ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/Coro/myhttpd/httpd.pl
Revision: 1.63
Committed: Wed Jan 23 03:07:05 2002 UTC (22 years, 5 months ago) by root
Content type: text/plain
Branch: MAIN
Changes since 1.62: +20 -9 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
7 use HTTP::Date;
8 use POSIX ();
9
10 use Compress::Zlib ();
11
12 no utf8;
13 use bytes;
14
15 # at least on my machine, this thingy serves files
16 # quite a bit faster than apache, ;)
17 # and quite a bit slower than thttpd :(
18
19 $SIG{PIPE} = 'IGNORE';
20
21 our $accesslog;
22 our $errorlog;
23
24 our $NOW;
25 our $HTTP_NOW;
26
27 Event->timer(interval => 1, hard => 1, cb => sub {
28 $NOW = time;
29 $HTTP_NOW = time2str $NOW;
30 })->now;
31
32 if ($ERROR_LOG) {
33 use IO::Handle;
34 open $errorlog, ">>$ERROR_LOG"
35 or die "$ERROR_LOG: $!";
36 $errorlog->autoflush(1);
37 }
38
39 if ($ACCESS_LOG) {
40 use IO::Handle;
41 open $accesslog, ">>$ACCESS_LOG"
42 or die "$ACCESS_LOG: $!";
43 $accesslog->autoflush(1);
44 }
45
46 sub slog {
47 my $level = shift;
48 my $format = shift;
49 my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW);
50 printf "$NOW: $format\n", @_;
51 printf $errorlog "$NOW: $format\n", @_ if $errorlog;
52 }
53
54 our $connections = new Coro::Semaphore $MAX_CONNECTS || 250;
55 our $httpevent = new Coro::Signal;
56
57 our $queue_file = new transferqueue $MAX_TRANSFERS;
58 our $queue_index = new transferqueue 10;
59
60 my @newcons;
61 my @pool;
62
63 # one "execution thread"
64 sub handler {
65 while () {
66 if (@newcons) {
67 eval {
68 conn->new(@{pop @newcons})->handle;
69 };
70 slog 1, "$@" if $@ && !ref $@;
71
72 $httpevent->broadcast; # only for testing, but doesn't matter much
73
74 $connections->up;
75 } else {
76 last if @pool >= $MAX_POOL;
77 push @pool, $Coro::current;
78 schedule;
79 }
80 }
81 }
82
83 sub listen_on {
84 my $listen = $_[0];
85
86 push @listen_sockets, $listen;
87
88 # the "main thread"
89 async {
90 slog 1, "accepting connections";
91 while () {
92 $connections->down;
93 push @newcons, [$listen->accept];
94 #slog 3, "accepted @$connections ".scalar(@pool);
95 if (@pool) {
96 (pop @pool)->ready;
97 } else {
98 async \&handler;
99 }
100
101 }
102 };
103 }
104
105 my $http_port = new Coro::Socket
106 LocalAddr => $SERVER_HOST,
107 LocalPort => $SERVER_PORT,
108 ReuseAddr => 1,
109 Listen => 50,
110 or die "unable to start server";
111
112 listen_on $http_port;
113
114 if ($SERVER_PORT2) {
115 my $http_port = new Coro::Socket
116 LocalAddr => $SERVER_HOST,
117 LocalPort => $SERVER_PORT2,
118 ReuseAddr => 1,
119 Listen => 50,
120 or die "unable to start server";
121
122 listen_on $http_port;
123 }
124
125 package conn;
126
127 use Socket;
128 use HTTP::Date;
129 use Convert::Scalar 'weaken';
130 use Linux::AIO;
131
132 Linux::AIO::min_parallel $::AIO_PARALLEL;
133
134 Event->io(fd => Linux::AIO::poll_fileno,
135 poll => 'r', async => 1,
136 cb => \&Linux::AIO::poll_cb);
137
138 our %conn; # $conn{ip}{self} => connobj
139 our %uri; # $uri{ip}{uri}{self}
140 our %blocked;
141 our %mimetype;
142
143 sub read_mimetypes {
144 local *M;
