Content-Range"} = "bytes */$length";
562	root	1.24	$hdr->{"Content-Length"} = $length;
563	root	1.81	$self->err (416, "not satisfiable", $hdr, "");
564	root	1.1
565			satisfiable:
566	root	1.4	# check for segmented downloads
567	root	1.10	if ($l && $::NO_SEGMENTED) {
568	root	1.95	my $timeout = $::NOW + 60;
569	root	1.58	while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
570	root	1.56	if ($timeout <= $::NOW) {
571	root	1.95	$self->err_segmented_download;
572	root	1.29	} else {
573	root	1.46	$httpevent->wait;
574	root	1.29	}
575	root	1.4	}
576			}
577
578	root	1.1	$hdr->{"Content-Range"} = "bytes $l-$h/$length";
579			@code = (206, "partial content");
580			$length = $h - $l + 1;
581
582			ignore:
583			} else {
584			($l, $h) = (0, $length - 1);
585			}
586
587	root	1.9	$self->{path} =~ /\.([^.]+)$/;
588			$hdr->{"Content-Type"} = $mimetype{lc $1} \|\| "application/octet-stream";
589	root	1.1	$hdr->{"Content-Length"} = $length;
590
591	root	1.81	$self->response (@code, $hdr, "");
592	root	1.1
593			if ($self->{method} eq "GET") {
594	root	1.32	$self->{time} = $::NOW;
595	root	1.71	$self->{written} = 0;
596	root	1.32
597	root	1.90	my $fh = Coro::AIO::aio_open $self->{path}, Fcntl::O_RDONLY, 0
598	root	1.1	or die "$self->{path}: late open failure ($!)";
599
600			$h -= $l - 1;
601
602	root	1.81	my $transfer = $queue->start_transfer ($h);
603	root	1.51	my $locked;
604	root	1.48	my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
605	root	1.19
606	root	1.1	while ($h > 0) {
607	root	1.90	Coro::cede;
608			my $transfer_lock = $TRANSFER_LOCK->guard ($self->{remote_id});
609
610	root	1.51	unless ($locked) {
611	root	1.81	if ($locked \|\|= $transfer->try ($::WAIT_INTERVAL)) {
612	root	1.48	$bufsize = $::BUFSIZE;
613			$self->{time} = $::NOW;
614	root	1.71	$self->{written} = 0;
615	root	1.48	}
616	root	1.56	}
617
618			if ($blocked{$self->{remote_id}}) {
619			$self->{h}{connection} = "close";
620	root	1.59	die bless {}, err::;
621	root	1.48	}
622
623	root	1.80	Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), my $buf, 0
624	root	1.79	or last;
625	root	1.67
626	root	1.95	# readahead to work around rijk disk issues
627			IO::AIO::aio_readahead $fh, $l + $bufsize, $bufsize;
628
629	root	1.79	$tbf->request (length $buf);
630	root	1.87	my $w = $self->{fh}->syswrite ($buf)
631	root	1.1	or last;
632	root	1.11	$::written += $w;
633			$self->{written} += $w;
634	root	1.80	$l += $w;
635	root	1.1	}
636	root	1.32
637			close $fh;
638	root	1.1	}
639	root	1.7	}
640
641	root	1.78	1
642	root	1.79