Coro/myhttpd/httpd.pl

use Coro;
use Coro::Semaphore;
use Coro::Event;
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;

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 strict;
use bytes;

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 $hash->{$_};
         }
      }
   }
}

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 $fh;

      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::;
         }

         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:	1.80
Committed:	Fri Dec 1 04:18:32 2006 UTC (17 years, 7 months ago) by root
Content type:	text/plain
Branch:	MAIN
Changes since 1.79:	+9 -6 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	1.79	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	1.56	Event->timer(interval => 1, hard => 1, cb => sub {
32			$NOW = time;
33			$HTTP_NOW = time2str $NOW;
34			})->now;
35
36			if ($ERROR_LOG) {
37			use IO::Handle;
38			open $errorlog, ">>$ERROR_LOG"
39			or die "$ERROR_LOG: $!";
40			$errorlog->autoflush(1);
41			}
42	root	1.27
43			if ($ACCESS_LOG) {
44			use IO::Handle;
45			open $accesslog, ">>$ACCESS_LOG"
46			or die "$ACCESS_LOG: $!";
47			$accesslog->autoflush(1);
48			}
49
50	root	1.1	sub slog {
51			my $level = shift;
52			my $format = shift;
53	root	1.56	my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
54			printf "$NOW: $format\n", @_;
55			printf $errorlog "$NOW: $format\n", @_ if $errorlog;
56	root	1.1	}
57
58	root	1.32	our $connections = new Coro::Semaphore $MAX_CONNECTS \|\| 250;
59	root	1.44	our $httpevent = new Coro::Signal;
60	root	1.34
61	root	1.63	our $queue_file = new transferqueue $MAX_TRANSFERS;
62			our $queue_index = new transferqueue 10;
63	root	1.62
64	root	1.72	our $tbf_top = new tbf rate => $TBF_RATE \|\| 100000;
65	root	1.67
66	root	1.68	my $unused_bytes = 0;
67			my $unused_last = time;
68
69			sub unused_bandwidth {
70			$unused_bytes += $_[0];
71			if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
72			$unused_last = $NOW;
73			$unused_bytes = 0;
74			$queue_file->force_wake_next;
75			slog 1, "forced filetransfer due to unused bandwidth";
76			}
77			}
78
79	root	1.6	my @newcons;
80	root	1.1	my @pool;
81
82	root	1.2	# one "execution thread"
83	root	1.1	sub handler {
84			while () {
85	root	1.38	if (@newcons) {
86	root	1.1	eval {
87	root	1.38	conn->new(@{pop @newcons})->handle;
88	root	1.1	};
89			slog 1, "$@" if $@ && !ref $@;
90	root	1.56
91			$httpevent->broadcast; # only for testing, but doesn't matter much
92
93	root	1.1	$connections->up;
94			} else {
95			last if @pool >= $MAX_POOL;
96			push @pool, $Coro::current;
97			schedule;
98			}
99			}
100			}
101
102	root	1.40	sub listen_on {
103			my $listen = $_[0];
104
105			push @listen_sockets, $listen;
106
107			# the "main thread"
108			async {
109			slog 1, "accepting connections";
110			while () {
111			$connections->down;
112			push @newcons, [$listen->accept];
113			#slog 3, "accepted @$connections ".scalar(@pool);
114			if (@pool) {
115			(pop @pool)->ready;
116			} else {
117			async \&handler;
118			}
119			}
120			};
121			}
122
123	root	1.4	my $http_port = new Coro::Socket
124			LocalAddr => $SERVER_HOST,
125			LocalPort => $SERVER_PORT,
126			ReuseAddr => 1,
127	root	1.13	Listen => 50,
128	root	1.4	or die "unable to start server";
129
130	root	1.40	listen_on $http_port;
131
132	root	1.41	if ($SERVER_PORT2) {
133			my $http_port = new Coro::Socket
134			LocalAddr => $SERVER_HOST,
135			LocalPort => $SERVER_PORT2,
136			ReuseAddr => 1,
137			Listen => 50,
138			or die "unable to start server";
139	root	1.40
140	root	1.41	listen_on $http_port;
