1 | use Coro; |
1 | use Coro; |
2 | use Coro::Semaphore; |
2 | use Coro::Semaphore; |
|
|
3 | use Coro::SemaphoreSet; |
3 | use Coro::Event; |
4 | use Coro::EV; |
4 | use Coro::Socket; |
5 | use Coro::Socket; |
|
|
6 | use Coro::Signal; |
|
|
7 | use Coro::AIO (); |
5 | |
8 | |
6 | no utf8; |
9 | use Fcntl; |
7 | use bytes; |
10 | use HTTP::Date; |
|
|
11 | use POSIX (); |
|
|
12 | |
|
|
13 | use Compress::Zlib (); |
|
|
14 | |
|
|
15 | use common::sense; |
8 | |
16 | |
9 | # at least on my machine, this thingy serves files |
17 | # at least on my machine, this thingy serves files |
10 | # quite a bit faster than apache, ;) |
18 | # quite a bit faster than apache, ;) |
11 | # and quite a bit slower than thttpd :( |
19 | # and quite a bit slower than thttpd :( |
12 | |
20 | |
13 | $SIG{PIPE} = 'IGNORE'; |
21 | $SIG{PIPE} = 'IGNORE'; |
14 | |
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 | |
15 | sub slog { |
54 | sub slog { |
16 | my $level = shift; |
55 | my $level = shift; |
17 | my $format = shift; |
56 | my $format = shift; |
|
|
57 | my $NOW = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW); |
18 | printf "---: $format\n", @_; |
58 | printf "$NOW: $format\n", @_; |
|
|
59 | printf $errorlog "$NOW: $format\n", @_ if $errorlog; |
19 | } |
60 | } |
20 | |
61 | |
21 | my $connections = new Coro::Semaphore $MAX_CONNECTS; |
62 | our $connections = new Coro::Semaphore $::MAX_CONNECTS || 250; |
|
|
63 | our $httpevent = new Coro::Signal; |
22 | |
64 | |
23 | my @newcons; |
65 | our $queue_file = new transferqueue $::MAX_TRANSFERS; |
24 | my @pool; |
66 | our $queue_index = new transferqueue 10; |
25 | |
67 | |
26 | # one "execution thread" |
68 | our $tbf_top = new tbf rate => $::TBF_RATE || 100000; |
27 | sub handler { |
69 | |
|
|
70 | my $unused_bytes = 0; |
|
|
71 | my $unused_last = time; |
|
|
72 | |
|
|
73 | sub unused_bandwidth { |
|
|
74 | $unused_bytes += $_[0]; |
|
|
75 | if ($unused_last < $NOW - 30 && $unused_bytes / ($NOW - $unused_last) > 50000) { |
|
|
76 | $unused_last = $NOW; |
|
|
77 | $unused_bytes = 0; |
|
|
78 | $queue_file->force_wake_next; |
|
|
79 | slog 1, "forced filetransfer due to unused bandwidth"; |
|
|
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"; |
28 | while () { |
91 | while () { |
29 | my $new = pop @newcons; |
92 | $connections->down; |
30 | if ($new) { |
93 | my @conn = $listen->accept; |
|
|
94 | #slog 3, "accepted @$connections ".scalar(@pool); |
|
|
95 | |
|
|
96 | async_pool { |
31 | eval { |
97 | eval { |
32 | conn->new(@$new)->handle; |
98 | conn->new (@conn)->handle; |
33 | }; |
99 | }; |
34 | slog 1, "$@" if $@ && !ref $@; |
100 | slog 1, "$@" if $@ && !ref $@; |
|
|
101 | |
|
|
102 | $httpevent->broadcast; # only for testing, but doesn't matter much |
|
|
103 | |
35 | $connections->up; |
104 | $connections->up; |
36 | } else { |
|
|
37 | last if @pool >= $MAX_POOL; |
|
|
38 | push @pool, $Coro::current; |
|
|
39 | schedule; |
|
|
40 | } |
105 | } |
|
|
106 | } |
41 | } |
107 | }; |
42 | } |
108 | } |
43 | |
109 | |
44 | my $http_port = new Coro::Socket |
110 | my $http_port = new Coro::Socket |
45 | LocalAddr => $SERVER_HOST, |
111 | LocalAddr => $::SERVER_HOST, |
46 | LocalPort => $SERVER_PORT, |
112 | LocalPort => $::SERVER_PORT, |
47 | ReuseAddr => 1, |
113 | ReuseAddr => 1, |
48 | Listen => 50, |
114 | Listen => 50, |
49 | or die "unable to start server"; |
115 | or die "unable to start server"; |
50 | |
116 | |
51 | push @listen_sockets, $http_port; |
117 | listen_on $http_port; |
52 | |
118 | |
53 | # the "main thread" |
