ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/Coro/myhttpd/httpd.pl
Revision: 1.93
Committed: Sun Mar 21 00:21:14 2010 UTC (14 years, 3 months ago) by root
Content type: text/plain
Branch: MAIN
CVS Tags: rel-6_0, rel-6_06, rel-6_05, rel-6_04, rel-6_03, rel-6_02, rel-6_01, rel-5_371, rel-5_372, rel-5_22, rel-5_23, rel-5_24, rel-5_25, rel-5_37, rel-5_36
Changes since 1.92: +1 -1 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 use Coro;
2 use Coro::Semaphore;
3 use Coro::SemaphoreSet;
4 use Coro::EV;
5 use Coro::Socket;
6 use Coro::Signal;
7 use Coro::AIO ();
8
9 use Fcntl;
10 use HTTP::Date;
11 use POSIX ();
12
13 use Compress::Zlib ();
14
15 use common::sense;
16
17 # at least on my machine, this thingy serves files
18 # quite a bit faster than apache, ;)
19 # and quite a bit slower than thttpd :(
20
21 $SIG{PIPE} = 'IGNORE';
22
23 our $accesslog;
24 our $errorlog;
25 our @listen_sockets;
26
27 our $NOW;
28 our $HTTP_NOW;
29
30 our $ERROR_LOG;
31 our $ACCESS_LOG;
32 our $TRANSFER_LOCK = new Coro::SemaphoreSet; # lock to be acquired per ip
33
34 our $update_time = EV::periodic 0, 1, undef, sub {
35 $NOW = time;
36 $HTTP_NOW = time2str $NOW;
37 };
38 $update_time->invoke;
39
40 if ($ERROR_LOG) {
41 use IO::Handle;
42 open $errorlog, ">>$ERROR_LOG"
43 or die "$ERROR_LOG: $!";
44 $errorlog->autoflush(1);
45 }
46
47 if ($ACCESS_LOG) {
48 use IO::Handle;
49 open $accesslog, ">>$ACCESS_LOG"
50 or die "$ACCESS_LOG: $!";
51 $accesslog->autoflush(1);
52 }
53
54 sub slog {
55 my $level = shift;
56 my $format = shift;
57
58 $format = sprintf $format, @_ if @_;
59
60 my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
61 print "$NOW: $format\n";
62 print $errorlog "$NOW: $format\n", @_ if $errorlog;
63 }
64
65 our $connections = new Coro::Semaphore $::MAX_CONNECTS || 250;
66 our $httpevent = new Coro::Signal;
67
68 our $queue_file = new transferqueue $::MAX_TRANSFERS;
69 our $queue_index = new transferqueue 10;
70
71 our $tbf_top = new tbf rate => $::TBF_RATE || 100000;
72
73 my $unused_bytes = 0;
74 my $unused_last = time;
75
76 sub unused_bandwidth {
77 $unused_bytes += $_[0];
78 if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) {
79 $unused_last = $NOW;
80 $unused_bytes = 0;
81 $queue_file->force_wake_next
82 and slog 1, "forced filetransfer due to unused bandwidth";
83 }
84 }
85
86 sub listen_on {
87 my $listen = $_[0];
88
89 push @listen_sockets, $listen;
90
91 # the "main thread"
92 async {
93 slog 1, "accepting connections";
94 while () {
95 $connections->down;
96 my @conn = $listen->accept;
97 #slog 3, "accepted @$connections ".scalar(@pool);
98
99 async_pool {
100 eval {
101 conn->new (@conn)->handle;
102 };
103 slog 1, "$@" if $@ && !ref $@;
104
105 $httpevent->broadcast; # only for testing, but doesn't matter much
106
107 $connections->up;
108 }
109 }
110 };
111 }
112
113 my $http_port = new Coro::Socket
114 LocalAddr => $::SERVER_HOST,
115 LocalPort => $::SERVER_PORT,
116 ReuseAddr => 1,
117 Listen => 50,
118 or die "unable to start server";
119
120 listen_on $http_port;
121
122 if ($::SERVER_PORT2) {
123 my $http_port = new Coro::Socket
124 LocalAddr => $::SERVER_HOST,
125 LocalPort => $::SERVER_PORT2,
126 ReuseAddr => 1,
127 Listen => 50,
128 or die "unable to start server";
129
130 listen_on $http_port;
131 }
132
133 package conn;
134
135 use common::sense;
136
137 use Socket;
138 use HTTP::Date;
139 use Convert::Scalar 'weaken';
140 use IO::AIO;
141
142 IO::AIO::min_parallel $::AIO_PARALLEL;
143
144 our $AIO_WATCHER = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
145
146 our %conn; # $conn{ip}{self} => connobj
147 our %uri; # $uri{ip}{uri}{self}
148 our %blocked;
149 our %mimetype;
150
151 sub read_mimetypes {
152 if (open my $fh, "<mime_types") {
153 while (<$fh>) {
154 if (/^([^#]\S+)\t+(\S+)$/) {
155 $mimetype{lc $1} = $2;
156 }
157 }
158 } else {
159 print "cannot open mime_types\n";
160 }
161 }
162
163 read_mimetypes;
164
165 sub new {
166 my $class = shift;
167 my $fh = shift;
168 my $peername = shift;
169 my $self = bless { fh => $fh }, $class;
170 my (undef, $iaddr) = unpack_sockaddr_in $peername
171 or $self->err (500, "unable to decode peername");
172
173 $self->{remote_addr} =
174 $self->{remote_id} = inet_ntoa $iaddr;
175
176 $self->{time} = $::NOW;
177
178 weaken ($Coro::current->{conn} = $self);
179
180 ++$::conns;
181 $::maxconns = $::conns if $::conns > $::maxconns;
182
183 $self
184 }
185
186 sub DESTROY {
187 my $self = shift;
188
189 --$::conns;
190 }
191
192 sub prune_cache {
193 my $hash = $_[0];
194
195 for (keys %$hash) {
196 if (ref $hash->{$_} eq HASH::) {
197 prune_cache($hash->{$_});
198 unless (scalar keys %{$hash->{$_}}) {
199 delete $hash->{$_};
200 }
201 }
202 }
203 }
204
205 sub prune_caches {
206 prune_cache \%conn;
207 prune_cache \%uri;
208
209 for (keys %blocked) {
210 delete $blocked{$_} unless $blocked{$_}[0] > $::NOW;
211 }
212 }
213
214 our $PRUNE_WATCHER = EV::timer 60, 60, \&prune_caches;
215
216 sub slog {
217 my $self = shift;
218 main::slog($_[0], "$self->{remote_id}> $_[1]");
219 }
220
221 sub response {
222 my ($self, $code, $msg, $hdr, $content) = @_;
223 my $res = "HTTP/1.1 $code $msg\015\012";
224 my $GZ = "";
225
226 if (exists $hdr->{Connection}) {
227 if ($hdr->{Connection} =~ /close/) {
228 $self->{h}{connection} = "close"
229 }
230 } else {
231 if ($self->{version} < 1.1) {
232 if ($self->{h}{connection} =~ /keep-alive/i) {
233 $hdr->{Connection} = "Keep-Alive";
234 } else {
235 $self->{h}{connection} = "close"
236 }
237 }
238 }
239
240 if ($self->{method} ne "HEAD"
241 && $self->{h}{"accept-encoding"} =~ /\bgzip\b/
242 && 400 < length $content
243 && $hdr->{"Content-Length"} == length $content
244 && !exists $hdr->{"Content-Encoding"}
245 ) {
246 my $orig = length $content;
247 $hdr->{"Content-Encoding"} = "gzip";
248 $content = Compress::Zlib::memGzip(\$content);
249 $hdr->{"Content-Length"} = length $content;
250 $GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
251 }
252
253 $res .= "Date: $HTTP_NOW\015\012";
254 $res .= "Server: $::NAME\015\012";
255
256 while (my ($h, $v) = each %$hdr) {
257 $res .= "$h: $v\015\012"
258 }
259 $res .= "\015\012";
260
261 $res .= $content if defined $content and $self->{method} ne "HEAD";
262
263 my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW).
264 " $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
265 " \"$self->{h}{referer}\"\n";
266
267 print $::accesslog $log if $::accesslog;
268 print STDERR $log;
269
270 $tbf_top->request(length $res, 1e6);
271 $self->{written} += print {$self->{fh}} $res;
272 }
273
274 sub err {
275 my $self = shift;
276 my ($code, $msg, $hdr, $content) = @_;
277
278 unless (defined $content) {
279 $content = "$code $msg\n";
280 $hdr->{"Content-Type"} = "text/plain";
281 $hdr->{"Content-Length"} = length $content;
282 }
283 $hdr->{"Connection"} = "close";
284
285 $self->response ($code, $msg, $hdr, $content);
286
287 die bless {}, err::
288 }
289
290 sub handle {
291 my $self = shift;
292 my $fh = $self->{fh};
293
294 my $host;
295
296 $fh->timeout($::REQ_TIMEOUT);
297 while () {
298 $self->{reqs}++;
299
300 # read request and parse first line
301 my $req = $fh->readline("\015\012\015\012");
302
303 unless (defined $req) {
304 if (exists $self->{version}) {
305 last;
306 } else {
307 $self->err(408, "request timeout");
308 }
309 }
310
311 $self->{h} = {};
312
313 $fh->timeout($::RES_TIMEOUT);
314
315 $req =~ /^(?:\015\012)?
316 (GET|HEAD) \040+
317 ([^\040]+) \040+
318 HTTP\/([0-9]+\.[0-9]+)
319 \015\012/gx
320 or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
321
322 $self->{method} = $1;
323 $self->{uri} = $2;
324 $self->{version} = $3;
325
326 $3 =~ /^1\./
327 or $self->err(506, "http protocol version $3 not supported");
328
329 # parse headers
330 {
331 my (%hdr, $h, $v);
332
333 $hdr{lc $1} .= ",$2"
334 while $req =~ /\G
335 ([^:\000-\040]+):
336 [\011\040]*
337 ((?: [^\015\012]+ | \015\012[\011\040] )*)
338 \015\012
339 /gxc;
340
341 $req =~ /\G\015\012$/
342 or $self->err(400, "bad request");
343
344 $self->{h}{$h} = substr $v, 1
345 while ($h, $v) = each %hdr;
346 }
347
348 # remote id should be unique per user
349 my $id = $self->{remote_addr};
350
351 if (exists $self->{h}{"client-ip"}) {
352 $id .= "[".$self->{h}{"client-ip"}."]";
353 } elsif (exists $self->{h}{"x-forwarded-for"}) {
354 $id .= "[".$self->{h}{"x-forwarded-for"}."]";
355 }
356
357 $self->{remote_id} = $id;
358
359 weaken (local $conn{$id}{$self*1} = $self);
360
361 if ($blocked{$id}) {
362 $self->err_blocked
363 if $blocked{$id}[0] > $::NOW;
364
365 delete $blocked{$id};
366 }
367
368 # find out server name and port
369 if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
370 $host = $1;
371 } else {
372 $host = $self->{h}{host};
373 }
374
375 if (defined $host) {
376 $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
377 } else {
378 ($self->{server_port}, $host)
379 = unpack_sockaddr_in $self->{fh}->sockname
380 or $self->err(500, "unable to get socket name");
381 $host = inet_ntoa $host;
382 }
383
384 $self->{server_name} = $host;
385
386 weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
387
388 eval {
389 $self->map_uri;
390 $self->respond;
391 };
392
393 die if $@ && !ref $@;
394
395 last if $self->{h}{connection} =~ /close/i;
396
397 $httpevent->broadcast;
398
399 $fh->timeout($::PER_TIMEOUT);
400 }
401 }
402
403 sub block {
404 my $self = shift;
405
406 $blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
407 $self->slog(2, "blocked ip $self->{remote_id}");
408 $self->err_blocked;
409 }
410
411 # uri => path mapping
412 sub map_uri {
413 my $self = shift;
414 my $host = $self->{server_name};
415 my $uri = $self->{uri};
416
417 $host =~ /[\/\\]/
418 and $self->err(400, "bad request");
419
420 # some massaging, also makes it more secure
421 $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
422 $uri =~ s%//+%/%g;
423 $uri =~ s%/\.(?=/|$)%%g;
424 1 while $uri =~ s%/[^/]+/\.\.(?=/|$)%%;
425
426 $uri =~ m%^/?\.\.(?=/|$)%
427 and $self->err(400, "bad request");
428
429 $self->{name} = $uri;
430
431 # now do the path mapping
432 $self->{path} = "$::DOCROOT/$host$uri";
433
434 $self->access_check;
435 }
436
437 sub _cgi {
438 my $self = shift;
439 my $path = shift;
440 my $fh;
441
442 # no two-way xxx supported
443 if (0 == fork) {
444 open STDOUT, ">&".fileno($self->{fh});
445 if (chdir $::DOCROOT) {
446 $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
447 $ENV{HTTP_HOST} = $self->{server_name};
448 $ENV{HTTP_PORT} = $self->{server_port};
449 $ENV{SCRIPT_NAME} = $self->{name};
450 exec $path;
451 }
452 Coro::State::_exit(0);
453 } else {
454 die;
455 }
456 }
457
458 sub server_hostport {
459 $_[0]{server_port} == 80
460 ? $_[0]{server_name}
461 : "$_[0]{server_name}:$_[0]{server_port}";
462 }
463
464 sub respond {
465 my $self = shift;
466 my $path = $self->{path};
467
468 if ($self->{name} =~ s%^/internal/([^/]+)%%) {
469 if ($::internal{$1}) {
470 $::internal{$1}->($self);
471 } else {
472 $self->err (404, "not found");
473 }
474 } else {
475
476 Coro::AIO::aio_stat $path
477 and $self->err (404, "not found");
478
479 $self->{stat} = [stat _];
480
481 # idiotic netscape sends idiotic headers AGAIN
482 my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
483 ? str2time $1 : 0;
484
485 if (-d _ && -r _) {