141			}
142	root	1.4
143	root	1.1	package conn;
144
145	root	1.80	use strict;
146			use bytes;
147
148	root	1.1	use Socket;
149			use HTTP::Date;
150	root	1.2	use Convert::Scalar 'weaken';
151	root	1.77	use IO::AIO;
152	root	1.16
153	root	1.77	IO::AIO::min_parallel $::AIO_PARALLEL;
154	root	1.16
155	root	1.79	Event->io (fd => IO::AIO::poll_fileno,
156			poll => 'r', async => 1,
157			cb => \&IO::AIO::poll_cb);
158	root	1.16
159	root	1.26	our %conn; # $conn{ip}{self} => connobj
160			our %uri; # $uri{ip}{uri}{self}
161	root	1.3	our %blocked;
162	root	1.9	our %mimetype;
163
164			sub read_mimetypes {
165			local *M;
166	root	1.10	if (open M, "<mime_types") {
167	root	1.9	while (<M>) {
168			if (/^([^#]\S+)\t+(\S+)$/) {
169			$mimetype{lc $1} = $2;
170			}
171			}
172			} else {
173	root	1.10	print "cannot open mime_types\n";
174	root	1.9	}
175			}
176	root	1.1
177	root	1.10	read_mimetypes;
178
179	root	1.1	sub new {
180			my $class = shift;
181	root	1.42	my $fh = shift;
182	root	1.6	my $peername = shift;
183	root	1.2	my $self = bless { fh => $fh }, $class;
184	root	1.6	my (undef, $iaddr) = unpack_sockaddr_in $peername
185			or $self->err(500, "unable to decode peername");
186	root	1.7
187	root	1.50	$self->{remote_addr} =
188			$self->{remote_id} = inet_ntoa $iaddr;
189	root	1.60
190	root	1.11	$self->{time} = $::NOW;
191	root	1.2
192	root	1.49	weaken ($Coro::current->{conn} = $self);
193
194	root	1.13	$::conns++;
195	root	1.53	$::maxconns = $::conns if $::conns > $::maxconns;
196	root	1.13
197	root	1.2	$self;
198			}
199
200			sub DESTROY {
201	root	1.56	#my $self = shift;
202	root	1.13	$::conns--;
203	root	1.1	}
204
205	root	1.73	sub prune_cache {
206			my $hash = $_[0];
207
208			for (keys %$hash) {
209			if (ref $hash->{$_} eq HASH::) {
210			prune_cache($hash->{$_});
211			unless (scalar keys %{$hash->{$_}}) {
212			delete $hash->{$_};
213			}
214			}
215			}
216			}
217
218			sub prune_caches {
219			prune_cache \%conn;
220			prune_cache \%uri;
221
222			for (keys %blocked) {
223	root	1.74	delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
224	root	1.73	}
225			}
226
227			Event->timer(interval => 60, cb => \&prune_caches);
228
229	root	1.1	sub slog {
230	root	1.4	my $self = shift;
231	root	1.50	main::slog($_[0], "$self->{remote_id}> $_[1]");
232	root	1.1	}
233
234	root	1.4	sub response {
235	root	1.1	my ($self, $code, $msg, $hdr, $content) = @_;
236	root	1.17	my $res = "HTTP/1.1 $code $msg\015\012";
237	root	1.63	my $GZ = "";
238	root	1.1
239	root	1.46	if (exists $hdr->{Connection}) {
240			if ($hdr->{Connection} =~ /close/) {
241			$self->{h}{connection} = "close"
242			}
243			} else {
244			if ($self->{version} < 1.1) {
245			if ($self->{h}{connection} =~ /keep-alive/i) {
246			$hdr->{Connection} = "Keep-Alive";
247			} else {
248			$self->{h}{connection} = "close"
249			}
250			}
251			}
252	root	1.28
253	root	1.63	if ($self->{method} ne "HEAD"
254			&& $self->{h}{"accept-encoding"} =~ /\bgzip\b/
255	root	1.64	&& 400 < length $content
256	root	1.63	&& $hdr->{"Content-Length"} == length $content
257			&& !exists $hdr->{"Content-Encoding"}
258			) {
259			my $orig = length $content;
260			$hdr->{"Content-Encoding"} = "gzip";
261			$content = Compress::Zlib::memGzip(\$content);
262			$hdr->{"Content-Length"} = length $content;
263			$GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
264			}
265
266	root	1.32	$res .= "Date: $HTTP_NOW\015\012";
267	root	1.71	$res .= "Server: $::NAME\015\012";
268	root	1.1
269			while (my ($h, $v) = each %$hdr) {
270			$res .= "$h: $v\015\012"
271			}