119 | if ($::SERVER_PORT2) { |
54 | async { |
120 | my $http_port = new Coro::Socket |
55 | slog 1, "accepting connections"; |
121 | LocalAddr => $::SERVER_HOST, |
56 | while () { |
122 | LocalPort => $::SERVER_PORT2, |
57 | $connections->down; |
123 | ReuseAddr => 1, |
58 | push @newcons, [$http_port->accept]; |
124 | Listen => 50, |
59 | #slog 3, "accepted @$connections ".scalar(@pool); |
125 | or die "unable to start server"; |
60 | $::NOW = time; |
|
|
61 | if (@pool) { |
|
|
62 | (pop @pool)->ready; |
|
|
63 | } else { |
|
|
64 | async \&handler; |
|
|
65 | } |
|
|
66 | |
126 | |
67 | } |
127 | listen_on $http_port; |
68 | }; |
128 | } |
69 | |
129 | |
70 | package conn; |
130 | package conn; |
|
|
131 | |
|
|
132 | use common::sense; |
71 | |
133 | |
72 | use Socket; |
134 | use Socket; |
73 | use HTTP::Date; |
135 | use HTTP::Date; |
74 | use Convert::Scalar 'weaken'; |
136 | use Convert::Scalar 'weaken'; |
|
|
137 | use IO::AIO; |
75 | |
138 | |
|
|
139 | IO::AIO::min_parallel $::AIO_PARALLEL; |
|
|
140 | |
|
|
141 | our $AIO_WATCHER = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb; |
|
|
142 | |
76 | our %conn; # $conn{ip}{fh} => connobj |
143 | our %conn; # $conn{ip}{self} => connobj |
|
|
144 | our %uri; # $uri{ip}{uri}{self} |
77 | our %blocked; |
145 | our %blocked; |
78 | our %mimetype; |
146 | our %mimetype; |
79 | |
147 | |
80 | sub read_mimetypes { |
148 | sub read_mimetypes { |
81 | local *M; |
|
|
82 | if (open M, "<mime_types") { |
149 | if (open my $fh, "<mime_types") { |
83 | while (<M>) { |
150 | while (<$fh>) { |
84 | if (/^([^#]\S+)\t+(\S+)$/) { |
151 | if (/^([^#]\S+)\t+(\S+)$/) { |
85 | $mimetype{lc $1} = $2; |
152 | $mimetype{lc $1} = $2; |
86 | } |
153 | } |
87 | } |
154 | } |
88 | } else { |
155 | } else { |
… | |
… | |
92 | |
159 | |
93 | read_mimetypes; |
160 | read_mimetypes; |
94 | |
161 | |
95 | sub new { |
162 | sub new { |
96 | my $class = shift; |
163 | my $class = shift; |
|
|
164 | my $fh = shift; |
97 | my $peername = shift; |
165 | my $peername = shift; |
98 | my $fh = shift; |
|
|
99 | my $self = bless { fh => $fh }, $class; |
166 | my $self = bless { fh => $fh }, $class; |
100 | my (undef, $iaddr) = unpack_sockaddr_in $peername |
167 | my (undef, $iaddr) = unpack_sockaddr_in $peername |
101 | or $self->err(500, "unable to decode peername"); |
168 | or $self->err (500, "unable to decode peername"); |
102 | |
169 | |
|
|
170 | $self->{remote_addr} = |
103 | $self->{remote_addr} = inet_ntoa $iaddr; |
171 | $self->{remote_id} = inet_ntoa $iaddr; |
|
|
172 | |
104 | $self->{time} = $::NOW; |
173 | $self->{time} = $::NOW; |
105 | |
174 | |
106 | # enter ourselves into various lists |
175 | weaken ($Coro::current->{conn} = $self); |
107 | weaken ($conn{$self->{remote_addr}}{$self*1} = $self); |
|
|
108 | |
176 | |
109 | $::conns++; |
177 | ++$::conns; |
|
|
178 | $::maxconns = $::conns if $::conns > $::maxconns; |
110 | |
179 | |
111 | $self; |
180 | $self |
112 | } |
181 | } |
113 | |
182 | |
114 | sub DESTROY { |
183 | sub DESTROY { |
115 | my $self = shift; |
184 | my $self = shift; |
116 | |
185 | |
117 | $::conns--; |
186 | --$::conns; |
118 | |
|
|
119 | delete $conn{$self->{remote_addr}}{$self*1}; |
|
|
120 | delete $uri{$self->{remote_addr}}{$self->{uri}}{$self*1}; |
|
|
121 | } |
187 | } |
|
|
188 | |
|
|
189 | sub prune_cache { |
|
|
190 | my $hash = $_[0]; |
|
|
191 | |
|
|
192 | for (keys %$hash) { |
|
|
193 | if (ref $hash->{$_} eq HASH::) { |
|
|
194 | prune_cache($hash->{$_}); |
|
|
195 | unless (scalar keys %{$hash->{$_}}) { |
