ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent/lib/AnyEvent/Handle.pm
(Generate patch)

Comparing AnyEvent/lib/AnyEvent/Handle.pm (file contents):
Revision 1.4 by elmex, Sun Apr 27 20:20:20 2008 UTC vs.
Revision 1.220 by root, Sun Jul 24 13:10:43 2011 UTC

1package AnyEvent::Handle;
2
3use warnings;
4use strict;
5
6use AnyEvent;
7use IO::Handle;
8use Errno qw/EAGAIN EINTR/;
9
10=head1 NAME 1=head1 NAME
11 2
12AnyEvent::Handle - non-blocking I/O on filehandles via AnyEvent 3AnyEvent::Handle - non-blocking I/O on streaming handles via AnyEvent
13
14=head1 VERSION
15
16Version 0.01
17
18=cut
19
20our $VERSION = '0.01';
21 4
22=head1 SYNOPSIS 5=head1 SYNOPSIS
23 6
24 use AnyEvent; 7 use AnyEvent;
25 use AnyEvent::Handle; 8 use AnyEvent::Handle;
26 9
27 my $cv = AnyEvent->condvar; 10 my $cv = AnyEvent->condvar;
28 11
29 my $ae_fh = AnyEvent::Handle->new (fh => \*STDIN); 12 my $hdl; $hdl = new AnyEvent::Handle
13 fh => \*STDIN,
14 on_error => sub {
15 my ($hdl, $fatal, $msg) = @_;
16 warn "got error $msg\n";
17 $hdl->destroy;
18 $cv->send;
19 };
30 20
31 $ae_fh->on_eof (sub { $cv->broadcast }); 21 # send some request line
22 $hdl->push_write ("getinfo\015\012");
32 23
33 $ae_fh->readlines (sub { 24 # read the response line
25 $hdl->push_read (line => sub {
34 my ($ae_fh, @lines) = @_; 26 my ($hdl, $line) = @_;
35 for (@lines) { 27 warn "got line <$line>\n";
28 $cv->send;
29 });
30
31 $cv->recv;
32
33=head1 DESCRIPTION
34
35This is a helper module to make it easier to do event-based I/O on
36stream-based filehandles (sockets, pipes, and other stream things).
37
38The L<AnyEvent::Intro> tutorial contains some well-documented
39AnyEvent::Handle examples.
40
41In the following, where the documentation refers to "bytes", it means
42characters. As sysread and syswrite are used for all I/O, their
43treatment of characters applies to this module as well.
44
45At the very minimum, you should specify C<fh> or C<connect>, and the
46C<on_error> callback.
47
48All callbacks will be invoked with the handle object as their first
49argument.
50
51=cut
52
53package AnyEvent::Handle;
54
55use Scalar::Util ();
56use List::Util ();
57use Carp ();
58use Errno qw(EAGAIN EINTR);
59
60use AnyEvent (); BEGIN { AnyEvent::common_sense }
61use AnyEvent::Util qw(WSAEWOULDBLOCK);
62
63our $VERSION = $AnyEvent::VERSION;
64
65sub _load_func($) {
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
80sub 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
88The constructor supports these arguments (all as C<< key => value >> pairs).
89
90=over 4
91
92=item fh => $filehandle [C<fh> or C<connect> MANDATORY]
93
94The filehandle this L<AnyEvent::Handle> object will operate on.
95NOTE: The filehandle will be set to non-blocking mode (using
96C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
97that mode.
98
99=item connect => [$host, $service] [C<fh> or C<connect> MANDATORY]
100
101Try to connect to the specified host and service (port), using
102C<AnyEvent::Socket::tcp_connect>. The C<$host> additionally becomes the
103default C<peername>.
104
105You have to specify either this parameter, or C<fh>, above.
106
107It is possible to push requests on the read and write queues, and modify
108properties of the stream, even while AnyEvent::Handle is connecting.
109
110When this parameter is specified, then the C<on_prepare>,
111C<on_connect_error> and C<on_connect> callbacks will be called under the
112appropriate circumstances:
113
114=over 4
115
116=item on_prepare => $cb->($handle)
117
118This (rarely used) callback is called before a new connection is
119attempted, but after the file handle has been created (you can access that
120file handle via C<< $handle->{fh} >>). It could be used to prepare the
121file handle with parameters required for the actual connect (as opposed to
122settings that can be changed when the connection is already established).
123
124The return value of this callback should be the connect timeout value in
125seconds (or C<0>, or C<undef>, or the empty list, to indicate that the
126default timeout is to be used).
127
128=item on_connect => $cb->($handle, $host, $port, $retry->())
129
130This callback is called when a connection has been successfully established.
131
132The peer's numeric host and port (the socket peername) are passed as
133parameters, together with a retry callback.
134
135If, for some reason, the handle is not acceptable, calling C<$retry>
136will continue with the next connection target (in case of multi-homed
137hosts or SRV records there can be multiple connection endpoints). At the
138time it is called the read and write queues, eof status, tls status and
139similar properties of the handle will have been reset.
140
141In most cases, you should ignore the C<$retry> parameter.
142
143=item on_connect_error => $cb->($handle, $message)
144
145This callback is called when the connection could not be
146established. C<$!> will contain the relevant error code, and C<$message> a
147message describing it (usually the same as C<"$!">).
148
149If this callback isn't specified, then C<on_error> will be called with a
150fatal error instead.
151
152=back
153
154=item on_error => $cb->($handle, $fatal, $message)
155
156This is the error callback, which is called when, well, some error
157occured, such as not being able to resolve the hostname, failure to
158connect, or a read error.
159
160Some errors are fatal (which is indicated by C<$fatal> being true). On
161fatal errors the handle object will be destroyed (by a call to C<< ->
162destroy >>) after invoking the error callback (which means you are free to
163examine the handle object). Examples of fatal errors are an EOF condition
164with active (but unsatisfiable) read watchers (C<EPIPE>) or I/O errors. In
165cases where the other side can close the connection at will, it is
166often easiest to not report C<EPIPE> errors in this callback.
167
168AnyEvent::Handle tries to find an appropriate error code for you to check
169against, but in some cases (TLS errors), this does not work well. It is
170recommended to always output the C<$message> argument in human-readable
171error messages (it's usually the same as C<"$!">).
172
173Non-fatal errors can be retried by returning, but it is recommended
174to simply ignore this parameter and instead abondon the handle object
175when this callback is invoked. Examples of non-fatal errors are timeouts
176C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
177
178On entry to the callback, the value of C<$!> contains the operating
179system error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
180C<EPROTO>).
181
182While not mandatory, it is I<highly> recommended to set this callback, as
183you will not be notified of errors otherwise. The default just calls
184C<croak>.
185
186=item on_read => $cb->($handle)
187
188This sets the default read callback, which is called when data arrives
189and no read request is in the queue (unlike read queue callbacks, this
190callback will only be called when at least one octet of data is in the
191read buffer).
192
193To access (and remove data from) the read buffer, use the C<< ->rbuf >>
194method or access the C<< $handle->{rbuf} >> member directly. Note that you
195must not enlarge or modify the read buffer, you can only remove data at
196the beginning from it.
197
198You can also call C<< ->push_read (...) >> or any other function that
199modifies the read queue. Or do both. Or ...
200
201When an EOF condition is detected, AnyEvent::Handle will first try to
202feed all the remaining data to the queued callbacks and C<on_read> before
203calling the C<on_eof> callback. If no progress can be made, then a fatal
204error will be raised (with C<$!> set to C<EPIPE>).
205
206Note that, unlike requests in the read queue, an C<on_read> callback
207doesn't mean you I<require> some data: if there is an EOF and there
208are outstanding read requests then an error will be flagged. With an
209C<on_read> callback, the C<on_eof> callback will be invoked.
210
211=item on_eof => $cb->($handle)
212
213Set the callback to be called when an end-of-file condition is detected,
214i.e. in the case of a socket, when the other side has closed the
215connection cleanly, and there are no outstanding read requests in the
216queue (if there are read requests, then an EOF counts as an unexpected
217connection close and will be flagged as an error).
