ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/Coro/myhttpd/httpd.pl
(Generate patch)

Comparing cvsroot/Coro/myhttpd/httpd.pl (file contents):
Revision 1.4 by root, Fri Aug 10 04:16:41 2001 UTC vs.
Revision 1.31 by root, Mon Aug 27 05:05:26 2001 UTC

9# at least on my machine, this thingy serves files 9# at least on my machine, this thingy serves files
10# quite a bit faster than apache, ;) 10# quite a bit faster than apache, ;)
11# and quite a bit slower than thttpd :( 11# and quite a bit slower than thttpd :(
12 12
13$SIG{PIPE} = 'IGNORE'; 13$SIG{PIPE} = 'IGNORE';
14 14
15our $accesslog;
16
17if ($ACCESS_LOG) {
18 use IO::Handle;
19 open $accesslog, ">>$ACCESS_LOG"
20 or die "$ACCESS_LOG: $!";
21 $accesslog->autoflush(1);
22}
23
15sub slog { 24sub slog {
16 my $level = shift; 25 my $level = shift;
17 my $format = shift; 26 my $format = shift;
18 printf "---: $format\n", @_; 27 printf "---: $format\n", @_;
19} 28}
20 29
21my $connections = new Coro::Semaphore $MAX_CONNECTS; 30my $connections = new Coro::Semaphore $MAX_CONNECTS;
22 31
23my @fh; 32my @newcons;
24my @pool; 33my @pool;
25 34
26# one "execution thread" 35# one "execution thread"
27sub handler { 36sub handler {
28 while () { 37 while () {
29 my $fh = pop @fh; 38 my $new = pop @newcons;
30 if ($fh) { 39 if ($new) {
31 eval { 40 eval {
32 conn->new($fh)->handle; 41 conn->new(@$new)->handle;
33 }; 42 };
34 close $fh;
35 slog 1, "$@" if $@ && !ref $@; 43 slog 1, "$@" if $@ && !ref $@;
36 $connections->up; 44 $connections->up;
37 } else { 45 } else {
38 last if @pool >= $MAX_POOL; 46 last if @pool >= $MAX_POOL;
39 push @pool, $Coro::current; 47 push @pool, $Coro::current;
44 52
45my $http_port = new Coro::Socket 53my $http_port = new Coro::Socket
46 LocalAddr => $SERVER_HOST, 54 LocalAddr => $SERVER_HOST,
47 LocalPort => $SERVER_PORT, 55 LocalPort => $SERVER_PORT,
48 ReuseAddr => 1, 56 ReuseAddr => 1,
49 Listen => 1, 57 Listen => 50,
50 or die "unable to start server"; 58 or die "unable to start server";
51 59
52push @listen_sockets, $http_port; 60push @listen_sockets, $http_port;
53 61
54# the "main thread" 62# the "main thread"
55async { 63async {
56 slog 1, "accepting connections"; 64 slog 1, "accepting connections";
57 while () { 65 while () {
58 $connections->down; 66 $connections->down;
59 push @fh, $http_port->accept; 67 push @newcons, [$http_port->accept];
60 #slog 3, "accepted @$connections ".scalar(@pool); 68 #slog 3, "accepted @$connections ".scalar(@pool);
61 $::NOW = time; 69 $::NOW = time;
62 if (@pool) { 70 if (@pool) {
63 (pop @pool)->ready; 71 (pop @pool)->ready;
64 } else { 72 } else {
71package conn; 79package conn;
72 80
73use Socket; 81use Socket;
74use HTTP::Date; 82use HTTP::Date;
75use Convert::Scalar 'weaken'; 83use Convert::Scalar 'weaken';
84use Linux::AIO;
76 85
86Linux::AIO::min_parallel $::AIO_PARALLEL;
87
88my $aio_requests = new Coro::Semaphore $::AIO_PARALLEL * 4;
89
90Event->io(fd => Linux::AIO::poll_fileno,
91 poll => 'r', async => 1,
92 cb => \&Linux::AIO::poll_cb);
93
77our %conn; # $conn{ip}{fh} => connobj 94our %conn; # $conn{ip}{self} => connobj
95our %uri; # $uri{ip}{uri}{self}
78our %blocked; 96our %blocked;
97our %mimetype;
98
99sub read_mimetypes {
100 local *M;
101 if (open M, "<mime_types") {
102 while (<M>) {
103 if (/^([^#]\S+)\t+(\S+)$/) {
104 $mimetype{lc $1} = $2;
105 }
106 }
107 } else {
108 print "cannot open mime_types\n";
109 }
110}
111
112read_mimetypes;
79 113
80sub new { 114sub new {
81 my $class = shift; 115 my $class = shift;
116 my $peername = shift;
82 my $fh = shift; 117 my $fh = shift;
83 my $self = bless { fh => $fh }, $class; 118 my $self = bless { fh => $fh }, $class;
