[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 { loop };

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;
   }

}

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:	1.9
Committed:	Wed Apr 11 02:49:39 2007 UTC (17 years, 2 months ago) by root
Branch:	MAIN
Changes since 1.8:	+11 -15 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	root	1.9	async { loop };
68	root	1.1
69	root	1.9	slog 1, "accepting connections";
70			while () {
71			$connections->down;
72			push @fh, $port->accept;
73			#slog 3, "accepted @$connections ".scalar(@pool);
74			if (@pool) {
75			(pop @pool)->ready;
76			} else {
77			async \&handler;
78	root	1.1	}
79
80	root	1.9	}
81	root	1.1
82			package conn;
83
84			use Socket;
85			use HTTP::Date;
86
87			sub new {
88			my $class = shift;
89			my $fh = shift;
90	root	1.3	my (undef, $iaddr) = unpack_sockaddr_in $fh->peername
91	root	1.1	or $self->err(500, "unable to get peername");
92			$self->{remote_address} = inet_ntoa $iaddr;
93			bless { fh => $fh }, $class;
94			}
95
96			sub slog {
97			main::slog(@_);
98			}
99
100			sub print_response {
101			my ($self, $code, $msg, $hdr, $content) = @_;
102			my $res = "HTTP/1.0 $code $msg\015\012";
103
104			$hdr->{Date} = time2str time; # slow? nah.
105
106			while (my ($h, $v) = each %$hdr) {
107			$res .= "$h: $v\015\012"
108			}
109			$res .= "\015\012$content" if defined $content;
110
111			print {$self->{fh}} $res;
112			}
113
114			sub err {
115			my $self = shift;
116			my ($code, $msg, $hdr, $content) = @_;
117
118			unless (defined $content) {
119			$content = "$code $msg";
120			$hdr->{"Content-Type"} = "text/plain";
121			$hdr->{"Content-Length"} = length $content;
122			}
123
124			$self->slog($msg) if $code;
125
126			$self->print_response($code, $msg, $hdr, $content);
127
128			die bless {}, err::;
129			}
130
131			sub handle {
132			my $self = shift;
133			my $fh = $self->{fh};
134
135			#while() {
136			$self->{h} = {};
137
138			# read request and parse first line
139			$fh->timeout($::REQ_TIMEOUT);
140			my $req = $fh->readline("\015\012\015\012");
141			$fh->timeout($::RES_TIMEOUT);
142
143			defined $req or
144			$self->err(408, "request timeout");
145
146			$req =~ /^(?:\015\012)?
147			(GET\|HEAD) \040+
148			([^\040]+) \040+
149			HTTP\/([0-9]+\.[0-9]+)
150			\015\012/gx
151			or $self->err(403, "method not allowed", { Allow => "GET,HEAD" });
152
153	root	1.6	$2 < 2
154	root	1.1	or $self->err(506, "http protocol version not supported");
155
156			$self->{method} = $1;
157			$self->{uri} = $2;
158
159			# parse headers
160			{
161			my (%hdr, $h, $v);
162
163			$hdr{lc $1} .= ",$2"
164			while $req =~ /\G
165			([^:\000-\040]+):
166	root	1.7	[\011\040]*
167			((?: [^\015\012]+ \| \015\012[\011\040] )*)
168	root	1.1	\015\012
169			/gxc;
170
171			$req =~ /\G\015\012$/
172			or $self->err(400, "bad request");
173
174			$self->{h}{$h} = substr $v, 1
175			while ($h, $v) = each %hdr;
176			}
177
178			$self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80;
179
180			$self->map_uri;
181			$self->respond;
182			#}
183			}
184
185			# uri => path mapping
186			sub map_uri {
187			my $self = shift;
188			my $host = $self->{h}{host} \|\| "default";
189			my $uri = $self->{uri};
190
191			# some massaging, also makes it more secure
192			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
193			$uri =~ s%//+%/%g;
194			$uri =~ s%/\.(?=/\|$)%%g;
195			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
196
197			$uri =~ m%^/?\.\.(?=/\|$)%
198			and $self->err(400, "bad request");
199
200			$self->{name} = $uri;
201
202			# now do the path mapping
203			$self->{path} = "$::DOCROOT/$host$uri";
204			}
205
206			sub server_address {
207			my $self = shift;
208	root	1.3	my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->sockname
209	root	1.1	or $self->err(500, "unable to get socket name");
210			((inet_ntoa $iaddr), $port);
211			}
212
213			sub server_host {
214			my $self = shift;
215			if (exists $self->{h}{host}) {
216			return $self->{h}{host};
217			} else {
218			return (($self->server_address)[0]);
219			}
220			}
221
222			sub server_hostport {
223			my $self = shift;
224			my ($host, $port);
225			if (exists $self->{h}{host}) {
226			($host, $port) = ($self->{h}{host}, $self->{server_port});
227			} else {
228			($host, $port) = $self->server_address;
229			}
230			$port = $port == 80 ? "" : ":$port";
231			$host.$port;
232			}
233
234	root	1.2	# no, this doesn't do cgi, but it's close enough
235			# for the no-longer-used directory indexing script.
236	root	1.1	sub _cgi {
237			my $self = shift;
238			my $path = shift;
239			my $fh;
240
241			# no two-way xxx supported
242			if (0 == fork) {
243			open STDOUT, ">&".fileno($self->{fh});
244			if (chdir $::DOCROOT) {
245			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
246			$ENV{HTTP_HOST} = $self->server_host;
247			$ENV{HTTP_PORT} = $self->{server_host};
248			$ENV{SCRIPT_NAME} = $self->{name};
249			exec $::INDEXPROG;
250			}
251			Coro::State::_exit(0);
252			} else {
253			}
254			}
255
256			sub respond {
257			my $self = shift;
258			my $path = $self->{path};
259
260			stat $path
261			or $self->err(404, "not found");
262
263			# idiotic netscape sends idiotic headers AGAIN
264			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
265			? str2time $1 : 0;
266
267			if (-d _ && -r _) {
268			# directory
269			if ($path !~ /\/$/) {
270			# create a redirect to get the trailing "/"
271			my $host = $self->server_hostport;
272			$self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" });
273			} else {
274			$ims < (stat _)[9]
275			or $self->err(304, "not modified");
276
277			if ($self->{method} eq "GET") {
278			if (-r "$path/index.html") {
279			$self->{path} .= "/index.html";
280			$self->handle_file;
281			} else {
282			$self->handle_dir;
283			}
284			}
285			}
286			} elsif (-f _ && -r _) {
287			-x _ and $self->err(403, "forbidden");
288			$self->handle_file;
289			} else {
290			$self->err(404, "not found");
291			}
292			}
293
294			sub handle_dir {
295			my $self = shift;
296			$self->_cgi($::INDEXPROG);
297			}
298
299			sub handle_file {
300			my $self = shift;
301			my $length = -s _;
302			my $hdr = {
303			"Last-Modified" => time2str ((stat _)[9]),
304			};
305
306			my @code = (200, "ok");
307			my ($l, $h);
308
309			if ($self->{h}{range} =~ /^bytes=(.*)$/) {
310			for (split /,/, $1) {
311			if (/^-(\d+)$/) {
312			($l, $h) = ($length - $1, $length - 1);
313			} elsif (/^(\d+)-(\d*)$/) {
314			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
315			} else {
316			($l, $h) = (0, $length - 1);
317			goto ignore;
318			}
319	root	1.7	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
320	root	1.1	}
321			$hdr->{"Content-Range"} = "bytes */$length";
322			$self->err(416, "not satisfiable", $hdr);
323
324			satisfiable:
325			$hdr->{"Content-Range"} = "bytes $l-$h/$length";
326			@code = (206, "partial content");
327			$length = $h - $l + 1;
328
329			ignore:
330			} else {
331			($l, $h) = (0, $length - 1);
332			}
333
334			if ($self->{path} =~ /\.html$/) {
335			$hdr->{"Content-Type"} = "text/html";
336			} else {
337			$hdr->{"Content-Type"} = "application/octet-stream";
338			}
339
340			$hdr->{"Content-Length"} = $length;
341
342			$self->print_response(@code, $hdr, "");
343
344			if ($self->{method} eq "GET") {
345			my ($fh, $buf);
346	root	1.7	open $fh, "<", $self->{path}
347	root	1.1	or die "$self->{path}: late open failure ($!)";
348
349			if ($l) {
350			sysseek $fh, $l, 0
351			or die "$self->{path}: cannot seek to $l ($!)";
352			}
353
354			$h -= $l - 1;
355
356			while ($h > 0) {
357			$h -= sysread $fh, $buf, $h > 4096 ? 4096 : $h;
358			print {$self->{fh}} $buf
359			or last;
360			}
361			}
362
363			close $fh;
364			}
365	root	1.8