218
219For sockets, this just means that the other side has stopped sending data,
220you can still try to write data, and, in fact, one can return from the EOF
221callback and continue writing data, as only the read part has been shut
222down.
223
224If an EOF condition has been detected but no C<on_eof> callback has been
225set, then a fatal error will be raised with C<$!> set to <0>.
226
227=item on_drain => $cb->($handle)
228
229This sets the callback that is called when the write buffer becomes empty
230(or immediately if the buffer is empty already).
231
232To append to the write buffer, use the C<< ->push_write >> method.
233
234This callback is useful when you don't want to put all of your write data
235into the queue at once, for example, when you want to write the contents
236of some file to the socket you might not want to read the whole file into
237memory and push it into the queue, but instead only read more data from
238the file when the write queue becomes empty.
239
240=item timeout => $fractional_seconds
241
242=item rtimeout => $fractional_seconds
243
244=item wtimeout => $fractional_seconds
245
246If non-zero, then these enables an "inactivity" timeout: whenever this
247many seconds pass without a successful read or write on the underlying
248file handle (or a call to C<timeout_reset>), the C<on_timeout> callback
249will be invoked (and if that one is missing, a non-fatal C<ETIMEDOUT>
250error will be raised).
251
252There are three variants of the timeouts that work independently of each
253other, for both read and write (triggered when nothing was read I<OR>
254written), just read (triggered when nothing was read), and just write:
255C<timeout>, C<rtimeout> and C<wtimeout>, with corresponding callbacks
256C<on_timeout>, C<on_rtimeout> and C<on_wtimeout>, and reset functions
257C<timeout_reset>, C<rtimeout_reset>, and C<wtimeout_reset>.
258
259Note that timeout processing is active even when you do not have any
260outstanding read or write requests: If you plan to keep the connection
261idle then you should disable the timeout temporarily or ignore the
262timeout in the corresponding C<on_timeout> callback, in which case
263AnyEvent::Handle will simply restart the timeout.
264
265Zero (the default) disables the corresponding timeout.
266
267=item on_timeout => $cb->($handle)
268
269=item on_rtimeout => $cb->($handle)
270
271=item on_wtimeout => $cb->($handle)
272
273Called whenever the inactivity timeout passes. If you return from this
274callback, then the timeout will be reset as if some activity had happened,
275so this condition is not fatal in any way.
276
277=item rbuf_max => <bytes>
278
279If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
280when the read buffer ever (strictly) exceeds this size. This is useful to
281avoid some forms of denial-of-service attacks.
282
283For example, a server accepting connections from untrusted sources should
284be configured to accept only so-and-so much data that it cannot act on
285(for example, when expecting a line, an attacker could send an unlimited
286amount of data without a callback ever being called as long as the line
287isn't finished).
288
289=item wbuf_max => <bytes>
290
291If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
292when the write buffer ever (strictly) exceeds this size. This is useful to
293avoid some forms of denial-of-service attacks.
294
295Although the units of this parameter is bytes, this is the I<raw> number
296of bytes not yet accepted by the kernel. This can make a difference when
297you e.g. use TLS, as TLS typically makes your write data larger (but it
298can also make it smaller due to compression).
299
300As an example of when this limit is useful, take a chat server that sends
301chat messages to a client. If the client does not read those in a timely
302manner then the send buffer in the server would grow unbounded.
303
304=item autocork => <boolean>
305
306When disabled (the default), C<push_write> will try to immediately
307write the data to the handle if possible. This avoids having to register
308a write watcher and wait for the next event loop iteration, but can
309be inefficient if you write multiple small chunks (on the wire, this
310disadvantage is usually avoided by your kernel's nagle algorithm, see
311C<no_delay>, but this option can save costly syscalls).
312
313When enabled, writes will always be queued till the next event loop
314iteration. This is efficient when you do many small writes per iteration,
315but less efficient when you do a single write only per iteration (or when
316the write buffer often is full). It also increases write latency.
317
318=item no_delay => <boolean>
319
320When doing small writes on sockets, your operating system kernel might
321wait a bit for more data before actually sending it out. This is called
322the Nagle algorithm, and usually it is beneficial.
323
324In some situations you want as low a delay as possible, which can be
325accomplishd by setting this option to a true value.
326
327The default is your operating system's default behaviour (most likely
328enabled). This option explicitly enables or disables it, if possible.
329
330=item keepalive => <boolean>
331
332Enables (default disable) the SO_KEEPALIVE option on the stream socket:
333normally, TCP connections have no time-out once established, so TCP
334connections, once established, can stay alive forever even when the other
335side has long gone. TCP keepalives are a cheap way to take down long-lived
336TCP connections when the other side becomes unreachable. While the default
337is OS-dependent, TCP keepalives usually kick in after around two hours,
338and, if the other side doesn't reply, take down the TCP connection some 10
339to 15 minutes later.
340
341It is harmless to specify this option for file handles that do not support
342keepalives, and enabling it on connections that are potentially long-lived
343is usually a good idea.
344
345=item oobinline => <boolean>
346
347BSD majorly fucked up the implementation of TCP urgent data. The result
348is that almost no OS implements TCP according to the specs, and every OS
349implements it slightly differently.
350
351If you want to handle TCP urgent data, then setting this flag (the default
352is enabled) gives you the most portable way of getting urgent data, by
353putting it into the stream.
354
355Since BSD emulation of OOB data on top of TCP's urgent data can have
356security implications, AnyEvent::Handle sets this flag automatically
357unless explicitly specified. Note that setting this flag after
358establishing a connection I<may> be a bit too late (data loss could
359already have occured on BSD systems), but at least it will protect you
360from most attacks.
361
362=item read_size => <bytes>
363
364The initial read block size, the number of bytes this module will try to
365read during each loop iteration. Each handle object will consume at least
366this amount of memory for the read buffer as well, so when handling many
367connections requirements). See also C<max_read_size>. Default: C<2048>.
368
369=item max_read_size => <bytes>
370
371The maximum read buffer size used by the dynamic adjustment
372algorithm: Each time AnyEvent::Handle can read C<read_size> bytes in
373one go it will double C<read_size> up to the maximum given by this
374option. Default: C<131072> or C<read_size>, whichever is higher.
375
376=item low_water_mark => <bytes>
377
378Sets the number of bytes (default: C<0>) that make up an "empty" write
379buffer: If the buffer reaches this size or gets even samller it is
380considered empty.
381
382Sometimes it can be beneficial (for performance reasons) to add data to
383the write buffer before it is fully drained, but this is a rare case, as
384the operating system kernel usually buffers data as well, so the default
385is good in almost all cases.
386
387=item linger => <seconds>
388
389If this is non-zero (default: C<3600>), the destructor of the
390AnyEvent::Handle object will check whether there is still outstanding
391write data and will install a watcher that will write this data to the
392socket. No errors will be reported (this mostly matches how the operating
393system treats outstanding data at socket close time).
394
395This will not work for partial TLS data that could not be encoded
396yet. This data will be lost. Calling the C<stoptls> method in time might
397help.
398
399=item peername => $string
400
401A string used to identify the remote site - usually the DNS hostname
402(I<not> IDN!) used to create the connection, rarely the IP address.
403
404Apart from being useful in error messages, this string is also used in TLS
405peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
406verification will be skipped when C<peername> is not specified or is
407C<undef>.
408
409=item tls => "accept" | "connect" | Net::SSLeay::SSL object
410
411When this parameter is given, it enables TLS (SSL) mode, that means
412AnyEvent will start a TLS handshake as soon as the connection has been
413established and will transparently encrypt/decrypt data afterwards.
414
415All TLS protocol errors will be signalled as C<EPROTO>, with an
416appropriate error message.
417
418TLS mode requires Net::SSLeay to be installed (it will be loaded
419automatically when you try to create a TLS handle): this module doesn't
420have a dependency on that module, so if your module requires it, you have
421to add the dependency yourself.
422
423Unlike TCP, TLS has a server and client side: for the TLS server side, use
424C<accept>, and for the TLS client side of a connection, use C<connect>
425mode.
426
427You can also provide your own TLS connection object, but you have
428to make sure that you call either C<Net::SSLeay::set_connect_state>
429or C<Net::SSLeay::set_accept_state> on it before you pass it to
430AnyEvent::Handle. Also, this module will take ownership of this connection
431object.
432
433At some future point, AnyEvent::Handle might switch to another TLS
434implementation, then the option to use your own session object will go
435away.
436
437B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
438passing in the wrong integer will lead to certain crash. This most often
439happens when one uses a stylish C<< tls => 1 >> and is surprised about the
440segmentation fault.
441
442Use the C<< ->starttls >> method if you need to start TLS negotiation later.
443
444=item tls_ctx => $anyevent_tls
445
446Use the given C<AnyEvent::TLS> object to create the new TLS connection
447(unless a connection object was specified directly). If this
448parameter is missing (or C<undef>), then AnyEvent::Handle will use
449C<AnyEvent::Handle::TLS_CTX>.
450
451Instead of an object, you can also specify a hash reference with C<< key
452=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
453new TLS context object.
454
455=item on_starttls => $cb->($handle, $success[, $error_message])
456
457This callback will be invoked when the TLS/SSL handshake has finished. If
458C<$success> is true, then the TLS handshake succeeded, otherwise it failed
459(C<on_stoptls> will not be called in this case).
460
461The session in C<< $handle->{tls} >> can still be examined in this
462callback, even when the handshake was not successful.
463
464TLS handshake failures will not cause C<on_error> to be invoked when this
465callback is in effect, instead, the error message will be passed to C<on_starttls>.
466
467Without this callback, handshake failures lead to C<on_error> being
468called as usual.
469
470Note that you cannot just call C<starttls> again in this callback. If you
471need to do that, start an zero-second timer instead whose callback can
472then call C<< ->starttls >> again.
473
474=item on_stoptls => $cb->($handle)
475
476When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
477set, then it will be invoked after freeing the TLS session. If it is not,
478then a TLS shutdown condition will be treated like a normal EOF condition
479on the handle.
480
481The session in C<< $handle->{tls} >> can still be examined in this
482callback.
483
484This callback will only be called on TLS shutdowns, not when the
485underlying handle signals EOF.
486
487=item json => JSON or JSON::XS object
488
489This is the json coder object used by the C<json> read and write types.
490
491If you don't supply it, then AnyEvent::Handle will create and use a
492suitable one (on demand), which will write and expect UTF-8 encoded JSON
493texts.
494
495Note that you are responsible to depend on the JSON module if you want to
496use this functionality, as AnyEvent does not have a dependency itself.
497
498=back
499
500=cut
501
502sub new {
503 my $class = shift;
504 my $self = bless { @_ }, $class;
505
506 if ($self->{fh}) {
507 $self->_start;
508 return unless $self->{fh}; # could be gone by now
509
510 } elsif ($self->{connect}) {
511 require AnyEvent::Socket;
512
513 $self->{peername} = $self->{connect}[0]
514 unless exists $self->{peername};
515
516 $self->{_skip_drain_rbuf} = 1;
517
518 {
519 Scalar::Util::weaken (my $self = $self);
520
521 $self->{_connect} =
522 AnyEvent::Socket::tcp_connect (
523 $self->{connect}[0],
524 $self->{connect}[1],
525 sub {
526 my ($fh, $host, $port, $retry) = @_;
527
528 delete $self->{_connect}; # no longer needed
529
530 if ($fh) {
531 $self->{fh} = $fh;
532
533 delete $self->{_skip_drain_rbuf};
534 $self->_start;
535
536 $self->{on_connect}
537 and $self->{on_connect}($self, $host, $port, sub {
538 delete @$self{qw(fh _tw _rtw _wtw _ww _rw _eof _queue rbuf _wbuf tls _tls_rbuf _tls_wbuf)};
539 $self->{_skip_drain_rbuf} = 1;
540 &$retry;
541 });
542
543 } else {
544 if ($self->{on_connect_error}) {
545 $self->{on_connect_error}($self, "$!");
546 $self->destroy if $self;
547 } else {
548 $self->_error ($!, 1);
549 }
550 }
551 },
552 sub {
553 local $self->{fh} = $_[0];
554
555 $self->{on_prepare}
556 ? $self->{on_prepare}->($self)
557 : ()
558 }
559 );
560 }
561
562 } else {
563 Carp::croak "AnyEvent::Handle: either an existing fh or the connect parameter must be specified";
564 }
565
566 $self
567}
568
569sub _start {
570 my ($self) = @_;
571
572 # too many clueless people try to use udp and similar sockets
573 # with AnyEvent::Handle, do them a favour.
574 my $type = getsockopt $self->{fh}, Socket::SOL_SOCKET (), Socket::SO_TYPE ();
575 Carp::croak "AnyEvent::Handle: only stream sockets supported, anything else will NOT work!"
576 if Socket::SOCK_STREAM () != (unpack "I", $type) && defined $type;
577
578 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
579
580 $self->{_activity} =
581 $self->{_ractivity} =
582 $self->{_wactivity} = AE::now;
583
584 $self->{read_size} ||= 2048;
585 $self->{max_read_size} = $self->{read_size}
586 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
587
588 $self->timeout (delete $self->{timeout} ) if $self->{timeout};
589 $self->rtimeout (delete $self->{rtimeout} ) if $self->{rtimeout};
590 $self->wtimeout (delete $self->{wtimeout} ) if $self->{wtimeout};
591
592 $self->no_delay (delete $self->{no_delay} ) if exists $self->{no_delay} && $self->{no_delay};
593 $self->keepalive (delete $self->{keepalive}) if exists $self->{keepalive} && $self->{keepalive};
594
595 $self->oobinline (exists $self->{oobinline} ? delete $self->{oobinline} : 1);
596
597 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
598 if $self->{tls};
599
600 $self->on_drain (delete $self->{on_drain} ) if $self->{on_drain};
601
602 $self->start_read
603 if $self->{on_read} || @{ $self->{_queue} };
604
605 $self->_drain_wbuf;
606}
607
608sub _error {
609 my ($self, $errno, $fatal, $message) = @_;
610
611 $! = $errno;
612 $message ||= "$!";
613
614 if ($self->{on_error}) {
615 $self->{on_error}($self, $fatal, $message);
616 $self->destroy if $fatal;
617 } elsif ($self->{fh} || $self->{connect}) {
618 $self->destroy;
619 Carp::croak "AnyEvent::Handle uncaught error: $message";
620 }
621}
622
623=item $fh = $handle->fh
624
625This method returns the file handle used to create the L<AnyEvent::Handle> object.
626
627=cut
628
629sub fh { $_[0]{fh} }
630
631=item $handle->on_error ($cb)
632
633Replace the current C<on_error> callback (see the C<on_error> constructor argument).
634
635=cut
636
637sub on_error {
638 $_[0]{on_error} = $_[1];
639}
640
641=item $handle->on_eof ($cb)
642
643Replace the current C<on_eof> callback (see the C<on_eof> constructor argument).
644
645=cut
646
647sub on_eof {
648 $_[0]{on_eof} = $_[1];
649}
650
651=item $handle->on_timeout ($cb)
652
653=item $handle->on_rtimeout ($cb)
654
655=item $handle->on_wtimeout ($cb)
656
657Replace the current C<on_timeout>, C<on_rtimeout> or C<on_wtimeout>
658callback, or disables the callback (but not the timeout) if C<$cb> =
659C<undef>. See the C<timeout> constructor argument and method.
660
661=cut
662
663# see below
664
665=item $handle->autocork ($boolean)
666
667Enables or disables the current autocork behaviour (see C<autocork>
668constructor argument). Changes will only take effect on the next write.
669
670=cut
671
672sub autocork {
673 $_[0]{autocork} = $_[1];
674}
675
676=item $handle->no_delay ($boolean)
677
678Enables or disables the C<no_delay> setting (see constructor argument of
679the same name for details).
680
681=cut
682
683sub no_delay {
684 $_[0]{no_delay} = $_[1];
685
686 setsockopt $_[0]{fh}, Socket::IPPROTO_TCP (), Socket::TCP_NODELAY (), int $_[1]
687 if $_[0]{fh};
688}
689
690=item $handle->keepalive ($boolean)
691
692Enables or disables the C<keepalive> setting (see constructor argument of
693the same name for details).
694
695=cut
696
697sub keepalive {
698 $_[0]{keepalive} = $_[1];
699
700 eval {
701 local $SIG{__DIE__};
702 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
703 if $_[0]{fh};
704 };
705}
706
707=item $handle->oobinline ($boolean)
708
709Enables or disables the C<oobinline> setting (see constructor argument of
710the same name for details).
711
712=cut
713
714sub oobinline {
715 $_[0]{oobinline} = $_[1];
716
717 eval {
718 local $SIG{__DIE__};
719 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_OOBINLINE (), int $_[1]
720 if $_[0]{fh};
721 };
722}
723
724=item $handle->keepalive ($boolean)
725
726Enables or disables the C<keepalive> setting (see constructor argument of
727the same name for details).
728
729=cut
730
731sub keepalive {
732 $_[0]{keepalive} = $_[1];
733
734 eval {
735 local $SIG{__DIE__};
736 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
737 if $_[0]{fh};
738 };
739}
740
741=item $handle->on_starttls ($cb)
742
743Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
744
745=cut
746
747sub on_starttls {
748 $_[0]{on_starttls} = $_[1];
749}
750
751=item $handle->on_stoptls ($cb)
752
753Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
754
755=cut
756
757sub on_stoptls {
758 $_[0]{on_stoptls} = $_[1];
759}
760
761=item $handle->rbuf_max ($max_octets)
762
763Configures the C<rbuf_max> setting (C<undef> disables it).
764
765=item $handle->wbuf_max ($max_octets)
766
767Configures the C<wbuf_max> setting (C<undef> disables it).
768
769=cut
770
771sub rbuf_max {
772 $_[0]{rbuf_max} = $_[1];
773}
774
775sub wbuf_max {
776 $_[0]{wbuf_max} = $_[1];
777}
778
779#############################################################################
780
781=item $handle->timeout ($seconds)
782
783=item $handle->rtimeout ($seconds)
784
785=item $handle->wtimeout ($seconds)
786
787Configures (or disables) the inactivity timeout.
788
789The timeout will be checked instantly, so this method might destroy the
790handle before it returns.
791
792=item $handle->timeout_reset
793
794=item $handle->rtimeout_reset
795
796=item $handle->wtimeout_reset
797
798Reset the activity timeout, as if data was received or sent.
799
800These methods are cheap to call.
801
802=cut
803
804for my $dir ("", "r", "w") {
805 my $timeout = "${dir}timeout";
806 my $tw = "_${dir}tw";
807 my $on_timeout = "on_${dir}timeout";
808 my $activity = "_${dir}activity";
809 my $cb;
810
811 *$on_timeout = sub {
812 $_[0]{$on_timeout} = $_[1];
813 };
814
815 *$timeout = sub {
816 my ($self, $new_value) = @_;
817
818 $new_value >= 0
819 or Carp::croak "AnyEvent::Handle->$timeout called with negative timeout ($new_value), caught";
820
821 $self->{$timeout} = $new_value;
822 delete $self->{$tw}; &$cb;
823 };
824
825 *{"${dir}timeout_reset"} = sub {
826 $_[0]{$activity} = AE::now;
827 };
828
829 # main workhorse:
830 # reset the timeout watcher, as neccessary
831 # also check for time-outs
832 $cb = sub {
833 my ($self) = @_;
834
835 if ($self->{$timeout} && $self->{fh}) {
836 my $NOW = AE::now;
837
838 # when would the timeout trigger?
839 my $after = $self->{$activity} + $self->{$timeout} - $NOW;
840
841 # now or in the past already?
842 if ($after <= 0) {
843 $self->{$activity} = $NOW;
844
845 if ($self->{$on_timeout}) {
846 $self->{$on_timeout}($self);
847 } else {
848 $self->_error (Errno::ETIMEDOUT);
849 }
850
851 # callback could have changed timeout value, optimise
852 return unless $self->{$timeout};
853
854 # calculate new after
855 $after = $self->{$timeout};
856 }
857
858 Scalar::Util::weaken $self;
859 return unless $self; # ->error could have destroyed $self
860
861 $self->{$tw} ||= AE::timer $after, 0, sub {
862 delete $self->{$tw};
863 $cb->($self);
36 chomp; 864 };
37 print "Line: $_"; 865 } else {
866 delete $self->{$tw};
867 }
868 }
869}
870
871#############################################################################
872
873=back
874
875=head2 WRITE QUEUE
876
877AnyEvent::Handle manages two queues per handle, one for writing and one
878for reading.
879
880The write queue is very simple: you can add data to its end, and
881AnyEvent::Handle will automatically try to get rid of it for you.
882
883When data could be written and the write buffer is shorter then the low
884water mark, the C<on_drain> callback will be invoked.
885
886=over 4
887
888=item $handle->on_drain ($cb)
889
890Sets the C<on_drain> callback or clears it (see the description of
891C<on_drain> in the constructor).
892
893This method may invoke callbacks (and therefore the handle might be
894destroyed after it returns).
895
896=cut
897
898sub on_drain {
899 my ($self, $cb) = @_;
900
901 $self->{on_drain} = $cb;
902
903 $cb->($self)
904 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
905}
906
907=item $handle->push_write ($data)
908
909Queues the given scalar to be written. You can push as much data as
910you want (only limited by the available memory and C<wbuf_max>), as
911C<AnyEvent::Handle> buffers it independently of the kernel.
912
913This method may invoke callbacks (and therefore the handle might be
914destroyed after it returns).
915
916=cut
917
918sub _drain_wbuf {
919 my ($self) = @_;
920
921 if (!$self->{_ww} && length $self->{wbuf}) {
922
923 Scalar::Util::weaken $self;
924
925 my $cb = sub {
926 my $len = syswrite $self->{fh}, $self->{wbuf};
927
928 if (defined $len) {
929 substr $self->{wbuf}, 0, $len, "";
930
931 $self->{_activity} = $self->{_wactivity} = AE::now;
932
933 $self->{on_drain}($self)
934 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
935 && $self->{on_drain};
936
937 delete $self->{_ww} unless length $self->{wbuf};
938 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
939 $self->_error ($!, 1);
940 }
941 };
942
943 # try to write data immediately
944 $cb->() unless $self->{autocork};
945
946 # if still data left in wbuf, we need to poll
947 $self->{_ww} = AE::io $self->{fh}, 1, $cb
948 if length $self->{wbuf};
949
950 if (
951 defined $self->{wbuf_max}
952 && $self->{wbuf_max} < length $self->{wbuf}
953 ) {
954 $self->_error (Errno::ENOSPC, 1), return;
955 }
956 };
957}
958
959our %WH;
960
961# deprecated
962sub register_write_type($$) {
963 $WH{$_[0]} = $_[1];
964}
965
966sub push_write {
967 my $self = shift;
968
969 if (@_ > 1) {
970 my $type = shift;
971
972 @_ = ($WH{$type} ||= _load_func "$type\::anyevent_write_type"
973 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_write")
974 ->($self, @_);
975 }
976
977 # we downgrade here to avoid hard-to-track-down bugs,
978 # and diagnose the problem earlier and better.
979
980 if ($self->{tls}) {
981 utf8::downgrade $self->{_tls_wbuf} .= $_[0];
982 &_dotls ($self) if $self->{fh};
983 } else {
984 utf8::downgrade $self->{wbuf} .= $_[0];
985 $self->_drain_wbuf if $self->{fh};
986 }
987}
988
989=item $handle->push_write (type => @args)
990
991Instead of formatting your data yourself, you can also let this module
992do the job by specifying a type and type-specific arguments. You
993can also specify the (fully qualified) name of a package, in which
994case AnyEvent tries to load the package and then expects to find the
995C<anyevent_write_type> function inside (see "custom write types", below).
996
997Predefined types are (if you have ideas for additional types, feel free to
998drop by and tell us):
999
1000=over 4
1001
1002=item netstring => $string
1003
1004Formats the given value as netstring
1005(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
1006
1007=cut
1008
1009register_write_type netstring => sub {
1010 my ($self, $string) = @_;
1011
1012 (length $string) . ":$string,"
1013};
1014
1015=item packstring => $format, $data
1016
1017An octet string prefixed with an encoded length. The encoding C<$format>
1018uses the same format as a Perl C<pack> format, but must specify a single
1019integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1020optional C<!>, C<< < >> or C<< > >> modifier).
