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 @fh;
my @pool;

# one "execution thread"
sub handler {
   while () {
      my $fh = pop @fh;
      if ($fh) {
         eval {
            conn->new($fh)->handle;
         };
         close $fh;
         slog 1, "$@" 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 @fh, $http_port->accept;
      #slog 3, "accepted @$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 $fh = shift;
   my $self = bless { fh => $fh }, $class;
   my (undef, $iaddr) = unpack_sockaddr_in $fh->getpeername
      or $self->err(500, "unable to get 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;

      Coro::Event::do_timer(after => 5);
      
      $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 > 16384 ? 16384 : $h;
         print {$self->{fh}} $buf
            or last;
         cede;
      }
   }

   close $fh;
}

1;
Revision:	1.4
Committed:	Fri Aug 10 04:16:41 2001 UTC (22 years, 11 months ago) by root
Content type:	text/plain
Branch:	MAIN
Changes since 1.3:	+40 -23 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			my @fh;
24			my @pool;
25
26	root	1.2	# one "execution thread"
27	root	1.1	sub handler {
28			while () {
29			my $fh = pop @fh;
30			if ($fh) {
31			eval {
32			conn->new($fh)->handle;
33			};
34			close $fh;
35			slog 1, "$@" if $@ && !ref $@;
36			$connections->up;
37			} else {
38			last if @pool >= $MAX_POOL;
39			push @pool, $Coro::current;
40			schedule;
41			}
42			}
43			}
44
45	root	1.4	my $http_port = new Coro::Socket
46			LocalAddr => $SERVER_HOST,
47			LocalPort => $SERVER_PORT,
48			ReuseAddr => 1,
49			Listen => 1,
50			or die "unable to start server";
51
52			push @listen_sockets, $http_port;
53
54	root	1.2	# the "main thread"
55	root	1.1	async {
56			slog 1, "accepting connections";
57			while () {
58			$connections->down;
59	root	1.4	push @fh, $http_port->accept;
60	root	1.1	#slog 3, "accepted @$connections ".scalar(@pool);
61	root	1.3	$::NOW = time;
62	root	1.1	if (@pool) {
63			(pop @pool)->ready;
64			} else {
65			async \&handler;
66			}
67
68			}
69			};
70
71			package conn;
72
73			use Socket;
74			use HTTP::Date;
75	root	1.2	use Convert::Scalar 'weaken';
76
77	root	1.3	our %conn; # $conn{ip}{fh} => connobj
78			our %blocked;
79	root	1.1
80			sub new {
81			my $class = shift;
82			my $fh = shift;
83	root	1.2	my $self = bless { fh => $fh }, $class;
84	root	1.1	my (undef, $iaddr) = unpack_sockaddr_in $fh->getpeername
85			or $self->err(500, "unable to get 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	root	1.2
227			Coro::Event::do_timer(after => 5);
228
229	root	1.1	$self->respond;
230			#}
231			}
232
233			# uri => path mapping
234			sub map_uri {
235			my $self = shift;
236			my $host = $self->{h}{host} \|\| "default";
237			my $uri = $self->{uri};
238
239			# some massaging, also makes it more secure
240			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
241			$uri =~ s%//+%/%g;
242			$uri =~ s%/\.(?=/\|$)%%g;
243			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
244
245			$uri =~ m%^/?\.\.(?=/\|$)%
246			and $self->err(400, "bad request");
247
248			$self->{name} = $uri;
249
250			# now do the path mapping
251			$self->{path} = "$::DOCROOT/$host$uri";
252			}
253
254			sub server_address {
255			my $self = shift;
256			my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->getsockname
257			or $self->err(500, "unable to get socket name");
258			((inet_ntoa $iaddr), $port);
259			}
260
261			sub server_host {
262			my $self = shift;
263			if (exists $self->{h}{host}) {
264			return $self->{h}{host};
265			} else {
266			return (($self->server_address)[0]);
267			}
268			}
269
270			sub server_hostport {
271			my $self = shift;
272			my ($host, $port);
273			if (exists $self->{h}{host}) {
