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 :(

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

$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;
      }
   }
}

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

   }
};

loop;
print "ende\n";#d#

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);

   print "$self->{remote_addr}: ".($self*1)." > ".%{$conn{$self->{remote_addr}}},"\n";
   $self;
}

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

sub slog {
   main::slog(@_);
}

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

   $hdr->{Date} = time2str $::NOW; # 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;
   }

   $self->slog($msg) if $code;

   $self->print_response($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",
              { "Retry-After" => $::BLOCKTIME },
              <<EOF);
You have been blocked because you opened too many connections. You
may retry at $time. Until then,
every new access will renew the block.
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:
      $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->print_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 > 4096 ? 4096 : $h;
         print {$self->{fh}} $buf
            or last;
      }
   }

   close $fh;
}

1;
Revision:	1.3
Committed:	Fri Aug 10 02:28:28 2001 UTC (22 years, 11 months ago) by root
Content type:	text/plain
Branch:	MAIN
Changes since 1.2:	+40 -7 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			my $port = new Coro::Socket
14			LocalAddr => $SERVER_HOST,
15			LocalPort => $SERVER_PORT,
16			ReuseAddr => 1,
17			Listen => 1,
18			or die "unable to start server";
19
20			$SIG{PIPE} = 'IGNORE';
21
22			sub slog {
23			my $level = shift;
24			my $format = shift;
25			printf "---: $format\n", @_;
26			}
27
28			my $connections = new Coro::Semaphore $MAX_CONNECTS;
29
30			my @fh;
31			my @pool;
32
33	root	1.2	# one "execution thread"
34	root	1.1	sub handler {
35			while () {
36			my $fh = pop @fh;
37			if ($fh) {
38			eval {
39			conn->new($fh)->handle;
40			};
41			close $fh;
42			slog 1, "$@" if $@ && !ref $@;
43			$connections->up;
44			} else {
45			last if @pool >= $MAX_POOL;
46			push @pool, $Coro::current;
47			schedule;
48			}
49			}
50			}
51
52	root	1.2	# the "main thread"
53	root	1.1	async {
54			slog 1, "accepting connections";
55			while () {
56			$connections->down;
57			push @fh, $port->accept;
58			#slog 3, "accepted @$connections ".scalar(@pool);
59	root	1.3	$::NOW = time;
60	root	1.1	if (@pool) {
61			(pop @pool)->ready;
62			} else {
63			async \&handler;
64			}
65
66			}
67			};
68
69			loop;
70			print "ende\n";#d#
71
72			package conn;
73
74			use Socket;
75			use HTTP::Date;
76	root	1.2	use Convert::Scalar 'weaken';
77
78	root	1.3	our %conn; # $conn{ip}{fh} => connobj
79			our %blocked;
80	root	1.1
81			sub new {
82			my $class = shift;
83			my $fh = shift;
84	root	1.2	my $self = bless { fh => $fh }, $class;
85	root	1.1	my (undef, $iaddr) = unpack_sockaddr_in $fh->getpeername
86			or $self->err(500, "unable to get peername");
87	root	1.3	$self->{remote_addr} = inet_ntoa $iaddr;
88	root	1.2
89			# enter ourselves into various lists
90	root	1.3	weaken ($conn{$self->{remote_addr}}{$self*1} = $self);
91
92			print "$self->{remote_addr}: ".($self*1)." > ".%{$conn{$self->{remote_addr}}},"\n";
93	root	1.2	$self;
94			}
95
96			sub DESTROY {
97			my $self = shift;
98	root	1.3	delete $conn{$self->{remote_addr}}{$self*1};
99			delete $uri{$self->{uri}}{$self*1};
100	root	1.1	}
101
102			sub slog {
103			main::slog(@_);
104			}
105
106			sub print_response {
107			my ($self, $code, $msg, $hdr, $content) = @_;
108			my $res = "HTTP/1.0 $code $msg\015\012";
109
110	root	1.3	$hdr->{Date} = time2str $::NOW; # slow? nah.
