ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/Coro/myhttpd/httpd.pl
Revision: 1.79
Committed: Fri Dec 1 03:53:33 2006 UTC (17 years, 7 months ago) by root
Content type: text/plain
Branch: MAIN
Changes since 1.78: +8 -16 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 use Coro;
2 use Coro::Semaphore;
3 use Coro::Event;
4 use Coro::Socket;
5 use Coro::Signal;
6 use Coro::AIO ();
7
8 use HTTP::Date;
9 use POSIX ();
10
11 use Compress::Zlib ();
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 $SIG{PIPE} = 'IGNORE';
21
22 our $accesslog;
23 our $errorlog;
24
25 our $NOW;
26 our $HTTP_NOW;
27
28 Event->timer(interval => 1, hard => 1, cb => sub {
29 $NOW = time;
30 $HTTP_NOW = time2str $NOW;
31 })->now;
32
33 if ($ERROR_LOG) {
34 use IO::Handle;
35 open $errorlog, ">>$ERROR_LOG"
36 or die "$ERROR_LOG: $!";
37 $errorlog->autoflush(1);
38 }
39
40 if ($ACCESS_LOG) {
41 use IO::Handle;
42 open $accesslog, ">>$ACCESS_LOG"
43 or die "$ACCESS_LOG: $!";
44 $accesslog->autoflush(1);
45 }
46
47 sub slog {
48 my $level = shift;
49 my $format = shift;
50 my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
51 printf "$NOW: $format\n", @_;
52 printf $errorlog "$NOW: $format\n", @_ if $errorlog;
53 }
54
55 our $connections = new Coro::Semaphore $MAX_CONNECTS || 250;
56 our $httpevent = new Coro::Signal;
57
58 our $queue_file = new transferqueue $MAX_TRANSFERS;
59 our $queue_index = new transferqueue 10;
60
61 our $tbf_top = new tbf rate => $TBF_RATE || 100000;
62
63 my $unused_bytes = 0;
64 my $unused_last = time;
65
66 sub unused_bandwidth {
67 $unused_bytes += $_[0];
68 if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
69 $unused_last = $NOW;
70 $unused_bytes = 0;
71 $queue_file->force_wake_next;
72 slog 1, "forced filetransfer due to unused bandwidth";
73 }
74 }
75
76 my @newcons;
77 my @pool;
78
79 # one "execution thread"
80 sub handler {
81 while () {
82 if (@newcons) {
83 eval {
84 conn->new(@{pop @newcons})->handle;
85 };
86 slog 1, "$@" if $@ && !ref $@;
87
88 $httpevent->broadcast; # only for testing, but doesn't matter much
89
90 $connections->up;
91 } else {
92 last if @pool >= $MAX_POOL;
93 push @pool, $Coro::current;
94 schedule;
95 }
96 }
97 }
98
99 sub listen_on {
100 my $listen = $_[0];
101
102 push @listen_sockets, $listen;
103
104 # the "main thread"
105 async {
106 slog 1, "accepting connections";
107 while () {
108 $connections->down;
109 push @newcons, [$listen->accept];
110 #slog 3, "accepted @$connections ".scalar(@pool);
111 if (@pool) {
112 (pop @pool)->ready;
113 } else {
114 async \&handler;
115 }
116 }
117 };
118 }
119
120 my $http_port = new Coro::Socket
121 LocalAddr => $SERVER_HOST,
122 LocalPort => $SERVER_PORT,
123 ReuseAddr => 1,
124 Listen => 50,
125 or die "unable to start server";
126
127 listen_on $http_port;
128
129 if ($SERVER_PORT2) {
130 my $http_port = new Coro::Socket
131 LocalAddr => $SERVER_HOST,
132 LocalPort => $SERVER_PORT2,
133 ReuseAddr => 1,
134 Listen => 50,
135 or die "unable to start server";
136
137 listen_on $http_port;
138 }
139
140 package conn;
141
142 use Socket;
143 use HTTP::Date;
144 use Convert::Scalar 'weaken';
145 use IO::AIO;
146
147 IO::AIO::min_parallel $::AIO_PARALLEL;
148
149 Event->io (fd => IO::AIO::poll_fileno,
150 poll => 'r', async => 1,
151 cb => \&IO::AIO::poll_cb);
152
153 our %conn; # $conn{ip}{self} => connobj
154 our %uri; # $uri{ip}{uri}{self}
155 our %blocked;
156 our %mimetype;
157
158 sub read_mimetypes {
159 local *M;
160 if (open M, "<mime_types") {
161 while (<M>) {
162 if (/^([^#]\S+)\t+(\S+)$/) {
163 $mimetype{lc $1} = $2;
164 }
165 }