145 if (open M, "<mime_types") {
146 while (<M>) {
147 if (/^([^#]\S+)\t+(\S+)$/) {
148 $mimetype{lc $1} = $2;
149 }
150 }
151 } else {
152 print "cannot open mime_types\n";
153 }
154 }
155
156 read_mimetypes;
157
158 sub new {
159 my $class = shift;
160 my $fh = shift;
161 my $peername = shift;
162 my $self = bless { fh => $fh }, $class;
163 my (undef, $iaddr) = unpack_sockaddr_in $peername
164 or $self->err(500, "unable to decode peername");
165
166 $self->{remote_addr} =
167 $self->{remote_id} = inet_ntoa $iaddr;
168
169 $self->{time} = $::NOW;
170
171 weaken ($Coro::current->{conn} = $self);
172
173 $::conns++;
174 $::maxconns = $::conns if $::conns > $::maxconns;
175
176 $self;
177 }
178
179 sub DESTROY {
180 #my $self = shift;
181 $::conns--;
182 }
183
184 sub slog {
185 my $self = shift;
186 main::slog($_[0], "$self->{remote_id}> $_[1]");
187 }
188
189 sub response {
190 my ($self, $code, $msg, $hdr, $content) = @_;
191 my $res = "HTTP/1.1 $code $msg\015\012";
192 my $GZ = "";
193
194 if (exists $hdr->{Connection}) {
195 if ($hdr->{Connection} =~ /close/) {
196 $self->{h}{connection} = "close"
197 }
198 } else {
199 if ($self->{version} < 1.1) {
200 if ($self->{h}{connection} =~ /keep-alive/i) {
201 $hdr->{Connection} = "Keep-Alive";
202 } else {
203 $self->{h}{connection} = "close"
204 }
205 }
206 }
207
208 if ($self->{method} ne "HEAD"
209 && $self->{h}{"accept-encoding"} =~ /\bgzip\b/
210 && $hdr->{"Content-Length"} == length $content
211 && !exists $hdr->{"Content-Encoding"}
212 ) {
213 my $orig = length $content;
214 $hdr->{"Content-Encoding"} = "gzip";
215 $content = Compress::Zlib::memGzip(\$content);
216 $hdr->{"Content-Length"} = length $content;
217 $GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig);
218 }
219
220 $res .= "Date: $HTTP_NOW\015\012";
221
222 while (my ($h, $v) = each %$hdr) {
223 $res .= "$h: $v\015\012"
224 }
225 $res .= "\015\012";
226
227 $res .= $content if defined $content and $self->{method} ne "HEAD";
228
229 my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $NOW).
230 " $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ.
231 " \"$self->{h}{referer}\"\n";
232
233 print $accesslog $log if $accesslog;
234 print STDERR $log;
235
236 $self->{written} +=
237 print {$self->{fh}} $res;
238 }
239
240 sub err {
241 my $self = shift;
242 my ($code, $msg, $hdr, $content) = @_;
243
244 unless (defined $content) {
245 $content = "$code $msg\n";
246 $hdr->{"Content-Type"} = "text/plain";
247 $hdr->{"Content-Length"} = length $content;
248 }
249 $hdr->{"Connection"} = "close";
250
251 $self->response($code, $msg, $hdr, $content);
252
253 die bless {}, err::;
254 }
255
256 sub handle {
257 my $self = shift;
258 my $fh = $self->{fh};
259
260 my $host;
261
262 $fh->timeout($::REQ_TIMEOUT);
263 while() {
264 $self->{reqs}++;
265
266 # read request and parse first line
267 my $req = $fh->readline("\015\012\015\012");
268
269 unless (defined $req) {
270 if (exists $self->{version}) {
271 last;
272 } else {
273 $self->err(408, "request timeout");
274 }
275 }
276
277 $self->{h} = {};
278
279 $fh->timeout($::RES_TIMEOUT);
280
281 $req =~ /^(?:\015\012)?