84 my (undef, $iaddr) = unpack_sockaddr_in $fh->getpeername 119 my (undef, $iaddr) = unpack_sockaddr_in $peername
85 or $self->err(500, "unable to get peername"); 120 or $self->err(500, "unable to decode peername");
121
86 $self->{remote_addr} = inet_ntoa $iaddr; 122 $self->{remote_addr} = inet_ntoa $iaddr;
123 $self->{time} = $::NOW;
87 124
88 # enter ourselves into various lists 125 # enter ourselves into various lists
89 weaken ($conn{$self->{remote_addr}}{$self*1} = $self); 126 weaken ($conn{$self->{remote_addr}}{$self*1} = $self);
90 127
128 $::conns++;
129
91 $self; 130 $self;
92} 131}
93 132
94sub DESTROY { 133sub DESTROY {
95 my $self = shift; 134 my $self = shift;
135
136 $::conns--;
137
138 $self->eoconn;
96 delete $conn{$self->{remote_addr}}{$self*1}; 139 delete $conn{$self->{remote_addr}}{$self*1};
140}
141
142# end of connection
143sub eoconn {
144 my $self = shift;
97 delete $uri{$self->{uri}}{$self*1}; 145 delete $uri{$self->{remote_addr}}{$self->{uri}}{$self*1};
98} 146}
99 147
100sub slog { 148sub slog {
101 my $self = shift; 149 my $self = shift;
102 main::slog($_[0], "$self->{remote_addr}> $_[1]"); 150 main::slog($_[0], ($self->{remote_id} || $self->{remote_addr}) ."> $_[1]");
103} 151}
104 152
105sub response { 153sub response {
106 my ($self, $code, $msg, $hdr, $content) = @_; 154 my ($self, $code, $msg, $hdr, $content) = @_;
107 my $res = "HTTP/1.0 $code $msg\015\012"; 155 my $res = "HTTP/1.1 $code $msg\015\012";
108 156
109 $res .= "Connection: close\015\012"; 157 $self->{h}{connection} ||= $hdr->{Connection};
158
110 $res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :( 159 $res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :(
111 160
112 while (my ($h, $v) = each %$hdr) { 161 while (my ($h, $v) = each %$hdr) {
113 $res .= "$h: $v\015\012" 162 $res .= "$h: $v\015\012"
114 } 163 }
164 $res .= "\015\012";
115 165
116 $res .= "\015\012$content" if defined $content; 166 $res .= $content if defined $content and $self->{method} ne "HEAD";
117 167
118 print STDERR "$self->{remote_addr} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}." \"$self->{h}{referer}\"\n";#d# 168 my $log = "$self->{remote_addr} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}." \"$self->{h}{referer}\"\n";
119 169
170 print $accesslog $log if $accesslog;
171 print STDERR $log;
172
173 $self->{written} +=
120 print {$self->{fh}} $res; 174 print {$self->{fh}} $res;
121} 175}
122 176
123sub err { 177sub err {
124 my $self = shift; 178 my $self = shift;
125 my ($code, $msg, $hdr, $content) = @_; 179 my ($code, $msg, $hdr, $content) = @_;
127 unless (defined $content) { 181 unless (defined $content) {
128 $content = "$code $msg"; 182 $content = "$code $msg";
129 $hdr->{"Content-Type"} = "text/plain"; 183 $hdr->{"Content-Type"} = "text/plain";
130 $hdr->{"Content-Length"} = length $content; 184 $hdr->{"Content-Length"} = length $content;
131 } 185 }
186 $hdr->{"Connection"} = "close";
132 187
133 $self->response($code, $msg, $hdr, $content); 188 $self->response($code, $msg, $hdr, $content);
134 189
135 die bless {}, err::; 190 die bless {}, err::;
136} 191}
137 192
138sub 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 {
144 "Content-Type" => "text/html",
145 "Retry-After" => $::BLOCKTIME
146 },
147 <<EOF);
148<html><p>
149You have been blocked because you opened too many connections. You
150may retry at</p>
151
152 <p><blockquote>$time.</blockquote></p>
153
154<p>Until then, each new access will renew the block. You might want to have a
155look at the <a href="http://www.goof.com/pcg/marc/animefaq.html">FAQ</a>.</p>
156</html>
157EOF
158}
159
160sub handle { 193sub handle {
161 my $self = shift; 194 my $self = shift;
162 my $fh = $self->{fh}; 195 my $fh = $self->{fh};