166 } else {
167 print "cannot open mime_types\n";
168 }
169 }
170
171 read_mimetypes;
172
173 sub new {
174 my $class = shift;
175 my $fh = shift;
176 my $peername = shift;
177 my $self = bless { fh => $fh }, $class;
178 my (undef, $iaddr) = unpack_sockaddr_in $peername
179 or $self->err(500, "unable to decode peername");
180
181 $self->{remote_addr} =
182 $self->{remote_id} = inet_ntoa $iaddr;
183
184 $self->{time} = $::NOW;
185
186 weaken ($Coro::current->{conn} = $self);
187
188 $::conns++;
189 $::maxconns = $::conns if $::conns > $::maxconns;
190
191 $self;
192 }
193
194 sub DESTROY {
195 #my $self = shift;
196 $::conns--;
197 }
198
199 sub prune_cache {
200 my $hash = $_[0];
201
202 for (keys %$hash) {
203 if (ref $hash->{$_} eq HASH::) {
204 prune_cache($hash->{$_});
205 unless (scalar keys %{$hash->{$_}}) {
206 delete $hash->{$_};
207 $d2++;
208 }
209 }
210 }
211 }
212
213 sub prune_caches {
214 prune_cache \%conn;
215 prune_cache \%uri;
216
217 for (keys %blocked) {
218 delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
219 }
220 }
221
222 Event->timer(interval => 60, cb => \&prune_caches);
223
224 sub slog {
225 my $self = shift;
226 main::slog($_[0], "$self->{remote_id}> $_[1]");
227 }
228
229 sub response {
230 my ($self, $code, $msg, $hdr, $content) = @_;
231 my $res = "HTTP/1.1 $code $msg\015\012";
232 my $GZ = "";
233
234 if (exists $hdr->{Connection}) {
235 if ($hdr->{Connection} =~ /close/) {
236 $self->{h}{connection} = "close"
237 }
238 } else {
239 if ($self->{version} < 1.1) {
240 if ($self->{h}{connection} =~ /keep-alive/i) {
241 $hdr->{Connection} = "Keep-Alive";
242 } else {
243 $self->{h}{connection} = "close"
244 }
245 }
246 }
247
248 if ($self->{method} ne "HEAD"
249 && $self->{h}{"accept-encoding"} =~ /\bgzip\b/
250 && 400 < length $content
251 && $hdr->{"Content-Length"} == length $content
252 && !exists $hdr->{"Content-Encoding"}
253 ) {
254 my $orig = length $content;
255 $hdr->{"Content-Encoding"} = "gzip";
256 $content = Compress::Zlib::memGzip(\$content);
257 $hdr->{"Content-Length"} = length $content;
258 $GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
259 }
260
261 $res .= "Date: $HTTP_NOW\015\012";
262 $res .= "Server: $::NAME\015\012";
263
264 while (my ($h, $v) = each %$hdr) {
265 $res .= "$h: $v\015\012"
266 }
267 $res .= "\015\012";
268
269 $res .= $content if defined $content and $self->{method} ne "HEAD";
270
271 my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
272 " $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
273 " \"$self->{h}{referer}\"\n";
274
275 print $::accesslog $log if $::accesslog;
276 print STDERR $log;
277
278 $tbf_top->request(length $res, 1e6);
279 $self->{written} += print {$self->{fh}} $res;
280 }
281
282 sub err {
283 my $self = shift;
284 my ($code, $msg, $hdr, $content) = @_;
285
286 unless (defined $content) {
287 $content = "$code $msg\n";
288 $hdr->{"Content-Type"} = "text/plain";
289 $hdr->{"Content-Length"} = length $content;
290 }
291 $hdr->{"Connection"} = "close";
292
293 $self->response($code, $msg, $hdr, $content);
294
295 die bless {}, err::;
296 }
297
298 sub handle {
299 my $self = shift;
300 my $fh = $self->{fh};
301
302 my $host;
303
304 $fh->timeout($::REQ_TIMEOUT);
305 while() {
306 $self->{reqs}++;
307
308 # read request and parse first line
309 my $req = $fh->readline("\015\012\015\012");
310
311 unless (defined $req) {
312 if (exists $self->{version}) {
313 last;
314 } else {
315 $self->err(408, "request timeout");
316 }
317 }
318
319 $self->{h} = {};
320
321 $fh->timeout($::RES_TIMEOUT);
322
323 $req =~ /^(?:\015\012)?