111	root	1.1
112			while (my ($h, $v) = each %$hdr) {
113			$res .= "$h: $v\015\012"
114			}
115			$res .= "\015\012$content" if defined $content;
116
117	root	1.3	print STDERR "$self->{remote_addr} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}." \"$self->{h}{referer}\"\n";#d#
118	root	1.2
119	root	1.1	print {$self->{fh}} $res;
120			}
121
122			sub err {
123			my $self = shift;
124			my ($code, $msg, $hdr, $content) = @_;
125
126			unless (defined $content) {
127			$content = "$code $msg";
128			$hdr->{"Content-Type"} = "text/plain";
129			$hdr->{"Content-Length"} = length $content;
130			}
131
132			$self->slog($msg) if $code;
133
134			$self->print_response($code, $msg, $hdr, $content);
135
136			die bless {}, err::;
137			}
138
139	root	1.3	sub err_blocked {
140			my $self = shift;
141			my $ip = $self->{remote_addr};
142			my $time = time2str $blocked{$ip} = $::NOW + $::BLOCKTIME;
143			$self->err(403, "too many connections",
144			{ "Retry-After" => $::BLOCKTIME },
145			<<EOF);
146			You have been blocked because you opened too many connections. You
147			may retry at $time. Until then,
148			every new access will renew the block.
149			EOF
150			}
151
152	root	1.1	sub handle {
153			my $self = shift;
154			my $fh = $self->{fh};
155
156			#while() {
157			$self->{h} = {};
158
159			# read request and parse first line
160			$fh->timeout($::REQ_TIMEOUT);
161			my $req = $fh->readline("\015\012\015\012");
162			$fh->timeout($::RES_TIMEOUT);
163
164			defined $req or
165			$self->err(408, "request timeout");
166
167	root	1.3	my $ip = $self->{remote_addr};
168
169			if ($blocked{$ip}) {
170			$self->err_blocked($blocked{$ip})
171			if $blocked{$ip} > $::NOW;
172
173			delete $blocked{$ip};
174			}
175
176			if (%{$conn{$ip}} > $::MAX_CONN_IP) {
177			$self->slog("blocked ip $ip");
178			$self->err_blocked;
179			}
180
181	root	1.1	$req =~ /^(?:\015\012)?
182			(GET\|HEAD) \040+
183			([^\040]+) \040+
184			HTTP\/([0-9]+\.[0-9]+)
185			\015\012/gx
186			or $self->err(403, "method not allowed", { Allow => "GET,HEAD" });
187
188			$2 ne "1.0"
189			or $self->err(506, "http protocol version not supported");
190
191			$self->{method} = $1;
192			$self->{uri} = $2;
193
194			# parse headers
195			{
196			my (%hdr, $h, $v);
197
198			$hdr{lc $1} .= ",$2"
199			while $req =~ /\G
200			([^:\000-\040]+):
201			[\008\040]*
202			((?: [^\015\012]+ \| \015\012[\008\040] )*)
203			\015\012
204			/gxc;
205
206			$req =~ /\G\015\012$/
207			or $self->err(400, "bad request");
208
209			$self->{h}{$h} = substr $v, 1
210			while ($h, $v) = each %hdr;
211			}
212
213			$self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80;
214	root	1.3
215			weaken ($uri{$self->{uri}}{$self*1} = $self);
216	root	1.1
217			$self->map_uri;
218	root	1.2
219			Coro::Event::do_timer(after => 5);
220
221	root	1.1	$self->respond;
222			#}
223			}
224
225			# uri => path mapping
226			sub map_uri {
227			my $self = shift;
228			my $host = $self->{h}{host} \|\| "default";
229			my $uri = $self->{uri};
230
231			# some massaging, also makes it more secure
232			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
233			$uri =~ s%//+%/%g;
234			$uri =~ s%/\.(?=/\|$)%%g;
235			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
236
237			$uri =~ m%^/?\.\.(?=/\|$)%
238			and $self->err(400, "bad request");
239
240			$self->{name} = $uri;
241
242			# now do the path mapping
243			$self->{path} = "$::DOCROOT/$host$uri";
244			}
245
246			sub server_address {
247			my $self = shift;
248			my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->getsockname
249			or $self->err(500, "unable to get socket name");
250			((inet_ntoa $iaddr), $port);
251			}
252
253			sub server_host {