163 196
197 my $host;
198
199 $fh->timeout($::REQ_TIMEOUT);
164 #while() { 200 while() {
201 $self->{reqs}++;
202
203 # read request and parse first line
204 my $req = $fh->readline("\015\012\015\012");
205
206 unless (defined $req) {
207 if (exists $self->{version}) {
208 last;
209 } else {
210 $self->err(408, "request timeout");
211 }
212 }
213
165 $self->{h} = {}; 214 $self->{h} = {};
166 215
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); 216 $fh->timeout($::RES_TIMEOUT);
171
172 defined $req or
173 $self->err(408, "request timeout");
174
175 my $ip = $self->{remote_addr}; 217 my $ip = $self->{remote_addr};
176 218
177 if ($blocked{$ip}) { 219 if ($blocked{$ip}) {
178 $self->err_blocked($blocked{$ip}) 220 $self->err_blocked($blocked{$ip})
179 if $blocked{$ip} > $::NOW; 221 if $blocked{$ip} > $::NOW;
180 222
181 delete $blocked{$ip}; 223 delete $blocked{$ip};
182 } 224 }
183 225
184 if (%{$conn{$ip}} > $::MAX_CONN_IP) { 226 if (%{$conn{$ip}} > $::MAX_CONN_IP) {
227 my $delay = 120;
228 while (%{$conn{$ip}} > $::MAX_CONN_IP) {
229 if ($delay <= 0) {
185 $self->slog("blocked ip $ip"); 230 $self->slog(2, "blocked ip $ip");
186 $self->err_blocked; 231 $self->err_blocked;
232 } else {
233 Coro::Event::do_timer(after => 3);
234 $delay -= 3;
235 }
236 }
187 } 237 }
188 238
189 $req =~ /^(?:\015\012)? 239 $req =~ /^(?:\015\012)?
190 (GET|HEAD) \040+ 240 (GET|HEAD) \040+
191 ([^\040]+) \040+ 241 ([^\040]+) \040+
192 HTTP\/([0-9]+\.[0-9]+) 242 HTTP\/([0-9]+\.[0-9]+)
193 \015\012/gx 243 \015\012/gx
194 or $self->err(403, "method not allowed", { Allow => "GET,HEAD" }); 244 or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
195
196 $2 ne "1.0"
197 or $self->err(506, "http protocol version not supported");
198 245
199 $self->{method} = $1; 246 $self->{method} = $1;
200 $self->{uri} = $2; 247 $self->{uri} = $2;
248 $self->{version} = $3;
249
250 $3 =~ /^1\./
251 or $self->err(506, "http protocol version $3 not supported");
201 252
202 # parse headers 253 # parse headers
203 { 254 {
204 my (%hdr, $h, $v); 255 my (%hdr, $h, $v);
205 256
216 267
217 $self->{h}{$h} = substr $v, 1 268 $self->{h}{$h} = substr $v, 1
218 while ($h, $v) = each %hdr; 269 while ($h, $v) = each %hdr;
219 } 270 }
220 271
272 # find out server name and port
273 if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
274 $host = $1;
275 } else {
276 $host = $self->{h}{host};
277 }
278
279 if (defined $host) {
221 $self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80; 280 $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
281 } else {
282 ($self->{server_port}, $host)
283 = unpack_sockaddr_in $self->{fh}->getsockname
284 or $self->err(500, "unable to get socket name");
285 $host = inet_ntoa $host;
286 }
222 287
288 $self->{server_name} = $host;
289
290 # remote id should be unique per user
291 $self->{remote_id} = $self->{remote_addr};
292
293 if (exists $self->{h}{"client-ip"}) {
294 $self->{remote_id} .= "[".$self->{h}{"client-ip"}."]";
295 } elsif (exists $self->{h}{"x-forwarded-for"}) {
296 $self->{remote_id} .= "[".$self->{h}{"x-forwarded-for"}."]";
297 }
298
223 weaken ($uri{$self->{uri}}{$self*1} = $self); 299 weaken ($uri{$self->{remote_addr}}{$self->{uri}}{$self*1} = $self);
224 300
301 eval {
225 $self->map_uri; 302 $self->map_uri;
226
227 Coro::Event::do_timer(after => 5);
228
229 $self->respond; 303 $self->respond;
304 };
305
306 $self->eoconn;
307
308 die if $@ && !ref $@;
309
310 last if $self->{h}{connection} =~ /close/ || $self->{version} < 1.1;
311
312 $fh->timeout($::PER_TIMEOUT);
230 #} 313 }
231} 314}
232 315
233# uri => path mapping 316# uri => path mapping
234sub map_uri { 317sub map_uri {