324 (GET|HEAD) \040+
325 ([^\040]+) \040+
326 HTTP\/([0-9]+\.[0-9]+)
327 \015\012/gx
328 or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
329
330 $self->{method} = $1;
331 $self->{uri} = $2;
332 $self->{version} = $3;
333
334 $3 =~ /^1\./
335 or $self->err(506, "http protocol version $3 not supported");
336
337 # parse headers
338 {
339 my (%hdr, $h, $v);
340
341 $hdr{lc $1} .= ",$2"
342 while $req =~ /\G
343 ([^:\000-\040]+):
344 [\011\040]*
345 ((?: [^\015\012]+ | \015\012[\011\040] )*)
346 \015\012
347 /gxc;
348
349 $req =~ /\G\015\012$/
350 or $self->err(400, "bad request");
351
352 $self->{h}{$h} = substr $v, 1
353 while ($h, $v) = each %hdr;
354 }
355
356 # remote id should be unique per user
357 my $id = $self->{remote_addr};
358
359 if (exists $self->{h}{"client-ip"}) {
360 $id .= "[".$self->{h}{"client-ip"}."]";
361 } elsif (exists $self->{h}{"x-forwarded-for"}) {
362 $id .= "[".$self->{h}{"x-forwarded-for"}."]";
363 }
364
365 $self->{remote_id} = $id;
366
367 weaken (local $conn{$id}{$self*1} = $self);
368
369 if ($blocked{$id}) {
370 $self->err_blocked
371 if $blocked{$id}[0] > $::NOW;
372
373 delete $blocked{$id};
374 }
375
376 # find out server name and port
377 if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
378 $host = $1;
379 } else {
380 $host = $self->{h}{host};
381 }
382
383 if (defined $host) {
384 $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
385 } else {
386 ($self->{server_port}, $host)
387 = unpack_sockaddr_in $self->{fh}->sockname
388 or $self->err(500, "unable to get socket name");
389 $host = inet_ntoa $host;
390 }
391
392 $self->{server_name} = $host;
393
394 weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
395
396 eval {
397 $self->map_uri;
398 $self->respond;
399 };
400
401 die if $@ && !ref $@;
402
403 last if $self->{h}{connection} =~ /close/i;
404
405 $httpevent->broadcast;
406
407 $fh->timeout($::PER_TIMEOUT);
408 }
409 }
410
411 sub block {
412 my $self = shift;
413
414 $blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
415 $self->slog(2, "blocked ip $self->{remote_id}");
416 $self->err_blocked;
417 }
418
419 # uri => path mapping
420 sub map_uri {
421 my $self = shift;
422 my $host = $self->{server_name};
423 my $uri = $self->{uri};
424
425 # some massaging, also makes it more secure
426 $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
427 $uri =~ s%//+%/%g;
428 $uri =~ s%/\.(?=/|$)%%g;
429 1 while $uri =~ s%/[^/]+/\.\.(?=/|$)%%;
430
431 $uri =~ m%^/?\.\.(?=/|$)%
432 and $self->err(400, "bad request");
433
434 $self->{name} = $uri;
435
436 # now do the path mapping
437 $self->{path} = "$::DOCROOT/$host$uri";
438
439 $self->access_check;
440 }
441
442 sub _cgi {
443 my $self = shift;
444 my $path = shift;
445 my $fh;
446
447 # no two-way xxx supported
448 if (0 == fork) {
449 open STDOUT, ">&".fileno($self->{fh});
450 if (chdir $::DOCROOT) {
451 $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
452 $ENV{HTTP_HOST} = $self->{server_name};
453 $ENV{HTTP_PORT} = $self->{server_port};
454 $ENV{SCRIPT_NAME} = $self->{name};
455 exec $path;
456 }
457 Coro::State::_exit(0);
458 } else {
459 die;
460 }
461 }
462
463 sub server_hostport {
464 $_[0]{server_port} == 80
465 ? $_[0]{server_name}
466 : "$_[0]{server_name}:$_[0]{server_port}";
467 }
468
469 sub respond {
470 my $self = shift;
471 my $path = $self->{path};
472
473 if ($self->{name} =~ s%^/internal/([^/]+)%%) {
474 if ($::internal{$1}) {
475 $::internal{$1}->($self);
476 } else {
477 $self->err(404, "not found");
478 }
479 } else {
480
481 stat $path
482 or $self->err(404, "not found");
483
484 $self->{stat} = [stat _];
485
486 # idiotic netscape sends idiotic headers AGAIN
487 my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
488 ? str2time $1 : 0;
489
490 if (-d _ && -r _) {
491 # directory
492 if ($path !~ /\/$/) {