|
|
196 | delete $hash->{$_}; |
|
|
197 | } |
|
|
198 | } |
|
|
199 | } |
|
|
200 | } |
|
|
201 | |
|
|
202 | sub prune_caches { |
|
|
203 | prune_cache \%conn; |
|
|
204 | prune_cache \%uri; |
|
|
205 | |
|
|
206 | for (keys %blocked) { |
|
|
207 | delete $blocked{$_} unless $blocked{$_}[0] > $::NOW; |
|
|
208 | } |
|
|
209 | } |
|
|
210 | |
|
|
211 | our $PRUNE_WATCHER = EV::timer 60, 60, \&prune_caches; |
122 | |
212 | |
123 | sub slog { |
213 | sub slog { |
124 | my $self = shift; |
214 | my $self = shift; |
125 | main::slog($_[0], "$self->{remote_addr}> $_[1]"); |
215 | main::slog($_[0], "$self->{remote_id}> $_[1]"); |
126 | } |
216 | } |
127 | |
217 | |
128 | sub response { |
218 | sub response { |
129 | my ($self, $code, $msg, $hdr, $content) = @_; |
219 | my ($self, $code, $msg, $hdr, $content) = @_; |
130 | my $res = "HTTP/1.0 $code $msg\015\012"; |
220 | my $res = "HTTP/1.1 $code $msg\015\012"; |
|
|
221 | my $GZ = ""; |
131 | |
222 | |
132 | $res .= "Connection: close\015\012"; |
223 | if (exists $hdr->{Connection}) { |
133 | $res .= "Date: ".(time2str $::NOW)."\015\012"; # slow? nah. :( |
224 | if ($hdr->{Connection} =~ /close/) { |
|
|
225 | $self->{h}{connection} = "close" |
|
|
226 | } |
|
|
227 | } else { |
|
|
228 | if ($self->{version} < 1.1) { |
|
|
229 | if ($self->{h}{connection} =~ /keep-alive/i) { |
|
|
230 | $hdr->{Connection} = "Keep-Alive"; |
|
|
231 | } else { |
|
|
232 | $self->{h}{connection} = "close" |
|
|
233 | } |
|
|
234 | } |
|
|
235 | } |
|
|
236 | |
|
|
237 | if ($self->{method} ne "HEAD" |
|
|
238 | && $self->{h}{"accept-encoding"} =~ /\bgzip\b/ |
|
|
239 | && 400 < length $content |
|
|
240 | && $hdr->{"Content-Length"} == length $content |
|
|
241 | && !exists $hdr->{"Content-Encoding"} |
|
|
242 | ) { |
|
|
243 | my $orig = length $content; |
|
|
244 | $hdr->{"Content-Encoding"} = "gzip"; |
|
|
245 | $content = Compress::Zlib::memGzip(\$content); |
|
|
246 | $hdr->{"Content-Length"} = length $content; |
|
|
247 | $GZ = sprintf "GZ%02d", 100 - 100*((length $content) / $orig); |
|
|
248 | } |
|
|
249 | |
|
|
250 | $res .= "Date: $HTTP_NOW\015\012"; |
|
|
251 | $res .= "Server: $::NAME\015\012"; |
134 | |
252 | |
135 | while (my ($h, $v) = each %$hdr) { |
253 | while (my ($h, $v) = each %$hdr) { |
136 | $res .= "$h: $v\015\012" |
254 | $res .= "$h: $v\015\012" |
137 | } |
255 | } |
138 | $res .= "\015\012"; |
256 | $res .= "\015\012"; |
139 | |
257 | |
140 | $res .= $content if defined $content and $self->{method} ne "HEAD"; |
258 | $res .= $content if defined $content and $self->{method} ne "HEAD"; |
141 | |
259 | |
142 | print STDERR "$self->{remote_addr} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}." \"$self->{h}{referer}\"\n";#d# |
260 | my $log = (POSIX::strftime "%Y-%m-%d %H:%M:%S", gmtime $::NOW). |
|
|
261 | " $self->{remote_id} \"$self->{uri}\" $code ".$hdr->{"Content-Length"}.$GZ. |
|
|
262 | " \"$self->{h}{referer}\"\n"; |
143 | |
263 | |
144 | $self->{written} += |
264 | print $::accesslog $log if $::accesslog; |
145 | print {$self->{fh}} $res; |
265 | print STDERR $log; |
|
|
266 | |
|
|
267 | $tbf_top->request(length $res, 1e6); |
|
|
268 | $self->{written} += print {$self->{fh}} $res; |
146 | } |
269 | } |
147 | |
270 | |
148 | sub err { |
271 | sub err { |
149 | my $self = shift; |
272 | my $self = shift; |
150 | my ($code, $msg, $hdr, $content) = @_; |
273 | my ($code, $msg, $hdr, $content) = @_; |
151 | |
274 | |
152 | unless (defined $content) { |
275 | unless (defined $content) { |
153 | $content = "$code $msg"; |
