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

#!/opt/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;

# move the event main loop into a coroutine
async { loop };

slog 1, "accepting connections";
while () {
   $connections->down;
   if (my $fh = $port->accept) {
      #slog 3, "accepted @$connections ".scalar(@pool);
      async_pool {
         eval {
            conn->new($fh)->handle;
         };
         close $fh;
         slog 1, "$@" if $@ && !ref $@;
         $connections->up;
      };
   }
}

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(405, "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};

   $host =~ /[\/\\]/
      and $self->err(400, "bad request");

   # 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 ($self->{h}{range} =~ /^bytes=(.*)$/i) {
      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:	1.14
Committed:	Tue Jun 3 13:35:12 2008 UTC (16 years ago) by root
Branch:	MAIN
CVS Tags:	rel-4_91, rel-5_151, rel-5_1, rel-5_0, rel-6_0, rel-6_5, rel-4_748, rel-4_8, rel-4_9, rel-4_743, rel-6_10, rel-4_744, rel-4_747, rel-6_13, rel-6_09, rel-6_08, rel-6_07, rel-6_06, rel-6_05, rel-6_04, rel-6_03, rel-6_02, rel-6_01, rel-5_161, rel-5_371, rel-5_372, rel-6_512, rel-6_513, rel-6_511, rel-6_514, rel-5_22, rel-5_23, rel-5_24, rel-5_25, rel-6_32, rel-6_33, rel-6_31, rel-6_36, rel-6_37, rel-5_162, rel-5_2, rel-6_38, rel-6_39, rel-4_802, rel-4_803, rel-4_801, rel-4_804, rel-5_37, rel-5_36, rel-4_479, rel-6_23, rel-6_29, rel-6_28, rel-6_46, rel-6_45, rel-6_51, rel-6_52, rel-6_53, rel-6_54, rel-6_55, rel-6_56, rel-6_57, rel-4_745, rel-4_901, rel-4_746, rel-5_11, rel-5_12, rel-5_15, rel-5_14, rel-5_17, rel-5_16, rel-6_43, rel-6_42, rel-6_41, rel-6_47, rel-5_132, rel-5_131, rel-6_44, rel-6_49, rel-6_48, rel-4_911, rel-4_912, HEAD
Changes since 1.13:	+1 -1 lines
Log Message:	fix ranges-header case sensitivity
#	User	Rev	Content
1	root	1.13	#!/opt/bin/perl
2	root	1.1
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
48	root	1.11	# move the event main loop into a coroutine
49			async { loop };
50
51			slog 1, "accepting connections";
52			while () {
53			$connections->down;
54			if (my $fh = $port->accept) {
55			#slog 3, "accepted @$connections ".scalar(@pool);
56			async_pool {
57	root	1.1	eval {
58			conn->new($fh)->handle;
59			};
60			close $fh;
61			slog 1, "$@" if $@ && !ref $@;
62			$connections->up;
63	root	1.11	};
64	root	1.1	}
65			}
66
67			package conn;
68
69			use Socket;
70			use HTTP::Date;
71
72			sub new {
73			my $class = shift;
74			my $fh = shift;
75	root	1.3	my (undef, $iaddr) = unpack_sockaddr_in $fh->peername
76	root	1.1	or $self->err(500, "unable to get peername");
77			$self->{remote_address} = inet_ntoa $iaddr;
78			bless { fh => $fh }, $class;
79			}
80
81			sub slog {
82			main::slog(@_);
83			}
84
85			sub print_response {
86			my ($self, $code, $msg, $hdr, $content) = @_;
87			my $res = "HTTP/1.0 $code $msg\015\012";
88
89			$hdr->{Date} = time2str time; # slow? nah.
90
91			while (my ($h, $v) = each %$hdr) {
92			$res .= "$h: $v\015\012"
93			}
94			$res .= "\015\012$content" if defined $content;
95
96			print {$self->{fh}} $res;
97			}
98
99			sub err {
100			my $self = shift;
101			my ($code, $msg, $hdr, $content) = @_;
102
103			unless (defined $content) {
104			$content = "$code $msg";
105			$hdr->{"Content-Type"} = "text/plain";
106			$hdr->{"Content-Length"} = length $content;
107			}
108
109			$self->slog($msg) if $code;
110
111			$self->print_response($code, $msg, $hdr, $content);
112
113			die bless {}, err::;
114			}
115
116			sub handle {
117			my $self = shift;
118			my $fh = $self->{fh};
119
120			#while() {
121			$self->{h} = {};
122
123			# read request and parse first line
124			$fh->timeout($::REQ_TIMEOUT);
125			my $req = $fh->readline("\015\012\015\012");
126			$fh->timeout($::RES_TIMEOUT);
127
128			defined $req or
129			$self->err(408, "request timeout");
130
131			$req =~ /^(?:\015\012)?
132			(GET\|HEAD) \040+
133			([^\040]+) \040+
134			HTTP\/([0-9]+\.[0-9]+)
135			\015\012/gx
136	root	1.12	or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
137	root	1.1
138	root	1.6	$2 < 2
139	root	1.1	or $self->err(506, "http protocol version not supported");
140
141			$self->{method} = $1;
142			$self->{uri} = $2;
143
144			# parse headers
145			{
146			my (%hdr, $h, $v);
147
148			$hdr{lc $1} .= ",$2"
149			while $req =~ /\G
150			([^:\000-\040]+):
151	root	1.7	[\011\040]*
152			((?: [^\015\012]+ \| \015\012[\011\040] )*)
153	root	1.1	\015\012
154			/gxc;
155
156			$req =~ /\G\015\012$/
157			or $self->err(400, "bad request");
158
159			$self->{h}{$h} = substr $v, 1
160			while ($h, $v) = each %hdr;
161			}
162
163			$self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80;
164
165			$self->map_uri;
166			$self->respond;
167			#}
168			}
169
