Coro/myhttpd/httpd.pl

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

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';
   
sub slog {
   my $level = shift;
   my $format = shift;
   printf "---: $format\n", @_;
}

my $connections = new Coro::Semaphore $MAX_CONNECTS;

my @newcons;
my @pool;

# one "execution thread"
sub handler {
   while () {
      my $new = pop @newcons;
      if ($new) {
         eval {
            conn->new(@$new)->handle;
         };
         slog 1, "$@" if $@ && !ref $@;
         $connections->up;
      } else {
         last if @pool >= $MAX_POOL;
         push @pool, $Coro::current;
         schedule;
      }
   }
}

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

push @listen_sockets, $http_port;

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

   }
};

package conn;

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

our %conn; # $conn{ip}{fh} => connobj
our %blocked;

sub new {
   my $class = shift;
   my $peername = shift;
   my $fh = 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} = inet_ntoa $iaddr;

   # enter ourselves into various lists
   weaken ($conn{$self->{remote_addr}}{$self*1} = $self);

   $self;
}

sub DESTROY {
   my $self = shift;
   delete $conn{$self->{remote_addr}}{$self*1};
   delete $uri{$self->{uri}}{$self*1};
}

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

sub response {
   my ($self, $code, $msg, $hdr, $content) = @_;
   my $res = "HTTP/1.0 $code $msg\015\012";

   $res .= "Connection: close\015\012";
   $res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :(

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

   $res .= "\015\012$content" if defined $content;

   print STDERR "$self->{remote_addr} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}." \"$self->{h}{referer}\"\n";#d#

   print {$self->{fh}} $res;
}

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

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

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

   die bless {}, err::;
}

sub err_blocked {
   my $self = shift;
   my $ip = $self->{remote_addr};
   my $time = time2str $blocked{$ip} = $::NOW + $::BLOCKTIME;
   $self->err(403, "too many connections",
              {
                 "Content-Type" => "text/html",
                 "Retry-After" => $::BLOCKTIME
              },
              <<EOF);
<html><p>
You have been blocked because you opened too many connections. You
may retry at</p>

   <p><blockquote>$time.</blockquote></p>
   
<p>Until then, each new access will renew the block. You might want to have a
look at the <a href="http://www.goof.com/pcg/marc/animefaq.html">FAQ</a>.</p>
</html>
EOF
}

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

   #while() {
      $self->{h} = {};

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

      defined $req or
         $self->err(408, "request timeout");

      my $ip = $self->{remote_addr};

      if ($blocked{$ip}) {
         $self->err_blocked($blocked{$ip})
            if $blocked{$ip} > $::NOW;

         delete $blocked{$ip};
      }

      if (%{$conn{$ip}} > $::MAX_CONN_IP) {
         $self->slog("blocked ip $ip");
         $self->err_blocked;
      }

      $req =~ /^(?:\015\012)?
                (GET|HEAD) \040+
                ([^\040]+) \040+
                HTTP\/([0-9]+\.[0-9]+)
                \015\012/gx
         or $self->err(403, "method not allowed", { Allow => "GET,HEAD" });

      $2 ne "1.0"
         or $self->err(506, "http protocol version not supported");

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

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

         $hdr{lc $1} .= ",$2"
            while $req =~ /\G
                  ([^:\000-\040]+):
                  [\008\040]*
                  ((?: [^\015\012]+ | \015\012[\008\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;
      }

      $self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80;

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

      $self->map_uri;
      $self->respond;
   #}
}

# uri => path mapping
sub map_uri {
   my $self = shift;
   my $host = $self->{h}{host} || "default";
   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";
}

sub server_address {
   my $self = shift;
   my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->getsockname
      or $self->err(500, "unable to get socket name");
   ((inet_ntoa $iaddr), $port);
}

sub server_host {
   my $self = shift;
   if (exists $self->{h}{host}) {
      return $self->{h}{host};
   } else {
      return (($self->server_address)[0]);
   }
}

sub server_hostport {
   my $self = shift;
   my ($host, $port);
   if (exists $self->{h}{host}) {
      ($host, $port) = ($self->{h}{host}, $self->{server_port});
   } else {
      ($host, $port) = $self->server_address;
   }
   $port = $port == 80 ? "" : ":$port";
   $host.$port;
}

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_host;
         $ENV{HTTP_PORT}       = $self->{server_host};
         $ENV{SCRIPT_NAME}     = $self->{name};
         exec $::INDEXPROG;
      }
      Coro::State::_exit(0);
   } else {
   }
}

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

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

   # 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 "/"
         my $host = $self->server_hostport;
         $self->err(301, "moved permanently", { Location =>  "http://$host$self->{uri}/" });
      } else {
         $ims < (stat _)[9]
            or $self->err(304, "not modified");

         if ($self->{method} eq "GET") {
            if (-r "$path/index.html") {
               $self->{path} .= "/index.html";
               $self->handle_file;
            } else {
               $self->handle_dir;
            }
         }
      }
   } elsif (-f _ && -r _) {
      -x _ and $self->err(403, "forbidden");
      $self->handle_file;
   } else {
      $self->err(404, "not found");
   }
}

sub handle_dir {
   my $self = shift;
   $self->_cgi($::INDEXPROG);
}

sub handle_file {
   my $self = shift;
   my $length = -s _;
   my $hdr = {
      "Last-Modified"  => time2str ((stat _)[9]),
   };

   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";
      $self->err(416, "not satisfiable", $hdr);

satisfiable:
      # check for segmented downloads
      if ($l && $NO_SEGMENTED) {
         if (%{$uri{$self->{uri}}} > 1) {
            $self->slog("segmented download refused\n");
            $self->err(400, "segmented downloads are not allowed");
         }
      }

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

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

   if ($self->{path} =~ /\.html$/) {
      $hdr->{"Content-Type"} = "text/html";
   } else {
      $hdr->{"Content-Type"} = "application/octet-stream";
   }

   $hdr->{"Content-Length"} = $length;

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

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

      if ($l) {
         sysseek $fh, $l, 0
            or die "$self->{path}: cannot seek to $l ($!)";
      }

      $h -= $l - 1;

      while ($h > 0) {
         $h -= sysread $fh, $buf, $h > $::BUFSIZE ? $::BUFSIZE : $h;
         $self->{fh}->syswrite($buf)
            or last;
      }
   }