276 | $content = "$code $msg\n"; |
154 | $hdr->{"Content-Type"} = "text/plain"; |
277 | $hdr->{"Content-Type"} = "text/plain"; |
155 | $hdr->{"Content-Length"} = length $content; |
278 | $hdr->{"Content-Length"} = length $content; |
156 | } |
279 | } |
|
|
280 | $hdr->{"Connection"} = "close"; |
157 | |
281 | |
158 | $self->response($code, $msg, $hdr, $content); |
282 | $self->response ($code, $msg, $hdr, $content); |
159 | |
283 | |
160 | die bless {}, err::; |
284 | die bless {}, err:: |
161 | } |
|
|
162 | |
|
|
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 | } |
285 | } |
187 | |
286 | |
188 | sub handle { |
287 | sub handle { |
189 | my $self = shift; |
288 | my $self = shift; |
190 | my $fh = $self->{fh}; |
289 | my $fh = $self->{fh}; |
191 | |
290 | |
|
|
291 | my $host; |
|
|
292 | |
|
|
293 | $fh->timeout($::REQ_TIMEOUT); |
192 | #while() { |
294 | while () { |
|
|
295 | $self->{reqs}++; |
|
|
296 | |
|
|
297 | # read request and parse first line |
|
|
298 | my $req = $fh->readline("\015\012\015\012"); |
|
|
299 | |
|
|
300 | unless (defined $req) { |
|
|
301 | if (exists $self->{version}) { |
|
|
302 | last; |
|
|
303 | } else { |
|
|
304 | $self->err(408, "request timeout"); |
|
|
305 | } |
|
|
306 | } |
|
|
307 | |
193 | $self->{h} = {}; |
308 | $self->{h} = {}; |
194 | |
309 | |
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); |
310 | $fh->timeout($::RES_TIMEOUT); |
199 | |
|
|
200 | defined $req or |
|
|
201 | $self->err(408, "request timeout"); |
|
|
202 | |
|
|
203 | my $ip = $self->{remote_addr}; |
|
|
204 | |
|
|
205 | if ($blocked{$ip}) { |
|
|
206 | $self->err_blocked($blocked{$ip}) |
|
|
207 | if $blocked{$ip} > $::NOW; |
|
|
208 | |
|
|
209 | delete $blocked{$ip}; |
|
|
210 | } |
|
|
211 | |
|
|
212 | if (%{$conn{$ip}} > $::MAX_CONN_IP) { |
|
|
213 | $self->slog(2, "blocked ip $ip"); |
|
|
214 | $self->err_blocked; |
|
|
215 | } |
|
|
216 | |
311 | |
217 | $req =~ /^(?:\015\012)? |
312 | $req =~ /^(?:\015\012)? |
218 | (GET|HEAD) \040+ |
313 | (GET|HEAD) \040+ |
219 | ([^\040]+) \040+ |
314 | ([^\040]+) \040+ |
220 | HTTP\/([0-9]+\.[0-9]+) |
315 | HTTP\/([0-9]+\.[0-9]+) |
221 | \015\012/gx |
316 | \015\012/gx |
222 | or $self->err(405, "method not allowed", { Allow => "GET,HEAD" }); |
317 | or $self->err(405, "method not allowed", { Allow => "GET,HEAD" }); |
223 | |
318 | |
224 | $2 ne "1.0" |
|
|
225 | or $self->err(506, "http protocol version not supported"); |
|
|
226 | |
|
|
227 | $self->{method} = $1; |
319 | $self->{method} = $1; |
228 | $self->{uri} = $2; |
320 | $self->{uri} = $2; |
|
|
321 | $self->{version} = $3; |
|
|
322 | |
|
|
323 | $3 =~ /^1\./ |
|
|
324 | or $self->err(506, "http protocol version $3 not supported"); |
229 | |
325 | |
230 | # parse headers |
326 | # parse headers |
231 | { |
327 | { |
232 | my (%hdr, $h, $v); |
328 | my (%hdr, $h, $v); |
233 | |
329 | |
234 | $hdr{lc $1} .= ",$2" |
330 | $hdr{lc $1} .= ",$2" |
235 | while $req =~ /\G |
331 | while $req =~ /\G |
236 | ([^:\000-\040]+): |
332 | ([^:\000-\040]+): |
237 | [\008\040]* |
333 | [\011\040]* |
238 | ((?: [^\015\012]+ | \015\012[\008\040] )*) |
334 | ((?: [^\015\012]+ | \015\012[\011\040] )*) |
239 | \015\012 |
335 | \015\012 |
240 | /gxc; |
336 | /gxc; |
241 | |
337 | |
242 | $req =~ /\G\015\012$/ |
338 | $req =~ /\G\015\012$/ |
243 | or $self->err(400, "bad request"); |
339 | or $self->err(400, "bad request"); |
244 | |
340 | |
245 | $self->{h}{$h} = substr $v, 1 |
341 | $self->{h}{$h} = substr $v, 1 |
246 | while ($h, $v) = each %hdr; |
342 | while ($h, $v) = each %hdr; |
247 | } |
343 | } |
248 | |
344 | |
|
|
345 | # remote id should be unique per user |
|
|
346 | my $id = $self->{remote_addr}; |
|
|
347 | |
|
|
348 | if (exists $self->{h}{"client-ip"}) { |
|
|
349 | $id .= "[".$self->{h}{"client-ip"}."]"; |
|
|
350 | } elsif (exists $self->{h}{"x-forwarded-for"}) { |
|
|
351 | $id .= "[".$self->{h}{"x-forwarded-for"}."]"; |
|
|
352 | } |
|
|
353 | |
|
|
354 | $self->{remote_id} = $id; |
|
|
355 | |
|
|
356 | weaken (local $conn{$id}{$self*1} = $self); |
|
|
357 | |
|
|
358 | if ($blocked{$id}) { |
|
|
359 | $self->err_blocked |
|
|
360 | if $blocked{$id}[0] > $::NOW; |
|
|
361 | |
|
|
362 | delete $blocked{$id}; |
|
|
363 | } |
|
|
364 | |
|
|
365 | # find out server name and port |
|
|
366 | if ($self->{uri} =~ s/^http:\/\/([^\/?#]*)//i) { |
|
|
367 | $host = $1; |
|
|
368 | } else { |
|
|
369 | $host = $self->{h}{host}; |
|
|
370 | } |
|
|
371 | |
|
|
372 | if (defined $host) { |
249 | $self->{server_port} = $self->{h}{host} =~ s/:([0-9]+)$// ? $1 : 80; |
373 | $self->{server_port} = $host =~ s/:([0-9]+)$// ? $1 : 80; |
|
|
374 | } else { |
|
|
375 | ($self->{server_port}, $host) |
|
|
376 | = unpack_sockaddr_in $self->{fh}->sockname |
|
|
377 | or $self->err(500, "unable to get socket name"); |
|
|
378 | $host = inet_ntoa $host; |
|
|
379 | } |
250 | |
380 | |
|
|
381 | $self->{server_name} = $host; |
|
|
382 | |
251 | weaken ($uri{$self->{remote_addr}}{$self->{uri}}{$self*1} = $self); |
383 | weaken (local $uri{$id}{$self->{uri}}{$self*1} = $self); |
252 | |
384 | |
|
|
385 | eval { |
253 | $self->map_uri; |
386 | $self->map_uri; |
254 | $self->respond; |
387 | $self->respond; |
|
|
388 | }; |
|
|
389 | |
|
|
390 | die if $@ && !ref $@; |
|
|
391 | |
|
|
392 | last if $self->{h}{connection} =~ /close/i; |
|
|
393 | |
|
|
394 | $httpevent->broadcast; |
|
|
395 | |
|
|
396 | $fh->timeout($::PER_TIMEOUT); |
255 | #} |
397 | } |
|
|
398 | } |
|
|
399 | |
|
|
400 | sub block { |
|
|
401 | my $self = shift; |
|
|
402 | |
|
|
403 | $blocked{$self->{remote_id}} = [$::NOW + $_[0], $_[1]]; |
|
|
404 | $self->slog(2, "blocked ip $self->{remote_id}"); |
|
|
405 | $self->err_blocked; |
256 | } |
406 | } |
257 | |
407 | |
258 | # uri => path mapping |
408 | # uri => path mapping |
259 | sub map_uri { |
409 | sub map_uri { |
260 | my $self = shift; |
410 | my $self = shift; |
261 | my $host = $self->{h}{host} || "default"; |
411 | my $host = $self->{server_name}; |
262 | my $uri = $self->{uri}; |
412 | my $uri = $self->{uri}; |
|
|
413 | |
|
|
414 | $host =~ /[\/\\]/ |
|
|
415 | and $self->err(400, "bad request"); |
263 | |
416 | |
264 | # some massaging, also makes it more secure |
417 | # some massaging, also makes it more secure |
265 | $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; |
418 | $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; |
266 | $uri =~ s%//+%/%g; |
419 | $uri =~ s%//+%/%g; |
267 | $uri =~ s%/\.(?=/|$)%%g; |
420 | $uri =~ s%/\.(?=/|$)%%g; |
… | |
… | |
276 | $self->{path} = "$::DOCROOT/$host$uri"; |
429 | $self->{path} = "$::DOCROOT/$host$uri"; |
277 | |
430 | |
278 | $self->access_check; |
431 | $self->access_check; |
279 | } |
432 | } |
280 | |
433 | |
281 | sub server_address { |
|
|
282 | my $self = shift; |
|
|
283 | my ($port, $iaddr) = unpack_sockaddr_in $self->{fh}->getsockname |
|
|
284 | or $self->err(500, "unable to get socket name"); |
|
|
285 | ((inet_ntoa $iaddr), $port); |
|
|
286 | } |
|
|
287 | |
|
|
288 | sub server_host { |
|
|
289 | my $self = shift; |
|
|
290 | if (exists $self->{h}{host}) { |
