Coro/myhttpd/httpd.pl

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

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;

our $ERROR_LOG;
our $ACCESS_LOG;

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;
   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 strict;
use bytes;

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

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

our $AIO_WATCHER = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;

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 = EV::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 {

      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;

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

      $h -= $l - 1;

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

      while ($h > 0) {
         unless ($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;

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

      close $fh;
   }
}

1

Revision:	1.85
Committed:	Mon Feb 18 17:48:00 2008 UTC (16 years, 4 months ago) by root
Content type:	text/plain
Branch:	MAIN
CVS Tags:	rel-4_741, rel-4_742, rel-4_74, rel-4_71, rel-4_72, rel-4_73, rel-4_50, rel-4_51, rel-4_45, rel-4_49, rel-4_48, rel-4_47, rel-4_46, rel-4_7
Changes since 1.84:	+3 -0 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	use Coro;
2			use Coro::Semaphore;
3	root	1.84	use Coro::EV;
4	root	1.1	use Coro::Socket;
5	root	1.44	use Coro::Signal;
6	root	1.79	use Coro::AIO ();
7	root	1.1
8	root	1.32	use HTTP::Date;
9	root	1.47	use POSIX ();
10	root	1.32
11	root	1.63	use Compress::Zlib ();
12
13	root	1.1	no utf8;
14			use bytes;
15
16			# at least on my machine, this thingy serves files
17			# quite a bit faster than apache, ;)
18			# and quite a bit slower than thttpd :(
19
20			$SIG{PIPE} = 'IGNORE';
21	root	1.27
22			our $accesslog;
23	root	1.56	our $errorlog;
24
25			our $NOW;
26			our $HTTP_NOW;
27
28	root	1.80	our $ERROR_LOG;
29			our $ACCESS_LOG;
30
31	root	1.84	our $update_time = EV::periodic 0, 1, undef, sub {
32	root	1.56	$NOW = time;
33			$HTTP_NOW = time2str $NOW;
34	root	1.84	};
35			$update_time->invoke;
36	root	1.56
37			if ($ERROR_LOG) {
38			use IO::Handle;
39			open $errorlog, ">>$ERROR_LOG"
40			or die "$ERROR_LOG: $!";
41			$errorlog->autoflush(1);
42			}
43	root	1.27
44			if ($ACCESS_LOG) {
45			use IO::Handle;
46			open $accesslog, ">>$ACCESS_LOG"
47			or die "$ACCESS_LOG: $!";
48			$accesslog->autoflush(1);
49			}
50
51	root	1.1	sub slog {
52			my $level = shift;
53			my $format = shift;
54	root	1.56	my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
55			printf "$NOW: $format\n", @_;
56			printf $errorlog "$NOW: $format\n", @_ if $errorlog;
57	root	1.1	}
58
59	root	1.32	our $connections = new Coro::Semaphore $MAX_CONNECTS \|\| 250;
60	root	1.44	our $httpevent = new Coro::Signal;
61	root	1.34
62	root	1.63	our $queue_file = new transferqueue $MAX_TRANSFERS;
63			our $queue_index = new transferqueue 10;
64	root	1.62
65	root	1.72	our $tbf_top = new tbf rate => $TBF_RATE \|\| 100000;
66	root	1.67
67	root	1.68	my $unused_bytes = 0;
68			my $unused_last = time;
69
70			sub unused_bandwidth {
71			$unused_bytes += $_[0];
72			if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
73			$unused_last = $NOW;
74			$unused_bytes = 0;
75			$queue_file->force_wake_next;
76			slog 1, "forced filetransfer due to unused bandwidth";
77			}
78			}
79
80	root	1.6	my @newcons;
81	root	1.1	my @pool;
82
83	root	1.2	# one "execution thread"
84	root	1.1	sub handler {
85			while () {
86	root	1.38	if (@newcons) {
87	root	1.1	eval {
88	root	1.81	conn->new (@{pop @newcons})->handle;
89	root	1.1	};
90			slog 1, "$@" if $@ && !ref $@;
91	root	1.56
92			$httpevent->broadcast; # only for testing, but doesn't matter much
93
94	root	1.1	$connections->up;
95			} else {
96			last if @pool >= $MAX_POOL;
97			push @pool, $Coro::current;