   close $fh;
}

1;
Revision:	1.6
Committed:	Fri Aug 10 21:03:41 2001 UTC (22 years, 11 months ago) by root
Content type:	text/plain
Branch:	MAIN
Changes since 1.5:	+8 -8 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
6			no utf8;
7			use bytes;
8
9			# at least on my machine, this thingy serves files
10			# quite a bit faster than apache, ;)
11			# and quite a bit slower than thttpd :(
12
13			$SIG{PIPE} = 'IGNORE';
14
15			sub slog {
16			my $level = shift;
17			my $format = shift;
18			printf "---: $format\n", @_;
19			}
20
21			my $connections = new Coro::Semaphore $MAX_CONNECTS;
22
23	root	1.6	my @newcons;
24	root	1.1	my @pool;
25
26	root	1.2	# one "execution thread"
27	root	1.1	sub handler {
28			while () {
29	root	1.6	my $new = pop @newcons;
30			if ($new) {
31	root	1.1	eval {
32	root	1.6	conn->new(@$new)->handle;
33	root	1.1	};
34			slog 1, "$@" if $@ && !ref $@;
35			$connections->up;
36			} else {
37			last if @pool >= $MAX_POOL;
38			push @pool, $Coro::current;
39			schedule;
40			}
41			}
42			}
43
44	root	1.4	my $http_port = new Coro::Socket
45			LocalAddr => $SERVER_HOST,
46			LocalPort => $SERVER_PORT,
47			ReuseAddr => 1,
48			Listen => 1,
49			or die "unable to start server";
50
51			push @listen_sockets, $http_port;
52
53	root	1.2	# the "main thread"
54	root	1.1	async {
55			slog 1, "accepting connections";
56			while () {
57			$connections->down;
58	root	1.6	push @newcons, [$http_port->accept];
59	root	1.1	#slog 3, "accepted @$connections ".scalar(@pool);
60	root	1.3	$::NOW = time;
61	root	1.1	if (@pool) {
62			(pop @pool)->ready;
63			} else {
64			async \&handler;
65			}
66
67			}
68			};
69
70			package conn;
71
72			use Socket;
73			use HTTP::Date;
74	root	1.2	use Convert::Scalar 'weaken';
75
76	root	1.3	our %conn; # $conn{ip}{fh} => connobj
77			our %blocked;
78	root	1.1
79			sub new {
80			my $class = shift;
81	root	1.6	my $peername = shift;
82	root	1.1	my $fh = shift;
83	root	1.2	my $self = bless { fh => $fh }, $class;
84	root	1.6	my (undef, $iaddr) = unpack_sockaddr_in $peername
85			or $self->err(500, "unable to decode peername");
86	root	1.3	$self->{remote_addr} = inet_ntoa $iaddr;
87	root	1.2
88			# enter ourselves into various lists
89	root	1.3	weaken ($conn{$self->{remote_addr}}{$self*1} = $self);
90
91	root	1.2	$self;
92			}
93
94			sub DESTROY {
95			my $self = shift;
96	root	1.3	delete $conn{$self->{remote_addr}}{$self*1};
97			delete $uri{$self->{uri}}{$self*1};
98	root	1.1	}
99
100			sub slog {
101	root	1.4	my $self = shift;
102			main::slog($_[0], "$self->{remote_addr}> $_[1]");
103	root	1.1	}
104
105	root	1.4	sub response {
106	root	1.1	my ($self, $code, $msg, $hdr, $content) = @_;
107			my $res = "HTTP/1.0 $code $msg\015\012";
108
109	root	1.4	$res .= "Connection: close\015\012";
110			$res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :(
111	root	1.1
112			while (my ($h, $v) = each %$hdr) {
113			$res .= "$h: $v\015\012"
114			}
115	root	1.4
116	root	1.1	$res .= "\015\012$content" if defined $content;
117
118	root	1.3	print STDERR "$self->{remote_addr} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}." \"$self->{h}{referer}\"\n";#d#
119	root	1.2
120	root	1.1	print {$self->{fh}} $res;
121			}
122
123			sub err {
124			my $self = shift;
125			my ($code, $msg, $hdr, $content) = @_;
126
127			unless (defined $content) {
128			$content = "$code $msg";
129			$hdr->{"Content-Type"} = "text/plain";
130			$hdr->{"Content-Length"} = length $content;
131			}
132
133	root	1.4	$self->response($code, $msg, $hdr, $content);
134	root	1.1
135			die bless {}, err::;
136			}
137
138	root	1.3	sub err_blocked {
139			my $self = shift;
140			my $ip = $self->{remote_addr};
141			my $time = time2str $blocked{$ip} = $::NOW + $::BLOCKTIME;
142			$self->err(403, "too many connections",
143	root	1.4	{
144			"Content-Type" => "text/html",
145			"Retry-After" => $::BLOCKTIME
146			},
147	root	1.3	<<EOF);
148	root	1.4	<html><p>
149	root	1.3	You have been blocked because you opened too many connections. You
150	root	1.4	may retry at</p>
151
152			<p><blockquote>$time.</blockquote></p>
153
154			<p>Until then, each new access will renew the block. You might want to have a
155			look at the <a href="http://www.goof.com/pcg/marc/animefaq.html">FAQ</a>.</p>
156			</html>
157	root	1.3	EOF
158			}
159
160	root	1.1	sub handle {
161			my $self = shift;
162			my $fh = $self->{fh};
163
164			#while() {
165			$self->{h} = {};
166
167			# read request and parse first line
168			$fh->timeout($::REQ_TIMEOUT);
169			my $req = $fh->readline("\015\012\015\012");
170			$fh->timeout($::RES_TIMEOUT);
171
172			defined $req or
173			$self->err(408, "request timeout");
174
175	root	1.3	my $ip = $self->{remote_addr};
176
177			if ($blocked{$ip}) {
178			$self->err_blocked($blocked{$ip})
179			if $blocked{$ip} > $::NOW;
180
181			delete $blocked{$ip};
182			}
183
184			if (%{$conn{$ip}} > $::MAX_CONN_IP) {
185			$self->slog("blocked ip $ip");
186			$self->err_blocked;
187			}
188
189	root	1.1	$req =~ /^(?:\015\012)?
190			(GET\|HEAD) \040+
191			([^\040]+) \040+
192			HTTP\/([0-9]+\.[0-9]+)
193			\015\012/gx
194			or $self->err(403, "method not allowed", { Allow => "GET,HEAD" });
195
196			$2 ne "1.0"
197			or $self->err(506, "http protocol version not supported");
198
199			$self->{method} = $1;
200			$self->{uri} = $2;
201
202			# parse headers
203			{
204			my (%hdr, $h, $v);
205