272	root	1.10	$res .= "\015\012";
273	root	1.4
274	root	1.13	$res .= $content if defined $content and $self->{method} ne "HEAD";
275	root	1.1
276	root	1.65	my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
277	root	1.63	" $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
278	root	1.50	" \"$self->{h}{referer}\"\n";
279	root	1.27
280	root	1.65	print $::accesslog $log if $::accesslog;
281	root	1.27	print STDERR $log;
282	root	1.2
283	root	1.67	$tbf_top->request(length $res, 1e6);
284			$self->{written} += print {$self->{fh}} $res;
285	root	1.1	}
286
287			sub err {
288			my $self = shift;
289			my ($code, $msg, $hdr, $content) = @_;
290
291			unless (defined $content) {
292	root	1.35	$content = "$code $msg\n";
293	root	1.1	$hdr->{"Content-Type"} = "text/plain";
294			$hdr->{"Content-Length"} = length $content;
295			}
296	root	1.17	$hdr->{"Connection"} = "close";
297	root	1.1
298	root	1.4	$self->response($code, $msg, $hdr, $content);
299	root	1.1
300			die bless {}, err::;
301			}
302
303			sub handle {
304			my $self = shift;
305			my $fh = $self->{fh};
306
307	root	1.29	my $host;
308
309	root	1.17	$fh->timeout($::REQ_TIMEOUT);
310			while() {
311			$self->{reqs}++;
312	root	1.1
313			# read request and parse first line
314			my $req = $fh->readline("\015\012\015\012");
315
316	root	1.17	unless (defined $req) {
317			if (exists $self->{version}) {
318			last;
319			} else {
320			$self->err(408, "request timeout");
321			}
322			}
323
324			$self->{h} = {};
325	root	1.1
326	root	1.17	$fh->timeout($::RES_TIMEOUT);
327	root	1.3
328	root	1.1	$req =~ /^(?:\015\012)?
329			(GET\|HEAD) \040+
330			([^\040]+) \040+
331			HTTP\/([0-9]+\.[0-9]+)
332			\015\012/gx
333	root	1.14	or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
334	root	1.1
335			$self->{method} = $1;
336			$self->{uri} = $2;
337	root	1.17	$self->{version} = $3;
338
339	root	1.20	$3 =~ /^1\./
340	root	1.17	or $self->err(506, "http protocol version $3 not supported");
341	root	1.1
342			# parse headers
343			{
344			my (%hdr, $h, $v);
345
346			$hdr{lc $1} .= ",$2"
347			while $req =~ /\G
348			([^:\000-\040]+):
349	root	1.66	[\011\040]*
350			((?: [^\015\012]+ \| \015\012[\011\040] )*)
351	root	1.1	\015\012
352			/gxc;
353
354			$req =~ /\G\015\012$/
355			or $self->err(400, "bad request");
356
357			$self->{h}{$h} = substr $v, 1
358			while ($h, $v) = each %hdr;
359			}
360
361	root	1.36	# remote id should be unique per user
362			my $id = $self->{remote_addr};
363
364			if (exists $self->{h}{"client-ip"}) {
365			$id .= "[".$self->{h}{"client-ip"}."]";
366			} elsif (exists $self->{h}{"x-forwarded-for"}) {
367			$id .= "[".$self->{h}{"x-forwarded-for"}."]";
368			}
369
370			$self->{remote_id} = $id;
371
372	root	1.56	weaken (local $conn{$id}{$self*1} = $self);
373
374	root	1.36	if ($blocked{$id}) {
375	root	1.56	$self->err_blocked
376			if $blocked{$id}[0] > $::NOW;
377	root	1.36
378			delete $blocked{$id};
379			}
380
381	root	1.29	# find out server name and port
382			if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
383			$host = $1;
384			} else {
385			$host = $self->{h}{host};
386			}
387
388			if (defined $host) {
389			$self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
390			} else {
391			($self->{server_port}, $host)
392	root	1.43	= unpack_sockaddr_in $self->{fh}->sockname
393	root	1.29	or $self->err(500, "unable to get socket name");
394			$host = inet_ntoa $host;
395			}
396
397			$self->{server_name} = $host;
398
399	root	1.56	weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
400	root	1.1
401	root	1.24	eval {
402			$self->map_uri;
403			$self->respond;
404			};
405
406			die if $@ && !ref $@;
407	root	1.17
408	root	1.56	last if $self->{h}{connection} =~ /close/i;
409	root	1.17
410	root	1.44	$httpevent->broadcast;
411
412	root	1.17	$fh->timeout($::PER_TIMEOUT);
413			}
414	root	1.1	}
415
416	root	1.56	sub block {
417			my $self = shift;
418
419			$blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
420			$self->slog(2, "blocked ip $self->{remote_id}");
421			$self->err_blocked;
422			}
423
424	root	1.1	# uri => path mapping
425			sub map_uri {
426			my $self = shift;
427	root	1.29	my $host = $self->{server_name};
428	root	1.1	my $uri = $self->{uri};
429
430			# some massaging, also makes it more secure
431			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
432			$uri =~ s%//+%/%g;
433			$uri =~ s%/\.(?=/\|$)%%g;
434			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
435
436			$uri =~ m%^/?\.\.(?=/\|$)%
437			and $self->err(400, "bad request");
438
439			$self->{name} = $uri;
440
441			# now do the path mapping
442			$self->{path} = "$::DOCROOT/$host$uri";
443	root	1.7
444			$self->access_check;
445	root	1.1	}
446
447			sub _cgi {
448			my $self = shift;
449			my $path = shift;
450			my $fh;
451
452			# no two-way xxx supported
453			if (0 == fork) {
454			open STDOUT, ">&".fileno($self->{fh});
455			if (chdir $::DOCROOT) {
456			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
457	root	1.29	$ENV{HTTP_HOST} = $self->{server_name};
458			$ENV{HTTP_PORT} = $self->{server_port};
459	root	1.1	$ENV{SCRIPT_NAME} = $self->{name};
460	root	1.10	exec $path;
461	root	1.1	}
462			Coro::State::_exit(0);
463			} else {
464	root	1.29	die;
465	root	1.1	}
466			}
467
468	root	1.29	sub server_hostport {
469			$_[0]{server_port} == 80
470			? $_[0]{server_name}
471			: "$_[0]{server_name}:$_[0]{server_port}";
472			}
473
474	root	1.1	sub respond {
475			my $self = shift;
476			my $path = $self->{path};
477
478	root	1.49	if ($self->{name} =~ s%^/internal/([^/]+)%%) {
479			if ($::internal{$1}) {
480			$::internal{$1}->($self);
481			} else {
482			$self->err(404, "not found");
483			}
484			} else {
485	root	1.1
486	root	1.49	stat $path
487			or $self->err(404, "not found");
488	root	1.10
489	root	1.49	$self->{stat} = [stat _];
490	root	1.1
491	root	1.49	# idiotic netscape sends idiotic headers AGAIN
492			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
493			? str2time $1 : 0;
494
495			if (-d _ && -r _) {
496			# directory
497			if ($path !~ /\/$/) {
498			# create a redirect to get the trailing "/"
499			# we don't try to avoid the :80
500			$self->err(301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
501	root	1.25	} else {
502	root	1.49	$ims < $self->{stat}[9]
503			or $self->err(304, "not modified");
504
505			if (-r "$path/index.html") {
506			# replace directory "size" by index.html filesize
507	root	1.55	$self->{stat} = [stat ($self->{path} .= "/index.html")];
508	root	1.67	$self->handle_file($queue_index, $tbf_top);
509	root	1.49	} else {
510			$self->handle_dir;
511			}
512	root	1.1	}
513	root	1.49	} elsif (-f _ && -r _) {
514			-x _ and $self->err(403, "forbidden");
515	root	1.56
516	root	1.58	if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
517	root	1.56	my $timeout = $::NOW + 10;
518	root	1.58	while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
519	root	1.56	if ($timeout < $::NOW) {
520			$self->block($::BLOCKTIME, "too many connections");
521			} else {
522			$httpevent->wait;
523			}
524			}
525			}
526
527	root	1.67	$self->handle_file($queue_file, $tbf_top);
528	root	1.49	} else {
529			$self->err(404, "not found");
