[ViewVC] Annotation of: cvs/Coro/eg/myhttpd

#!/usr/bin/perl

# this is a relatively small web-server, using coroutines for connections.
# play around with it but do not use it in production without checking it
# works for you. ask myhttpd@plan9.de in case of problems, or if you are
# interested in a newer version (more useless features).

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 :(

$MAX_CONNECTS = 500;                    # maximum simult. connects
$REQ_TIMEOUT  =  60;                    # request timeout
$RES_TIMEOUT  = 180;                    # response timeout
$MAX_POOL     =  20;                    # max. number of idle workers
$DOCROOT      = "/usr/www/htdocs";      # document root
$INDEXPROG    = "/usr/www/bin/dols";    # indexing program (nph-cgi script)
$SERVER_HOST  = "0.0.0.0";              # host to bind on
$SERVER_PORT  = 80;                     # port to listen on

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;

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;
      }
   }
}

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

   }
};

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

package conn;

use Socket;
use HTTP::Date;

sub new {
   my $class = shift;
   my $fh = shift;
   my (undef, $iaddr) = unpack_sockaddr_in $fh->peername
      or $self->err(500, "unable to get peername");
   $self->{remote_address} = inet_ntoa $iaddr;
   bless { fh => $fh }, $class;
}

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 time; # slow? nah.

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

   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 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");

      $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 < 2
         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]+):
                  [\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;
      }

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

      $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}->sockname
      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;
}

# no, this doesn't do cgi, but it's close enough
# for the no-longer-used directory indexing script.
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;
}