|
|
291 | return $self->{h}{host}; |
|
|
292 | } else { |
|
|
293 | return (($self->server_address)[0]); |
|
|
294 | } |
|
|
295 | } |
|
|
296 | |
|
|
297 | sub server_hostport { |
|
|
298 | my $self = shift; |
|
|
299 | my ($host, $port); |
|
|
300 | if (exists $self->{h}{host}) { |
|
|
301 | ($host, $port) = ($self->{h}{host}, $self->{server_port}); |
|
|
302 | } else { |
|
|
303 | ($host, $port) = $self->server_address; |
|
|
304 | } |
|
|
305 | $port = $port == 80 ? "" : ":$port"; |
|
|
306 | $host.$port; |
|
|
307 | } |
|
|
308 | |
|
|
309 | sub _cgi { |
434 | sub _cgi { |
310 | my $self = shift; |
435 | my $self = shift; |
311 | my $path = shift; |
436 | my $path = shift; |
312 | my $fh; |
437 | my $fh; |
313 | |
438 | |
314 | # no two-way xxx supported |
439 | # no two-way xxx supported |
315 | if (0 == fork) { |
440 | if (0 == fork) { |
316 | open STDOUT, ">&".fileno($self->{fh}); |
441 | open STDOUT, ">&".fileno($self->{fh}); |
317 | if (chdir $::DOCROOT) { |
442 | if (chdir $::DOCROOT) { |
318 | $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike |
443 | $ENV{SERVER_SOFTWARE} = "thttpd-myhttpd"; # we are thttpd-alike |
319 | $ENV{HTTP_HOST} = $self->server_host; |
444 | $ENV{HTTP_HOST} = $self->{server_name}; |
320 | $ENV{HTTP_PORT} = $self->{server_host}; |
445 | $ENV{HTTP_PORT} = $self->{server_port}; |
321 | $ENV{SCRIPT_NAME} = $self->{name}; |
446 | $ENV{SCRIPT_NAME} = $self->{name}; |
322 | exec $path; |
447 | exec $path; |
323 | } |
448 | } |
324 | Coro::State::_exit(0); |
449 | Coro::State::_exit(0); |
325 | } else { |
450 | } else { |
|
|
451 | die; |
326 | } |
452 | } |
|
|
453 | } |
|
|
454 | |
|
|
455 | sub server_hostport { |
|
|
456 | $_[0]{server_port} == 80 |
|
|
457 | ? $_[0]{server_name} |
|
|
458 | : "$_[0]{server_name}:$_[0]{server_port}"; |
327 | } |
459 | } |
328 | |
460 | |
329 | sub respond { |
461 | sub respond { |
330 | my $self = shift; |
462 | my $self = shift; |
331 | my $path = $self->{path}; |
463 | my $path = $self->{path}; |
332 | |
464 | |
333 | stat $path |
465 | if ($self->{name} =~ s%^/internal/([^/]+)%%) { |
334 | or $self->err(404, "not found"); |
466 | if ($::internal{$1}) { |
335 | |
467 | $::internal{$1}->($self); |
336 | $self->{stat} = [stat _]; |
|
|
337 | |
|
|
338 | # idiotic netscape sends idiotic headers AGAIN |
|
|
339 | my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/ |
|
|
340 | ? str2time $1 : 0; |
|
|
341 | |
|
|
342 | if (-d _ && -r _) { |
|
|
343 | # directory |
|
|
344 | if ($path !~ /\/$/) { |
|
|
345 | # create a redirect to get the trailing "/" |
|
|
346 | my $host = $self->server_hostport; |
|
|
347 | $self->err(301, "moved permanently", { Location => "http://$host$self->{uri}/" }); |
|
|
348 | } else { |
468 | } else { |
|
|
469 | $self->err (404, "not found"); |
|
|
470 | } |
|
|
471 | } else { |
|
|
472 | |
|
|
473 | Coro::AIO::aio_stat $path |
|
|
474 | and $self->err (404, "not found"); |
|
|
475 | |
|
|
476 | $self->{stat} = [stat _]; |
|
|
477 | |
|
|
478 | # idiotic netscape sends idiotic headers AGAIN |
|
|
479 | my $ims = $self->{h}{"if-modified-since"} =~ /^([^;]+)/ |
|
|
480 | ? str2time $1 : 0; |
|
|
481 | |
|
|
482 | if (-d _ && -r _) { |
|
|
483 | # directory |
|
|
484 | if ($path !~ /\/$/) { |
|
|
485 | # create a redirect to get the trailing "/" |
|
|
486 | # we don't try to avoid the :80 |
|
|
487 | $self->err (301, "moved permanently", { Location => "http://".$self->server_hostport."$self->{uri}/" }); |
|
|
488 | } else { |
349 | $ims < $self->{stat}[9] |
489 | $ims < $self->{stat}[9] |