235 my $self = shift; 318 my $self = shift;
236 my $host = $self->{h}{host} || "default"; 319 my $host = $self->{server_name};
237 my $uri = $self->{uri}; 320 my $uri = $self->{uri};
238 321
239 # some massaging, also makes it more secure 322 # some massaging, also makes it more secure
240 $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; 323 $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
241 $uri =~ s%//+%/%g; 324 $uri =~ s%//+%/%g;
247 330
248 $self->{name} = $uri; 331 $self->{name} = $uri;
249 332
250 # now do the path mapping 333 # now do the path mapping
251 $self->{path} = "$::DOCROOT/$host$uri"; 334 $self->{path} = "$::DOCROOT/$host$uri";
252}
253 335
254sub server_address { 336 $self->access_check;
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
261sub 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
270sub 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} 337}
281 338
282sub _cgi { 339sub _cgi {
283 my $self = shift; 340 my $self = shift;
284 my $path = shift; 341 my $path = shift;
287 # no two-way xxx supported 344 # no two-way xxx supported
288 if (0 == fork) { 345 if (0 == fork) {
289 open STDOUT, ">&".fileno($self->{fh}); 346 open STDOUT, ">&".fileno($self->{fh});
290 if (chdir $::DOCROOT) { 347 if (chdir $::DOCROOT) {
291 $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike 348 $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
292 $ENV{HTTP_HOST} = $self->server_host; 349 $ENV{HTTP_HOST} = $self->{server_name};
293 $ENV{HTTP_PORT} = $self->{server_host}; 350 $ENV{HTTP_PORT} = $self->{server_port};
294 $ENV{SCRIPT_NAME} = $self->{name}; 351 $ENV{SCRIPT_NAME} = $self->{name};
295 exec $::INDEXPROG; 352 exec $path;
296 } 353 }
297 Coro::State::_exit(0); 354 Coro::State::_exit(0);
298 } else { 355 } else {
356 die;
299 } 357 }
358}
359
360sub server_hostport {
361 $_[0]{server_port} == 80
362 ? $_[0]{server_name}
363 : "$_[0]{server_name}:$_[0]{server_port}";
300} 364}
301 365
302sub respond { 366sub respond {
303 my $self = shift; 367 my $self = shift;
304 my $path = $self->{path}; 368 my $path = $self->{path};
305 369
306 stat $path 370 stat $path
307 or $self->err(404, "not found"); 371 or $self->err(404, "not found");
372
373 $self->{stat} = [stat _];
308 374
309 # idiotic netscape sends idiotic headers AGAIN 375 # idiotic netscape sends idiotic headers AGAIN
310 my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/ 376 my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
311 ? str2time $1 : 0; 377 ? str2time $1 : 0;
312 378
313 if (-d _ && -r _) { 379 if (-d _ && -r _) {
314 # directory 380 # directory
315 if ($path !~ /\/$/) { 381 if ($path !~ /\/$/) {
316 # create a redirect to get the trailing "/" 382 # create a redirect to get the trailing "/"
317 my $host = $self->server_hostport; 383 # we don't try to avoid the :80
318 $self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" }); 384 $self->err(301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
319 } else { 385 } else {
320 $ims < (stat _)[9] 386 $ims < $self->{stat}[9]
321 or $self->err(304, "not modified"); 387 or $self->err(304, "not modified");
322 388
323 if ($self->{method} eq "GET") {
324 if (-r "$path/index.html") { 389 if (-r "$path/index.html") {
325 $self->{path} .= "/index.html"; 390 $self->{path} .= "/index.html";
326 $self->handle_file; 391 $self->handle_file;
327 } else { 392 } else {
328 $self->handle_dir; 393 $self->handle_dir;
329 }
330 } 394 }
331 } 395 }
332 } elsif (-f _ && -r _) { 396 } elsif (-f _ && -r _) {
333 -x _ and $self->err(403, "forbidden"); 397 -x _ and $self->err(403, "forbidden");
334 $self->handle_file; 398 $self->handle_file;
337 } 401 }
338} 402}
339 403
340sub handle_dir { 404sub handle_dir {