Revision:	1.8
Committed:	Wed Apr 11 02:48:20 2007 UTC (17 years, 2 months ago) by root
Branch:	MAIN
Changes since 1.7:	+2 -0 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	#!/usr/bin/perl
2
3	root	1.7	# this is a relatively small web-server, using coroutines for connections.
4			# play around with it but do not use it in production without checking it
5			# works for you. ask myhttpd@plan9.de in case of problems, or if you are
6			# interested in a newer version (more useless features).
7	root	1.2
8	root	1.1	use Coro;
9			use Coro::Semaphore;
10			use Coro::Event;
11			use Coro::Socket;
12
13			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			$MAX_CONNECTS = 500; # maximum simult. connects
21			$REQ_TIMEOUT = 60; # request timeout
22			$RES_TIMEOUT = 180; # response timeout
23			$MAX_POOL = 20; # max. number of idle workers
24			$DOCROOT = "/usr/www/htdocs"; # document root
25			$INDEXPROG = "/usr/www/bin/dols"; # indexing program (nph-cgi script)
26			$SERVER_HOST = "0.0.0.0"; # host to bind on
27			$SERVER_PORT = 80; # port to listen on
28
29			my $port = new Coro::Socket
30			LocalAddr => $SERVER_HOST,
31			LocalPort => $SERVER_PORT,
32			ReuseAddr => 1,
33			Listen => 1,
34			or die "unable to start server";
35
36			$SIG{PIPE} = 'IGNORE';
37
38			sub slog {
39			my $level = shift;
40			my $format = shift;
41			#printf "---: $format\n", @_;
42			}
43
44			my $connections = new Coro::Semaphore $MAX_CONNECTS;
45
46			my @fh;
47			my @pool;
48
49			sub handler {
50			while () {
51			my $fh = pop @fh;
52			if ($fh) {
53			eval {
54			conn->new($fh)->handle;
55			};
56			close $fh;
57			slog 1, "$@" if $@ && !ref $@;
58			$connections->up;
59			} else {
60			last if @pool >= $MAX_POOL;
61			push @pool, $Coro::current;
62			schedule;
63			}
64			}
65			}
66
67			async {
68			slog 1, "accepting connections";
69			while () {
70			$connections->down;
71			push @fh, $port->accept;
72			#slog 3, "accepted @$connections ".scalar(@pool);
73			if (@pool) {
74			(pop @pool)->ready;
75			} else {
76			async \&handler;
77			}
78
79			}
80			};
81
82	root	1.8	cede;
83	root	1.1	loop;
84			print "ende\n";#d#
85
86			package conn;
87
88			use Socket;
89			use HTTP::Date;
90
91			sub new {
92			my $class = shift;
93			my $fh = shift;
94	root	1.3	my (undef, $iaddr) = unpack_sockaddr_in $fh->peername
95	root	1.1	or $self->err(500, "unable to get peername");
96			$self->{remote_address} = inet_ntoa $iaddr;
97			bless { fh => $fh }, $class;
98			}
99
100			sub slog {
101			main::slog(@_);
102			}
103
104			sub print_response {
105			my ($self, $code, $msg, $hdr, $content) = @_;
106			my $res = "HTTP/1.0 $code $msg\015\012";
107
108			$hdr->{Date} = time2str time; # slow? nah.
109
110			while (my ($h, $v) = each %$hdr) {
111			$res .= "$h: $v\015\012"
112			}
113			$res .= "\015\012$content" if defined $content;
114
115			print {$self->{fh}} $res;
116			}
117
118			sub err {
119			my $self = shift;
120			my ($code, $msg, $hdr, $content) = @_;
121
122			unless (defined $content) {
123			$content = "$code $msg";
124			$hdr->{"Content-Type"} = "text/plain";
125			$hdr->{"Content-Length"} = length $content;
126			}
127
128			$self->slog($msg) if $code;
129
130			$self->print_response($code, $msg, $hdr, $content);
131
132			die bless {}, err::;
133			}
134
135			sub handle {
136			my $self = shift;
137			my $fh = $self->{fh};
138
139			#while() {
140			$self->{h} = {};
141
142			# read request and parse first line
143			$fh->timeout($::REQ_TIMEOUT);
144			my $req = $fh->readline("\015\012\015\012");
145			$fh->timeout($::RES_TIMEOUT);
146
147			defined $req or
148			$self->err(408, "request timeout");
149
150			$req =~ /^(?:\015\012)?
151			(GET\|HEAD) \040+
152			([^\040]+) \040+
153			HTTP\/([0-9]+\.[0-9]+)
154			\015\012/gx
155			or $self->err(403, "method not allowed", { Allow => "GET,HEAD" });
156
157	root	1.6	$2 < 2
158	root	1.1	or $self->err(506, "http protocol version not supported");
159
160			$self->{method} = $1;
161			$self->{uri} = $2;
162
163			# parse headers
164			{
165			my (%hdr, $h, $v);
166
167			$hdr{lc $1} .= ",$2"
168			while $req =~ /\G
169			([^:\000-\040]+):
170	root	1.7	[\011\040]*
171			((?: [^\015\012]+ \| \015\012[\011\040] )*)
172	root	1.1	\015\012
173			/gxc;
174
175			$req =~ /\G\015\012$/
176			or $self->err(400, "bad request");
177
178			$self->{h}{$h} = substr $v, 1
179			while ($h, $v) = each %hdr;
180			}
181
182			$self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80;
183
184			$self->map_uri;
185			$self->respond;
186			#}
187			}
188
189			# uri => path mapping
190			sub map_uri {
191			my $self = shift;