493 # create a redirect to get the trailing "/"
494 # we don't try to avoid the :80
495 $self->err(301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
496 } else {
497 $ims < $self->{stat}[9]
498 or $self->err(304, "not modified");
499
500 if (-r "$path/index.html") {
501 # replace directory "size" by index.html filesize
502 $self->{stat} = [stat ($self->{path} .= "/index.html")];
503 $self->handle_file($queue_index, $tbf_top);
504 } else {
505 $self->handle_dir;
506 }
507 }
508 } elsif (-f _ && -r _) {
509 -x _ and $self->err(403, "forbidden");
510
511 if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
512 my $timeout = $::NOW + 10;
513 while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
514 if ($timeout < $::NOW) {
515 $self->block($::BLOCKTIME, "too many connections");
516 } else {
517 $httpevent->wait;
518 }
519 }
520 }
521
522 $self->handle_file($queue_file, $tbf_top);
523 } else {
524 $self->err(404, "not found");
525 }
526 }
527 }
528
529 sub handle_dir {
530 my $self = shift;
531 my $idx = $self->diridx;
532
533 $self->response(200, "ok",
534 {
535 "Content-Type" => "text/html; charset=utf-8",
536 "Content-Length" => length $idx,
537 "Last-Modified" => time2str ($self->{stat}[9]),
538 },
539 $idx);
540 }
541
542 sub handle_file {
543 my ($self, $queue, $tbf) = @_;
544 my $length = $self->{stat}[7];
545 my $hdr = {
546 "Last-Modified" => time2str ((stat _)[9]),
547 "Accept-Ranges" => "bytes",
548 };
549
550 my @code = (200, "ok");
551 my ($l, $h);
552
553 if ($self->{h}{range} =~ /^bytes=(.*)$/) {
554 for (split /,/, $1) {
555 if (/^-(\d+)$/) {
556 ($l, $h) = ($length - $1, $length - 1);
557 } elsif (/^(\d+)-(\d*)$/) {
558 ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1);
559 } else {
560 ($l, $h) = (0, $length - 1);
561 goto ignore;
562 }
563 goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
564 }
565 $hdr->{"Content-Range"} = "bytes */$length";
566 $hdr->{"Content-Length"} = $length;
567 $self->err(416, "not satisfiable", $hdr, "");
568
569 satisfiable:
570 # check for segmented downloads
571 if ($l && $::NO_SEGMENTED) {
572 my $timeout = $::NOW + 15;
573 while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
574 if ($timeout <= $::NOW) {
575 $self->block($::BLOCKTIME, "segmented downloads are forbidden");
576 #$self->err_segmented_download;
577 } else {
578 $httpevent->wait;
579 }
580 }
581 }
582
583 $hdr->{"Content-Range"} = "bytes $l-$h/$length";
584 @code = (206, "partial content");
585 $length = $h - $l + 1;
586
587 ignore:
588 } else {
589 ($l, $h) = (0, $length - 1);
590 }
591
592 $self->{path} =~ /\.([^.]+)$/;
593 $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream";
594 $hdr->{"Content-Length"} = $length;
595
596 $self->response(@code, $hdr, "");
597
598 if ($self->{method} eq "GET") {
599 $self->{time} = $::NOW;
600 $self->{written} = 0;
601
602 my $current = $Coro::current;
603
604 my ($fh, $buf, $r);
605
606 open $fh, "<", $self->{path}
607 or die "$self->{path}: late open failure ($!)";
608
609 $h -= $l - 1;
610
611 if (0) { # !AIO
612 if ($l) {
613 sysseek $fh, $l, 0;
614 }
615 }
616
617 my $transfer = $queue->start_transfer($h);
618 my $locked;
619 my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
620
621 while ($h > 0) {
622 unless ($locked) {
623 if ($locked ||= $transfer->try($::WAIT_INTERVAL)) {
624 $bufsize = $::BUFSIZE;
625 $self->{time} = $::NOW;
626 $self->{written} = 0;
627 }
628 }
629
630 if ($blocked{$self->{remote_id}}) {
631 $self->{h}{connection} = "close";
632 die bless {}, err::;
633 }
634
635 Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), $buf, 0
636 or last;
637
638 $tbf->request (length $buf);
639 my $w = syswrite $self->{fh}, $buf
640 or last;
641 $::written += $w;
642 $self->{written} += $w;
643 $l += $r;
644 }
645
646 close $fh;
647 }
648 }
649
650 1
651