1021
1022=cut
1023
1024register_write_type packstring => sub {
1025 my ($self, $format, $string) = @_;
1026
1027 pack "$format/a*", $string
1028};
1029
1030=item json => $array_or_hashref
1031
1032Encodes the given hash or array reference into a JSON object. Unless you
1033provide your own JSON object, this means it will be encoded to JSON text
1034in UTF-8.
1035
1036JSON objects (and arrays) are self-delimiting, so you can write JSON at
1037one end of a handle and read them at the other end without using any
1038additional framing.
1039
1040The generated JSON text is guaranteed not to contain any newlines: While
1041this module doesn't need delimiters after or between JSON texts to be
1042able to read them, many other languages depend on that.
1043
1044A simple RPC protocol that interoperates easily with others is to send
1045JSON arrays (or objects, although arrays are usually the better choice as
1046they mimic how function argument passing works) and a newline after each
1047JSON text:
1048
1049 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
1050 $handle->push_write ("\012");
1051
1052An AnyEvent::Handle receiver would simply use the C<json> read type and
1053rely on the fact that the newline will be skipped as leading whitespace:
1054
1055 $handle->push_read (json => sub { my $array = $_[1]; ... });
1056
1057Other languages could read single lines terminated by a newline and pass
1058this line into their JSON decoder of choice.
1059
1060=cut
1061
1062sub json_coder() {
1063 eval { require JSON::XS; JSON::XS->new->utf8 }
1064 || do { require JSON; JSON->new->utf8 }
1065}
1066
1067register_write_type json => sub {
1068 my ($self, $ref) = @_;
1069
1070 my $json = $self->{json} ||= json_coder;
1071
1072 $json->encode ($ref)
1073};
1074
1075=item storable => $reference
1076
1077Freezes the given reference using L<Storable> and writes it to the
1078handle. Uses the C<nfreeze> format.
1079
1080=cut
1081
1082register_write_type storable => sub {
1083 my ($self, $ref) = @_;
1084
1085 require Storable;
1086
1087 pack "w/a*", Storable::nfreeze ($ref)
1088};
1089
1090=back
1091
1092=item $handle->push_shutdown
1093
1094Sometimes you know you want to close the socket after writing your data
1095before it was actually written. One way to do that is to replace your
1096C<on_drain> handler by a callback that shuts down the socket (and set
1097C<low_water_mark> to C<0>). This method is a shorthand for just that, and
1098replaces the C<on_drain> callback with:
1099
1100 sub { shutdown $_[0]{fh}, 1 }
1101
1102This simply shuts down the write side and signals an EOF condition to the
1103the peer.
1104
1105You can rely on the normal read queue and C<on_eof> handling
1106afterwards. This is the cleanest way to close a connection.
1107
1108This method may invoke callbacks (and therefore the handle might be
1109destroyed after it returns).
1110
1111=cut
1112
1113sub push_shutdown {
1114 my ($self) = @_;
1115
1116 delete $self->{low_water_mark};
1117 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
1118}
1119
1120=item custom write types - Package::anyevent_write_type $handle, @args
1121
1122Instead of one of the predefined types, you can also specify the name of
1123a package. AnyEvent will try to load the package and then expects to find
1124a function named C<anyevent_write_type> inside. If it isn't found, it
1125progressively tries to load the parent package until it either finds the
1126function (good) or runs out of packages (bad).
1127
1128Whenever the given C<type> is used, C<push_write> will the function with
1129the handle object and the remaining arguments.
1130
1131The function is supposed to return a single octet string that will be
1132appended to the write buffer, so you cna mentally treat this function as a
1133"arguments to on-the-wire-format" converter.
1134
1135Example: implement a custom write type C<join> that joins the remaining
1136arguments using the first one.
1137
1138 $handle->push_write (My::Type => " ", 1,2,3);
1139
1140 # uses the following package, which can be defined in the "My::Type" or in
1141 # the "My" modules to be auto-loaded, or just about anywhere when the
1142 # My::Type::anyevent_write_type is defined before invoking it.
1143
1144 package My::Type;
1145
1146 sub anyevent_write_type {
1147 my ($handle, $delim, @args) = @_;
1148
1149 join $delim, @args
1150 }
1151
1152=cut
1153
1154#############################################################################
1155
1156=back
1157
1158=head2 READ QUEUE
1159
1160AnyEvent::Handle manages two queues per handle, one for writing and one
1161for reading.
1162
1163The read queue is more complex than the write queue. It can be used in two
1164ways, the "simple" way, using only C<on_read> and the "complex" way, using
1165a queue.
1166
1167In the simple case, you just install an C<on_read> callback and whenever
1168new data arrives, it will be called. You can then remove some data (if
1169enough is there) from the read buffer (C<< $handle->rbuf >>). Or you can
1170leave the data there if you want to accumulate more (e.g. when only a
1171partial message has been received so far), or change the read queue with
1172e.g. C<push_read>.
1173
1174In the more complex case, you want to queue multiple callbacks. In this
1175case, AnyEvent::Handle will call the first queued callback each time new
1176data arrives (also the first time it is queued) and remove it when it has
1177done its job (see C<push_read>, below).
1178
1179This way you can, for example, push three line-reads, followed by reading
1180a chunk of data, and AnyEvent::Handle will execute them in order.
1181
1182Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
1183the specified number of bytes which give an XML datagram.
1184
1185 # in the default state, expect some header bytes
1186 $handle->on_read (sub {
1187 # some data is here, now queue the length-header-read (4 octets)
1188 shift->unshift_read (chunk => 4, sub {
1189 # header arrived, decode
1190 my $len = unpack "N", $_[1];
1191
1192 # now read the payload
1193 shift->unshift_read (chunk => $len, sub {
1194 my $xml = $_[1];
1195 # handle xml
1196 });
1197 });
1198 });
1199
1200Example 2: Implement a client for a protocol that replies either with "OK"
1201and another line or "ERROR" for the first request that is sent, and 64
1202bytes for the second request. Due to the availability of a queue, we can
1203just pipeline sending both requests and manipulate the queue as necessary
1204in the callbacks.
1205
1206When the first callback is called and sees an "OK" response, it will
1207C<unshift> another line-read. This line-read will be queued I<before> the
120864-byte chunk callback.
1209
1210 # request one, returns either "OK + extra line" or "ERROR"
1211 $handle->push_write ("request 1\015\012");
1212
1213 # we expect "ERROR" or "OK" as response, so push a line read
1214 $handle->push_read (line => sub {
1215 # if we got an "OK", we have to _prepend_ another line,
1216 # so it will be read before the second request reads its 64 bytes
1217 # which are already in the queue when this callback is called
1218 # we don't do this in case we got an error
1219 if ($_[1] eq "OK") {
1220 $_[0]->unshift_read (line => sub {
1221 my $response = $_[1];
1222 ...
1223 });
38 } 1224 }
39 }); 1225 });
40 1226
41 # or use the constructor to pass the callback: 1227 # request two, simply returns 64 octets
1228 $handle->push_write ("request 2\015\012");
42 1229
43 my $ae_fh2 = 1230 # simply read 64 bytes, always
44 AnyEvent::Handle->new ( 1231 $handle->push_read (chunk => 64, sub {
45 fh => \*STDIN, 1232 my $response = $_[1];
46 on_eof => sub { 1233 ...
47 $cv->broadcast; 1234 });
48 }, 1235
49 on_readline => sub { 1236=over 4
50 my ($ae_fh, @lines) = @_; 1237
51 for (@lines) { 1238=cut
52 chomp; 1239
53 print "Line: $_"; 1240sub _drain_rbuf {
1241 my ($self) = @_;
1242
1243 # avoid recursion
1244 return if $self->{_skip_drain_rbuf};
1245 local $self->{_skip_drain_rbuf} = 1;
1246
1247 while () {
1248 # we need to use a separate tls read buffer, as we must not receive data while
1249 # we are draining the buffer, and this can only happen with TLS.