350 | or $self->err(304, "not modified"); |
490 | or $self->err (304, "not modified"); |
351 | |
491 | |
352 | if ($self->{method} eq "GET") { |
|
|
353 | if (-r "$path/index.html") { |
492 | if (-r "$path/index.html") { |
|
|
493 | # replace directory "size" by index.html filesize |
354 | $self->{path} .= "/index.html"; |
494 | $self->{stat} = [stat ($self->{path} .= "/index.html")]; |
355 | $self->handle_file; |
495 | $self->handle_file ($queue_index, $tbf_top); |
356 | } else { |
496 | } else { |
357 | $self->handle_dir; |
497 | $self->handle_dir; |
358 | } |
498 | } |
359 | } |
499 | } |
360 | } |
|
|
361 | } elsif (-f _ && -r _) { |
500 | } elsif (-f _ && -r _) { |
362 | -x _ and $self->err(403, "forbidden"); |
501 | -x _ and $self->err (403, "forbidden"); |
363 | $self->handle_file; |
502 | |
|
|
503 | if (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) { |
|
|
504 | my $timeout = $::NOW + 10; |
|
|
505 | while (keys %{$conn{$self->{remote_id}}} > $::MAX_TRANSFERS_IP) { |
|
|
506 | if ($timeout < $::NOW) { |
|
|
507 | $self->block($::BLOCKTIME, "too many connections"); |
|
|
508 | } else { |
|
|
509 | $httpevent->wait; |
|
|
510 | } |
|
|
511 | } |
|
|
512 | } |
|
|
513 | |
|
|
514 | $self->handle_file ($queue_file, $tbf_top); |
364 | } else { |
515 | } else { |
365 | $self->err(404, "not found"); |
516 | $self->err (404, "not found"); |
|
|
517 | } |
366 | } |
518 | } |
367 | } |
519 | } |
368 | |
520 | |
369 | sub handle_dir { |
521 | sub handle_dir { |
370 | my $self = shift; |
522 | my $self = shift; |
371 | my $idx = $self->diridx; |
523 | my $idx = $self->diridx; |
372 | |
524 | |
373 | $self->response(200, "ok", |
525 | $self->response (200, "ok", |
374 | { |
526 | { |
375 | "Content-Type" => "text/html", |
527 | "Content-Type" => "text/html; charset=utf-8", |
376 | "Content-Length" => length $idx, |
528 | "Content-Length" => length $idx, |
|
|
529 | "Last-Modified" => time2str ($self->{stat}[9]), |
377 | }, |
530 | }, |
378 | $idx); |
531 | $idx); |
379 | } |
532 | } |
380 | |
533 | |
381 | sub handle_file { |
534 | sub handle_file { |
382 | my $self = shift; |
535 | my ($self, $queue, $tbf) = @_; |
383 | my $length = -s _; |
536 | my $length = $self->{stat}[7]; |
384 | my $hdr = { |
537 | my $hdr = { |
385 | "Last-Modified" => time2str ((stat _)[9]), |
538 | "Last-Modified" => time2str ((stat _)[9]), |
|
|
539 | "Accept-Ranges" => "bytes", |
386 | }; |
540 | }; |
387 | |
541 | |
388 | my @code = (200, "ok"); |
542 | my @code = (200, "ok"); |
389 | my ($l, $h); |
543 | my ($l, $h); |
390 | |
544 | |
391 | if ($self->{h}{range} =~ /^bytes=(.*)$/) { |
545 | if ($self->{h}{range} =~ /^bytes=(.*)$/i) { |
392 | for (split /,/, $1) { |
546 | for (split /,/, $1) { |
393 | if (/^-(\d+)$/) { |
547 | if (/^-(\d+)$/) { |
394 | ($l, $h) = ($length - $1, $length - 1); |
548 | ($l, $h) = ($length - $1, $length - 1); |
395 | } elsif (/^(\d+)-(\d*)$/) { |
549 | } elsif (/^(\d+)-(\d*)$/) { |
396 | ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1); |
550 | ($l, $h) = ($1, ($2 ne "" || $2 >= $length) ? $2 : $length - 1); |
397 | } else { |
551 | } else { |
398 | ($l, $h) = (0, $length - 1); |
552 | ($l, $h) = (0, $length - 1); |
399 | goto ignore; |
553 | goto ignore; |
400 | } |
554 | } |
401 | goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h > $l; |
555 | goto satisfiable if $l >= 0 && $l < $length && $h >= 0 && $h >= $l; |
402 | } |
556 | } |
403 | $hdr->{"Content-Range"} = "bytes */$length"; |
557 | $hdr->{"Content-Range"} = "bytes */$length"; |
|
|
558 | $hdr->{"Content-Length"} = $length; |
404 | $self->err(416, "not satisfiable", $hdr); |
559 | $self->err (416, "not satisfiable", $hdr, ""); |
405 | |
560 | |
406 | satisfiable: |
561 | satisfiable: |
407 | # check for segmented downloads |
562 | # check for segmented downloads |
408 | if ($l && $::NO_SEGMENTED) { |
563 | if ($l && $::NO_SEGMENTED) { |
|
|
564 | my $timeout = $::NOW + 15; |
409 | if (%{$uri{$self->{remote_addr}}{$self->{uri}}} > 1) { |
565 | while (keys %{$uri{$self->{remote_id}}{$self->{uri}}} > 1) { |
410 | Coro::Event::do_timer(after => 15); |
566 | if ($timeout <= $::NOW) { |
411 | |
567 | $self->block ($::BLOCKTIME, "segmented downloads are forbidden"); |
412 | $self->err(400, "segmented downloads are not allowed"); |
568 | #$self->err_segmented_download; |
|
|
569 | } else { |
|
|
570 | $httpevent->wait; |
|
|
571 | } |
413 | } |
572 | } |
414 | } |
573 | } |
415 | |
574 | |
416 | $hdr->{"Content-Range"} = "bytes $l-$h/$length"; |
575 | $hdr->{"Content-Range"} = "bytes $l-$h/$length"; |
417 | @code = (206, "partial content"); |
576 | @code = (206, "partial content"); |
… | |
… | |
424 | |
583 | |
425 | $self->{path} =~ /\.([^.]+)$/; |
584 | $self->{path} =~ /\.([^.]+)$/; |
426 | $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream"; |
585 | $hdr->{"Content-Type"} = $mimetype{lc $1} || "application/octet-stream"; |
427 | $hdr->{"Content-Length"} = $length; |
586 | $hdr->{"Content-Length"} = $length; |
428 | |
587 | |
429 | $self->response(@code, $hdr, ""); |
588 | $self->response (@code, $hdr, ""); |
430 | |
589 | |
431 | if ($self->{method} eq "GET") { |
590 | if ($self->{method} eq "GET") { |
432 | my ($fh, $buf); |
591 | $self->{time} = $::NOW; |
433 | open $fh, "<", $self->{path} |
592 | $self->{written} = 0; |
|
|
593 | |
|
|
594 | my $fh = Coro::AIO::aio_open $self->{path}, Fcntl::O_RDONLY, 0 |
434 | or die "$self->{path}: late open failure ($!)"; |
595 | or die "$self->{path}: late open failure ($!)"; |
435 | |
596 | |
436 | if ($l) { |
|
|
437 | sysseek $fh, $l, 0 |
|
|
438 | or die "$self->{path}: cannot seek to $l ($!)"; |
|
|
439 | } |
|
|
440 | |
|
|
441 | $h -= $l - 1; |
597 | $h -= $l - 1; |
442 | |
598 | |
|
|
599 | my $transfer = $queue->start_transfer ($h); |
|
|
600 | my $locked; |
|
|
601 | my $bufsize = $::WAIT_BUFSIZE; # initial buffer size |
|
|
602 | |
443 | while ($h > 0) { |
603 | while ($h > 0) { |
444 | $h -= sysread $fh, $buf, $h > $::BUFSIZE ? $::BUFSIZE : $h; |
604 | Coro::cede; |
|
|
605 | my $transfer_lock = $TRANSFER_LOCK->guard ($self->{remote_id}); |
|
|
606 | |
|
|
607 | unless ($locked) { |
|
|
608 | if ($locked ||= $transfer->try ($::WAIT_INTERVAL)) { |
|
|
609 | $bufsize = $::BUFSIZE; |
|
|
610 | $self->{time} = $::NOW; |
|
|
611 | $self->{written} = 0; |
|
|
612 | } |
|
|
613 | } |
|
|
614 | |
|
|
615 | if ($blocked{$self->{remote_id}}) { |
|
|
616 | $self->{h}{connection} = "close"; |
|
|
617 | die bless {}, err::; |
|
|
618 | } |
|
|
619 | |
|
|
620 | Coro::AIO::aio_read $fh, $l, ($h > $bufsize ? $bufsize : $h), my $buf, 0 |
|
|
621 | or last; |
|
|
622 | |
|
|
623 | $tbf->request (length $buf); |
445 | my $w = $self->{fh}->syswrite($buf) |
624 | my $w = $self->{fh}->syswrite ($buf) |
446 | or last; |
625 | or last; |
447 | $::written += $w; |
626 | $::written += $w; |
448 | $self->{written} += $w; |
627 | $self->{written} += $w; |
|
|
628 | $l += $w; |
449 | } |
629 | } |
450 | } |
|
|
451 | |
630 | |
452 | close $fh; |
631 | close $fh; |
|
|
632 | } |
453 | } |
633 | } |
454 | |
634 | |
455 | 1; |
635 | 1 |
|
|
636 | |