98			schedule;
99			}
100			}
101			}
102
103	root	1.40	sub listen_on {
104			my $listen = $_[0];
105
106			push @listen_sockets, $listen;
107
108			# the "main thread"
109			async {
110			slog 1, "accepting connections";
111			while () {
112			$connections->down;
113			push @newcons, [$listen->accept];
114			#slog 3, "accepted @$connections ".scalar(@pool);
115			if (@pool) {
116			(pop @pool)->ready;
117			} else {
118			async \&handler;
119			}
120			}
121			};
122			}
123
124	root	1.4	my $http_port = new Coro::Socket
125			LocalAddr => $SERVER_HOST,
126			LocalPort => $SERVER_PORT,
127			ReuseAddr => 1,
128	root	1.81	Listen => 50,
129	root	1.4	or die "unable to start server";
130
131	root	1.40	listen_on $http_port;
132
133	root	1.41	if ($SERVER_PORT2) {
134			my $http_port = new Coro::Socket
135			LocalAddr => $SERVER_HOST,
136			LocalPort => $SERVER_PORT2,
137			ReuseAddr => 1,
138	root	1.81	Listen => 50,
139	root	1.41	or die "unable to start server";
140	root	1.40
141	root	1.41	listen_on $http_port;
142			}
143	root	1.4
144	root	1.1	package conn;
145
146	root	1.80	use strict;
147			use bytes;
148
149	root	1.1	use Socket;
150			use HTTP::Date;
151	root	1.2	use Convert::Scalar 'weaken';
152	root	1.77	use IO::AIO;
153	root	1.16
154	root	1.77	IO::AIO::min_parallel $::AIO_PARALLEL;
155	root	1.16
156	root	1.84	our $AIO_WATCHER = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
157	root	1.16
158	root	1.26	our %conn; # $conn{ip}{self} => connobj
159			our %uri; # $uri{ip}{uri}{self}
160	root	1.3	our %blocked;
161	root	1.9	our %mimetype;
162
163			sub read_mimetypes {
164	root	1.81	if (open my $fh, "<mime_types") {
165			while (<$fh>) {
166	root	1.9	if (/^([^#]\S+)\t+(\S+)$/) {
167			$mimetype{lc $1} = $2;
168			}
169			}
170			} else {
171	root	1.10	print "cannot open mime_types\n";
172	root	1.9	}
173			}
174	root	1.1
175	root	1.10	read_mimetypes;
176
177	root	1.1	sub new {
178			my $class = shift;
179	root	1.42	my $fh = shift;
180	root	1.6	my $peername = shift;
181	root	1.2	my $self = bless { fh => $fh }, $class;
182	root	1.6	my (undef, $iaddr) = unpack_sockaddr_in $peername
183	root	1.81	or $self->err (500, "unable to decode peername");
184	root	1.7
185	root	1.50	$self->{remote_addr} =
186			$self->{remote_id} = inet_ntoa $iaddr;
187	root	1.60
188	root	1.11	$self->{time} = $::NOW;
189	root	1.2
190	root	1.49	weaken ($Coro::current->{conn} = $self);
191
192	root	1.82	++$::conns;
193	root	1.53	$::maxconns = $::conns if $::conns > $::maxconns;
194	root	1.13
195	root	1.81	$self
196	root	1.2	}
197
198			sub DESTROY {
199	root	1.82	my $self = shift;
200
201	root	1.81	--$::conns;
202	root	1.1	}
203
204	root	1.73	sub prune_cache {
205			my $hash = $_[0];
206
207			for (keys %$hash) {
208			if (ref $hash->{$_} eq HASH::) {
209			prune_cache($hash->{$_});
210			unless (scalar keys %{$hash->{$_}}) {
211			delete $hash->{$_};
212			}
213			}
214			}
215			}
216
217			sub prune_caches {
218			prune_cache \%conn;
219			prune_cache \%uri;
220
221			for (keys %blocked) {
222	root	1.74	delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
223	root	1.73	}
224			}
225
226	root	1.84	our $PRUNE_WATCHER = EV::timer 60, 60, \&prune_caches;
227	root	1.73
228	root	1.1	sub slog {
229	root	1.4	my $self = shift;
230	root	1.50	main::slog($_[0], "$self->{remote_id}> $_[1]");
231	root	1.1	}
232
233	root	1.4	sub response {
234	root	1.1	my ($self, $code, $msg, $hdr, $content) = @_;
235	root	1.17	my $res = "HTTP/1.1 $code $msg\015\012";
236	root	1.63	my $GZ = "";
237	root	1.1
238	root	1.46	if (exists $hdr->{Connection}) {
239			if ($hdr->{Connection} =~ /close/) {
240			$self->{h}{connection} = "close"
241			}
242			} else {
243			if ($self->{version} < 1.1) {
244			if ($self->{h}{connection} =~ /keep-alive/i) {
245			$hdr->{Connection} = "Keep-Alive";
246			} else {
247			$self->{h}{connection} = "close"
248			}