486 # directory
487 if ($path !~ /\/$/) {
488 # create a redirect to get the trailing "/"
489 # we don't try to avoid the :80
490 $self->err (301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
491 } else {
492 $ims < $self->{stat}[9]
493 or $self->err (304, "not modified");
494
495 if (-r "$path/index.html") {
496 # replace directory "size" by index.html filesize
497 $self->{stat} = [stat ($self->{path} .= "/index.html")];
498 $self->handle_file ($queue_index, $tbf_top);
499 } else {
500 $self->handle_dir;
501 }
502 }
503 } elsif (-f _ && -r _) {
504 -x _ and $self->err (403, "forbidden");
505
506 if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
507 my $timeout = $::NOW + 10;
508 while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
509 if ($timeout < $::NOW) {
510 $self->block($::BLOCKTIME, "too many connections");
511 } else {
512 $httpevent->wait;
513 }
514 }
515 }
516
517 $self->handle_file ($queue_file, $tbf_top);
518 } else {
519 $self->err (404, "not found");
520 }
521 }
522 }
523
524 sub handle_dir {
525 my $self = shift;
526 my $idx = $self->diridx;
527
528 $self->response (200, "ok",
529 {
530 "Content-Type" => "text/html; charset=utf-8",
531 "Content-Length" => length $idx,
532 "Last-Modified" => time2str ($self->{stat}[9]),
533 },
534 $idx);
535 }
536
537 sub handle_file {
538 my ($self, $queue, $tbf) = @_;
539 my $length = $self->{stat}[7];
540 my $hdr = {
541 "Last-Modified" => time2str ((stat _)[9]),
542 "Accept-Ranges" => "bytes",
543 };
544
545 my @code = (200, "ok");
546 my ($l, $h);
547
548 if ($self->{h}{range} =~ /^bytes=(.*)$/i) {
549 for (split /,/, $1) {
550 if (/^-(\d+)$/) {
551 ($l, $h) = ($length - $1, $length - 1);
552 } elsif (/^(\d+)-(\d*)$/) {
553 ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1);
554 } else {
555 ($l, $h) = (0, $length - 1);
556 goto ignore;
557 }
558 goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
559 }
560 $hdr->{"Content-Range"} = "bytes */$length";
561 $hdr->{"Content-Length"} = $length;
562 $self->err (416, "not satisfiable", $hdr, "");
563
564 satisfiable:
565 # check for segmented downloads
566 if ($l && $::NO_SEGMENTED) {
567 my $timeout = $::NOW + 15;
568 while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
569 if ($timeout <= $::NOW) {
570 $self->block ($::BLOCKTIME, "segmented downloads are forbidden");
571 #$self->err_segmented_download;
572 } else {
573 $httpevent->wait;
574 }
575 }
576 }
577
578 $hdr->{"Content-Range"} = "bytes $l-$h/$length";
579 @code = (206, "partial content");
580 $length = $h - $l + 1;
581
582 ignore:
583 } else {
584 ($l, $h) = (0, $length - 1);
585 }
586
587 $self->{path} =~ /\.([^.]+)$/;
588 $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream";
589 $hdr->{"Content-Length"} = $length;
590
591 $self->response (@code, $hdr, "");
592
593 if ($self->{method} eq "GET") {
594 $self->{time} = $::NOW;
595 $self->{written} = 0;
596
597 my $fh = Coro::AIO::aio_open $self->{path}, Fcntl::O_RDONLY, 0
598 or die "$self->{path}: late open failure ($!)";
599
600 $h -= $l - 1;
601
602 my $transfer = $queue->start_transfer ($h);
603 my $locked;
604 my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
605
606 while ($h > 0) {
607 Coro::cede;
608 my $transfer_lock = $TRANSFER_LOCK->guard ($self->{remote_id});
609
610 unless ($locked) {
611 if ($locked ||= $transfer->try ($::WAIT_INTERVAL)) {
612 $bufsize = $::BUFSIZE;
613 $self->{time} = $::NOW;
614 $self->{written} = 0;
615 }
616 }
617
618 if ($blocked{$self->{remote_id}}) {
619 $self->{h}{connection} = "close";
620 die bless {}, err::;
621 }
622
623 Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), my $buf, 0
624 or last;
625
626 $tbf->request (length $buf);
627 my $w = $self->{fh}->syswrite ($buf)
628 or last;
629 $::written += $w;
630 $self->{written} += $w;
631 $l += $w;
632 }
633
634 close $fh;
635 }
636 }
637
638 1
639