… | |
… | |
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 | |
|
|
15 | our $accesslog; |
|
|
16 | |
|
|
17 | if ($ACCESS_LOG) { |
|
|
18 | use IO::Handle; |
|
|
19 | open $accesslog, ">>$ACCESS_LOG" |
|
|
20 | or die "$ACCESS_LOG: $!"; |
|
|
21 | $accesslog->autoflush(1); |
|
|
22 | } |
|
|
23 | |
15 | sub slog { |
24 | sub 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 | } |
… | |
… | |
70 | package conn; |
79 | package conn; |
71 | |
80 | |
72 | use Socket; |
81 | use Socket; |
73 | use HTTP::Date; |
82 | use HTTP::Date; |
74 | use Convert::Scalar 'weaken'; |
83 | use Convert::Scalar 'weaken'; |
|
|
84 | use Linux::AIO; |
75 | |
85 | |
|
|
86 | Linux::AIO::min_parallel $::AIO_PARALLEL; |
|
|
87 | |
|
|
88 | my $aio_requests = new Coro::Semaphore $::AIO_PARALLEL * 4; |
|
|
89 | |
|
|
90 | Event->io(fd => Linux::AIO::poll_fileno, |
|
|
91 | poll => 'r', async => 1, |
|
|
92 | cb => \&Linux::AIO::poll_cb); |
|
|
93 | |
76 | our %conn; # $conn{ip}{fh} => connobj |
94 | our %conn; # $conn{ip}{self} => connobj |
|
|
95 | our %uri; # $uri{ip}{uri}{self} |
77 | our %blocked; |
96 | our %blocked; |
78 | our %mimetype; |
97 | our %mimetype; |
79 | |
98 | |
80 | sub read_mimetypes { |
99 | sub read_mimetypes { |
81 | local *M; |
100 | local *M; |
… | |
… | |
114 | sub DESTROY { |
133 | sub DESTROY { |
115 | my $self = shift; |
134 | my $self = shift; |
116 | |
135 | |
117 | $::conns--; |
136 | $::conns--; |
118 | |
137 | |
|
|
138 | $self->eoconn; |
119 | delete $conn{$self->{remote_addr}}{$self*1}; |
139 | delete $conn{$self->{remote_addr}}{$self*1}; |
|
|
140 | } |
|
|
141 | |
|
|
142 | # end of connection |
|
|
143 | sub eoconn { |
|
|
144 | my $self = shift; |
120 | delete $uri{$self->{remote_addr}}{$self->{uri}}{$self*1}; |
145 | delete $uri{$self->{remote_addr}}{$self->{uri}}{$self*1}; |
121 | } |
146 | } |
122 | |
147 | |
123 | sub slog { |
148 | sub slog { |
124 | my $self = shift; |
149 | my $self = shift; |
125 | main::slog($_[0], "$self->{remote_addr}> $_[1]"); |
150 | main::slog($_[0], ($self->{remote_id} || $self->{remote_addr}) ."> $_[1]"); |
126 | } |
151 | } |
127 | |
152 | |
128 | sub response { |
153 | sub response { |
129 | my ($self, $code, $msg, $hdr, $content) = @_; |
154 | my ($self, $code, $msg, $hdr, $content) = @_; |
130 | my $res = "HTTP/1.0 $code $msg\015\012"; |
155 | my $res = "HTTP/1.1 $code $msg\015\012"; |
131 | |
156 | |
132 | $res .= "Connection: close\015\012"; |
157 | $self->{h}{connection} ||= $hdr->{Connection}; |
|
|
158 | |
133 | $res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :( |
159 | $res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :( |
134 | |
160 | |
135 | while (my ($h, $v) = each %$hdr) { |
161 | while (my ($h, $v) = each %$hdr) { |
136 | $res .= "$h: $v\015\012" |
162 | $res .= "$h: $v\015\012" |
137 | } |
163 | } |
138 | $res .= "\015\012"; |
164 | $res .= "\015\012"; |
139 | |
165 | |
140 | $res .= $content if defined $content and $self->{method} ne "HEAD"; |
166 | $res .= $content if defined $content and $self->{method} ne "HEAD"; |
141 | |
167 | |
142 | print STDERR "$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"; |
|
|
169 | |
|
|
170 | print $accesslog $log if $accesslog; |
|
|
171 | print STDERR $log; |
143 | |
172 | |
144 | $self->{written} += |
173 | $self->{written} += |
145 | print {$self->{fh}} $res; |
174 | print {$self->{fh}} $res; |
146 | } |
175 | } |
147 | |
176 | |
… | |
… | |
152 | unless (defined $content) { |
181 | unless (defined $content) { |
153 | $content = "$code $msg"; |
182 | $content = "$code $msg"; |
154 | $hdr->{"Content-Type"} = "text/plain"; |
183 | $hdr->{"Content-Type"} = "text/plain"; |
155 | $hdr->{"Content-Length"} = length $content; |
184 | $hdr->{"Content-Length"} = length $content; |
156 | } |
185 | } |
|
|
186 | $hdr->{"Connection"} = "close"; |
157 | |
187 | |
158 | $self->response($code, $msg, $hdr, $content); |
188 | $self->response($code, $msg, $hdr, $content); |
159 | |
189 | |
160 | die bless {}, err::; |
190 | die bless {}, err::; |
161 | } |
191 | } |
162 | |
192 | |
163 | sub err_blocked { |
|
|
164 | my $self = shift; |
|
|
165 | my $ip = $self->{remote_addr}; |
|
|
166 | my $time = time2str $blocked{$ip} = $::NOW + $::BLOCKTIME; |
|
|
167 | |
|
|
168 | Coro::Event::do_timer(after => 15); |
|
|
169 | |
|
|
170 | $self->err(401, "too many connections", |
|
|
171 | { |
|
|
172 | "Content-Type" => "text/html", |
|
|
173 | "Retry-After" => $::BLOCKTIME |
|
|
174 | }, |
|
|
175 | <<EOF); |
|
|
176 | <html><p> |
|
|
177 | You have been blocked because you opened too many connections. You |
|
|
178 | may retry at</p> |
|
|
179 | |
|
|
180 | <p><blockquote>$time.</blockquote></p> |
|
|
181 | |
|
|
182 | <p>Until then, each new access will renew the block. You might want to have a |
|
|
183 | look at the <a href="http://www.goof.com/pcg/marc/animefaq.html">FAQ</a>.</p> |
|
|
184 | </html> |
|
|
185 | EOF |
|
|
186 | } |
|
|
187 | |
|
|
188 | sub handle { |
193 | sub handle { |
189 | my $self = shift; |
194 | my $self = shift; |
190 | my $fh = $self->{fh}; |
195 | my $fh = $self->{fh}; |
191 | |
196 | |
|
|
197 | my $host; |
|
|
198 | |
|
|
199 | $fh->timeout($::REQ_TIMEOUT); |
192 | #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 | |
193 | $self->{h} = {}; |
214 | $self->{h} = {}; |
194 | |
215 | |
195 | # read request and parse first line |
|
|
196 | $fh->timeout($::REQ_TIMEOUT); |
|
|
197 | my $req = $fh->readline("\015\012\015\012"); |
|
|
198 | $fh->timeout($::RES_TIMEOUT); |
216 | $fh->timeout($::RES_TIMEOUT); |
199 | |
|
|
200 | defined $req or |
|
|
201 | $self->err(408, "request timeout"); |
|
|
202 | |
|
|
203 | my $ip = $self->{remote_addr}; |
217 | my $ip = $self->{remote_addr}; |
204 | |
218 | |
205 | $self->err_blocked($blocked{$ip}); |
|
|
206 | if ($blocked{$ip}) { |
219 | if ($blocked{$ip}) { |
207 | $self->err_blocked($blocked{$ip}) |
220 | $self->err_blocked($blocked{$ip}) |
208 | if $blocked{$ip} > $::NOW; |
221 | if $blocked{$ip} > $::NOW; |
209 | |
222 | |
210 | delete $blocked{$ip}; |
223 | delete $blocked{$ip}; |
… | |
… | |
220 | ([^\040]+) \040+ |
233 | ([^\040]+) \040+ |
221 | HTTP\/([0-9]+\.[0-9]+) |
234 | HTTP\/([0-9]+\.[0-9]+) |
222 | \015\012/gx |
235 | \015\012/gx |
223 | or $self->err(405, "method not allowed", { Allow => "GET,HEAD" }); |
236 | or $self->err(405, "method not allowed", { Allow => "GET,HEAD" }); |
224 | |
237 | |
225 | $2 ne "1.0" |
|
|
226 | or $self->err(506, "http protocol version not supported"); |
|
|
227 | |
|
|
228 | $self->{method} = $1; |
238 | $self->{method} = $1; |
229 | $self->{uri} = $2; |
239 | $self->{uri} = $2; |
|
|
240 | $self->{version} = $3; |
|
|
241 | |
|
|
242 | $3 =~ /^1\./ |
|
|
243 | or $self->err(506, "http protocol version $3 not supported"); |
230 | |
244 | |
231 | # parse headers |
245 | # parse headers |
232 | { |
246 | { |
233 | my (%hdr, $h, $v); |
247 | my (%hdr, $h, $v); |
234 | |
248 | |
… | |
… | |
245 | |
259 | |
246 | $self->{h}{$h} = substr $v, 1 |
260 | $self->{h}{$h} = substr $v, 1 |
247 | while ($h, $v) = each %hdr; |
261 | while ($h, $v) = each %hdr; |
248 | } |
262 | } |
249 | |
263 | |
|
|
264 | # find out server name and port |
|
|
265 | if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) { |
|
|
266 | $host = $1; |
|
|
267 | } else { |
|
|
268 | $host = $self->{h}{host}; |
|
|
269 | } |
|
|
270 | |
|
|
271 | if (defined $host) { |
250 | $self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80; |
272 | $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80; |
|
|
273 | } else { |
|
|
274 | ($self->{server_port}, $host) |
|
|
275 | = unpack_sockaddr_in $self->{fh}->getsockname |
|
|
276 | or $self->err(500, "unable to get socket name"); |
|
|
277 | $host = inet_ntoa $host; |
|
|
278 | } |
|
|
279 | |
|
|
280 | $self->{server_name} = $host; |
|
|
281 | |
|
|
282 | # remote id should be unique per user |
|
|
283 | $self->{remote_id} = $self->{remote_addr}; |
|
|
284 | |
|
|
285 | if (exists $self->{h}{"client-ip"}) { |
|
|
286 | $self->{remote_id} .= "[".$self->{h}{"client-ip"}."]"; |
|
|
287 | } elsif (exists $self->{h}{"x-forwarded-for"}) { |
|
|
288 | $self->{remote_id} .= "[".$self->{h}{"x-forwarded-for"}."]"; |
|
|
289 | } |
251 | |
290 | |
252 | weaken ($uri{$self->{remote_addr}}{$self->{uri}}{$self*1} = $self); |
291 | weaken ($uri{$self->{remote_addr}}{$self->{uri}}{$self*1} = $self); |
253 | |
292 | |
|
|
293 | eval { |
254 | $self->map_uri; |
294 | $self->map_uri; |
255 | $self->respond; |
295 | $self->respond; |
|
|
296 | }; |
|
|
297 | |
|
|
298 | $self->eoconn; |
|
|
299 | |
|
|
300 | die if $@ && !ref $@; |
|
|
301 | |
|
|
302 | last if $self->{h}{connection} =~ /close/ || $self->{version} < 1.1; |
|
|
303 | |
|
|
304 | $fh->timeout($::PER_TIMEOUT); |
256 | #} |
305 | } |
257 | } |
306 | } |
258 | |
307 | |
259 | # uri => path mapping |
308 | # uri => path mapping |
260 | sub map_uri { |
309 | sub map_uri { |
261 | my $self = shift; |
310 | my $self = shift; |
262 | my $host = $self->{h}{host} || "default"; |
311 | my $host = $self->{server_name}; |
263 | my $uri = $self->{uri}; |
312 | my $uri = $self->{uri}; |
264 | |
313 | |
265 | # some massaging, also makes it more secure |
314 | # some massaging, also makes it more secure |
266 | $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; |
315 | $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; |
267 | $uri =~ s%//+%/%g; |
316 | $uri =~ s%//+%/%g; |
… | |
… | |
277 | $self->{path} = "$::DOCROOT/$host$uri"; |
326 | $self->{path} = "$::DOCROOT/$host$uri"; |
278 | |
327 | |
279 | $self->access_check; |
328 | $self->access_check; |
280 | } |
329 | } |
281 | |
330 | |
282 | sub server_address { |
|
|
283 | my $self = shift; |
|
|
284 | my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->getsockname |
|
|
285 | or $self->err(500, "unable to get socket name"); |
|
|
286 | ((inet_ntoa $iaddr), $port); |
|
|
287 | } |
|
|
288 | |
|
|
289 | sub server_host { |
|
|
290 | my $self = shift; |
|
|
291 | if (exists $self->{h}{host}) { |
|
|
292 | return $self->{h}{host}; |
|
|
293 | } else { |
|
|
294 | return (($self->server_address)[0]); |
|
|
295 | } |
|
|
296 | } |
|
|
297 | |
|
|
298 | sub server_hostport { |
|
|
299 | my $self = shift; |
|
|
300 | my ($host, $port); |
|
|
301 | if (exists $self->{h}{host}) { |
|
|
302 | ($host, $port) = ($self->{h}{host}, $self->{server_port}); |
|
|
303 | } else { |
|
|
304 | ($host, $port) = $self->server_address; |
|
|
305 | } |
|
|
306 | $port = $port == 80 ? "" : ":$port"; |
|
|
307 | $host.$port; |
|
|
308 | } |
|
|
309 | |
|
|
310 | sub _cgi { |
331 | sub _cgi { |
311 | my $self = shift; |
332 | my $self = shift; |
312 | my $path = shift; |
333 | my $path = shift; |
313 | my $fh; |
334 | my $fh; |
314 | |
335 | |
315 | # no two-way xxx supported |
336 | # no two-way xxx supported |
316 | if (0 == fork) { |
337 | if (0 == fork) { |
317 | open STDOUT, ">&".fileno($self->{fh}); |
338 | open STDOUT, ">&".fileno($self->{fh}); |
318 | if (chdir $::DOCROOT) { |
339 | if (chdir $::DOCROOT) { |
319 | $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike |
340 | $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike |
320 | $ENV{HTTP_HOST} = $self->server_host; |
341 | $ENV{HTTP_HOST} = $self->{server_name}; |
321 | $ENV{HTTP_PORT} = $self->{server_host}; |
342 | $ENV{HTTP_PORT} = $self->{server_port}; |
322 | $ENV{SCRIPT_NAME} = $self->{name}; |
343 | $ENV{SCRIPT_NAME} = $self->{name}; |
323 | exec $path; |
344 | exec $path; |
324 | } |
345 | } |
325 | Coro::State::_exit(0); |
346 | Coro::State::_exit(0); |
326 | } else { |
347 | } else { |
|
|
348 | die; |
327 | } |
349 | } |
|
|
350 | } |
|
|
351 | |
|
|
352 | sub server_hostport { |
|
|
353 | $_[0]{server_port} == 80 |
|
|
354 | ? $_[0]{server_name} |
|
|
355 | : "$_[0]{server_name}:$_[0]{server_port}"; |
328 | } |
356 | } |
329 | |
357 | |
330 | sub respond { |
358 | sub respond { |
331 | my $self = shift; |
359 | my $self = shift; |
332 | my $path = $self->{path}; |
360 | my $path = $self->{path}; |
… | |
… | |
342 | |
370 | |
343 | if (-d _ && -r _) { |
371 | if (-d _ && -r _) { |
344 | # directory |
372 | # directory |
345 | if ($path !~ /\/$/) { |
373 | if ($path !~ /\/$/) { |
346 | # create a redirect to get the trailing "/" |
374 | # create a redirect to get the trailing "/" |
347 | my $host = $self->server_hostport; |
375 | # we don't try to avoid the :80 |
348 | $self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" }); |
376 | $self->err(301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" }); |
349 | } else { |
377 | } else { |
350 | $ims < $self->{stat}[9] |
378 | $ims < $self->{stat}[9] |
351 | or $self->err(304, "not modified"); |
379 | or $self->err(304, "not modified"); |
352 | |
380 | |
353 | if ($self->{method} eq "GET") { |
|
|
354 | if (-r "$path/index.html") { |
381 | if (-r "$path/index.html") { |
355 | $self->{path} .= "/index.html"; |
382 | $self->{path} .= "/index.html"; |
356 | $self->handle_file; |
383 | $self->handle_file; |
357 | } else { |
384 | } else { |
358 | $self->handle_dir; |
385 | $self->handle_dir; |
359 | } |
|
|
360 | } |
386 | } |
361 | } |
387 | } |
362 | } elsif (-f _ && -r _) { |
388 | } elsif (-f _ && -r _) { |
363 | -x _ and $self->err(403, "forbidden"); |
389 | -x _ and $self->err(403, "forbidden"); |
364 | $self->handle_file; |
390 | $self->handle_file; |
… | |
… | |
397 | ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1); |
423 | ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1); |
398 | } else { |
424 | } else { |
399 | ($l, $h) = (0, $length - 1); |
425 | ($l, $h) = (0, $length - 1); |
400 | goto ignore; |
426 | goto ignore; |
401 | } |
427 | } |
402 | goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h > $l; |
428 | goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l; |
403 | } |
429 | } |
404 | $hdr->{"Content-Range"} = "bytes */$length"; |
430 | $hdr->{"Content-Range"} = "bytes */$length"; |
|
|
431 | $hdr->{"Content-Length"} = $length; |
|
|
432 | $self->slog(9, "not satisfiable($self->{h}{range}|".$self->{h}{"user-agent"}.")"); |
405 | $self->err(416, "not satisfiable", $hdr); |
433 | $self->err(416, "not satisfiable", $hdr, ""); |
406 | |
434 | |
407 | satisfiable: |
435 | satisfiable: |
408 | # check for segmented downloads |
436 | # check for segmented downloads |
409 | if ($l && $::NO_SEGMENTED) { |
437 | if ($l && $::NO_SEGMENTED) { |
|
|
438 | my $delay = 180; |
410 | if (%{$uri{$self->{remote_addr}}{$self->{uri}}} > 1) { |
439 | while (%{$uri{$self->{remote_addr}}{$self->{uri}}} > 1) { |
|
|
440 | if ($delay <= 0) { |
|
|
441 | $self->err_segmented_download; |
|
|
442 | } else { |
411 | Coro::Event::do_timer(after => 15); |
443 | Coro::Event::do_timer(after => 3); $delay -= 3; |
412 | |
444 | } |
413 | $self->err(400, "segmented downloads are not allowed"); |
|
|
414 | } |
445 | } |
415 | } |
446 | } |
416 | |
447 | |
417 | $hdr->{"Content-Range"} = "bytes $l-$h/$length"; |
448 | $hdr->{"Content-Range"} = "bytes $l-$h/$length"; |
418 | @code = (206, "partial content"); |
449 | @code = (206, "partial content"); |
… | |
… | |
428 | $hdr->{"Content-Length"} = $length; |
459 | $hdr->{"Content-Length"} = $length; |
429 | |
460 | |
430 | $self->response(@code, $hdr, ""); |
461 | $self->response(@code, $hdr, ""); |
431 | |
462 | |
432 | if ($self->{method} eq "GET") { |
463 | if ($self->{method} eq "GET") { |
433 | my ($fh, $buf); |
464 | my ($fh, $buf, $r); |
|
|
465 | my $current = $Coro::current; |
434 | open $fh, "<", $self->{path} |
466 | open $fh, "<", $self->{path} |
435 | or die "$self->{path}: late open failure ($!)"; |
467 | or die "$self->{path}: late open failure ($!)"; |
436 | |
468 | |
437 | if ($l) { |
|
|
438 | sysseek $fh, $l, 0 |
|
|
439 | or die "$self->{path}: cannot seek to $l ($!)"; |
|
|
440 | } |
|
|
441 | |
|
|
442 | $h -= $l - 1; |
469 | $h -= $l - 1; |
443 | |
470 | |
|
|
471 | if (0) { |
|
|
472 | if ($l) { |
|
|
473 | sysseek $fh, $l, 0; |
|
|
474 | } |
|
|
475 | } |
|
|
476 | |
444 | while ($h > 0) { |
477 | while ($h > 0) { |
|
|
478 | if (0) { |
445 | $h -= sysread $fh, $buf, $h > $::BUFSIZE ? $::BUFSIZE : $h; |
479 | sysread $fh, $buf, $h > $::BUFSIZE ? $::BUFSIZE : $h |
|
|
480 | or last; |
|
|
481 | } else { |
|
|
482 | undef $buf; |
|
|
483 | $aio_requests->down; |
|
|
484 | aio_read($fh, $l, ($h > $::BUFSIZE ? $::BUFSIZE : $h), |
|
|
485 | $buf, 0, sub { |
|
|
486 | $r = $_[0]; |
|
|
487 | $current->ready; |
|
|
488 | }); |
|
|
489 | &Coro::schedule; |
|
|
490 | $aio_requests->up; |
|
|
491 | last unless $r; |
|
|
492 | } |
446 | my $w = $self->{fh}->syswrite($buf) |
493 | my $w = $self->{fh}->syswrite($buf) |
447 | or last; |
494 | or last; |
448 | $::written += $w; |
495 | $::written += $w; |
449 | $self->{written} += $w; |
496 | $self->{written} += $w; |
|
|
497 | $l += $r; |
450 | } |
498 | } |
451 | } |
499 | } |
452 | |
500 | |
453 | close $fh; |
501 | close $fh; |
454 | } |
502 | } |