282 (GET|HEAD) \040+
283 ([^\040]+) \040+
284 HTTP\/([0-9]+\.[0-9]+)
285 \015\012/gx
286 or $self->err(405, "method not allowed", { Allow => "GET,HEAD" });
287
288 $self->{method} = $1;
289 $self->{uri} = $2;
290 $self->{version} = $3;
291
292 $3 =~ /^1\./
293 or $self->err(506, "http protocol version $3 not supported");
294
295 # parse headers
296 {
297 my (%hdr, $h, $v);
298
299 $hdr{lc $1} .= ",$2"
300 while $req =~ /\G
301 ([^:\000-\040]+):
302 [\010\040]*
303 ((?: [^\015\012]+ | \015\012[\010\040] )*)
304 \015\012
305 /gxc;
306
307 $req =~ /\G\015\012$/
308 or $self->err(400, "bad request");
309
310 $self->{h}{$h} = substr $v, 1
311 while ($h, $v) = each %hdr;
312 }
313
314 # remote id should be unique per user
315 my $id = $self->{remote_addr};
316
317 if (exists $self->{h}{"client-ip"}) {
318 $id .= "[".$self->{h}{"client-ip"}."]";
319 } elsif (exists $self->{h}{"x-forwarded-for"}) {
320 $id .= "[".$self->{h}{"x-forwarded-for"}."]";
321 }
322
323 $self->{remote_id} = $id;
324
325 weaken (local $conn{$id}{$self*1} = $self);
326
327 if ($blocked{$id}) {
328 $self->err_blocked
329 if $blocked{$id}[0] > $::NOW;
330
331 delete $blocked{$id};
332 }
333
334 # find out server name and port
335 if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) {
336 $host = $1;
337 } else {
338 $host = $self->{h}{host};
339 }
340
341 if (defined $host) {
342 $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80;
343 } else {
344 ($self->{server_port}, $host)
345 = unpack_sockaddr_in $self->{fh}->sockname
346 or $self->err(500, "unable to get socket name");
347 $host = inet_ntoa $host;
348 }
349
350 $self->{server_name} = $host;
351
352 weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self);
353
354 eval {
355 $self->map_uri;
356 $self->respond;
357 };
358
359 die if $@ && !ref $@;
360
361 last if $self->{h}{connection} =~ /close/i;
362
363 $httpevent->broadcast;
364
365 $fh->timeout($::PER_TIMEOUT);
366 }
367 }
368
369 sub block {
370 my $self = shift;
371
372 $blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]];
373 $self->slog(2, "blocked ip $self->{remote_id}");
374 $self->err_blocked;
375 }
376
377 # uri => path mapping
378 sub map_uri {
379 my $self = shift;
380 my $host = $self->{server_name};
381 my $uri = $self->{uri};
382
383 # some massaging, also makes it more secure
384 $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge;
385 $uri =~ s%//+%/%g;
386 $uri =~ s%/\.(?=/|$)%%g;
387 1 while $uri =~ s%/[^/]+/\.\.(?=/|$)%%;
388
389 $uri =~ m%^/?\.\.(?=/|$)%
390 and $self->err(400, "bad request");
391
392 $self->{name} = $uri;
393
394 # now do the path mapping
395 $self->{path} = "$::DOCROOT/$host$uri";
396
397 $self->access_check;
398 }
399
400 sub _cgi {
401 my $self = shift;
402 my $path = shift;
403 my $fh;
404
405 # no two-way xxx supported
406 if (0 == fork) {
407 open STDOUT, ">&".fileno($self->{fh});
408 if (chdir $::DOCROOT) {
409 $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike
410 $ENV{HTTP_HOST} = $self->{server_name};
411 $ENV{HTTP_PORT} = $self->{server_port};
412 $ENV{SCRIPT_NAME} = $self->{name};
413 exec $path;
414 }
415 Coro::State::_exit(0);
416 } else {
417 die;
418 }
419 }
420
421 sub server_hostport {
422 $_[0]{server_port} == 80
423 ? $_[0]{server_name}
424 : "$_[0]{server_name}:$_[0]{server_port}";
425 }
426
427 sub respond {
428 my $self = shift;
429 my $path = $self->{path};
430
431 if ($self->{name} =~ s%^/internal/([^/]+)%%) {
432 if ($::internal{$1}) {
433 $::internal{$1}->($self);
434 } else {
435 $self->err(404, "not found");
436 }
437 } else {
438
439 stat $path
440 or $self->err(404, "not found");
441
442 $self->{stat} = [stat _];
443
444 # idiotic netscape sends idiotic headers AGAIN
445 my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/
446 ? str2time $1 : 0;
447
448 if (-d _ && -r _) {
449 # directory
450 if ($path !~ /\/$/) {