1250 $self->{rbuf} .= delete $self->{_tls_rbuf}
1251 if exists $self->{_tls_rbuf};
1252
1253 my $len = length $self->{rbuf};
1254
1255 if (my $cb = shift @{ $self->{_queue} }) {
1256 unless ($cb->($self)) {
1257 # no progress can be made
1258 # (not enough data and no data forthcoming)
1259 $self->_error (Errno::EPIPE, 1), return
1260 if $self->{_eof};
1261
1262 unshift @{ $self->{_queue} }, $cb;
54 } 1263 last;
55 } 1264 }
56 );
57
58 $cv->wait;
59
60=head1 DESCRIPTION
61
62This module is a helper module to make it easier to do non-blocking I/O
63on filehandles (and sockets, see L<AnyEvent::Socket>).
64
65The event loop is provided by L<AnyEvent>.
66
67=head1 METHODS
68
69=over 4
70
71=item B<new (%args)>
72
73The constructor has these arguments:
74
75=over 4
76
77=item fh => $filehandle
78
79The filehandle this L<AnyEvent::Handle> object will operate on.
80
81NOTE: The filehandle will be set to non-blocking.
82
83=item read_block_size => $size
84
85The default read block size use for reads via the C<on_read>
86method.
87
88=item on_read => $cb
89
90=item on_eof => $cb
91
92=item on_error => $cb
93
94These are shortcuts, that will call the corresponding method and set the callback to C<$cb>.
95
96=item on_readline => $cb
97
98The C<readlines> method is called with the default seperator and C<$cb> as callback
99for you.
100
101=back
102
103=cut
104
105sub new {
106 my $this = shift;
107 my $class = ref($this) || $this;
108 my $self = {
109 read_block_size => 4096,
110 rbuf => '',
111 @_
112 };
113 bless $self, $class;
114
115 $self->{fh}->blocking (0) if $self->{fh};
116
117 if ($self->{on_read}) {
118 $self->on_read ($self->{on_read});
119
120 } elsif ($self->{on_readline}) { 1265 } elsif ($self->{on_read}) {
121 $self->readlines ($self->{on_readline}); 1266 last unless $len;
122 1267
1268 $self->{on_read}($self);
1269
1270 if (
1271 $len == length $self->{rbuf} # if no data has been consumed
1272 && !@{ $self->{_queue} } # and the queue is still empty
1273 && $self->{on_read} # but we still have on_read
1274 ) {
1275 # no further data will arrive
1276 # so no progress can be made
1277 $self->_error (Errno::EPIPE, 1), return
1278 if $self->{_eof};
1279
1280 last; # more data might arrive
1281 }
1282 } else {
1283 # read side becomes idle
1284 delete $self->{_rw} unless $self->{tls};
1285 last;
1286 }
1287 }
1288
123 } elsif ($self->{on_eof}) { 1289 if ($self->{_eof}) {
124 $self->on_eof ($self->{on_eof}); 1290 $self->{on_eof}
1291 ? $self->{on_eof}($self)
1292 : $self->_error (0, 1, "Unexpected end-of-file");
125 1293
126 } elsif ($self->{on_error}) { 1294 return;
127 $self->on_eof ($self->{on_error});
128 } 1295 }
129 1296
130 return $self 1297 if (
131} 1298 defined $self->{rbuf_max}
1299 && $self->{rbuf_max} < length $self->{rbuf}
1300 ) {
1301 $self->_error (Errno::ENOSPC, 1), return;
1302 }
132 1303
133=item B<fh> 1304 # may need to restart read watcher
1305 unless ($self->{_rw}) {
1306 $self->start_read
1307 if $self->{on_read} || @{ $self->{_queue} };
1308 }
1309}
134 1310
135This method returns the filehandle of the L<AnyEvent::Handle> object. 1311=item $handle->on_read ($cb)
136 1312
137=cut 1313This replaces the currently set C<on_read> callback, or clears it (when
1314the new callback is C<undef>). See the description of C<on_read> in the
1315constructor.
138 1316
139sub fh { $_[0]->{fh} } 1317This method may invoke callbacks (and therefore the handle might be
140 1318destroyed after it returns).
141=item B<on_read ($callback)>
142
143This method installs a C<$callback> that will be called
144when new data arrived. You can access the read buffer via the C<rbuf>
145method (see below).
146
147The first argument of the C<$callback> will be the L<AnyEvent::Handle> object.
148 1319
149=cut 1320=cut
150 1321
151sub on_read { 1322sub on_read {
152 my ($self, $cb) = @_; 1323 my ($self, $cb) = @_;
1324
153 $self->{on_read} = $cb; 1325 $self->{on_read} = $cb;
1326 $self->_drain_rbuf if $cb;
1327}
154 1328
155 unless (defined $self->{on_read}) { 1329=item $handle->rbuf
156 delete $self->{on_read_w}; 1330
1331Returns the read buffer (as a modifiable lvalue). You can also access the
1332read buffer directly as the C<< ->{rbuf} >> member, if you want (this is
1333much faster, and no less clean).
1334
1335The only operation allowed on the read buffer (apart from looking at it)
1336is removing data from its beginning. Otherwise modifying or appending to
1337it is not allowed and will lead to hard-to-track-down bugs.
1338
1339NOTE: The read buffer should only be used or modified in the C<on_read>
1340callback or when C<push_read> or C<unshift_read> are used with a single
1341callback (i.e. untyped). Typed C<push_read> and C<unshift_read> methods
1342will manage the read buffer on their own.
1343
1344=cut
1345
1346sub rbuf : lvalue {
1347 $_[0]{rbuf}
1348}
1349
1350=item $handle->push_read ($cb)
1351
1352=item $handle->unshift_read ($cb)
1353
1354Append the given callback to the end of the queue (C<push_read>) or
1355prepend it (C<unshift_read>).
1356
1357The callback is called each time some additional read data arrives.
1358
1359It must check whether enough data is in the read buffer already.
1360
1361If not enough data is available, it must return the empty list or a false
1362value, in which case it will be called repeatedly until enough data is
1363available (or an error condition is detected).
1364
1365If enough data was available, then the callback must remove all data it is
1366interested in (which can be none at all) and return a true value. After returning
1367true, it will be removed from the queue.
1368
1369These methods may invoke callbacks (and therefore the handle might be
1370destroyed after it returns).
1371
1372=cut
1373
1374our %RH;
1375
1376sub register_read_type($$) {
1377 $RH{$_[0]} = $_[1];
1378}
1379
1380sub push_read {
1381 my $self = shift;
1382 my $cb = pop;
1383
1384 if (@_) {
1385 my $type = shift;
1386
1387 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1388 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_read")
1389 ->($self, $cb, @_);
1390 }
1391
1392 push @{ $self->{_queue} }, $cb;
1393 $self->_drain_rbuf;
1394}
1395
1396sub unshift_read {
1397 my $self = shift;
1398 my $cb = pop;
1399
1400 if (@_) {
1401 my $type = shift;
1402
1403 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1404 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::unshift_read")
1405 ->($self, $cb, @_);
1406 }
1407
1408 unshift @{ $self->{_queue} }, $cb;
1409 $self->_drain_rbuf;
1410}
1411
1412=item $handle->push_read (type => @args, $cb)
1413
1414=item $handle->unshift_read (type => @args, $cb)
1415
1416Instead of providing a callback that parses the data itself you can chose
1417between a number of predefined parsing formats, for chunks of data, lines
1418etc. You can also specify the (fully qualified) name of a package, in
1419which case AnyEvent tries to load the package and then expects to find the
1420C<anyevent_read_type> function inside (see "custom read types", below).
1421
1422Predefined types are (if you have ideas for additional types, feel free to
1423drop by and tell us):
1424
1425=over 4
1426
1427=item chunk => $octets, $cb->($handle, $data)
1428
1429Invoke the callback only once C<$octets> bytes have been read. Pass the
1430data read to the callback. The callback will never be called with less
1431data.
1432
1433Example: read 2 bytes.