249			}
250			}
251	root	1.28
252	root	1.63	if ($self->{method} ne "HEAD"
253			&& $self->{h}{"accept-encoding"} =~ /\bgzip\b/
254	root	1.64	&& 400 < length $content
255	root	1.63	&& $hdr->{"Content-Length"} == length $content
256			&& !exists $hdr->{"Content-Encoding"}
257			) {
258			my $orig = length $content;
259			$hdr->{"Content-Encoding"} = "gzip";
260			$content = Compress::Zlib::memGzip(\$content);
261			$hdr->{"Content-Length"} = length $content;
262			$GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
263			}
264
265	root	1.32	$res .= "Date: $HTTP_NOW\015\012";
266	root	1.71	$res .= "Server: $::NAME\015\012";
267	root	1.1
268			while (my ($h, $v) = each %$hdr) {
269			$res .= "$h: $v\015\012"
270			}
271	root	1.10	$res .= "\015\012";
272	root	1.4
273	root	1.13	$res .= $content if defined $content and $self->{method} ne "HEAD";
274	root	1.1
275	root	1.65	my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
276	root	1.63	" $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
277	root	1.50	" \"$self->{h}{referer}\"\n";
278	root	1.27
279	root	1.65	print $::accesslog $log if $::accesslog;
280	root	1.27	print STDERR $log;
281	root	1.2
282	root	1.67	$tbf_top->request(length $res, 1e6);
283			$self->{written} += print {$self->{fh}} $res;
284	root	1.1	}
285
286			sub err {
287			my $self = shift;
288			my ($code, $msg, $hdr, $content) = @_;
289
290			unless (defined $content) {
291	root	1.35	$content = "$code $msg\n";
292	root	1.1	$hdr->{"Content-Type"} = "text/plain";
293			$hdr->{"Content-Length"} = length $content;
294			}
295	root	1.17	$hdr->{"Connection"} = "close";
296	root	1.1
297	root	1.81	$self->response ($code, $msg, $hdr, $content);
298	root	1.1
299	root	1.81	die bless {}, err::
300	root	1.1	}
301
302			sub handle {
303			my $self = shift;
304			my $fh = $self->{fh};
305
306	root	1.29	my $host;
307
308	root	1.17	$fh->timeout($::REQ_TIMEOUT);
309	root	1.81	while () {
310	root	1.17	$self->{reqs}++;
311	root	1.1
312			# read request and parse first line
313			my $req = $fh->readline("\015\012\015\012");
314
315	root	1.17	unless (defined $req) {
316			if (exists $self->{version}) {
317			last;
318			} else {
319			$self->err(408, "request timeout");
320			}
321			}
322
323			$self->{h} = {};
324	root	1.1
325	root	1.17	$fh->timeout($::RES_TIMEOUT);
326	root	1.3
327	root	1.1	$req =~ /^(?:\015\012)?
328			(GET\|HEAD) \040+
329			([^\040]+) \040+
330			HTTP\/([0-9]+\.[0-9]+)
331			\015\012/gx
332	root	1.14	or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
333	root	1.1
334			$self->{method} = $1;
335			$self->{uri} = $2;
336	root	1.17	$self->{version} = $3;
337
338	root	1.20	$3 =~ /^1\./
339	root	1.17	or $self->err(506, "http protocol version $3 not supported");
340	root	1.1
341			# parse headers
342			{
343			my (%hdr, $h, $v);
344
345			$hdr{lc $1} .= ",$2"
346			while $req =~ /\G
347			([^:\000-\040]+):
348	root	1.66	[\011\040]*
349			((?: [^\015\012]+ \| \015\012[\011\040] )*)
350	root	1.1	\015\012
351			/gxc;
352
353			$req =~ /\G\015\012$/
354			or $self->err(400, "bad request");
355
356			$self->{h}{$h} = substr $v, 1
357			while ($h, $v) = each %hdr;
358			}
359
360	root	1.36	# remote id should be unique per user
361			my $id = $self->{remote_addr};
362
363			if (exists $self->{h}{"client-ip"}) {
364			$id .= "[".$self->{h}{"client-ip"}."]";
365			} elsif (exists $self->{h}{"x-forwarded-for"}) {
366			$id .= "[".$self->{h}{"x-forwarded-for"}."]";
367			}
368
369			$self->{remote_id} = $id;
370
371	root	1.56	weaken (local $conn{$id}{$self*1} = $self);
372
373	root	1.36	if ($blocked{$id}) {
374	root	1.56	$self->err_blocked
375			if $blocked{$id}[0] > $::NOW;
376	root	1.36
377			delete $blocked{$id};
378			}
379
380	root	1.29	# find out server name and port
381			if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
382			$host = $1;
383			} else {
384			$host = $self->{h}{host};