451 # create a redirect to get the trailing "/"
452 # we don't try to avoid the :80
453 $self->err(301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" });
454 } else {
455 $ims < $self->{stat}[9]
456 or $self->err(304, "not modified");
457
458 if (-r "$path/index.html") {
459 # replace directory "size" by index.html filesize
460 $self->{stat} = [stat ($self->{path} .= "/index.html")];
461 $self->handle_file($queue_index);
462 } else {
463 $self->handle_dir;
464 }
465 }
466 } elsif (-f _ && -r _) {
467 -x _ and $self->err(403, "forbidden");
468
469 if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
470 my $timeout = $::NOW + 10;
471 while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) {
472 if ($timeout < $::NOW) {
473 $self->block($::BLOCKTIME, "too many connections");
474 } else {
475 $httpevent->wait;
476 }
477 }
478 }
479
480 $self->handle_file($queue_file);
481 } else {
482 $self->err(404, "not found");
483 }
484 }
485 }
486
487 sub handle_dir {
488 my $self = shift;
489 my $idx = $self->diridx;
490
491 $self->response(200, "ok",
492 {
493 "Content-Type" => "text/html",
494 "Content-Length" => length $idx,
495 "Last-Modified" => time2str ($self->{stat}[9]),
496 },
497 $idx);
498 }
499
500 sub handle_file {
501 my ($self, $queue) = @_;
502 my $length = $self->{stat}[7];
503 my $hdr = {
504 "Last-Modified" => time2str ((stat _)[9]),
505 };
506
507 my @code = (200, "ok");
508 my ($l, $h);
509
510 if ($self->{h}{range} =~ /^bytes=(.*)$/) {
511 for (split /,/, $1) {
512 if (/^-(\d+)$/) {
513 ($l, $h) = ($length - $1, $length - 1);
514 } elsif (/^(\d+)-(\d*)$/) {
515 ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1);
516 } else {
517 ($l, $h) = (0, $length - 1);
518 goto ignore;
519 }
520 goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l;
521 }
522 $hdr->{"Content-Range"} = "bytes */$length";
523 $hdr->{"Content-Length"} = $length;
524 $self->err(416, "not satisfiable", $hdr, "");
525
526 satisfiable:
527 # check for segmented downloads
528 if ($l && $::NO_SEGMENTED) {
529 my $timeout = $::NOW + 15;
530 while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) {
531 if ($timeout <= $::NOW) {
532 $self->block($::BLOCKTIME, "segmented downloads are forbidden");
533 #$self->err_segmented_download;
534 } else {
535 $httpevent->wait;
536 }
537 }
538 }
539
540 $hdr->{"Content-Range"} = "bytes $l-$h/$length";
541 @code = (206, "partial content");
542 $length = $h - $l + 1;
543
544 ignore:
545 } else {
546 ($l, $h) = (0, $length - 1);
547 }
548
549 $self->{path} =~ /\.([^.]+)$/;
550 $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream";
551 $hdr->{"Content-Length"} = $length;
552
553 $self->response(@code, $hdr, "");
554
555 if ($self->{method} eq "GET") {
556 $self->{time} = $::NOW;
557
558 my $current = $Coro::current;
559
560 my ($fh, $buf, $r);
561
562 open $fh, "<", $self->{path}
563 or die "$self->{path}: late open failure ($!)";
564
565 $h -= $l - 1;
566
567 if (0) { # !AIO
568 if ($l) {
569 sysseek $fh, $l, 0;
570 }
571 }
572
573 my $transfer = $queue->start_transfer($h);
574 my $locked;
575 my $bufsize = $::WAIT_BUFSIZE; # initial buffer size
576
577 while ($h > 0) {
578 unless ($locked) {
579 if ($locked ||= $transfer->try($::WAIT_INTERVAL)) {
580 $bufsize = $::BUFSIZE;
581 $self->{time} = $::NOW;
582 }
583 }
584
585 if ($blocked{$self->{remote_id}}) {
586 $self->{h}{connection} = "close";
587 die bless {}, err::;
588 }
589
590 if (0) { # !AIO
591 sysread $fh, $buf, $h > $bufsize ? $bufsize : $h
592 or last;
593 } else {
594 aio_read($fh, $l, ($h > $bufsize ? $bufsize : $h),
595 $buf, 0, sub {
596 $r = $_[0];
597 Coro::ready($current);
598 });
599 &Coro::schedule;
600 last unless $r;
601 }
602 my $w = syswrite $self->{fh}, $buf
603 or last;
604 $::written += $w;
605 $self->{written} += $w;
606 $l += $r;
607 }
608
609 close $fh;
610 }
611 }
612
613 1;