254			my $self = shift;
255			if (exists $self->{h}{host}) {
256			return $self->{h}{host};
257			} else {
258			return (($self->server_address)[0]);
259			}
260			}
261
262			sub server_hostport {
263			my $self = shift;
264			my ($host, $port);
265			if (exists $self->{h}{host}) {
266			($host, $port) = ($self->{h}{host}, $self->{server_port});
267			} else {
268			($host, $port) = $self->server_address;
269			}
270			$port = $port == 80 ? "" : ":$port";
271			$host.$port;
272			}
273
274			sub _cgi {
275			my $self = shift;
276			my $path = shift;
277			my $fh;
278
279			# no two-way xxx supported
280			if (0 == fork) {
281			open STDOUT, ">&".fileno($self->{fh});
282			if (chdir $::DOCROOT) {
283			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
284			$ENV{HTTP_HOST} = $self->server_host;
285			$ENV{HTTP_PORT} = $self->{server_host};
286			$ENV{SCRIPT_NAME} = $self->{name};
287			exec $::INDEXPROG;
288			}
289			Coro::State::_exit(0);
290			} else {
291			}
292			}
293
294			sub respond {
295			my $self = shift;
296			my $path = $self->{path};
297
298			stat $path
299			or $self->err(404, "not found");
300
301			# idiotic netscape sends idiotic headers AGAIN
302			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
303			? str2time $1 : 0;
304
305			if (-d _ && -r _) {
306			# directory
307			if ($path !~ /\/$/) {
308			# create a redirect to get the trailing "/"
309			my $host = $self->server_hostport;
310			$self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" });
311			} else {
312			$ims < (stat _)[9]
313			or $self->err(304, "not modified");
314
315			if ($self->{method} eq "GET") {
316			if (-r "$path/index.html") {
317			$self->{path} .= "/index.html";
318			$self->handle_file;
319			} else {
320			$self->handle_dir;
321			}
322			}
323			}
324			} elsif (-f _ && -r _) {
325			-x _ and $self->err(403, "forbidden");
326			$self->handle_file;
327			} else {
328			$self->err(404, "not found");
329			}
330			}
331
332			sub handle_dir {
333			my $self = shift;
334			$self->_cgi($::INDEXPROG);
335			}
336
337			sub handle_file {
338			my $self = shift;
339			my $length = -s _;
340			my $hdr = {
341			"Last-Modified" => time2str ((stat _)[9]),
342			};
343
344			my @code = (200, "ok");
345			my ($l, $h);
346
347			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
348			for (split /,/, $1) {
349			if (/^-(\d+)$/) {
350			($l, $h) = ($length - $1, $length - 1);
351			} elsif (/^(\d+)-(\d*)$/) {
352			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
353			} else {
354			($l, $h) = (0, $length - 1);
355			goto ignore;
356			}
357			goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h > $l;
358			}
359			$hdr->{"Content-Range"} = "bytes */$length";
360			$self->err(416, "not satisfiable", $hdr);
361
362			satisfiable:
363			$hdr->{"Content-Range"} = "bytes $l-$h/$length";
364			@code = (206, "partial content");
365			$length = $h - $l + 1;
366
367			ignore:
368			} else {
369			($l, $h) = (0, $length - 1);
370			}
371
372			if ($self->{path} =~ /\.html$/) {
373			$hdr->{"Content-Type"} = "text/html";
374			} else {
375			$hdr->{"Content-Type"} = "application/octet-stream";
376			}
377
378			$hdr->{"Content-Length"} = $length;
379
380			$self->print_response(@code, $hdr, "");
381
382			if ($self->{method} eq "GET") {
383			my ($fh, $buf);
384			open $fh, "<", $self->{path}
385			or die "$self->{path}: late open failure ($!)";
386
387			if ($l) {
388			sysseek $fh, $l, 0
389			or die "$self->{path}: cannot seek to $l ($!)";
390			}
391
392			$h -= $l - 1;
393
394			while ($h > 0) {
395			$h -= sysread $fh, $buf, $h > 4096 ? 4096 : $h;
396			print {$self->{fh}} $buf
397			or last;
398			}
399			}
400
401			close $fh;
402			}
403	root	1.2
404			1;