385			}
386
387			if (defined $host) {
388			$self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
389			} else {
390			($self->{server_port}, $host)
391	root	1.43	= unpack_sockaddr_in $self->{fh}->sockname
392	root	1.29	or $self->err(500, "unable to get socket name");
393			$host = inet_ntoa $host;
394			}
395
396			$self->{server_name} = $host;
397
398	root	1.56	weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
399	root	1.1
400	root	1.24	eval {
401			$self->map_uri;
402			$self->respond;
403			};
404
405			die if $@ && !ref $@;
406	root	1.17
407	root	1.56	last if $self->{h}{connection} =~ /close/i;
408	root	1.17
409	root	1.44	$httpevent->broadcast;
410
411	root	1.17	$fh->timeout($::PER_TIMEOUT);
412			}
413	root	1.1	}
414
415	root	1.56	sub block {
416			my $self = shift;
417
418			$blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
419			$self->slog(2, "blocked ip $self->{remote_id}");
420			$self->err_blocked;
421			}
422
423	root	1.1	# uri => path mapping
424			sub map_uri {
425			my $self = shift;
426	root	1.29	my $host = $self->{server_name};
427	root	1.1	my $uri = $self->{uri};
428
429	root	1.85	$host =~ /[\/\\]/
430			and $self->err(400, "bad request");
431
432	root	1.1	# some massaging, also makes it more secure
433			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
434			$uri =~ s%//+%/%g;
435			$uri =~ s%/\.(?=/\|$)%%g;
436			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
437
438			$uri =~ m%^/?\.\.(?=/\|$)%
439			and $self->err(400, "bad request");
440
441			$self->{name} = $uri;
442
443			# now do the path mapping
444			$self->{path} = "$::DOCROOT/$host$uri";
445	root	1.7
446			$self->access_check;
447	root	1.1	}
448
449			sub _cgi {
450			my $self = shift;
451			my $path = shift;
452			my $fh;
453
454			# no two-way xxx supported
455			if (0 == fork) {
456			open STDOUT, ">&".fileno($self->{fh});
457			if (chdir $::DOCROOT) {
458			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
459	root	1.29	$ENV{HTTP_HOST} = $self->{server_name};
460			$ENV{HTTP_PORT} = $self->{server_port};
461	root	1.1	$ENV{SCRIPT_NAME} = $self->{name};
462	root	1.10	exec $path;
463	root	1.1	}
464			Coro::State::_exit(0);
465			} else {
466	root	1.29	die;
467	root	1.1	}
468			}
469
470	root	1.29	sub server_hostport {
471			$_[0]{server_port} == 80
472			? $_[0]{server_name}
473			: "$_[0]{server_name}:$_[0]{server_port}";
474			}
475
476	root	1.1	sub respond {
477			my $self = shift;
478			my $path = $self->{path};
479
480	root	1.49	if ($self->{name} =~ s%^/internal/([^/]+)%%) {
481			if ($::internal{$1}) {
482			$::internal{$1}->($self);
483			} else {
484	root	1.81	$self->err (404, "not found");
485	root	1.49	}
486			} else {
487	root	1.1
488	root	1.49	stat $path
489	root	1.81	or $self->err (404, "not found");
490	root	1.10
491	root	1.49	$self->{stat} = [stat _];
492	root	1.1
493	root	1.49	# idiotic netscape sends idiotic headers AGAIN
494			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
495			? str2time $1 : 0;
496
497			if (-d _ && -r _) {
498			# directory
499			if ($path !~ /\/$/) {
500			# create a redirect to get the trailing "/"
501			# we don't try to avoid the :80
502	root	1.81	$self->err (301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
503	root	1.25	} else {
504	root	1.49	$ims < $self->{stat}[9]
505	root	1.81	or $self->err (304, "not modified");
506	root	1.49
507			if (-r "$path/index.html") {
508			# replace directory "size" by index.html filesize
509	root	1.55	$self->{stat} = [stat ($self->{path} .= "/index.html")];
510	root	1.81	$self->handle_file ($queue_index, $tbf_top);
511	root	1.49	} else {
512			$self->handle_dir;
513			}
514	root	1.1	}
515	root	1.49	} elsif (-f _ && -r _) {
516	root	1.81	-x _ and $self->err (403, "forbidden");
517	root	1.56
518	root	1.58	if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
519	root	1.56	my $timeout = $::NOW + 10;