1434
1435 $handle->push_read (chunk => 2, sub {
1436 warn "yay ", unpack "H*", $_[1];
1437 });
1438
1439=cut
1440
1441register_read_type chunk => sub {
1442 my ($self, $cb, $len) = @_;
1443
1444 sub {
1445 $len <= length $_[0]{rbuf} or return;
1446 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
1447 1
1448 }
1449};
1450
1451=item line => [$eol, ]$cb->($handle, $line, $eol)
1452
1453The callback will be called only once a full line (including the end of
1454line marker, C<$eol>) has been read. This line (excluding the end of line
1455marker) will be passed to the callback as second argument (C<$line>), and
1456the end of line marker as the third argument (C<$eol>).
1457
1458The end of line marker, C<$eol>, can be either a string, in which case it
1459will be interpreted as a fixed record end marker, or it can be a regex
1460object (e.g. created by C<qr>), in which case it is interpreted as a
1461regular expression.
1462
1463The end of line marker argument C<$eol> is optional, if it is missing (NOT
1464undef), then C<qr|\015?\012|> is used (which is good for most internet
1465protocols).
1466
1467Partial lines at the end of the stream will never be returned, as they are
1468not marked by the end of line marker.
1469
1470=cut
1471
1472register_read_type line => sub {
1473 my ($self, $cb, $eol) = @_;
1474
1475 if (@_ < 3) {
1476 # this is more than twice as fast as the generic code below
1477 sub {
1478 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1479
1480 $cb->($_[0], $1, $2);
1481 1
1482 }
1483 } else {
1484 $eol = quotemeta $eol unless ref $eol;
1485 $eol = qr|^(.*?)($eol)|s;
1486
1487 sub {
1488 $_[0]{rbuf} =~ s/$eol// or return;
1489
1490 $cb->($_[0], $1, $2);
1491 1
1492 }
1493 }
1494};
1495
1496=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1497
1498Makes a regex match against the regex object C<$accept> and returns
1499everything up to and including the match.
1500
1501Example: read a single line terminated by '\n'.
1502
1503 $handle->push_read (regex => qr<\n>, sub { ... });
1504
1505If C<$reject> is given and not undef, then it determines when the data is
1506to be rejected: it is matched against the data when the C<$accept> regex
1507does not match and generates an C<EBADMSG> error when it matches. This is
1508useful to quickly reject wrong data (to avoid waiting for a timeout or a
1509receive buffer overflow).
1510
1511Example: expect a single decimal number followed by whitespace, reject
1512anything else (not the use of an anchor).
1513
1514 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1515
1516If C<$skip> is given and not C<undef>, then it will be matched against
1517the receive buffer when neither C<$accept> nor C<$reject> match,
1518and everything preceding and including the match will be accepted
1519unconditionally. This is useful to skip large amounts of data that you
1520know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1521have to start matching from the beginning. This is purely an optimisation
1522and is usually worth it only when you expect more than a few kilobytes.
1523
1524Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1525expect the header to be very large (it isn't in practice, but...), we use
1526a skip regex to skip initial portions. The skip regex is tricky in that
1527it only accepts something not ending in either \015 or \012, as these are
1528required for the accept regex.
1529
1530 $handle->push_read (regex =>
1531 qr<\015\012\015\012>,
1532 undef, # no reject
1533 qr<^.*[^\015\012]>,
1534 sub { ... });
1535
1536=cut
1537
1538register_read_type regex => sub {
1539 my ($self, $cb, $accept, $reject, $skip) = @_;
1540
1541 my $data;
1542 my $rbuf = \$self->{rbuf};
1543
1544 sub {
1545 # accept
1546 if ($$rbuf =~ $accept) {
1547 $data .= substr $$rbuf, 0, $+[0], "";
1548 $cb->($_[0], $data);
1549 return 1;
1550 }
1551
1552 # reject
1553 if ($reject && $$rbuf =~ $reject) {
1554 $_[0]->_error (Errno::EBADMSG);
1555 }
1556
1557 # skip
1558 if ($skip && $$rbuf =~ $skip) {
1559 $data .= substr $$rbuf, 0, $+[0], "";
1560 }
1561
1562 ()
1563 }
1564};
1565
1566=item netstring => $cb->($handle, $string)
1567
1568A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
1569
1570Throws an error with C<$!> set to EBADMSG on format violations.
1571
1572=cut
1573
1574register_read_type netstring => sub {
1575 my ($self, $cb) = @_;
1576
1577 sub {
1578 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1579 if ($_[0]{rbuf} =~ /[^0-9]/) {
1580 $_[0]->_error (Errno::EBADMSG);
1581 }
157 return; 1582 return;
1583 }
1584
1585 my $len = $1;
1586
1587 $_[0]->unshift_read (chunk => $len, sub {
1588 my $string = $_[1];
1589 $_[0]->unshift_read (chunk => 1, sub {
1590 if ($_[1] eq ",") {
1591 $cb->($_[0], $string);
1592 } else {
1593 $_[0]->_error (Errno::EBADMSG);
1594 }
1595 });
1596 });
1597
1598 1
1599 }
1600};
1601
1602=item packstring => $format, $cb->($handle, $string)
1603
1604An octet string prefixed with an encoded length. The encoding C<$format>
1605uses the same format as a Perl C<pack> format, but must specify a single
1606integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1607optional C<!>, C<< < >> or C<< > >> modifier).
1608
1609For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1610EPP uses a prefix of C<N> (4 octtes).
1611
1612Example: read a block of data prefixed by its length in BER-encoded
1613format (very efficient).
1614
1615 $handle->push_read (packstring => "w", sub {
1616 my ($handle, $data) = @_;
158 } 1617 });
159 1618
160 $self->{on_read_w} = 1619=cut
161 AnyEvent->io (poll => 'r', fh => $self->{fh}, cb => sub { 1620
162 #d# warn "READ:[$self->{read_size}] $self->{read_block_size} : ".length ($self->{rbuf})."\n"; 1621register_read_type packstring => sub {
1622 my ($self, $cb, $format) = @_;
1623
1624 sub {
1625 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1626 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1627 or return;
1628
1629 $format = length pack $format, $len;
1630
1631 # bypass unshift if we already have the remaining chunk
1632 if ($format + $len <= length $_[0]{rbuf}) {
1633 my $data = substr $_[0]{rbuf}, $format, $len;
1634 substr $_[0]{rbuf}, 0, $format + $len, "";
1635 $cb->($_[0], $data);
1636 } else {
1637 # remove prefix
1638 substr $_[0]{rbuf}, 0, $format, "";
1639
1640 # read remaining chunk
1641 $_[0]->unshift_read (chunk => $len, $cb);
1642 }
1643
1644 1
1645 }
1646};
1647
1648=item json => $cb->($handle, $hash_or_arrayref)
1649
1650Reads a JSON object or array, decodes it and passes it to the
1651callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1652
1653If a C<json> object was passed to the constructor, then that will be used
1654for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1655
1656This read type uses the incremental parser available with JSON version
16572.09 (and JSON::XS version 2.2) and above. You have to provide a
1658dependency on your own: this module will load the JSON module, but
1659AnyEvent does not depend on it itself.
1660
1661Since JSON texts are fully self-delimiting, the C<json> read and write
1662types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1663the C<json> write type description, above, for an actual example.
1664
1665=cut
1666
1667register_read_type json => sub {
1668 my ($self, $cb) = @_;
1669
1670 my $json = $self->{json} ||= json_coder;
1671
1672 my $data;
163 my $rbuf_len = length $self->{rbuf}; 1673 my $rbuf = \$self->{rbuf};
164 my $l; 1674
1675 sub {
1676 my $ref = eval { $json->incr_parse ($_[0]{rbuf}) };
1677
1678 if ($ref) {
1679 $_[0]{rbuf} = $json->incr_text;
1680 $json->incr_text = "";
1681 $cb->($_[0], $ref);
1682
1683 1
1684 } elsif ($@) {
1685 # error case
1686 $json->incr_skip;
1687
1688 $_[0]{rbuf} = $json->incr_text;
1689 $json->incr_text = "";
1690
1691 $_[0]->_error (Errno::EBADMSG);
1692
1693 ()
1694 } else {
1695 $_[0]{rbuf} = "";
1696
1697 ()
1698 }
1699 }
1700};
1701
1702=item storable => $cb->($handle, $ref)
1703
1704Deserialises a L<Storable> frozen representation as written by the
1705C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1706data).