170			# uri => path mapping
171			sub map_uri {
172			my $self = shift;
173			my $host = $self->{h}{host} \|\| "default";
174			my $uri = $self->{uri};
175
176	root	1.13	$host =~ /[\/\\]/
177			and $self->err(400, "bad request");
178
179	root	1.1	# some massaging, also makes it more secure
180			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
181			$uri =~ s%//+%/%g;
182			$uri =~ s%/\.(?=/\|$)%%g;
183			1 while $uri =~ s%/[^/]+/\.\.(?=/\|$)%%;
184
185			$uri =~ m%^/?\.\.(?=/\|$)%
186			and $self->err(400, "bad request");
187
188			$self->{name} = $uri;
189
190			# now do the path mapping
191			$self->{path} = "$::DOCROOT/$host$uri";
192			}
193
194			sub server_address {
195			my $self = shift;
196	root	1.3	my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->sockname
197	root	1.1	or $self->err(500, "unable to get socket name");
198			((inet_ntoa $iaddr), $port);
199			}
200
201			sub server_host {
202			my $self = shift;
203			if (exists $self->{h}{host}) {
204			return $self->{h}{host};
205			} else {
206			return (($self->server_address)[0]);
207			}
208			}
209
210			sub server_hostport {
211			my $self = shift;
212			my ($host, $port);
213			if (exists $self->{h}{host}) {
214			($host, $port) = ($self->{h}{host}, $self->{server_port});
215			} else {
216			($host, $port) = $self->server_address;
217			}
218			$port = $port == 80 ? "" : ":$port";
219			$host.$port;
220			}
221
222	root	1.2	# no, this doesn't do cgi, but it's close enough
223			# for the no-longer-used directory indexing script.
224	root	1.1	sub _cgi {
225			my $self = shift;
226			my $path = shift;
227			my $fh;
228
229			# no two-way xxx supported
230			if (0 == fork) {
231			open STDOUT, ">&".fileno($self->{fh});
232			if (chdir $::DOCROOT) {
233			$ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
234			$ENV{HTTP_HOST} = $self->server_host;
235			$ENV{HTTP_PORT} = $self->{server_host};
236			$ENV{SCRIPT_NAME} = $self->{name};
237			exec $::INDEXPROG;
238			}
239			Coro::State::_exit(0);
240			} else {
241			}
242			}
243
244			sub respond {
245			my $self = shift;
246			my $path = $self->{path};
247
248			stat $path
249			or $self->err(404, "not found");
250
251			# idiotic netscape sends idiotic headers AGAIN
252			my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
253			? str2time $1 : 0;
254
255			if (-d _ && -r _) {
256			# directory
257			if ($path !~ /\/$/) {
258			# create a redirect to get the trailing "/"
259			my $host = $self->server_hostport;
260			$self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" });
261			} else {
262			$ims < (stat _)[9]
263			or $self->err(304, "not modified");
264
265			if ($self->{method} eq "GET") {
266			if (-r "$path/index.html") {
267			$self->{path} .= "/index.html";
268			$self->handle_file;
269			} else {
270			$self->handle_dir;
271			}
272			}
273			}
274			} elsif (-f _ && -r _) {
275			-x _ and $self->err(403, "forbidden");
276			$self->handle_file;
277			} else {
278			$self->err(404, "not found");
279			}
280			}
281
282			sub handle_dir {
283			my $self = shift;
284			$self->_cgi($::INDEXPROG);
285			}
286
287			sub handle_file {
288			my $self = shift;
289			my $length = -s _;
290			my $hdr = {
291			"Last-Modified" => time2str ((stat _)[9]),
292			};
293
294			my @code = (200, "ok");
295			my ($l, $h);
296
297	root	1.14	if ($self->{h}{range} =~ /^bytes=(.*)$/i) {
298	root	1.1	for (split /,/, $1) {
299			if (/^-(\d+)$/) {
300			($l, $h) = ($length - $1, $length - 1);
301			} elsif (/^(\d+)-(\d*)$/) {
302			($l, $h) = ($1, ($2 ne "" \|\| $2 >= $length) ? $2 : $length - 1);
303			} else {
304			($l, $h) = (0, $length - 1);
305			goto ignore;
306			}
307	root	1.7	goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
308	root	1.1	}
309			$hdr->{"Content-Range"} = "bytes */$length";
310			$self->err(416, "not satisfiable", $hdr);
311
312			satisfiable:
313			$hdr->{"Content-Range"} = "bytes $l-$h/$length";
314			@code = (206, "partial content");
315			$length = $h - $l + 1;
316
317			ignore:
318			} else {
319			($l, $h) = (0, $length - 1);
320			}
321
322			if ($self->{path} =~ /\.html$/) {
323			$hdr->{"Content-Type"} = "text/html";
324			} else {
325			$hdr->{"Content-Type"} = "application/octet-stream";
326			}
327
328			$hdr->{"Content-Length"} = $length;
329
330			$self->print_response(@code, $hdr, "");
331
332			if ($self->{method} eq "GET") {
333			my ($fh, $buf);
334	root	1.7	open $fh, "<", $self->{path}
335	root	1.1	or die "$self->{path}: late open failure ($!)";
336
337			if ($l) {
338			sysseek $fh, $l, 0
339			or die "$self->{path}: cannot seek to $l ($!)";
340			}
341
342			$h -= $l - 1;
343
344			while ($h > 0) {
345			$h -= sysread $fh, $buf, $h > 4096 ? 4096 : $h;
346			print {$self->{fh}} $buf
347			or last;
348			}
349			}
350
351			close $fh;
352			}
353	root	1.8