206			$hdr{lc $1} .= ",$2"
207			while $req =~ /\G
208			([^:\000-\040]+):
209			[\008\040]*
210			((?: [^\015\012]+ \| \015\012[\008\040] )*)
211			\015\012
212			/gxc;
213
214			$req =~ /\G\015\012$/
215			or $self->err(400, "bad request");
216
217			$self->{h}{$h} = substr $v, 1
218			while ($h, $v) = each %hdr;
219			}
220
221			$self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80;
222	root	1.3
223			weaken ($uri{$self->{uri}}{$self*1} = $self);
224	root	1.1
225			$self->map_uri;
226			$self->respond;
227			#}
228			}
229
230			# uri => path mapping
231			sub map_uri {
232			my $self = shift;
233			my $host = $self->{h}{host} \|\| "default";
234			my $uri = $self->{uri};
235
236			# some massaging, also makes it more secure
237			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
238			$uri =~ s%//+%/%g;
239			$uri =~ s%/\.(?=/\|$)%%g;
240			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
241
242			$uri =~ m%^/?\.\.(?=/\|$)%
243			and $self->err(400, "bad request");
244
245			$self->{name} = $uri;
246
247			# now do the path mapping
248			$self->{path} = "$::DOCROOT/$host$uri";
249			}
250
251			sub server_address {
252			my $self = shift;
253			my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->getsockname
254			or $self->err(500, "unable to get socket name");
255			((inet_ntoa $iaddr), $port);
256			}
257
258			sub server_host {
259			my $self = shift;
260			if (exists $self->{h}{host}) {
261			return $self->{h}{host};
262			} else {
263			return (($self->server_address)[0]);
264			}
265			}
266
267			sub server_hostport {
268			my $self = shift;
269			my ($host, $port);
270			if (exists $self->{h}{host}) {
271			($host, $port) = ($self->{h}{host}, $self->{server_port});
272			} else {
273			($host, $port) = $self->server_address;
274			}
275			$port = $port == 80 ? "" : ":$port";
276			$host.$port;
277			}
278
279			sub _cgi {
280			my $self = shift;
281			my $path = shift;
282			my $fh;
283
284			# no two-way xxx supported
285			if (0 == fork) {
286			open STDOUT, ">&".fileno($self->{fh});
287			if (chdir $::DOCROOT) {
288			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
289			$ENV{HTTP_HOST} = $self->server_host;
290			$ENV{HTTP_PORT} = $self->{server_host};
291			$ENV{SCRIPT_NAME} = $self->{name};
292			exec $::INDEXPROG;
293			}
294			Coro::State::_exit(0);
295			} else {
296			}
297			}
298
299			sub respond {
300			my $self = shift;
301			my $path = $self->{path};
302
303			stat $path
304			or $self->err(404, "not found");
305
306			# idiotic netscape sends idiotic headers AGAIN
307			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
308			? str2time $1 : 0;
309
310			if (-d _ && -r _) {
311			# directory
312			if ($path !~ /\/$/) {
313			# create a redirect to get the trailing "/"
314			my $host = $self->server_hostport;
315			$self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" });
316			} else {
317			$ims < (stat _)[9]
318			or $self->err(304, "not modified");
319
320			if ($self->{method} eq "GET") {
321			if (-r "$path/index.html") {
322			$self->{path} .= "/index.html";
323			$self->handle_file;
324			} else {
325			$self->handle_dir;
326			}
327			}
328			}
329			} elsif (-f _ && -r _) {
330			-x _ and $self->err(403, "forbidden");
331			$self->handle_file;
332			} else {
333			$self->err(404, "not found");
334			}
335			}
336
337			sub handle_dir {
338			my $self = shift;
339			$self->_cgi($::INDEXPROG);
340			}
341
342			sub handle_file {
343			my $self = shift;
344			my $length = -s _;
345			my $hdr = {
346			"Last-Modified" => time2str ((stat _)[9]),
347			};
348
349			my @code = (200, "ok");
350			my ($l, $h);
351
352			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
353			for (split /,/, $1) {
354			if (/^-(\d+)$/) {
355			($l, $h) = ($length - $1, $length - 1);
356			} elsif (/^(\d+)-(\d*)$/) {
357			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
358			} else {
359			($l, $h) = (0, $length - 1);
360			goto ignore;
361			}
362			goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h > $l;
363			}
364			$hdr->{"Content-Range"} = "bytes */$length";
365			$self->err(416, "not satisfiable", $hdr);
366
367			satisfiable:
368	root	1.4	# check for segmented downloads
369			if ($l && $NO_SEGMENTED) {
370			if (%{$uri{$self->{uri}}} > 1) {
371			$self->slog("segmented download refused\n");
372			$self->err(400, "segmented downloads are not allowed");
373			}
374			}
375
376	root	1.1	$hdr->{"Content-Range"} = "bytes $l-$h/$length";
377			@code = (206, "partial content");
378			$length = $h - $l + 1;
379
380			ignore:
381			} else {
382			($l, $h) = (0, $length - 1);
383			}
384
385			if ($self->{path} =~ /\.html$/) {
386			$hdr->{"Content-Type"} = "text/html";
387			} else {
388			$hdr->{"Content-Type"} = "application/octet-stream";
389			}
390
391			$hdr->{"Content-Length"} = $length;
392
393	root	1.4	$self->response(@code, $hdr, "");
394	root	1.1
395			if ($self->{method} eq "GET") {
396			my ($fh, $buf);
397			open $fh, "<", $self->{path}
398			or die "$self->{path}: late open failure ($!)";
399
400			if ($l) {
401			sysseek $fh, $l, 0
402			or die "$self->{path}: cannot seek to $l ($!)";
403			}
404
405			$h -= $l - 1;
406
407			while ($h > 0) {
408	root	1.5	$h -= sysread $fh, $buf, $h > $::BUFSIZE ? $::BUFSIZE : $h;
409			$self->{fh}->syswrite($buf)
410	root	1.1	or last;
411			}
412			}
413
414			close $fh;
415			}
416	root	1.2
417			1;