341 my $self = shift; 405 my $self = shift;
342 $self->_cgi($::INDEXPROG); 406 my $idx = $self->diridx;
407
408 $self->response(200, "ok",
409 {
410 "Content-Type" => "text/html",
411 "Content-Length" => length $idx,
412 },
413 $idx);
343} 414}
344 415
345sub handle_file { 416sub handle_file {
346 my $self = shift; 417 my $self = shift;
347 my $length = -s _; 418 my $length = -s _;
360 ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1); 431 ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1);
361 } else { 432 } else {
362 ($l, $h) = (0, $length - 1); 433 ($l, $h) = (0, $length - 1);
363 goto ignore; 434 goto ignore;
364 } 435 }
365 goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h > $l; 436 goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
366 } 437 }
367 $hdr->{"Content-Range"} = "bytes */$length"; 438 $hdr->{"Content-Range"} = "bytes */$length";
439 $hdr->{"Content-Length"} = $length;
440 $self->slog(9, "not satisfiable($self->{h}{range}|".$self->{h}{"user-agent"}.")");
368 $self->err(416, "not satisfiable", $hdr); 441 $self->err(416, "not satisfiable", $hdr, "");
369 442
370satisfiable: 443satisfiable:
371 # check for segmented downloads 444 # check for segmented downloads
372 if ($l && $NO_SEGMENTED) { 445 if ($l && $::NO_SEGMENTED) {
373 if (%{$uri{$self->{uri}}} > 1) { 446 my $delay = 180;
374 $self->slog("segmented download refused\n"); 447 while (%{$uri{$self->{remote_addr}}{$self->{uri}}} > 1) {
375 $self->err(400, "segmented downloads are not allowed"); 448 if ($delay <= 0) {
449 $self->err_segmented_download;
450 } else {
451 Coro::Event::do_timer(after => 3); $delay -= 3;
452 }
376 } 453 }
377 } 454 }
378 455
379 $hdr->{"Content-Range"} = "bytes $l-$h/$length"; 456 $hdr->{"Content-Range"} = "bytes $l-$h/$length";
380 @code = (206, "partial content"); 457 @code = (206, "partial content");
383ignore: 460ignore:
384 } else { 461 } else {
385 ($l, $h) = (0, $length - 1); 462 ($l, $h) = (0, $length - 1);
386 } 463 }
387 464
388 if ($self->{path} =~ /\.html$/) { 465 $self->{path} =~ /\.([^.]+)$/;
389 $hdr->{"Content-Type"} = "text/html";
390 } else {
391 $hdr->{"Content-Type"} = "application/octet-stream"; 466 $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream";
392 }
393
394 $hdr->{"Content-Length"} = $length; 467 $hdr->{"Content-Length"} = $length;
395 468
396 $self->response(@code, $hdr, ""); 469 $self->response(@code, $hdr, "");
397 470
398 if ($self->{method} eq "GET") { 471 if ($self->{method} eq "GET") {
399 my ($fh, $buf); 472 my ($fh, $buf, $r);
473 my $current = $Coro::current;
400 open $fh, "<", $self->{path} 474 open $fh, "<", $self->{path}
401 or die "$self->{path}: late open failure ($!)"; 475 or die "$self->{path}: late open failure ($!)";
402 476
403 if ($l) {
404 sysseek $fh, $l, 0
405 or die "$self->{path}: cannot seek to $l ($!)";
406 }
407
408 $h -= $l - 1; 477 $h -= $l - 1;
409 478
479 if (0) {
480 if ($l) {
481 sysseek $fh, $l, 0;
482 }
483 }
484
410 while ($h > 0) { 485 while ($h > 0) {
411 $h -= sysread $fh, $buf, $h > 16384 ? 16384 : $h; 486 if (0) {
412 print {$self->{fh}} $buf 487 sysread $fh, $buf, $h > $::BUFSIZE ? $::BUFSIZE : $h
488 or last;
489 } else {
490 undef $buf;
491 $aio_requests->down;
492 aio_read($fh, $l, ($h > $::BUFSIZE ? $::BUFSIZE : $h),
493 $buf, 0, sub {
494 $r = $_[0];
495 $current->ready;
496 });
497 &Coro::schedule;
498 $aio_requests->up;
499 last unless $r;
500 }
501 my $w = $self->{fh}->syswrite($buf)
413 or last; 502 or last;
414 cede; 503 $::written += $w;
504 $self->{written} += $w;
505 $l += $r;
415 } 506 }
416 } 507 }
417 508
418 close $fh; 509 close $fh;
419} 510}

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines