ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent/lib/AnyEvent/Handle.pm
Revision: 1.242
Committed: Wed Dec 10 04:29:33 2014 UTC (9 years, 6 months ago) by root
Branch: MAIN
CVS Tags: rel-7_08
Changes since 1.241: +0 -3 lines
Log Message:
7.08

File Contents

# Content
1 =head1 NAME
2
3 AnyEvent::Handle - non-blocking I/O on streaming handles via AnyEvent
4
5 =head1 SYNOPSIS
6
7 use AnyEvent;
8 use AnyEvent::Handle;
9
10 my $cv = AnyEvent->condvar;
11
12 my $hdl; $hdl = new AnyEvent::Handle
13 fh => \*STDIN,
14 on_error => sub {
15 my ($hdl, $fatal, $msg) = @_;
16 AE::log error => $msg;
17 $hdl->destroy;
18 $cv->send;
19 };
20
21 # send some request line
22 $hdl->push_write ("getinfo\015\012");
23
24 # read the response line
25 $hdl->push_read (line => sub {
26 my ($hdl, $line) = @_;
27 say "got line <$line>";
28 $cv->send;
29 });
30
31 $cv->recv;
32
33 =head1 DESCRIPTION
34
35 This is a helper module to make it easier to do event-based I/O on
36 stream-based filehandles (sockets, pipes, and other stream things).
37
38 The L<AnyEvent::Intro> tutorial contains some well-documented
39 AnyEvent::Handle examples.
40
41 In the following, where the documentation refers to "bytes", it means
42 characters. As sysread and syswrite are used for all I/O, their
43 treatment of characters applies to this module as well.
44
45 At the very minimum, you should specify C<fh> or C<connect>, and the
46 C<on_error> callback.
47
48 All callbacks will be invoked with the handle object as their first
49 argument.
50
51 =cut
52
53 package AnyEvent::Handle;
54
55 use Scalar::Util ();
56 use List::Util ();
57 use Carp ();
58 use Errno qw(EAGAIN EINTR);
59
60 use AnyEvent (); BEGIN { AnyEvent::common_sense }
61 use AnyEvent::Util qw(WSAEWOULDBLOCK);
62
63 our $VERSION = $AnyEvent::VERSION;
64
65 sub _load_func($) {
66 my $func = $_[0];
67
68 unless (defined &$func) {
69 my $pkg = $func;
70 do {
71 $pkg =~ s/::[^:]+$//
72 or return;
73 eval "require $pkg";
74 } until defined &$func;
75 }
76
77 \&$func
78 }
79
80 sub MAX_READ_SIZE() { 131072 }
81
82 =head1 METHODS
83
84 =over 4
85
86 =item $handle = B<new> AnyEvent::Handle fh => $filehandle, key => value...
87
88 The constructor supports these arguments (all as C<< key => value >> pairs).
89
90 =over 4
91
92 =item fh => $filehandle [C<fh> or C<connect> MANDATORY]
93
94 The filehandle this L<AnyEvent::Handle> object will operate on.
95 NOTE: The filehandle will be set to non-blocking mode (using
96 C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
97 that mode.
98
99 =item connect => [$host, $service] [C<fh> or C<connect> MANDATORY]
100
101 Try to connect to the specified host and service (port), using
102 C<AnyEvent::Socket::tcp_connect>. The C<$host> additionally becomes the
103 default C<peername>.
104
105 You have to specify either this parameter, or C<fh>, above.
106
107 It is possible to push requests on the read and write queues, and modify
108 properties of the stream, even while AnyEvent::Handle is connecting.
109
110 When this parameter is specified, then the C<on_prepare>,
111 C<on_connect_error> and C<on_connect> callbacks will be called under the
112 appropriate circumstances:
113
114 =over 4
115
116 =item on_prepare => $cb->($handle)
117
118 This (rarely used) callback is called before a new connection is
119 attempted, but after the file handle has been created (you can access that
120 file handle via C<< $handle->{fh} >>). It could be used to prepare the
121 file handle with parameters required for the actual connect (as opposed to
122 settings that can be changed when the connection is already established).
123
124 The return value of this callback should be the connect timeout value in
125 seconds (or C<0>, or C<undef>, or the empty list, to indicate that the
126 default timeout is to be used).
127
128 =item on_connect => $cb->($handle, $host, $port, $retry->())
129
130 This callback is called when a connection has been successfully established.
131
132 The peer's numeric host and port (the socket peername) are passed as
133 parameters, together with a retry callback. At the time it is called the
134 read and write queues, EOF status, TLS status and similar properties of
135 the handle will have been reset.
136
137 If, for some reason, the handle is not acceptable, calling C<$retry> will
138 continue with the next connection target (in case of multi-homed hosts or
139 SRV records there can be multiple connection endpoints). The C<$retry>
140 callback can be invoked after the connect callback returns, i.e. one can
141 start a handshake and then decide to retry with the next host if the
142 handshake fails.
143
144 In most cases, you should ignore the C<$retry> parameter.
145
146 =item on_connect_error => $cb->($handle, $message)
147
148 This callback is called when the connection could not be
149 established. C<$!> will contain the relevant error code, and C<$message> a
150 message describing it (usually the same as C<"$!">).
151
152 If this callback isn't specified, then C<on_error> will be called with a
153 fatal error instead.
154
155 =back
156
157 =item on_error => $cb->($handle, $fatal, $message)
158
159 This is the error callback, which is called when, well, some error
160 occured, such as not being able to resolve the hostname, failure to
161 connect, or a read error.
162
163 Some errors are fatal (which is indicated by C<$fatal> being true). On
164 fatal errors the handle object will be destroyed (by a call to C<< ->
165 destroy >>) after invoking the error callback (which means you are free to
166 examine the handle object). Examples of fatal errors are an EOF condition
167 with active (but unsatisfiable) read watchers (C<EPIPE>) or I/O errors. In
168 cases where the other side can close the connection at will, it is
169 often easiest to not report C<EPIPE> errors in this callback.
170
171 AnyEvent::Handle tries to find an appropriate error code for you to check
172 against, but in some cases (TLS errors), this does not work well.
173
174 If you report the error to the user, it is recommended to always output
175 the C<$message> argument in human-readable error messages (you don't need
176 to report C<"$!"> if you report C<$message>).
177
178 If you want to react programmatically to the error, then looking at C<$!>
179 and comparing it against some of the documented C<Errno> values is usually
180 better than looking at the C<$message>.
181
182 Non-fatal errors can be retried by returning, but it is recommended
183 to simply ignore this parameter and instead abondon the handle object
184 when this callback is invoked. Examples of non-fatal errors are timeouts
185 C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
186
187 On entry to the callback, the value of C<$!> contains the operating
188 system error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
189 C<EPROTO>).
190
191 While not mandatory, it is I<highly> recommended to set this callback, as
192 you will not be notified of errors otherwise. The default just calls
193 C<croak>.
194
195 =item on_read => $cb->($handle)
196
197 This sets the default read callback, which is called when data arrives
198 and no read request is in the queue (unlike read queue callbacks, this
199 callback will only be called when at least one octet of data is in the
200 read buffer).
201
202 To access (and remove data from) the read buffer, use the C<< ->rbuf >>
203 method or access the C<< $handle->{rbuf} >> member directly. Note that you
204 must not enlarge or modify the read buffer, you can only remove data at
205 the beginning from it.
206
207 You can also call C<< ->push_read (...) >> or any other function that
208 modifies the read queue. Or do both. Or ...
209
210 When an EOF condition is detected, AnyEvent::Handle will first try to
211 feed all the remaining data to the queued callbacks and C<on_read> before
212 calling the C<on_eof> callback. If no progress can be made, then a fatal
213 error will be raised (with C<$!> set to C<EPIPE>).
214
215 Note that, unlike requests in the read queue, an C<on_read> callback
216 doesn't mean you I<require> some data: if there is an EOF and there
217 are outstanding read requests then an error will be flagged. With an
218 C<on_read> callback, the C<on_eof> callback will be invoked.
219
220 =item on_eof => $cb->($handle)
221
222 Set the callback to be called when an end-of-file condition is detected,
223 i.e. in the case of a socket, when the other side has closed the
224 connection cleanly, and there are no outstanding read requests in the
225 queue (if there are read requests, then an EOF counts as an unexpected
226 connection close and will be flagged as an error).
227
228 For sockets, this just means that the other side has stopped sending data,
229 you can still try to write data, and, in fact, one can return from the EOF
230 callback and continue writing data, as only the read part has been shut
231 down.
232
233 If an EOF condition has been detected but no C<on_eof> callback has been
234 set, then a fatal error will be raised with C<$!> set to <0>.
235
236 =item on_drain => $cb->($handle)
237
238 This sets the callback that is called once when the write buffer becomes
239 empty (and immediately when the handle object is created).
240
241 To append to the write buffer, use the C<< ->push_write >> method.
242
243 This callback is useful when you don't want to put all of your write data
244 into the queue at once, for example, when you want to write the contents
245 of some file to the socket you might not want to read the whole file into
246 memory and push it into the queue, but instead only read more data from
247 the file when the write queue becomes empty.
248
249 =item timeout => $fractional_seconds
250
251 =item rtimeout => $fractional_seconds
252
253 =item wtimeout => $fractional_seconds
254
255 If non-zero, then these enables an "inactivity" timeout: whenever this
256 many seconds pass without a successful read or write on the underlying
257 file handle (or a call to C<timeout_reset>), the C<on_timeout> callback
258 will be invoked (and if that one is missing, a non-fatal C<ETIMEDOUT>
259 error will be raised).
260
261 There are three variants of the timeouts that work independently of each
262 other, for both read and write (triggered when nothing was read I<OR>
263 written), just read (triggered when nothing was read), and just write:
264 C<timeout>, C<rtimeout> and C<wtimeout>, with corresponding callbacks
265 C<on_timeout>, C<on_rtimeout> and C<on_wtimeout>, and reset functions
266 C<timeout_reset>, C<rtimeout_reset>, and C<wtimeout_reset>.
267
268 Note that timeout processing is active even when you do not have any
269 outstanding read or write requests: If you plan to keep the connection
270 idle then you should disable the timeout temporarily or ignore the
271 timeout in the corresponding C<on_timeout> callback, in which case
272 AnyEvent::Handle will simply restart the timeout.
273
274 Zero (the default) disables the corresponding timeout.
275
276 =item on_timeout => $cb->($handle)
277
278 =item on_rtimeout => $cb->($handle)
279
280 =item on_wtimeout => $cb->($handle)
281
282 Called whenever the inactivity timeout passes. If you return from this
283 callback, then the timeout will be reset as if some activity had happened,
284 so this condition is not fatal in any way.
285
286 =item rbuf_max => <bytes>
287
288 If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
289 when the read buffer ever (strictly) exceeds this size. This is useful to
290 avoid some forms of denial-of-service attacks.
291
292 For example, a server accepting connections from untrusted sources should
293 be configured to accept only so-and-so much data that it cannot act on
294 (for example, when expecting a line, an attacker could send an unlimited
295 amount of data without a callback ever being called as long as the line
296 isn't finished).
297
298 =item wbuf_max => <bytes>
299
300 If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
301 when the write buffer ever (strictly) exceeds this size. This is useful to
302 avoid some forms of denial-of-service attacks.
303
304 Although the units of this parameter is bytes, this is the I<raw> number
305 of bytes not yet accepted by the kernel. This can make a difference when
306 you e.g. use TLS, as TLS typically makes your write data larger (but it
307 can also make it smaller due to compression).
308
309 As an example of when this limit is useful, take a chat server that sends
310 chat messages to a client. If the client does not read those in a timely
311 manner then the send buffer in the server would grow unbounded.
312
313 =item autocork => <boolean>
314
315 When disabled (the default), C<push_write> will try to immediately
316 write the data to the handle if possible. This avoids having to register
317 a write watcher and wait for the next event loop iteration, but can
318 be inefficient if you write multiple small chunks (on the wire, this
319 disadvantage is usually avoided by your kernel's nagle algorithm, see
320 C<no_delay>, but this option can save costly syscalls).
321
322 When enabled, writes will always be queued till the next event loop
323 iteration. This is efficient when you do many small writes per iteration,
324 but less efficient when you do a single write only per iteration (or when
325 the write buffer often is full). It also increases write latency.
326
327 =item no_delay => <boolean>
328
329 When doing small writes on sockets, your operating system kernel might
330 wait a bit for more data before actually sending it out. This is called
331 the Nagle algorithm, and usually it is beneficial.
332
333 In some situations you want as low a delay as possible, which can be
334 accomplishd by setting this option to a true value.
335
336 The default is your operating system's default behaviour (most likely
337 enabled). This option explicitly enables or disables it, if possible.
338
339 =item keepalive => <boolean>
340
341 Enables (default disable) the SO_KEEPALIVE option on the stream socket:
342 normally, TCP connections have no time-out once established, so TCP
343 connections, once established, can stay alive forever even when the other
344 side has long gone. TCP keepalives are a cheap way to take down long-lived
345 TCP connections when the other side becomes unreachable. While the default
346 is OS-dependent, TCP keepalives usually kick in after around two hours,
347 and, if the other side doesn't reply, take down the TCP connection some 10
348 to 15 minutes later.
349
350 It is harmless to specify this option for file handles that do not support
351 keepalives, and enabling it on connections that are potentially long-lived
352 is usually a good idea.
353
354 =item oobinline => <boolean>
355
356 BSD majorly fucked up the implementation of TCP urgent data. The result
357 is that almost no OS implements TCP according to the specs, and every OS
358 implements it slightly differently.
359
360 If you want to handle TCP urgent data, then setting this flag (the default
361 is enabled) gives you the most portable way of getting urgent data, by
362 putting it into the stream.
363
364 Since BSD emulation of OOB data on top of TCP's urgent data can have
365 security implications, AnyEvent::Handle sets this flag automatically
366 unless explicitly specified. Note that setting this flag after
367 establishing a connection I<may> be a bit too late (data loss could
368 already have occured on BSD systems), but at least it will protect you
369 from most attacks.
370
371 =item read_size => <bytes>
372
373 The initial read block size, the number of bytes this module will try
374 to read during each loop iteration. Each handle object will consume
375 at least this amount of memory for the read buffer as well, so when
376 handling many connections watch out for memory requirements). See also
377 C<max_read_size>. Default: C<2048>.
378
379 =item max_read_size => <bytes>
380
381 The maximum read buffer size used by the dynamic adjustment
382 algorithm: Each time AnyEvent::Handle can read C<read_size> bytes in
383 one go it will double C<read_size> up to the maximum given by this
384 option. Default: C<131072> or C<read_size>, whichever is higher.
385
386 =item low_water_mark => <bytes>
387
388 Sets the number of bytes (default: C<0>) that make up an "empty" write
389 buffer: If the buffer reaches this size or gets even samller it is
390 considered empty.
391
392 Sometimes it can be beneficial (for performance reasons) to add data to
393 the write buffer before it is fully drained, but this is a rare case, as
394 the operating system kernel usually buffers data as well, so the default
395 is good in almost all cases.
396
397 =item linger => <seconds>
398
399 If this is non-zero (default: C<3600>), the destructor of the
400 AnyEvent::Handle object will check whether there is still outstanding
401 write data and will install a watcher that will write this data to the
402 socket. No errors will be reported (this mostly matches how the operating
403 system treats outstanding data at socket close time).
404
405 This will not work for partial TLS data that could not be encoded
406 yet. This data will be lost. Calling the C<stoptls> method in time might
407 help.
408
409 =item peername => $string
410
411 A string used to identify the remote site - usually the DNS hostname
412 (I<not> IDN!) used to create the connection, rarely the IP address.
413
414 Apart from being useful in error messages, this string is also used in TLS
415 peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
416 verification will be skipped when C<peername> is not specified or is
417 C<undef>.
418
419 =item tls => "accept" | "connect" | Net::SSLeay::SSL object
420
421 When this parameter is given, it enables TLS (SSL) mode, that means
422 AnyEvent will start a TLS handshake as soon as the connection has been
423 established and will transparently encrypt/decrypt data afterwards.
424
425 All TLS protocol errors will be signalled as C<EPROTO>, with an
426 appropriate error message.
427
428 TLS mode requires Net::SSLeay to be installed (it will be loaded
429 automatically when you try to create a TLS handle): this module doesn't
430 have a dependency on that module, so if your module requires it, you have
431 to add the dependency yourself. If Net::SSLeay cannot be loaded or is too
432 old, you get an C<EPROTO> error.
433
434 Unlike TCP, TLS has a server and client side: for the TLS server side, use
435 C<accept>, and for the TLS client side of a connection, use C<connect>
436 mode.
437
438 You can also provide your own TLS connection object, but you have
439 to make sure that you call either C<Net::SSLeay::set_connect_state>
440 or C<Net::SSLeay::set_accept_state> on it before you pass it to
441 AnyEvent::Handle. Also, this module will take ownership of this connection
442 object.
443
444 At some future point, AnyEvent::Handle might switch to another TLS
445 implementation, then the option to use your own session object will go
446 away.
447
448 B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
449 passing in the wrong integer will lead to certain crash. This most often
450 happens when one uses a stylish C<< tls => 1 >> and is surprised about the
451 segmentation fault.
452
453 Use the C<< ->starttls >> method if you need to start TLS negotiation later.
454
455 =item tls_ctx => $anyevent_tls
456
457 Use the given C<AnyEvent::TLS> object to create the new TLS connection
458 (unless a connection object was specified directly). If this
459 parameter is missing (or C<undef>), then AnyEvent::Handle will use
460 C<AnyEvent::Handle::TLS_CTX>.
461
462 Instead of an object, you can also specify a hash reference with C<< key
463 => value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
464 new TLS context object.
465
466 =item on_starttls => $cb->($handle, $success[, $error_message])
467
468 This callback will be invoked when the TLS/SSL handshake has finished. If
469 C<$success> is true, then the TLS handshake succeeded, otherwise it failed
470 (C<on_stoptls> will not be called in this case).
471
472 The session in C<< $handle->{tls} >> can still be examined in this
473 callback, even when the handshake was not successful.
474
475 TLS handshake failures will not cause C<on_error> to be invoked when this
476 callback is in effect, instead, the error message will be passed to C<on_starttls>.
477
478 Without this callback, handshake failures lead to C<on_error> being
479 called as usual.
480
481 Note that you cannot just call C<starttls> again in this callback. If you
482 need to do that, start an zero-second timer instead whose callback can
483 then call C<< ->starttls >> again.
484
485 =item on_stoptls => $cb->($handle)
486
487 When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
488 set, then it will be invoked after freeing the TLS session. If it is not,
489 then a TLS shutdown condition will be treated like a normal EOF condition
490 on the handle.
491
492 The session in C<< $handle->{tls} >> can still be examined in this
493 callback.
494
495 This callback will only be called on TLS shutdowns, not when the
496 underlying handle signals EOF.
497
498 =item json => L<JSON>, L<JSON::PP> or L<JSON::XS> object
499
500 This is the json coder object used by the C<json> read and write types.
501
502 If you don't supply it, then AnyEvent::Handle will create and use a
503 suitable one (on demand), which will write and expect UTF-8 encoded
504 JSON texts (either using L<JSON::XS> or L<JSON>). The written texts are
505 guaranteed not to contain any newline character.
506
507 For security reasons, this encoder will likely I<not> handle numbers and
508 strings, only arrays and objects/hashes. The reason is that originally
509 JSON was self-delimited, but Dougles Crockford thought it was a splendid
510 idea to redefine JSON incompatibly, so this is no longer true.
511
512 For protocols that used back-to-back JSON texts, this might lead to
513 run-ins, where two or more JSON texts will be interpreted as one JSON
514 text.
515
516 For this reason, if the default encoder uses L<JSON::XS>, it will default
517 to not allowing anything but arrays and objects/hashes, at least for the
518 forseeable future (it will change at some point). This might or might not
519 be true for the L<JSON> module, so this might cause a security issue.
520
521 If you depend on either behaviour, you should create your own json object
522 and pass it in explicitly.
523
524 =item cbor => L<CBOR::XS> object
525
526 This is the cbor coder object used by the C<cbor> read and write types.
527
528 If you don't supply it, then AnyEvent::Handle will create and use a
529 suitable one (on demand), which will write CBOR without using extensions,
530 if possible.
531
532 Note that you are responsible to depend on the L<CBOR::XS> module if you
533 want to use this functionality, as AnyEvent does not have a dependency on
534 it itself.
535
536 =back
537
538 =cut
539
540 sub new {
541 my $class = shift;
542 my $self = bless { @_ }, $class;
543
544 if ($self->{fh}) {
545 $self->_start;
546 return unless $self->{fh}; # could be gone by now
547
548 } elsif ($self->{connect}) {
549 require AnyEvent::Socket;
550
551 $self->{peername} = $self->{connect}[0]
552 unless exists $self->{peername};
553
554 $self->{_skip_drain_rbuf} = 1;
555
556 {
557 Scalar::Util::weaken (my $self = $self);
558
559 $self->{_connect} =
560 AnyEvent::Socket::tcp_connect (
561 $self->{connect}[0],
562 $self->{connect}[1],
563 sub {
564 my ($fh, $host, $port, $retry) = @_;
565
566 delete $self->{_connect}; # no longer needed
567
568 if ($fh) {
569 $self->{fh} = $fh;
570
571 delete $self->{_skip_drain_rbuf};
572 $self->_start;
573
574 $self->{on_connect}
575 and $self->{on_connect}($self, $host, $port, sub {
576 delete @$self{qw(fh _tw _rtw _wtw _ww _rw _eof _queue rbuf _wbuf tls _tls_rbuf _tls_wbuf)};
577 $self->{_skip_drain_rbuf} = 1;
578 &$retry;
579 });
580
581 } else {
582 if ($self->{on_connect_error}) {
583 $self->{on_connect_error}($self, "$!");
584 $self->destroy if $self;
585 } else {
586 $self->_error ($!, 1);
587 }
588 }
589 },
590 sub {
591 local $self->{fh} = $_[0];
592
593 $self->{on_prepare}
594 ? $self->{on_prepare}->($self)
595 : ()
596 }
597 );
598 }
599
600 } else {
601 Carp::croak "AnyEvent::Handle: either an existing fh or the connect parameter must be specified";
602 }
603
604 $self
605 }
606
607 sub _start {
608 my ($self) = @_;
609
610 # too many clueless people try to use udp and similar sockets
611 # with AnyEvent::Handle, do them a favour.
612 my $type = getsockopt $self->{fh}, Socket::SOL_SOCKET (), Socket::SO_TYPE ();
613 Carp::croak "AnyEvent::Handle: only stream sockets supported, anything else will NOT work!"
614 if Socket::SOCK_STREAM () != (unpack "I", $type) && defined $type;
615
616 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
617
618 $self->{_activity} =
619 $self->{_ractivity} =
620 $self->{_wactivity} = AE::now;
621
622 $self->{read_size} ||= 2048;
623 $self->{max_read_size} = $self->{read_size}
624 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
625
626 $self->timeout (delete $self->{timeout} ) if $self->{timeout};
627 $self->rtimeout (delete $self->{rtimeout} ) if $self->{rtimeout};
628 $self->wtimeout (delete $self->{wtimeout} ) if $self->{wtimeout};
629
630 $self->no_delay (delete $self->{no_delay} ) if exists $self->{no_delay} && $self->{no_delay};
631 $self->keepalive (delete $self->{keepalive}) if exists $self->{keepalive} && $self->{keepalive};
632
633 $self->oobinline (exists $self->{oobinline} ? delete $self->{oobinline} : 1);
634
635 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
636 if $self->{tls};
637
638 $self->on_drain (delete $self->{on_drain} ) if $self->{on_drain};
639
640 $self->start_read
641 if $self->{on_read} || @{ $self->{_queue} };
642
643 $self->_drain_wbuf;
644 }
645
646 sub _error {
647 my ($self, $errno, $fatal, $message) = @_;
648
649 $! = $errno;
650 $message ||= "$!";
651
652 if ($self->{on_error}) {
653 $self->{on_error}($self, $fatal, $message);
654 $self->destroy if $fatal;
655 } elsif ($self->{fh} || $self->{connect}) {
656 $self->destroy;
657 Carp::croak "AnyEvent::Handle uncaught error: $message";
658 }
659 }
660
661 =item $fh = $handle->fh
662
663 This method returns the file handle used to create the L<AnyEvent::Handle> object.
664
665 =cut
666
667 sub fh { $_[0]{fh} }
668
669 =item $handle->on_error ($cb)
670
671 Replace the current C<on_error> callback (see the C<on_error> constructor argument).
672
673 =cut
674
675 sub on_error {
676 $_[0]{on_error} = $_[1];
677 }
678
679 =item $handle->on_eof ($cb)
680
681 Replace the current C<on_eof> callback (see the C<on_eof> constructor argument).
682
683 =cut
684
685 sub on_eof {
686 $_[0]{on_eof} = $_[1];
687 }
688
689 =item $handle->on_timeout ($cb)
690
691 =item $handle->on_rtimeout ($cb)
692
693 =item $handle->on_wtimeout ($cb)
694
695 Replace the current C<on_timeout>, C<on_rtimeout> or C<on_wtimeout>
696 callback, or disables the callback (but not the timeout) if C<$cb> =
697 C<undef>. See the C<timeout> constructor argument and method.
698
699 =cut
700
701 # see below
702
703 =item $handle->autocork ($boolean)
704
705 Enables or disables the current autocork behaviour (see C<autocork>
706 constructor argument). Changes will only take effect on the next write.
707
708 =cut
709
710 sub autocork {
711 $_[0]{autocork} = $_[1];
712 }
713
714 =item $handle->no_delay ($boolean)
715
716 Enables or disables the C<no_delay> setting (see constructor argument of
717 the same name for details).
718
719 =cut
720
721 sub no_delay {
722 $_[0]{no_delay} = $_[1];
723
724 setsockopt $_[0]{fh}, Socket::IPPROTO_TCP (), Socket::TCP_NODELAY (), int $_[1]
725 if $_[0]{fh};
726 }
727
728 =item $handle->keepalive ($boolean)
729
730 Enables or disables the C<keepalive> setting (see constructor argument of
731 the same name for details).
732
733 =cut
734
735 sub keepalive {
736 $_[0]{keepalive} = $_[1];
737
738 eval {
739 local $SIG{__DIE__};
740 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
741 if $_[0]{fh};
742 };
743 }
744
745 =item $handle->oobinline ($boolean)
746
747 Enables or disables the C<oobinline> setting (see constructor argument of
748 the same name for details).
749
750 =cut
751
752 sub oobinline {
753 $_[0]{oobinline} = $_[1];
754
755 eval {
756 local $SIG{__DIE__};
757 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_OOBINLINE (), int $_[1]
758 if $_[0]{fh};
759 };
760 }
761
762 =item $handle->keepalive ($boolean)
763
764 Enables or disables the C<keepalive> setting (see constructor argument of
765 the same name for details).
766
767 =cut
768
769 sub keepalive {
770 $_[0]{keepalive} = $_[1];
771
772 eval {
773 local $SIG{__DIE__};
774 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
775 if $_[0]{fh};
776 };
777 }
778
779 =item $handle->on_starttls ($cb)
780
781 Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
782
783 =cut
784
785 sub on_starttls {
786 $_[0]{on_starttls} = $_[1];
787 }
788
789 =item $handle->on_stoptls ($cb)
790
791 Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
792
793 =cut
794
795 sub on_stoptls {
796 $_[0]{on_stoptls} = $_[1];
797 }
798
799 =item $handle->rbuf_max ($max_octets)
800
801 Configures the C<rbuf_max> setting (C<undef> disables it).
802
803 =item $handle->wbuf_max ($max_octets)
804
805 Configures the C<wbuf_max> setting (C<undef> disables it).
806
807 =cut
808
809 sub rbuf_max {
810 $_[0]{rbuf_max} = $_[1];
811 }
812
813 sub wbuf_max {
814 $_[0]{wbuf_max} = $_[1];
815 }
816
817 #############################################################################
818
819 =item $handle->timeout ($seconds)
820
821 =item $handle->rtimeout ($seconds)
822
823 =item $handle->wtimeout ($seconds)
824
825 Configures (or disables) the inactivity timeout.
826
827 The timeout will be checked instantly, so this method might destroy the
828 handle before it returns.
829
830 =item $handle->timeout_reset
831
832 =item $handle->rtimeout_reset
833
834 =item $handle->wtimeout_reset
835
836 Reset the activity timeout, as if data was received or sent.
837
838 These methods are cheap to call.
839
840 =cut
841
842 for my $dir ("", "r", "w") {
843 my $timeout = "${dir}timeout";
844 my $tw = "_${dir}tw";
845 my $on_timeout = "on_${dir}timeout";
846 my $activity = "_${dir}activity";
847 my $cb;
848
849 *$on_timeout = sub {
850 $_[0]{$on_timeout} = $_[1];
851 };
852
853 *$timeout = sub {
854 my ($self, $new_value) = @_;
855
856 $new_value >= 0
857 or Carp::croak "AnyEvent::Handle->$timeout called with negative timeout ($new_value), caught";
858
859 $self->{$timeout} = $new_value;
860 delete $self->{$tw}; &$cb;
861 };
862
863 *{"${dir}timeout_reset"} = sub {
864 $_[0]{$activity} = AE::now;
865 };
866
867 # main workhorse:
868 # reset the timeout watcher, as neccessary
869 # also check for time-outs
870 $cb = sub {
871 my ($self) = @_;
872
873 if ($self->{$timeout} && $self->{fh}) {
874 my $NOW = AE::now;
875
876 # when would the timeout trigger?
877 my $after = $self->{$activity} + $self->{$timeout} - $NOW;
878
879 # now or in the past already?
880 if ($after <= 0) {
881 $self->{$activity} = $NOW;
882
883 if ($self->{$on_timeout}) {
884 $self->{$on_timeout}($self);
885 } else {
886 $self->_error (Errno::ETIMEDOUT);
887 }
888
889 # callback could have changed timeout value, optimise
890 return unless $self->{$timeout};
891
892 # calculate new after
893 $after = $self->{$timeout};
894 }
895
896 Scalar::Util::weaken $self;
897 return unless $self; # ->error could have destroyed $self
898
899 $self->{$tw} ||= AE::timer $after, 0, sub {
900 delete $self->{$tw};
901 $cb->($self);
902 };
903 } else {
904 delete $self->{$tw};
905 }
906 }
907 }
908
909 #############################################################################
910
911 =back
912
913 =head2 WRITE QUEUE
914
915 AnyEvent::Handle manages two queues per handle, one for writing and one
916 for reading.
917
918 The write queue is very simple: you can add data to its end, and
919 AnyEvent::Handle will automatically try to get rid of it for you.
920
921 When data could be written and the write buffer is shorter then the low
922 water mark, the C<on_drain> callback will be invoked once.
923
924 =over 4
925
926 =item $handle->on_drain ($cb)
927
928 Sets the C<on_drain> callback or clears it (see the description of
929 C<on_drain> in the constructor).
930
931 This method may invoke callbacks (and therefore the handle might be
932 destroyed after it returns).
933
934 =cut
935
936 sub on_drain {
937 my ($self, $cb) = @_;
938
939 $self->{on_drain} = $cb;
940
941 $cb->($self)
942 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
943 }
944
945 =item $handle->push_write ($data)
946
947 Queues the given scalar to be written. You can push as much data as
948 you want (only limited by the available memory and C<wbuf_max>), as
949 C<AnyEvent::Handle> buffers it independently of the kernel.
950
951 This method may invoke callbacks (and therefore the handle might be
952 destroyed after it returns).
953
954 =cut
955
956 sub _drain_wbuf {
957 my ($self) = @_;
958
959 if (!$self->{_ww} && length $self->{wbuf}) {
960
961 Scalar::Util::weaken $self;
962
963 my $cb = sub {
964 my $len = syswrite $self->{fh}, $self->{wbuf};
965
966 if (defined $len) {
967 substr $self->{wbuf}, 0, $len, "";
968
969 $self->{_activity} = $self->{_wactivity} = AE::now;
970
971 $self->{on_drain}($self)
972 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
973 && $self->{on_drain};
974
975 delete $self->{_ww} unless length $self->{wbuf};
976 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
977 $self->_error ($!, 1);
978 }
979 };
980
981 # try to write data immediately
982 $cb->() unless $self->{autocork};
983
984 # if still data left in wbuf, we need to poll
985 $self->{_ww} = AE::io $self->{fh}, 1, $cb
986 if length $self->{wbuf};
987
988 if (
989 defined $self->{wbuf_max}
990 && $self->{wbuf_max} < length $self->{wbuf}
991 ) {
992 $self->_error (Errno::ENOSPC, 1), return;
993 }
994 };
995 }
996
997 our %WH;
998
999 # deprecated
1000 sub register_write_type($$) {
1001 $WH{$_[0]} = $_[1];
1002 }
1003
1004 sub push_write {
1005 my $self = shift;
1006
1007 if (@_ > 1) {
1008 my $type = shift;
1009
1010 @_ = ($WH{$type} ||= _load_func "$type\::anyevent_write_type"
1011 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_write")
1012 ->($self, @_);
1013 }
1014
1015 # we downgrade here to avoid hard-to-track-down bugs,
1016 # and diagnose the problem earlier and better.
1017
1018 if ($self->{tls}) {
1019 utf8::downgrade $self->{_tls_wbuf} .= $_[0];
1020 &_dotls ($self) if $self->{fh};
1021 } else {
1022 utf8::downgrade $self->{wbuf} .= $_[0];
1023 $self->_drain_wbuf if $self->{fh};
1024 }
1025 }
1026
1027 =item $handle->push_write (type => @args)
1028
1029 Instead of formatting your data yourself, you can also let this module
1030 do the job by specifying a type and type-specific arguments. You
1031 can also specify the (fully qualified) name of a package, in which
1032 case AnyEvent tries to load the package and then expects to find the
1033 C<anyevent_write_type> function inside (see "custom write types", below).
1034
1035 Predefined types are (if you have ideas for additional types, feel free to
1036 drop by and tell us):
1037
1038 =over 4
1039
1040 =item netstring => $string
1041
1042 Formats the given value as netstring
1043 (http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
1044
1045 =cut
1046
1047 register_write_type netstring => sub {
1048 my ($self, $string) = @_;
1049
1050 (length $string) . ":$string,"
1051 };
1052
1053 =item packstring => $format, $data
1054
1055 An octet string prefixed with an encoded length. The encoding C<$format>
1056 uses the same format as a Perl C<pack> format, but must specify a single
1057 integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1058 optional C<!>, C<< < >> or C<< > >> modifier).
1059
1060 =cut
1061
1062 register_write_type packstring => sub {
1063 my ($self, $format, $string) = @_;
1064
1065 pack "$format/a*", $string
1066 };
1067
1068 =item json => $array_or_hashref
1069
1070 Encodes the given hash or array reference into a JSON object. Unless you
1071 provide your own JSON object, this means it will be encoded to JSON text
1072 in UTF-8.
1073
1074 The default encoder might or might not handle every type of JSON value -
1075 it might be limited to arrays and objects for security reasons. See the
1076 C<json> constructor attribute for more details.
1077
1078 JSON objects (and arrays) are self-delimiting, so if you only use arrays
1079 and hashes, you can write JSON at one end of a handle and read them at the
1080 other end without using any additional framing.
1081
1082 The JSON text generated by the default encoder is guaranteed not to
1083 contain any newlines: While this module doesn't need delimiters after or
1084 between JSON texts to be able to read them, many other languages depend on
1085 them.
1086
1087 A simple RPC protocol that interoperates easily with other languages is
1088 to send JSON arrays (or objects, although arrays are usually the better
1089 choice as they mimic how function argument passing works) and a newline
1090 after each JSON text:
1091
1092 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
1093 $handle->push_write ("\012");
1094
1095 An AnyEvent::Handle receiver would simply use the C<json> read type and
1096 rely on the fact that the newline will be skipped as leading whitespace:
1097
1098 $handle->push_read (json => sub { my $array = $_[1]; ... });
1099
1100 Other languages could read single lines terminated by a newline and pass
1101 this line into their JSON decoder of choice.
1102
1103 =item cbor => $perl_scalar
1104
1105 Encodes the given scalar into a CBOR value. Unless you provide your own
1106 L<CBOR::XS> object, this means it will be encoded to a CBOR string not
1107 using any extensions, if possible.
1108
1109 CBOR values are self-delimiting, so you can write CBOR at one end of
1110 a handle and read them at the other end without using any additional
1111 framing.
1112
1113 A simple nd very very fast RPC protocol that interoperates with
1114 other languages is to send CBOR and receive CBOR values (arrays are
1115 recommended):
1116
1117 $handle->push_write (cbor => ["method", "arg1", "arg2"]); # whatever
1118
1119 An AnyEvent::Handle receiver would simply use the C<cbor> read type:
1120
1121 $handle->push_read (cbor => sub { my $array = $_[1]; ... });
1122
1123 =cut
1124
1125 sub json_coder() {
1126 eval { require JSON::XS; JSON::XS->new->utf8 }
1127 || do { require JSON::PP; JSON::PP->new->utf8 }
1128 }
1129
1130 register_write_type json => sub {
1131 my ($self, $ref) = @_;
1132
1133 ($self->{json} ||= json_coder)
1134 ->encode ($ref)
1135 };
1136
1137 sub cbor_coder() {
1138 require CBOR::XS;
1139 CBOR::XS->new
1140 }
1141
1142 register_write_type cbor => sub {
1143 my ($self, $scalar) = @_;
1144
1145 ($self->{cbor} ||= cbor_coder)
1146 ->encode ($scalar)
1147 };
1148
1149 =item storable => $reference
1150
1151 Freezes the given reference using L<Storable> and writes it to the
1152 handle. Uses the C<nfreeze> format.
1153
1154 =cut
1155
1156 register_write_type storable => sub {
1157 my ($self, $ref) = @_;
1158
1159 require Storable unless $Storable::VERSION;
1160
1161 pack "w/a*", Storable::nfreeze ($ref)
1162 };
1163
1164 =back
1165
1166 =item $handle->push_shutdown
1167
1168 Sometimes you know you want to close the socket after writing your data
1169 before it was actually written. One way to do that is to replace your
1170 C<on_drain> handler by a callback that shuts down the socket (and set
1171 C<low_water_mark> to C<0>). This method is a shorthand for just that, and
1172 replaces the C<on_drain> callback with:
1173
1174 sub { shutdown $_[0]{fh}, 1 }
1175
1176 This simply shuts down the write side and signals an EOF condition to the
1177 the peer.
1178
1179 You can rely on the normal read queue and C<on_eof> handling
1180 afterwards. This is the cleanest way to close a connection.
1181
1182 This method may invoke callbacks (and therefore the handle might be
1183 destroyed after it returns).
1184
1185 =cut
1186
1187 sub push_shutdown {
1188 my ($self) = @_;
1189
1190 delete $self->{low_water_mark};
1191 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
1192 }
1193
1194 =item custom write types - Package::anyevent_write_type $handle, @args
1195
1196 Instead of one of the predefined types, you can also specify the name of
1197 a package. AnyEvent will try to load the package and then expects to find
1198 a function named C<anyevent_write_type> inside. If it isn't found, it
1199 progressively tries to load the parent package until it either finds the
1200 function (good) or runs out of packages (bad).
1201
1202 Whenever the given C<type> is used, C<push_write> will the function with
1203 the handle object and the remaining arguments.
1204
1205 The function is supposed to return a single octet string that will be
1206 appended to the write buffer, so you can mentally treat this function as a
1207 "arguments to on-the-wire-format" converter.
1208
1209 Example: implement a custom write type C<join> that joins the remaining
1210 arguments using the first one.
1211
1212 $handle->push_write (My::Type => " ", 1,2,3);
1213
1214 # uses the following package, which can be defined in the "My::Type" or in
1215 # the "My" modules to be auto-loaded, or just about anywhere when the
1216 # My::Type::anyevent_write_type is defined before invoking it.
1217
1218 package My::Type;
1219
1220 sub anyevent_write_type {
1221 my ($handle, $delim, @args) = @_;
1222
1223 join $delim, @args
1224 }
1225
1226 =cut
1227
1228 #############################################################################
1229
1230 =back
1231
1232 =head2 READ QUEUE
1233
1234 AnyEvent::Handle manages two queues per handle, one for writing and one
1235 for reading.
1236
1237 The read queue is more complex than the write queue. It can be used in two
1238 ways, the "simple" way, using only C<on_read> and the "complex" way, using
1239 a queue.
1240
1241 In the simple case, you just install an C<on_read> callback and whenever
1242 new data arrives, it will be called. You can then remove some data (if
1243 enough is there) from the read buffer (C<< $handle->rbuf >>). Or you can
1244 leave the data there if you want to accumulate more (e.g. when only a
1245 partial message has been received so far), or change the read queue with
1246 e.g. C<push_read>.
1247
1248 In the more complex case, you want to queue multiple callbacks. In this
1249 case, AnyEvent::Handle will call the first queued callback each time new
1250 data arrives (also the first time it is queued) and remove it when it has
1251 done its job (see C<push_read>, below).
1252
1253 This way you can, for example, push three line-reads, followed by reading
1254 a chunk of data, and AnyEvent::Handle will execute them in order.
1255
1256 Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
1257 the specified number of bytes which give an XML datagram.
1258
1259 # in the default state, expect some header bytes
1260 $handle->on_read (sub {
1261 # some data is here, now queue the length-header-read (4 octets)
1262 shift->unshift_read (chunk => 4, sub {
1263 # header arrived, decode
1264 my $len = unpack "N", $_[1];
1265
1266 # now read the payload
1267 shift->unshift_read (chunk => $len, sub {
1268 my $xml = $_[1];
1269 # handle xml
1270 });
1271 });
1272 });
1273
1274 Example 2: Implement a client for a protocol that replies either with "OK"
1275 and another line or "ERROR" for the first request that is sent, and 64
1276 bytes for the second request. Due to the availability of a queue, we can
1277 just pipeline sending both requests and manipulate the queue as necessary
1278 in the callbacks.
1279
1280 When the first callback is called and sees an "OK" response, it will
1281 C<unshift> another line-read. This line-read will be queued I<before> the
1282 64-byte chunk callback.
1283
1284 # request one, returns either "OK + extra line" or "ERROR"
1285 $handle->push_write ("request 1\015\012");
1286
1287 # we expect "ERROR" or "OK" as response, so push a line read
1288 $handle->push_read (line => sub {
1289 # if we got an "OK", we have to _prepend_ another line,
1290 # so it will be read before the second request reads its 64 bytes
1291 # which are already in the queue when this callback is called
1292 # we don't do this in case we got an error
1293 if ($_[1] eq "OK") {
1294 $_[0]->unshift_read (line => sub {
1295 my $response = $_[1];
1296 ...
1297 });
1298 }
1299 });
1300
1301 # request two, simply returns 64 octets
1302 $handle->push_write ("request 2\015\012");
1303
1304 # simply read 64 bytes, always
1305 $handle->push_read (chunk => 64, sub {
1306 my $response = $_[1];
1307 ...
1308 });
1309
1310 =over 4
1311
1312 =cut
1313
1314 sub _drain_rbuf {
1315 my ($self) = @_;
1316
1317 # avoid recursion
1318 return if $self->{_skip_drain_rbuf};
1319 local $self->{_skip_drain_rbuf} = 1;
1320
1321 while () {
1322 # we need to use a separate tls read buffer, as we must not receive data while
1323 # we are draining the buffer, and this can only happen with TLS.
1324 $self->{rbuf} .= delete $self->{_tls_rbuf}
1325 if exists $self->{_tls_rbuf};
1326
1327 my $len = length $self->{rbuf};
1328
1329 if (my $cb = shift @{ $self->{_queue} }) {
1330 unless ($cb->($self)) {
1331 # no progress can be made
1332 # (not enough data and no data forthcoming)
1333 $self->_error (Errno::EPIPE, 1), return
1334 if $self->{_eof};
1335
1336 unshift @{ $self->{_queue} }, $cb;
1337 last;
1338 }
1339 } elsif ($self->{on_read}) {
1340 last unless $len;
1341
1342 $self->{on_read}($self);
1343
1344 if (
1345 $len == length $self->{rbuf} # if no data has been consumed
1346 && !@{ $self->{_queue} } # and the queue is still empty
1347 && $self->{on_read} # but we still have on_read
1348 ) {
1349 # no further data will arrive
1350 # so no progress can be made
1351 $self->_error (Errno::EPIPE, 1), return
1352 if $self->{_eof};
1353
1354 last; # more data might arrive
1355 }
1356 } else {
1357 # read side becomes idle
1358 delete $self->{_rw} unless $self->{tls};
1359 last;
1360 }
1361 }
1362
1363 if ($self->{_eof}) {
1364 $self->{on_eof}
1365 ? $self->{on_eof}($self)
1366 : $self->_error (0, 1, "Unexpected end-of-file");
1367
1368 return;
1369 }
1370
1371 if (
1372 defined $self->{rbuf_max}
1373 && $self->{rbuf_max} < length $self->{rbuf}
1374 ) {
1375 $self->_error (Errno::ENOSPC, 1), return;
1376 }
1377
1378 # may need to restart read watcher
1379 unless ($self->{_rw}) {
1380 $self->start_read
1381 if $self->{on_read} || @{ $self->{_queue} };
1382 }
1383 }
1384
1385 =item $handle->on_read ($cb)
1386
1387 This replaces the currently set C<on_read> callback, or clears it (when
1388 the new callback is C<undef>). See the description of C<on_read> in the
1389 constructor.
1390
1391 This method may invoke callbacks (and therefore the handle might be
1392 destroyed after it returns).
1393
1394 =cut
1395
1396 sub on_read {
1397 my ($self, $cb) = @_;
1398
1399 $self->{on_read} = $cb;
1400 $self->_drain_rbuf if $cb;
1401 }
1402
1403 =item $handle->rbuf
1404
1405 Returns the read buffer (as a modifiable lvalue). You can also access the
1406 read buffer directly as the C<< ->{rbuf} >> member, if you want (this is
1407 much faster, and no less clean).
1408
1409 The only operation allowed on the read buffer (apart from looking at it)
1410 is removing data from its beginning. Otherwise modifying or appending to
1411 it is not allowed and will lead to hard-to-track-down bugs.
1412
1413 NOTE: The read buffer should only be used or modified in the C<on_read>
1414 callback or when C<push_read> or C<unshift_read> are used with a single
1415 callback (i.e. untyped). Typed C<push_read> and C<unshift_read> methods
1416 will manage the read buffer on their own.
1417
1418 =cut
1419
1420 sub rbuf : lvalue {
1421 $_[0]{rbuf}
1422 }
1423
1424 =item $handle->push_read ($cb)
1425
1426 =item $handle->unshift_read ($cb)
1427
1428 Append the given callback to the end of the queue (C<push_read>) or
1429 prepend it (C<unshift_read>).
1430
1431 The callback is called each time some additional read data arrives.
1432
1433 It must check whether enough data is in the read buffer already.
1434
1435 If not enough data is available, it must return the empty list or a false
1436 value, in which case it will be called repeatedly until enough data is
1437 available (or an error condition is detected).
1438
1439 If enough data was available, then the callback must remove all data it is
1440 interested in (which can be none at all) and return a true value. After returning
1441 true, it will be removed from the queue.
1442
1443 These methods may invoke callbacks (and therefore the handle might be
1444 destroyed after it returns).
1445
1446 =cut
1447
1448 our %RH;
1449
1450 sub register_read_type($$) {
1451 $RH{$_[0]} = $_[1];
1452 }
1453
1454 sub push_read {
1455 my $self = shift;
1456 my $cb = pop;
1457
1458 if (@_) {
1459 my $type = shift;
1460
1461 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1462 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_read")
1463 ->($self, $cb, @_);
1464 }
1465
1466 push @{ $self->{_queue} }, $cb;
1467 $self->_drain_rbuf;
1468 }
1469
1470 sub unshift_read {
1471 my $self = shift;
1472 my $cb = pop;
1473
1474 if (@_) {
1475 my $type = shift;
1476
1477 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1478 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::unshift_read")
1479 ->($self, $cb, @_);
1480 }
1481
1482 unshift @{ $self->{_queue} }, $cb;
1483 $self->_drain_rbuf;
1484 }
1485
1486 =item $handle->push_read (type => @args, $cb)
1487
1488 =item $handle->unshift_read (type => @args, $cb)
1489
1490 Instead of providing a callback that parses the data itself you can chose
1491 between a number of predefined parsing formats, for chunks of data, lines
1492 etc. You can also specify the (fully qualified) name of a package, in
1493 which case AnyEvent tries to load the package and then expects to find the
1494 C<anyevent_read_type> function inside (see "custom read types", below).
1495
1496 Predefined types are (if you have ideas for additional types, feel free to
1497 drop by and tell us):
1498
1499 =over 4
1500
1501 =item chunk => $octets, $cb->($handle, $data)
1502
1503 Invoke the callback only once C<$octets> bytes have been read. Pass the
1504 data read to the callback. The callback will never be called with less
1505 data.
1506
1507 Example: read 2 bytes.
1508
1509 $handle->push_read (chunk => 2, sub {
1510 say "yay " . unpack "H*", $_[1];
1511 });
1512
1513 =cut
1514
1515 register_read_type chunk => sub {
1516 my ($self, $cb, $len) = @_;
1517
1518 sub {
1519 $len <= length $_[0]{rbuf} or return;
1520 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
1521 1
1522 }
1523 };
1524
1525 =item line => [$eol, ]$cb->($handle, $line, $eol)
1526
1527 The callback will be called only once a full line (including the end of
1528 line marker, C<$eol>) has been read. This line (excluding the end of line
1529 marker) will be passed to the callback as second argument (C<$line>), and
1530 the end of line marker as the third argument (C<$eol>).
1531
1532 The end of line marker, C<$eol>, can be either a string, in which case it
1533 will be interpreted as a fixed record end marker, or it can be a regex
1534 object (e.g. created by C<qr>), in which case it is interpreted as a
1535 regular expression.
1536
1537 The end of line marker argument C<$eol> is optional, if it is missing (NOT
1538 undef), then C<qr|\015?\012|> is used (which is good for most internet
1539 protocols).
1540
1541 Partial lines at the end of the stream will never be returned, as they are
1542 not marked by the end of line marker.
1543
1544 =cut
1545
1546 register_read_type line => sub {
1547 my ($self, $cb, $eol) = @_;
1548
1549 if (@_ < 3) {
1550 # this is faster then the generic code below
1551 sub {
1552 (my $pos = index $_[0]{rbuf}, "\012") >= 0
1553 or return;
1554
1555 (my $str = substr $_[0]{rbuf}, 0, $pos + 1, "") =~ s/(\015?\012)\Z// or die;
1556 $cb->($_[0], $str, "$1");
1557 1
1558 }
1559 } else {
1560 $eol = quotemeta $eol unless ref $eol;
1561 $eol = qr|^(.*?)($eol)|s;
1562
1563 sub {
1564 $_[0]{rbuf} =~ s/$eol// or return;
1565
1566 $cb->($_[0], "$1", "$2");
1567 1
1568 }
1569 }
1570 };
1571
1572 =item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1573
1574 Makes a regex match against the regex object C<$accept> and returns
1575 everything up to and including the match.
1576
1577 Example: read a single line terminated by '\n'.
1578
1579 $handle->push_read (regex => qr<\n>, sub { ... });
1580
1581 If C<$reject> is given and not undef, then it determines when the data is
1582 to be rejected: it is matched against the data when the C<$accept> regex
1583 does not match and generates an C<EBADMSG> error when it matches. This is
1584 useful to quickly reject wrong data (to avoid waiting for a timeout or a
1585 receive buffer overflow).
1586
1587 Example: expect a single decimal number followed by whitespace, reject
1588 anything else (not the use of an anchor).
1589
1590 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1591
1592 If C<$skip> is given and not C<undef>, then it will be matched against
1593 the receive buffer when neither C<$accept> nor C<$reject> match,
1594 and everything preceding and including the match will be accepted
1595 unconditionally. This is useful to skip large amounts of data that you
1596 know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1597 have to start matching from the beginning. This is purely an optimisation
1598 and is usually worth it only when you expect more than a few kilobytes.
1599
1600 Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1601 expect the header to be very large (it isn't in practice, but...), we use
1602 a skip regex to skip initial portions. The skip regex is tricky in that
1603 it only accepts something not ending in either \015 or \012, as these are
1604 required for the accept regex.
1605
1606 $handle->push_read (regex =>
1607 qr<\015\012\015\012>,
1608 undef, # no reject
1609 qr<^.*[^\015\012]>,
1610 sub { ... });
1611
1612 =cut
1613
1614 register_read_type regex => sub {
1615 my ($self, $cb, $accept, $reject, $skip) = @_;
1616
1617 my $data;
1618 my $rbuf = \$self->{rbuf};
1619
1620 sub {
1621 # accept
1622 if ($$rbuf =~ $accept) {
1623 $data .= substr $$rbuf, 0, $+[0], "";
1624 $cb->($_[0], $data);
1625 return 1;
1626 }
1627
1628 # reject
1629 if ($reject && $$rbuf =~ $reject) {
1630 $_[0]->_error (Errno::EBADMSG);
1631 }
1632
1633 # skip
1634 if ($skip && $$rbuf =~ $skip) {
1635 $data .= substr $$rbuf, 0, $+[0], "";
1636 }
1637
1638 ()
1639 }
1640 };
1641
1642 =item netstring => $cb->($handle, $string)
1643
1644 A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
1645
1646 Throws an error with C<$!> set to EBADMSG on format violations.
1647
1648 =cut
1649
1650 register_read_type netstring => sub {
1651 my ($self, $cb) = @_;
1652
1653 sub {
1654 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1655 if ($_[0]{rbuf} =~ /[^0-9]/) {
1656 $_[0]->_error (Errno::EBADMSG);
1657 }
1658 return;
1659 }
1660
1661 my $len = $1;
1662
1663 $_[0]->unshift_read (chunk => $len, sub {
1664 my $string = $_[1];
1665 $_[0]->unshift_read (chunk => 1, sub {
1666 if ($_[1] eq ",") {
1667 $cb->($_[0], $string);
1668 } else {
1669 $_[0]->_error (Errno::EBADMSG);
1670 }
1671 });
1672 });
1673
1674 1
1675 }
1676 };
1677
1678 =item packstring => $format, $cb->($handle, $string)
1679
1680 An octet string prefixed with an encoded length. The encoding C<$format>
1681 uses the same format as a Perl C<pack> format, but must specify a single
1682 integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1683 optional C<!>, C<< < >> or C<< > >> modifier).
1684
1685 For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1686 EPP uses a prefix of C<N> (4 octtes).
1687
1688 Example: read a block of data prefixed by its length in BER-encoded
1689 format (very efficient).
1690
1691 $handle->push_read (packstring => "w", sub {
1692 my ($handle, $data) = @_;
1693 });
1694
1695 =cut
1696
1697 register_read_type packstring => sub {
1698 my ($self, $cb, $format) = @_;
1699
1700 sub {
1701 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1702 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1703 or return;
1704
1705 $format = length pack $format, $len;
1706
1707 # bypass unshift if we already have the remaining chunk
1708 if ($format + $len <= length $_[0]{rbuf}) {
1709 my $data = substr $_[0]{rbuf}, $format, $len;
1710 substr $_[0]{rbuf}, 0, $format + $len, "";
1711 $cb->($_[0], $data);
1712 } else {
1713 # remove prefix
1714 substr $_[0]{rbuf}, 0, $format, "";
1715
1716 # read remaining chunk
1717 $_[0]->unshift_read (chunk => $len, $cb);
1718 }
1719
1720 1
1721 }
1722 };
1723
1724 =item json => $cb->($handle, $hash_or_arrayref)
1725
1726 Reads a JSON object or array, decodes it and passes it to the
1727 callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1728
1729 If a C<json> object was passed to the constructor, then that will be
1730 used for the final decode, otherwise it will create a L<JSON::XS> or
1731 L<JSON::PP> coder object expecting UTF-8.
1732
1733 This read type uses the incremental parser available with JSON version
1734 2.09 (and JSON::XS version 2.2) and above.
1735
1736 Since JSON texts are fully self-delimiting, the C<json> read and write
1737 types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1738 the C<json> write type description, above, for an actual example.
1739
1740 =cut
1741
1742 register_read_type json => sub {
1743 my ($self, $cb) = @_;
1744
1745 my $json = $self->{json} ||= json_coder;
1746
1747 my $data;
1748
1749 sub {
1750 my $ref = eval { $json->incr_parse ($_[0]{rbuf}) };
1751
1752 if ($ref) {
1753 $_[0]{rbuf} = $json->incr_text;
1754 $json->incr_text = "";
1755 $cb->($_[0], $ref);
1756
1757 1
1758 } elsif ($@) {
1759 # error case
1760 $json->incr_skip;
1761
1762 $_[0]{rbuf} = $json->incr_text;
1763 $json->incr_text = "";
1764
1765 $_[0]->_error (Errno::EBADMSG);
1766
1767 ()
1768 } else {
1769 $_[0]{rbuf} = "";
1770
1771 ()
1772 }
1773 }
1774 };
1775
1776 =item cbor => $cb->($handle, $scalar)
1777
1778 Reads a CBOR value, decodes it and passes it to the callback. When a parse
1779 error occurs, an C<EBADMSG> error will be raised.
1780
1781 If a L<CBOR::XS> object was passed to the constructor, then that will be
1782 used for the final decode, otherwise it will create a CBOR coder without
1783 enabling any options.
1784
1785 You have to provide a dependency to L<CBOR::XS> on your own: this module
1786 will load the L<CBOR::XS> module, but AnyEvent does not depend on it
1787 itself.
1788
1789 Since CBOR values are fully self-delimiting, the C<cbor> read and write
1790 types are an ideal simple RPC protocol: just exchange CBOR datagrams. See
1791 the C<cbor> write type description, above, for an actual example.
1792
1793 =cut
1794
1795 register_read_type cbor => sub {
1796 my ($self, $cb) = @_;
1797
1798 my $cbor = $self->{cbor} ||= cbor_coder;
1799
1800 my $data;
1801
1802 sub {
1803 my (@value) = eval { $cbor->incr_parse ($_[0]{rbuf}) };
1804
1805 if (@value) {
1806 $cb->($_[0], @value);
1807
1808 1
1809 } elsif ($@) {
1810 # error case
1811 $cbor->incr_reset;
1812
1813 $_[0]->_error (Errno::EBADMSG);
1814
1815 ()
1816 } else {
1817 ()
1818 }
1819 }
1820 };
1821
1822 =item storable => $cb->($handle, $ref)
1823
1824 Deserialises a L<Storable> frozen representation as written by the
1825 C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1826 data).
1827
1828 Raises C<EBADMSG> error if the data could not be decoded.
1829
1830 =cut
1831
1832 register_read_type storable => sub {
1833 my ($self, $cb) = @_;
1834
1835 require Storable unless $Storable::VERSION;
1836
1837 sub {
1838 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1839 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1840 or return;
1841
1842 my $format = length pack "w", $len;
1843
1844 # bypass unshift if we already have the remaining chunk
1845 if ($format + $len <= length $_[0]{rbuf}) {
1846 my $data = substr $_[0]{rbuf}, $format, $len;
1847 substr $_[0]{rbuf}, 0, $format + $len, "";
1848
1849 eval { $cb->($_[0], Storable::thaw ($data)); 1 }
1850 or return $_[0]->_error (Errno::EBADMSG);
1851 } else {
1852 # remove prefix
1853 substr $_[0]{rbuf}, 0, $format, "";
1854
1855 # read remaining chunk
1856 $_[0]->unshift_read (chunk => $len, sub {
1857 eval { $cb->($_[0], Storable::thaw ($_[1])); 1 }
1858 or $_[0]->_error (Errno::EBADMSG);
1859 });
1860 }
1861
1862 1
1863 }
1864 };
1865
1866 =item tls_detect => $cb->($handle, $detect, $major, $minor)
1867
1868 Checks the input stream for a valid SSL or TLS handshake TLSPaintext
1869 record without consuming anything. Only SSL version 3 or higher
1870 is handled, up to the fictituous protocol 4.x (but both SSL3+ and
1871 SSL2-compatible framing is supported).
1872
1873 If it detects that the input data is likely TLS, it calls the callback
1874 with a true value for C<$detect> and the (on-wire) TLS version as second
1875 and third argument (C<$major> is C<3>, and C<$minor> is 0..3 for SSL
1876 3.0, TLS 1.0, 1.1 and 1.2, respectively). If it detects the input to
1877 be definitely not TLS, it calls the callback with a false value for
1878 C<$detect>.
1879
1880 The callback could use this information to decide whether or not to start
1881 TLS negotiation.
1882
1883 In all cases the data read so far is passed to the following read
1884 handlers.
1885
1886 Usually you want to use the C<tls_autostart> read type instead.
1887
1888 If you want to design a protocol that works in the presence of TLS
1889 dtection, make sure that any non-TLS data doesn't start with the octet 22
1890 (ASCII SYN, 16 hex) or 128-255 (i.e. highest bit set). The checks this
1891 read type does are a bit more strict, but might losen in the future to
1892 accomodate protocol changes.
1893
1894 This read type does not rely on L<AnyEvent::TLS> (and thus, not on
1895 L<Net::SSLeay>).
1896
1897 =item tls_autostart => $tls[, $tls_ctx]
1898
1899 Tries to detect a valid SSL or TLS handshake. If one is detected, it tries
1900 to start tls by calling C<starttls> with the given arguments.
1901
1902 In practise, C<$tls> must be C<accept>, or a Net::SSLeay context that has
1903 been configured to accept, as servers do not normally send a handshake on
1904 their own and ths cannot be detected in this way.
1905
1906 See C<tls_detect> above for more details.
1907
1908 Example: give the client a chance to start TLS before accepting a text
1909 line.
1910
1911 $hdl->push_read (tls_detect => "accept");
1912 $hdl->push_read (line => sub {
1913 print "received ", ($_[0]{tls} ? "encrypted" : "cleartext"), " <$_[1]>\n";
1914 });
1915
1916 =cut
1917
1918 register_read_type tls_detect => sub {
1919 my ($self, $cb) = @_;
1920
1921 sub {
1922 # this regex matches a full or partial tls record
1923 if (
1924 # ssl3+: type(22=handshake) major(=3) minor(any) length_hi
1925 $self->{rbuf} =~ /^(?:\z| \x16 (\z| [\x03\x04] (?:\z| . (?:\z| [\x00-\x40] ))))/xs
1926 # ssl2 comapatible: len_hi len_lo type(1) major minor dummy(forlength)
1927 or $self->{rbuf} =~ /^(?:\z| [\x80-\xff] (?:\z| . (?:\z| \x01 (\z| [\x03\x04] (?:\z| . (?:\z| . ))))))/xs
1928 ) {
1929 return if 3 != length $1; # partial match, can't decide yet
1930
1931 # full match, valid TLS record
1932 my ($major, $minor) = unpack "CC", $1;
1933 $cb->($self, "accept", $major + $minor * 0.1);
1934 } else {
1935 # mismatch == guaranteed not TLS
1936 $cb->($self, undef);
1937 }
1938
1939 1
1940 }
1941 };
1942
1943 register_read_type tls_autostart => sub {
1944 my ($self, @tls) = @_;
1945
1946 $RH{tls_detect}($self, sub {
1947 return unless $_[1];
1948 $_[0]->starttls (@tls);
1949 })
1950 };
1951
1952 =back
1953
1954 =item custom read types - Package::anyevent_read_type $handle, $cb, @args
1955
1956 Instead of one of the predefined types, you can also specify the name
1957 of a package. AnyEvent will try to load the package and then expects to
1958 find a function named C<anyevent_read_type> inside. If it isn't found, it
1959 progressively tries to load the parent package until it either finds the
1960 function (good) or runs out of packages (bad).
1961
1962 Whenever this type is used, C<push_read> will invoke the function with the
1963 handle object, the original callback and the remaining arguments.
1964
1965 The function is supposed to return a callback (usually a closure) that
1966 works as a plain read callback (see C<< ->push_read ($cb) >>), so you can
1967 mentally treat the function as a "configurable read type to read callback"
1968 converter.
1969
1970 It should invoke the original callback when it is done reading (remember
1971 to pass C<$handle> as first argument as all other callbacks do that,
1972 although there is no strict requirement on this).
1973
1974 For examples, see the source of this module (F<perldoc -m
1975 AnyEvent::Handle>, search for C<register_read_type>)).
1976
1977 =item $handle->stop_read
1978
1979 =item $handle->start_read
1980
1981 In rare cases you actually do not want to read anything from the
1982 socket. In this case you can call C<stop_read>. Neither C<on_read> nor
1983 any queued callbacks will be executed then. To start reading again, call
1984 C<start_read>.
1985
1986 Note that AnyEvent::Handle will automatically C<start_read> for you when
1987 you change the C<on_read> callback or push/unshift a read callback, and it
1988 will automatically C<stop_read> for you when neither C<on_read> is set nor
1989 there are any read requests in the queue.
1990
1991 In older versions of this module (<= 5.3), these methods had no effect,
1992 as TLS does not support half-duplex connections. In current versions they
1993 work as expected, as this behaviour is required to avoid certain resource
1994 attacks, where the program would be forced to read (and buffer) arbitrary
1995 amounts of data before being able to send some data. The drawback is that
1996 some readings of the the SSL/TLS specifications basically require this
1997 attack to be working, as SSL/TLS implementations might stall sending data
1998 during a rehandshake.
1999
2000 As a guideline, during the initial handshake, you should not stop reading,
2001 and as a client, it might cause problems, depending on your application.
2002
2003 =cut
2004
2005 sub stop_read {
2006 my ($self) = @_;
2007
2008 delete $self->{_rw};
2009 }
2010
2011 sub start_read {
2012 my ($self) = @_;
2013
2014 unless ($self->{_rw} || $self->{_eof} || !$self->{fh}) {
2015 Scalar::Util::weaken $self;
2016
2017 $self->{_rw} = AE::io $self->{fh}, 0, sub {
2018 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
2019 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size}, length $$rbuf;
2020
2021 if ($len > 0) {
2022 $self->{_activity} = $self->{_ractivity} = AE::now;
2023
2024 if ($self->{tls}) {
2025 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
2026
2027 &_dotls ($self);
2028 } else {
2029 $self->_drain_rbuf;
2030 }
2031
2032 if ($len == $self->{read_size}) {
2033 $self->{read_size} *= 2;
2034 $self->{read_size} = $self->{max_read_size} || MAX_READ_SIZE
2035 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
2036 }
2037
2038 } elsif (defined $len) {
2039 delete $self->{_rw};
2040 $self->{_eof} = 1;
2041 $self->_drain_rbuf;
2042
2043 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
2044 return $self->_error ($!, 1);
2045 }
2046 };
2047 }
2048 }
2049
2050 our $ERROR_SYSCALL;
2051 our $ERROR_WANT_READ;
2052
2053 sub _tls_error {
2054 my ($self, $err) = @_;
2055
2056 return $self->_error ($!, 1)
2057 if $err == Net::SSLeay::ERROR_SYSCALL ();
2058
2059 my $err = Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
2060
2061 # reduce error string to look less scary
2062 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
2063
2064 if ($self->{_on_starttls}) {
2065 (delete $self->{_on_starttls})->($self, undef, $err);
2066 &_freetls;
2067 } else {
2068 &_freetls;
2069 $self->_error (Errno::EPROTO, 1, $err);
2070 }
2071 }
2072
2073 # poll the write BIO and send the data if applicable
2074 # also decode read data if possible
2075 # this is basiclaly our TLS state machine
2076 # more efficient implementations are possible with openssl,
2077 # but not with the buggy and incomplete Net::SSLeay.
2078 sub _dotls {
2079 my ($self) = @_;
2080
2081 my $tmp;
2082
2083 while (length $self->{_tls_wbuf}) {
2084 if (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) <= 0) {
2085 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
2086
2087 return $self->_tls_error ($tmp)
2088 if $tmp != $ERROR_WANT_READ
2089 && ($tmp != $ERROR_SYSCALL || $!);
2090
2091 last;
2092 }
2093
2094 substr $self->{_tls_wbuf}, 0, $tmp, "";
2095 }
2096
2097 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
2098 unless (length $tmp) {
2099 $self->{_on_starttls}
2100 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
2101 &_freetls;
2102
2103 if ($self->{on_stoptls}) {
2104 $self->{on_stoptls}($self);
2105 return;
2106 } else {
2107 # let's treat SSL-eof as we treat normal EOF
2108 delete $self->{_rw};
2109 $self->{_eof} = 1;
2110 }
2111 }
2112
2113 $self->{_tls_rbuf} .= $tmp;
2114 $self->_drain_rbuf;
2115 $self->{tls} or return; # tls session might have gone away in callback
2116 }
2117
2118 $tmp = Net::SSLeay::get_error ($self->{tls}, -1); # -1 is not neccessarily correct, but Net::SSLeay doesn't tell us
2119 return $self->_tls_error ($tmp)
2120 if $tmp != $ERROR_WANT_READ
2121 && ($tmp != $ERROR_SYSCALL || $!);
2122
2123 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
2124 $self->{wbuf} .= $tmp;
2125 $self->_drain_wbuf;
2126 $self->{tls} or return; # tls session might have gone away in callback
2127 }
2128
2129 $self->{_on_starttls}
2130 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
2131 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
2132 }
2133
2134 =item $handle->starttls ($tls[, $tls_ctx])
2135
2136 Instead of starting TLS negotiation immediately when the AnyEvent::Handle
2137 object is created, you can also do that at a later time by calling
2138 C<starttls>. See the C<tls> constructor argument for general info.
2139
2140 Starting TLS is currently an asynchronous operation - when you push some
2141 write data and then call C<< ->starttls >> then TLS negotiation will start
2142 immediately, after which the queued write data is then sent. This might
2143 change in future versions, so best make sure you have no outstanding write
2144 data when calling this method.
2145
2146 The first argument is the same as the C<tls> constructor argument (either
2147 C<"connect">, C<"accept"> or an existing Net::SSLeay object).
2148
2149 The second argument is the optional C<AnyEvent::TLS> object that is used
2150 when AnyEvent::Handle has to create its own TLS connection object, or
2151 a hash reference with C<< key => value >> pairs that will be used to
2152 construct a new context.
2153
2154 The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
2155 context in C<< $handle->{tls_ctx} >> after this call and can be used or
2156 changed to your liking. Note that the handshake might have already started
2157 when this function returns.
2158
2159 Due to bugs in OpenSSL, it might or might not be possible to do multiple
2160 handshakes on the same stream. It is best to not attempt to use the
2161 stream after stopping TLS.
2162
2163 This method may invoke callbacks (and therefore the handle might be
2164 destroyed after it returns).
2165
2166 =cut
2167
2168 our %TLS_CACHE; #TODO not yet documented, should we?
2169
2170 sub starttls {
2171 my ($self, $tls, $ctx) = @_;
2172
2173 Carp::croak "It is an error to call starttls on an AnyEvent::Handle object while TLS is already active, caught"
2174 if $self->{tls};
2175
2176 unless (defined $AnyEvent::TLS::VERSION) {
2177 eval {
2178 require Net::SSLeay;
2179 require AnyEvent::TLS;
2180 1
2181 } or return $self->_error (Errno::EPROTO, 1, "TLS support not available on this system");
2182 }
2183
2184 $self->{tls} = $tls;
2185 $self->{tls_ctx} = $ctx if @_ > 2;
2186
2187 return unless $self->{fh};
2188
2189 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
2190 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
2191
2192 $tls = delete $self->{tls};
2193 $ctx = $self->{tls_ctx};
2194
2195 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context or session
2196
2197 if ("HASH" eq ref $ctx) {
2198 if ($ctx->{cache}) {
2199 my $key = $ctx+0;
2200 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
2201 } else {
2202 $ctx = new AnyEvent::TLS %$ctx;
2203 }
2204 }
2205
2206 $self->{tls_ctx} = $ctx || TLS_CTX ();
2207 $self->{tls} = $tls = $self->{tls_ctx}->_get_session ($tls, $self, $self->{peername});
2208
2209 # basically, this is deep magic (because SSL_read should have the same issues)
2210 # but the openssl maintainers basically said: "trust us, it just works".
2211 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
2212 # and mismaintained ssleay-module doesn't even offer them).
2213 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
2214 #
2215 # in short: this is a mess.
2216 #
2217 # note that we do not try to keep the length constant between writes as we are required to do.
2218 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
2219 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
2220 # have identity issues in that area.
2221 # Net::SSLeay::CTX_set_mode ($ssl,
2222 # (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
2223 # | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
2224 Net::SSLeay::CTX_set_mode ($tls, 1|2);
2225
2226 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
2227 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
2228
2229 Net::SSLeay::BIO_write ($self->{_rbio}, $self->{rbuf});
2230 $self->{rbuf} = "";
2231
2232 Net::SSLeay::set_bio ($tls, $self->{_rbio}, $self->{_wbio});
2233
2234 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
2235 if $self->{on_starttls};
2236
2237 &_dotls; # need to trigger the initial handshake
2238 $self->start_read; # make sure we actually do read
2239 }
2240
2241 =item $handle->stoptls
2242
2243 Shuts down the SSL connection - this makes a proper EOF handshake by
2244 sending a close notify to the other side, but since OpenSSL doesn't
2245 support non-blocking shut downs, it is not guaranteed that you can re-use
2246 the stream afterwards.
2247
2248 This method may invoke callbacks (and therefore the handle might be
2249 destroyed after it returns).
2250
2251 =cut
2252
2253 sub stoptls {
2254 my ($self) = @_;
2255
2256 if ($self->{tls} && $self->{fh}) {
2257 Net::SSLeay::shutdown ($self->{tls});
2258
2259 &_dotls;
2260
2261 # # we don't give a shit. no, we do, but we can't. no...#d#
2262 # # we, we... have to use openssl :/#d#
2263 # &_freetls;#d#
2264 }
2265 }
2266
2267 sub _freetls {
2268 my ($self) = @_;
2269
2270 return unless $self->{tls};
2271
2272 $self->{tls_ctx}->_put_session (delete $self->{tls})
2273 if $self->{tls} > 0;
2274
2275 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
2276 }
2277
2278 =item $handle->resettls
2279
2280 This rarely-used method simply resets and TLS state on the handle, usually
2281 causing data loss.
2282
2283 One case where it may be useful is when you want to skip over the data in
2284 the stream but you are not interested in interpreting it, so data loss is
2285 no concern.
2286
2287 =cut
2288
2289 *resettls = \&_freetls;
2290
2291 sub DESTROY {
2292 my ($self) = @_;
2293
2294 &_freetls;
2295
2296 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
2297
2298 if ($linger && length $self->{wbuf} && $self->{fh}) {
2299 my $fh = delete $self->{fh};
2300 my $wbuf = delete $self->{wbuf};
2301
2302 my @linger;
2303
2304 push @linger, AE::io $fh, 1, sub {
2305 my $len = syswrite $fh, $wbuf, length $wbuf;
2306
2307 if ($len > 0) {
2308 substr $wbuf, 0, $len, "";
2309 } elsif (defined $len || ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK)) {
2310 @linger = (); # end
2311 }
2312 };
2313 push @linger, AE::timer $linger, 0, sub {
2314 @linger = ();
2315 };
2316 }
2317 }
2318
2319 =item $handle->destroy
2320
2321 Shuts down the handle object as much as possible - this call ensures that
2322 no further callbacks will be invoked and as many resources as possible
2323 will be freed. Any method you will call on the handle object after
2324 destroying it in this way will be silently ignored (and it will return the
2325 empty list).
2326
2327 Normally, you can just "forget" any references to an AnyEvent::Handle
2328 object and it will simply shut down. This works in fatal error and EOF
2329 callbacks, as well as code outside. It does I<NOT> work in a read or write
2330 callback, so when you want to destroy the AnyEvent::Handle object from
2331 within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
2332 that case.
2333
2334 Destroying the handle object in this way has the advantage that callbacks
2335 will be removed as well, so if those are the only reference holders (as
2336 is common), then one doesn't need to do anything special to break any
2337 reference cycles.
2338
2339 The handle might still linger in the background and write out remaining
2340 data, as specified by the C<linger> option, however.
2341
2342 =cut
2343
2344 sub destroy {
2345 my ($self) = @_;
2346
2347 $self->DESTROY;
2348 %$self = ();
2349 bless $self, "AnyEvent::Handle::destroyed";
2350 }
2351
2352 sub AnyEvent::Handle::destroyed::AUTOLOAD {
2353 #nop
2354 }
2355
2356 =item $handle->destroyed
2357
2358 Returns false as long as the handle hasn't been destroyed by a call to C<<
2359 ->destroy >>, true otherwise.
2360
2361 Can be useful to decide whether the handle is still valid after some
2362 callback possibly destroyed the handle. For example, C<< ->push_write >>,
2363 C<< ->starttls >> and other methods can call user callbacks, which in turn
2364 can destroy the handle, so work can be avoided by checking sometimes:
2365
2366 $hdl->starttls ("accept");
2367 return if $hdl->destroyed;
2368 $hdl->push_write (...
2369
2370 Note that the call to C<push_write> will silently be ignored if the handle
2371 has been destroyed, so often you can just ignore the possibility of the
2372 handle being destroyed.
2373
2374 =cut
2375
2376 sub destroyed { 0 }
2377 sub AnyEvent::Handle::destroyed::destroyed { 1 }
2378
2379 =item AnyEvent::Handle::TLS_CTX
2380
2381 This function creates and returns the AnyEvent::TLS object used by default
2382 for TLS mode.
2383
2384 The context is created by calling L<AnyEvent::TLS> without any arguments.
2385
2386 =cut
2387
2388 our $TLS_CTX;
2389
2390 sub TLS_CTX() {
2391 $TLS_CTX ||= do {
2392 require AnyEvent::TLS;
2393
2394 new AnyEvent::TLS
2395 }
2396 }
2397
2398 =back
2399
2400
2401 =head1 NONFREQUENTLY ASKED QUESTIONS
2402
2403 =over 4
2404
2405 =item I C<undef> the AnyEvent::Handle reference inside my callback and
2406 still get further invocations!
2407
2408 That's because AnyEvent::Handle keeps a reference to itself when handling
2409 read or write callbacks.
2410
2411 It is only safe to "forget" the reference inside EOF or error callbacks,
2412 from within all other callbacks, you need to explicitly call the C<<
2413 ->destroy >> method.
2414
2415 =item Why is my C<on_eof> callback never called?
2416
2417 Probably because your C<on_error> callback is being called instead: When
2418 you have outstanding requests in your read queue, then an EOF is
2419 considered an error as you clearly expected some data.
2420
2421 To avoid this, make sure you have an empty read queue whenever your handle
2422 is supposed to be "idle" (i.e. connection closes are O.K.). You can set
2423 an C<on_read> handler that simply pushes the first read requests in the
2424 queue.
2425
2426 See also the next question, which explains this in a bit more detail.
2427
2428 =item How can I serve requests in a loop?
2429
2430 Most protocols consist of some setup phase (authentication for example)
2431 followed by a request handling phase, where the server waits for requests
2432 and handles them, in a loop.
2433
2434 There are two important variants: The first (traditional, better) variant
2435 handles requests until the server gets some QUIT command, causing it to
2436 close the connection first (highly desirable for a busy TCP server). A
2437 client dropping the connection is an error, which means this variant can
2438 detect an unexpected detection close.
2439
2440 To handle this case, always make sure you have a non-empty read queue, by
2441 pushing the "read request start" handler on it:
2442
2443 # we assume a request starts with a single line
2444 my @start_request; @start_request = (line => sub {
2445 my ($hdl, $line) = @_;
2446
2447 ... handle request
2448
2449 # push next request read, possibly from a nested callback
2450 $hdl->push_read (@start_request);
2451 });
2452
2453 # auth done, now go into request handling loop
2454 # now push the first @start_request
2455 $hdl->push_read (@start_request);
2456
2457 By always having an outstanding C<push_read>, the handle always expects
2458 some data and raises the C<EPIPE> error when the connction is dropped
2459 unexpectedly.
2460
2461 The second variant is a protocol where the client can drop the connection
2462 at any time. For TCP, this means that the server machine may run out of
2463 sockets easier, and in general, it means you cannot distinguish a protocl
2464 failure/client crash from a normal connection close. Nevertheless, these
2465 kinds of protocols are common (and sometimes even the best solution to the
2466 problem).
2467
2468 Having an outstanding read request at all times is possible if you ignore
2469 C<EPIPE> errors, but this doesn't help with when the client drops the
2470 connection during a request, which would still be an error.
2471
2472 A better solution is to push the initial request read in an C<on_read>
2473 callback. This avoids an error, as when the server doesn't expect data
2474 (i.e. is idly waiting for the next request, an EOF will not raise an
2475 error, but simply result in an C<on_eof> callback. It is also a bit slower
2476 and simpler:
2477
2478 # auth done, now go into request handling loop
2479 $hdl->on_read (sub {
2480 my ($hdl) = @_;
2481
2482 # called each time we receive data but the read queue is empty
2483 # simply start read the request
2484
2485 $hdl->push_read (line => sub {
2486 my ($hdl, $line) = @_;
2487
2488 ... handle request
2489
2490 # do nothing special when the request has been handled, just
2491 # let the request queue go empty.
2492 });
2493 });
2494
2495 =item I get different callback invocations in TLS mode/Why can't I pause
2496 reading?
2497
2498 Unlike, say, TCP, TLS connections do not consist of two independent
2499 communication channels, one for each direction. Or put differently, the
2500 read and write directions are not independent of each other: you cannot
2501 write data unless you are also prepared to read, and vice versa.
2502
2503 This means that, in TLS mode, you might get C<on_error> or C<on_eof>
2504 callback invocations when you are not expecting any read data - the reason
2505 is that AnyEvent::Handle always reads in TLS mode.
2506
2507 During the connection, you have to make sure that you always have a
2508 non-empty read-queue, or an C<on_read> watcher. At the end of the
2509 connection (or when you no longer want to use it) you can call the
2510 C<destroy> method.
2511
2512 =item How do I read data until the other side closes the connection?
2513
2514 If you just want to read your data into a perl scalar, the easiest way
2515 to achieve this is by setting an C<on_read> callback that does nothing,
2516 clearing the C<on_eof> callback and in the C<on_error> callback, the data
2517 will be in C<$_[0]{rbuf}>:
2518
2519 $handle->on_read (sub { });
2520 $handle->on_eof (undef);
2521 $handle->on_error (sub {
2522 my $data = delete $_[0]{rbuf};
2523 });
2524
2525 Note that this example removes the C<rbuf> member from the handle object,
2526 which is not normally allowed by the API. It is expressly permitted in
2527 this case only, as the handle object needs to be destroyed afterwards.
2528
2529 The reason to use C<on_error> is that TCP connections, due to latencies
2530 and packets loss, might get closed quite violently with an error, when in
2531 fact all data has been received.
2532
2533 It is usually better to use acknowledgements when transferring data,
2534 to make sure the other side hasn't just died and you got the data
2535 intact. This is also one reason why so many internet protocols have an
2536 explicit QUIT command.
2537
2538 =item I don't want to destroy the handle too early - how do I wait until
2539 all data has been written?
2540
2541 After writing your last bits of data, set the C<on_drain> callback
2542 and destroy the handle in there - with the default setting of
2543 C<low_water_mark> this will be called precisely when all data has been
2544 written to the socket:
2545
2546 $handle->push_write (...);
2547 $handle->on_drain (sub {
2548 AE::log debug => "All data submitted to the kernel.";
2549 undef $handle;
2550 });
2551
2552 If you just want to queue some data and then signal EOF to the other side,
2553 consider using C<< ->push_shutdown >> instead.
2554
2555 =item I want to contact a TLS/SSL server, I don't care about security.
2556
2557 If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
2558 connect to it and then create the AnyEvent::Handle with the C<tls>
2559 parameter:
2560
2561 tcp_connect $host, $port, sub {
2562 my ($fh) = @_;
2563
2564 my $handle = new AnyEvent::Handle
2565 fh => $fh,
2566 tls => "connect",
2567 on_error => sub { ... };
2568
2569 $handle->push_write (...);
2570 };
2571
2572 =item I want to contact a TLS/SSL server, I do care about security.
2573
2574 Then you should additionally enable certificate verification, including
2575 peername verification, if the protocol you use supports it (see
2576 L<AnyEvent::TLS>, C<verify_peername>).
2577
2578 E.g. for HTTPS:
2579
2580 tcp_connect $host, $port, sub {
2581 my ($fh) = @_;
2582
2583 my $handle = new AnyEvent::Handle
2584 fh => $fh,
2585 peername => $host,
2586 tls => "connect",
2587 tls_ctx => { verify => 1, verify_peername => "https" },
2588 ...
2589
2590 Note that you must specify the hostname you connected to (or whatever
2591 "peername" the protocol needs) as the C<peername> argument, otherwise no
2592 peername verification will be done.
2593
2594 The above will use the system-dependent default set of trusted CA
2595 certificates. If you want to check against a specific CA, add the
2596 C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
2597
2598 tls_ctx => {
2599 verify => 1,
2600 verify_peername => "https",
2601 ca_file => "my-ca-cert.pem",
2602 },
2603
2604 =item I want to create a TLS/SSL server, how do I do that?
2605
2606 Well, you first need to get a server certificate and key. You have
2607 three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
2608 self-signed certificate (cheap. check the search engine of your choice,
2609 there are many tutorials on the net) or c) make your own CA (tinyca2 is a
2610 nice program for that purpose).
2611
2612 Then create a file with your private key (in PEM format, see
2613 L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
2614 file should then look like this:
2615
2616 -----BEGIN RSA PRIVATE KEY-----
2617 ...header data
2618 ... lots of base64'y-stuff
2619 -----END RSA PRIVATE KEY-----
2620
2621 -----BEGIN CERTIFICATE-----
2622 ... lots of base64'y-stuff
2623 -----END CERTIFICATE-----
2624
2625 The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
2626 specify this file as C<cert_file>:
2627
2628 tcp_server undef, $port, sub {
2629 my ($fh) = @_;
2630
2631 my $handle = new AnyEvent::Handle
2632 fh => $fh,
2633 tls => "accept",
2634 tls_ctx => { cert_file => "my-server-keycert.pem" },
2635 ...
2636
2637 When you have intermediate CA certificates that your clients might not
2638 know about, just append them to the C<cert_file>.
2639
2640 =back
2641
2642 =head1 SUBCLASSING AnyEvent::Handle
2643
2644 In many cases, you might want to subclass AnyEvent::Handle.
2645
2646 To make this easier, a given version of AnyEvent::Handle uses these
2647 conventions:
2648
2649 =over 4
2650
2651 =item * all constructor arguments become object members.
2652
2653 At least initially, when you pass a C<tls>-argument to the constructor it
2654 will end up in C<< $handle->{tls} >>. Those members might be changed or
2655 mutated later on (for example C<tls> will hold the TLS connection object).
2656
2657 =item * other object member names are prefixed with an C<_>.
2658
2659 All object members not explicitly documented (internal use) are prefixed
2660 with an underscore character, so the remaining non-C<_>-namespace is free
2661 for use for subclasses.
2662
2663 =item * all members not documented here and not prefixed with an underscore
2664 are free to use in subclasses.
2665
2666 Of course, new versions of AnyEvent::Handle may introduce more "public"
2667 member variables, but that's just life. At least it is documented.
2668
2669 =back
2670
2671 =head1 AUTHOR
2672
2673 Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
2674
2675 =cut
2676
2677 1
2678