274			($host, $port) = ($self->{h}{host}, $self->{server_port});
275			} else {
276			($host, $port) = $self->server_address;
277			}
278			$port = $port == 80 ? "" : ":$port";
279			$host.$port;
280			}
281
282			sub _cgi {
283			my $self = shift;
284			my $path = shift;
285			my $fh;
286
287			# no two-way xxx supported
288			if (0 == fork) {
289			open STDOUT, ">&".fileno($self->{fh});
290			if (chdir $::DOCROOT) {
291			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
292			$ENV{HTTP_HOST} = $self->server_host;
293			$ENV{HTTP_PORT} = $self->{server_host};
294			$ENV{SCRIPT_NAME} = $self->{name};
295			exec $::INDEXPROG;
296			}
297			Coro::State::_exit(0);
298			} else {
299			}
300			}
301
302			sub respond {
303			my $self = shift;
304			my $path = $self->{path};
305
306			stat $path
307			or $self->err(404, "not found");
308
309			# idiotic netscape sends idiotic headers AGAIN
310			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
311			? str2time $1 : 0;
312
313			if (-d _ && -r _) {
314			# directory
315			if ($path !~ /\/$/) {
316			# create a redirect to get the trailing "/"
317			my $host = $self->server_hostport;
318			$self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" });
319			} else {
320			$ims < (stat _)[9]
321			or $self->err(304, "not modified");
322
323			if ($self->{method} eq "GET") {
324			if (-r "$path/index.html") {
325			$self->{path} .= "/index.html";
326			$self->handle_file;
327			} else {
328			$self->handle_dir;
329			}
330			}
331			}
332			} elsif (-f _ && -r _) {
333			-x _ and $self->err(403, "forbidden");
334			$self->handle_file;
335			} else {
336			$self->err(404, "not found");
337			}
338			}
339
340			sub handle_dir {
341			my $self = shift;
342			$self->_cgi($::INDEXPROG);
343			}
344
345			sub handle_file {
346			my $self = shift;
347			my $length = -s _;
348			my $hdr = {
349			"Last-Modified" => time2str ((stat _)[9]),
350			};
351
352			my @code = (200, "ok");
353			my ($l, $h);
354
355			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
356			for (split /,/, $1) {
357			if (/^-(\d+)$/) {
358			($l, $h) = ($length - $1, $length - 1);
359			} elsif (/^(\d+)-(\d*)$/) {
360			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
361			} else {
362			($l, $h) = (0, $length - 1);
363			goto ignore;
364			}
365			goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h > $l;
366			}
367			$hdr->{"Content-Range"} = "bytes */$length";
368			$self->err(416, "not satisfiable", $hdr);
369
370			satisfiable:
371	root	1.4	# check for segmented downloads
372			if ($l && $NO_SEGMENTED) {
373			if (%{$uri{$self->{uri}}} > 1) {
374			$self->slog("segmented download refused\n");
375			$self->err(400, "segmented downloads are not allowed");
376			}
377			}
378
379	root	1.1	$hdr->{"Content-Range"} = "bytes $l-$h/$length";
380			@code = (206, "partial content");
381			$length = $h - $l + 1;
382
383			ignore:
384			} else {
385			($l, $h) = (0, $length - 1);
386			}
387
388			if ($self->{path} =~ /\.html$/) {
389			$hdr->{"Content-Type"} = "text/html";
390			} else {
391			$hdr->{"Content-Type"} = "application/octet-stream";
392			}
393
394			$hdr->{"Content-Length"} = $length;
395
396	root	1.4	$self->response(@code, $hdr, "");
397	root	1.1
398			if ($self->{method} eq "GET") {
399			my ($fh, $buf);
400			open $fh, "<", $self->{path}
401			or die "$self->{path}: late open failure ($!)";
402
403			if ($l) {
404			sysseek $fh, $l, 0
405			or die "$self->{path}: cannot seek to $l ($!)";
406			}
407
408			$h -= $l - 1;
409
410			while ($h > 0) {
411	root	1.4	$h -= sysread $fh, $buf, $h > 16384 ? 16384 : $h;
412	root	1.1	print {$self->{fh}} $buf
413			or last;
414	root	1.4	cede;
415	root	1.1	}
416			}
417
418			close $fh;
419			}
420	root	1.2
421			1;