192			my $host = $self->{h}{host} \|\| "default";
193			my $uri = $self->{uri};
194
195			# some massaging, also makes it more secure
196			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
197			$uri =~ s%//+%/%g;
198			$uri =~ s%/\.(?=/\|$)%%g;
199			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
200
201			$uri =~ m%^/?\.\.(?=/\|$)%
202			and $self->err(400, "bad request");
203
204			$self->{name} = $uri;
205
206			# now do the path mapping
207			$self->{path} = "$::DOCROOT/$host$uri";
208			}
209
210			sub server_address {
211			my $self = shift;
212	root	1.3	my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->sockname
213	root	1.1	or $self->err(500, "unable to get socket name");
214			((inet_ntoa $iaddr), $port);
215			}
216
217			sub server_host {
218			my $self = shift;
219			if (exists $self->{h}{host}) {
220			return $self->{h}{host};
221			} else {
222			return (($self->server_address)[0]);
223			}
224			}
225
226			sub server_hostport {
227			my $self = shift;
228			my ($host, $port);
229			if (exists $self->{h}{host}) {
230			($host, $port) = ($self->{h}{host}, $self->{server_port});
231			} else {
232			($host, $port) = $self->server_address;
233			}
234			$port = $port == 80 ? "" : ":$port";
235			$host.$port;
236			}
237
238	root	1.2	# no, this doesn't do cgi, but it's close enough
239			# for the no-longer-used directory indexing script.
240	root	1.1	sub _cgi {
241			my $self = shift;
242			my $path = shift;
243			my $fh;
244
245			# no two-way xxx supported
246			if (0 == fork) {
247			open STDOUT, ">&".fileno($self->{fh});
248			if (chdir $::DOCROOT) {
249			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
250			$ENV{HTTP_HOST} = $self->server_host;
251			$ENV{HTTP_PORT} = $self->{server_host};
252			$ENV{SCRIPT_NAME} = $self->{name};
253			exec $::INDEXPROG;
254			}
255			Coro::State::_exit(0);
256			} else {
257			}
258			}
259
260			sub respond {
261			my $self = shift;
262			my $path = $self->{path};
263
264			stat $path
265			or $self->err(404, "not found");
266
267			# idiotic netscape sends idiotic headers AGAIN
268			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
269			? str2time $1 : 0;
270
271			if (-d _ && -r _) {
272			# directory
273			if ($path !~ /\/$/) {
274			# create a redirect to get the trailing "/"
275			my $host = $self->server_hostport;
276			$self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" });
277			} else {
278			$ims < (stat _)[9]
279			or $self->err(304, "not modified");
280
281			if ($self->{method} eq "GET") {
282			if (-r "$path/index.html") {
283			$self->{path} .= "/index.html";
284			$self->handle_file;
285			} else {
286			$self->handle_dir;
287			}
288			}
289			}
290			} elsif (-f _ && -r _) {
291			-x _ and $self->err(403, "forbidden");
292			$self->handle_file;
293			} else {
294			$self->err(404, "not found");
295			}
296			}
297
298			sub handle_dir {
299			my $self = shift;
300			$self->_cgi($::INDEXPROG);
301			}
302
303			sub handle_file {
304			my $self = shift;
305			my $length = -s _;
306			my $hdr = {
307			"Last-Modified" => time2str ((stat _)[9]),
308			};
309
310			my @code = (200, "ok");
311			my ($l, $h);
312
313			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
314			for (split /,/, $1) {
315			if (/^-(\d+)$/) {
316			($l, $h) = ($length - $1, $length - 1);
317			} elsif (/^(\d+)-(\d*)$/) {
318			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
319			} else {
320			($l, $h) = (0, $length - 1);
321			goto ignore;
322			}
323	root	1.7	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
324	root	1.1	}
325			$hdr->{"Content-Range"} = "bytes */$length";
326			$self->err(416, "not satisfiable", $hdr);
327
328			satisfiable:
329			$hdr->{"Content-Range"} = "bytes $l-$h/$length";
330			@code = (206, "partial content");
331			$length = $h - $l + 1;
332
333			ignore:
334			} else {
335			($l, $h) = (0, $length - 1);
336			}
337
338			if ($self->{path} =~ /\.html$/) {
339			$hdr->{"Content-Type"} = "text/html";
340			} else {
341			$hdr->{"Content-Type"} = "application/octet-stream";
342			}
343
344			$hdr->{"Content-Length"} = $length;
345
346			$self->print_response(@code, $hdr, "");
347
348			if ($self->{method} eq "GET") {
349			my ($fh, $buf);
350	root	1.7	open $fh, "<", $self->{path}
351	root	1.1	or die "$self->{path}: late open failure ($!)";
352
353			if ($l) {
354			sysseek $fh, $l, 0
355			or die "$self->{path}: cannot seek to $l ($!)";
356			}
357
358			$h -= $l - 1;
359
360			while ($h > 0) {
361			$h -= sysread $fh, $buf, $h > 4096 ? 4096 : $h;
362			print {$self->{fh}} $buf
363			or last;
364			}
365			}
366
367			close $fh;
368			}
369	root	1.8