530	root	1.1	}
531			}
532			}
533
534			sub handle_dir {
535			my $self = shift;
536	root	1.10	my $idx = $self->diridx;
537
538			$self->response(200, "ok",
539			{
540	root	1.76	"Content-Type" => "text/html; charset=utf-8",
541	root	1.10	"Content-Length" => length $idx,
542	root	1.55	"Last-Modified" => time2str ($self->{stat}[9]),
543	root	1.10	},
544			$idx);
545	root	1.1	}
546
547			sub handle_file {
548	root	1.67	my ($self, $queue, $tbf) = @_;
549	root	1.34	my $length = $self->{stat}[7];
550	root	1.1	my $hdr = {
551	root	1.75	"Last-Modified" => time2str ((stat _)[9]),
552			"Accept-Ranges" => "bytes",
553	root	1.1	};
554
555			my @code = (200, "ok");
556			my ($l, $h);
557
558			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
559			for (split /,/, $1) {
560			if (/^-(\d+)$/) {
561			($l, $h) = ($length - $1, $length - 1);
562			} elsif (/^(\d+)-(\d*)$/) {
563			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
564			} else {
565			($l, $h) = (0, $length - 1);
566			goto ignore;
567			}
568	root	1.26	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
569	root	1.1	}
570			$hdr->{"Content-Range"} = "bytes */$length";
571	root	1.24	$hdr->{"Content-Length"} = $length;
572			$self->err(416, "not satisfiable", $hdr, "");
573	root	1.1
574			satisfiable:
575	root	1.4	# check for segmented downloads
576	root	1.10	if ($l && $::NO_SEGMENTED) {
577	root	1.56	my $timeout = $::NOW + 15;
578	root	1.58	while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
579	root	1.56	if ($timeout <= $::NOW) {
580			$self->block($::BLOCKTIME, "segmented downloads are forbidden");
581			#$self->err_segmented_download;
582	root	1.29	} else {
583	root	1.46	$httpevent->wait;
584	root	1.29	}
585	root	1.4	}
586			}
587
588	root	1.1	$hdr->{"Content-Range"} = "bytes $l-$h/$length";
589			@code = (206, "partial content");
590			$length = $h - $l + 1;
591
592			ignore:
593			} else {
594			($l, $h) = (0, $length - 1);
595			}
596
597	root	1.9	$self->{path} =~ /\.([^.]+)$/;
598			$hdr->{"Content-Type"} = $mimetype{lc $1} \|\| "application/octet-stream";
599	root	1.1	$hdr->{"Content-Length"} = $length;
600
601	root	1.4	$self->response(@code, $hdr, "");
602	root	1.1
603			if ($self->{method} eq "GET") {
604	root	1.32	$self->{time} = $::NOW;
605	root	1.71	$self->{written} = 0;
606	root	1.32
607	root	1.80	my $fh;
608	root	1.32
609	root	1.1	open $fh, "<", $self->{path}
610			or die "$self->{path}: late open failure ($!)";
611
612			$h -= $l - 1;
613
614	root	1.49	if (0) { # !AIO
615	root	1.19	if ($l) {
616			sysseek $fh, $l, 0;
617			}
618			}
619	root	1.48
620	root	1.56	my $transfer = $queue->start_transfer($h);
621	root	1.51	my $locked;
622	root	1.48	my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
623	root	1.19
624	root	1.1	while ($h > 0) {
625	root	1.51	unless ($locked) {
626			if ($locked \|\|= $transfer->try($::WAIT_INTERVAL)) {
627	root	1.48	$bufsize = $::BUFSIZE;
628			$self->{time} = $::NOW;
629	root	1.71	$self->{written} = 0;
630	root	1.48	}
631	root	1.56	}
632
633			if ($blocked{$self->{remote_id}}) {
634			$self->{h}{connection} = "close";
635	root	1.59	die bless {}, err::;
636	root	1.48	}
637
638	root	1.80	Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), my $buf, 0
639	root	1.79	or last;
640	root	1.67
641	root	1.79	$tbf->request (length $buf);
642	root	1.37	my $w = syswrite $self->{fh}, $buf
643	root	1.1	or last;
644	root	1.11	$::written += $w;
645			$self->{written} += $w;
646	root	1.80	$l += $w;
647	root	1.1	}
648	root	1.32
649			close $fh;
650	root	1.1	}
651	root	1.7	}
652
653	root	1.78	1
654	root	1.79