520	root	1.58	while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
521	root	1.56	if ($timeout < $::NOW) {
522			$self->block($::BLOCKTIME, "too many connections");
523			} else {
524			$httpevent->wait;
525			}
526			}
527			}
528
529	root	1.81	$self->handle_file ($queue_file, $tbf_top);
530	root	1.49	} else {
531	root	1.81	$self->err (404, "not found");
532	root	1.1	}
533			}
534			}
535
536			sub handle_dir {
537			my $self = shift;
538	root	1.10	my $idx = $self->diridx;
539
540	root	1.81	$self->response (200, "ok",
541	root	1.10	{
542	root	1.76	"Content-Type" => "text/html; charset=utf-8",
543	root	1.10	"Content-Length" => length $idx,
544	root	1.55	"Last-Modified" => time2str ($self->{stat}[9]),
545	root	1.10	},
546			$idx);
547	root	1.1	}
548
549			sub handle_file {
550	root	1.67	my ($self, $queue, $tbf) = @_;
551	root	1.34	my $length = $self->{stat}[7];
552	root	1.1	my $hdr = {
553	root	1.75	"Last-Modified" => time2str ((stat _)[9]),
554			"Accept-Ranges" => "bytes",
555	root	1.1	};
556
557			my @code = (200, "ok");
558			my ($l, $h);
559
560			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
561			for (split /,/, $1) {
562			if (/^-(\d+)$/) {
563			($l, $h) = ($length - $1, $length - 1);
564			} elsif (/^(\d+)-(\d*)$/) {
565			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
566			} else {
567			($l, $h) = (0, $length - 1);
568			goto ignore;
569			}
570	root	1.26	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
571	root	1.1	}
572			$hdr->{"Content-Range"} = "bytes */$length";
573	root	1.24	$hdr->{"Content-Length"} = $length;
574	root	1.81	$self->err (416, "not satisfiable", $hdr, "");
575	root	1.1
576			satisfiable:
577	root	1.4	# check for segmented downloads
578	root	1.10	if ($l && $::NO_SEGMENTED) {
579	root	1.56	my $timeout = $::NOW + 15;
580	root	1.58	while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
581	root	1.56	if ($timeout <= $::NOW) {
582	root	1.81	$self->block ($::BLOCKTIME, "segmented downloads are forbidden");
583	root	1.56	#$self->err_segmented_download;
584	root	1.29	} else {
585	root	1.46	$httpevent->wait;
586	root	1.29	}
587	root	1.4	}
588			}
589
590	root	1.1	$hdr->{"Content-Range"} = "bytes $l-$h/$length";
591			@code = (206, "partial content");
592			$length = $h - $l + 1;
593
594			ignore:
595			} else {
596			($l, $h) = (0, $length - 1);
597			}
598
599	root	1.9	$self->{path} =~ /\.([^.]+)$/;
600			$hdr->{"Content-Type"} = $mimetype{lc $1} \|\| "application/octet-stream";
601	root	1.1	$hdr->{"Content-Length"} = $length;
602
603	root	1.81	$self->response (@code, $hdr, "");
604	root	1.1
605			if ($self->{method} eq "GET") {
606	root	1.32	$self->{time} = $::NOW;
607	root	1.71	$self->{written} = 0;
608	root	1.32
609	root	1.81	open my $fh, "<", $self->{path}
610	root	1.1	or die "$self->{path}: late open failure ($!)";
611
612			$h -= $l - 1;
613
614	root	1.81	my $transfer = $queue->start_transfer ($h);
615	root	1.51	my $locked;
616	root	1.48	my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
617	root	1.19
618	root	1.1	while ($h > 0) {
619	root	1.51	unless ($locked) {
620	root	1.81	if ($locked \|\|= $transfer->try ($::WAIT_INTERVAL)) {
621	root	1.48	$bufsize = $::BUFSIZE;
622			$self->{time} = $::NOW;
623	root	1.71	$self->{written} = 0;
624	root	1.48	}
625	root	1.56	}
626
627			if ($blocked{$self->{remote_id}}) {
628			$self->{h}{connection} = "close";
629	root	1.59	die bless {}, err::;
630	root	1.48	}
631
632	root	1.80	Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), my $buf, 0
633	root	1.79	or last;
634	root	1.67
635	root	1.79	$tbf->request (length $buf);
636	root	1.37	my $w = syswrite $self->{fh}, $buf
637	root	1.1	or last;
638	root	1.11	$::written += $w;
639			$self->{written} += $w;
640	root	1.80	$l += $w;
641	root	1.1	}
642	root	1.32
643			close $fh;
644	root	1.1	}
645	root	1.7	}
646
647	root	1.78	1
648	root	1.79