1707
1708Raises C<EBADMSG> error if the data could not be decoded.
1709
1710=cut
1711
1712register_read_type storable => sub {
1713 my ($self, $cb) = @_;
1714
1715 require Storable;
1716
1717 sub {
1718 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1719 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1720 or return;
1721
1722 my $format = length pack "w", $len;
1723
1724 # bypass unshift if we already have the remaining chunk
1725 if ($format + $len <= length $_[0]{rbuf}) {
1726 my $data = substr $_[0]{rbuf}, $format, $len;
1727 substr $_[0]{rbuf}, 0, $format + $len, "";
1728 $cb->($_[0], Storable::thaw ($data));
1729 } else {
1730 # remove prefix
1731 substr $_[0]{rbuf}, 0, $format, "";
1732
1733 # read remaining chunk
1734 $_[0]->unshift_read (chunk => $len, sub {
1735 if (my $ref = eval { Storable::thaw ($_[1]) }) {
1736 $cb->($_[0], $ref);
1737 } else {
1738 $_[0]->_error (Errno::EBADMSG);
1739 }
1740 });
1741 }
1742
1743 1
1744 }
1745};
1746
1747=back
1748
1749=item custom read types - Package::anyevent_read_type $handle, $cb, @args
1750
1751Instead of one of the predefined types, you can also specify the name
1752of a package. AnyEvent will try to load the package and then expects to
1753find a function named C<anyevent_read_type> inside. If it isn't found, it
1754progressively tries to load the parent package until it either finds the
1755function (good) or runs out of packages (bad).
1756
1757Whenever this type is used, C<push_read> will invoke the function with the
1758handle object, the original callback and the remaining arguments.
1759
1760The function is supposed to return a callback (usually a closure) that
1761works as a plain read callback (see C<< ->push_read ($cb) >>), so you can
1762mentally treat the function as a "configurable read type to read callback"
1763converter.
1764
1765It should invoke the original callback when it is done reading (remember
1766to pass C<$handle> as first argument as all other callbacks do that,
1767although there is no strict requirement on this).
1768
1769For examples, see the source of this module (F<perldoc -m
1770AnyEvent::Handle>, search for C<register_read_type>)).
1771
1772=item $handle->stop_read
1773
1774=item $handle->start_read
1775
1776In rare cases you actually do not want to read anything from the
1777socket. In this case you can call C<stop_read>. Neither C<on_read> nor
1778any queued callbacks will be executed then. To start reading again, call
1779C<start_read>.
1780
1781Note that AnyEvent::Handle will automatically C<start_read> for you when
1782you change the C<on_read> callback or push/unshift a read callback, and it
1783will automatically C<stop_read> for you when neither C<on_read> is set nor
1784there are any read requests in the queue.
1785
1786In older versions of this module (<= 5.3), these methods had no effect,
1787as TLS does not support half-duplex connections. In current versions they
1788work as expected, as this behaviour is required to avoid certain resource
1789attacks, where the program would be forced to read (and buffer) arbitrary
1790amounts of data before being able to send some data. The drawback is that
1791some readings of the the SSL/TLS specifications basically require this
1792attack to be working, as SSL/TLS implementations might stall sending data
1793during a rehandshake.
1794
1795As a guideline, during the initial handshake, you should not stop reading,
1796and as a client, it might cause problems, depending on your applciation.
1797
1798=cut
1799
1800sub stop_read {
1801 my ($self) = @_;
1802
1803 delete $self->{_rw};
1804}
1805
1806sub start_read {
1807 my ($self) = @_;
1808
1809 unless ($self->{_rw} || $self->{_eof} || !$self->{fh}) {
1810 Scalar::Util::weaken $self;
1811
1812 $self->{_rw} = AE::io $self->{fh}, 0, sub {
1813 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
1814 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size}, length $$rbuf;
1815
1816 if ($len > 0) {
1817 $self->{_activity} = $self->{_ractivity} = AE::now;
1818
1819 if ($self->{tls}) {
1820 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1821
1822 &_dotls ($self);
1823 } else {
1824 $self->_drain_rbuf;
1825 }
1826
165 if (defined $self->{read_size}) { 1827 if ($len == $self->{read_size}) {
166 $l = sysread $self->{fh}, $self->{rbuf}, 1828 $self->{read_size} *= 2;
167 ($self->{read_size} - $rbuf_len), $rbuf_len; 1829 $self->{read_size} = $self->{max_read_size} || MAX_READ_SIZE
1830 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
1831 }
1832
1833 } elsif (defined $len) {
1834 delete $self->{_rw};
1835 $self->{_eof} = 1;
1836 $self->_drain_rbuf;
1837
1838 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
1839 return $self->_error ($!, 1);
1840 }
1841 };
1842 }
1843}
1844
1845our $ERROR_SYSCALL;
1846our $ERROR_WANT_READ;
1847
1848sub _tls_error {
1849 my ($self, $err) = @_;
1850
1851 return $self->_error ($!, 1)
1852 if $err == Net::SSLeay::ERROR_SYSCALL ();
1853
1854 my $err =Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
1855
1856 # reduce error string to look less scary
1857 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
1858
1859 if ($self->{_on_starttls}) {
1860 (delete $self->{_on_starttls})->($self, undef, $err);
1861 &_freetls;
1862 } else {
1863 &_freetls;
1864 $self->_error (Errno::EPROTO, 1, $err);
1865 }
1866}
1867
1868# poll the write BIO and send the data if applicable
1869# also decode read data if possible
1870# this is basiclaly our TLS state machine
1871# more efficient implementations are possible with openssl,
1872# but not with the buggy and incomplete Net::SSLeay.
1873sub _dotls {
1874 my ($self) = @_;
1875
1876 my $tmp;
1877
1878 if (length $self->{_tls_wbuf}) {
1879 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
1880 substr $self->{_tls_wbuf}, 0, $tmp, "";
1881 }
1882
1883 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1884 return $self->_tls_error ($tmp)
1885 if $tmp != $ERROR_WANT_READ
1886 && ($tmp != $ERROR_SYSCALL || $!);
1887 }
1888
1889 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1890 unless (length $tmp) {
1891 $self->{_on_starttls}
1892 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1893 &_freetls;
1894
1895 if ($self->{on_stoptls}) {
1896 $self->{on_stoptls}($self);
1897 return;
168 } else { 1898 } else {
169 $l = sysread $self->{fh}, $self->{rbuf}, $self->{read_block_size}, $rbuf_len; 1899 # let's treat SSL-eof as we treat normal EOF
170 }
171 #d# warn "READL $l [$self->{rbuf}]\n";
172
173 if (not defined $l) {
174 return if $! == EAGAIN || $! == EINTR;
175 $self->{on_error}->($self) if $self->{on_error};
176 delete $self->{on_read_w}; 1900 delete $self->{_rw};
177
178 } elsif ($l == 0) {
179 $self->{on_eof}->($self) if $self->{on_eof};
180 delete $self->{on_read_w};
181
182 } else {
183 $self->{on_read}->($self);
184 }
185 });
186}
187
188=item B<on_error ($callback)>
189
190Whenever a read or write operation resulted in an error the C<$callback>
191will be called.
192
193The first argument of C<$callback> will be the L<AnyEvent::Handle> object itself.
194The error is given as errno in C<$!>.
195
196=cut
197
198sub on_error {
199 $_[0]->{on_error} = $_[1];
200}
201
202=item B<on_eof ($callback)>
203
204Installs the C<$callback> that will be called when the end of file is
205encountered in a read operation this C<$callback> will be called. The first
206argument will be the L<AnyEvent::Handle> object itself.
207
208=cut
209
210sub on_eof {
211 $_[0]->{on_eof} = $_[1];
212}
213
214=item B<rbuf>
215
216Returns a reference to the read buffer.
217
218NOTE: The read buffer should only be used or modified if the C<on_read>
219method is used directly. The C<read> and C<readlines> methods will provide
220the read data to their callbacks.
221
222=cut
223
224sub rbuf : lvalue {
225 $_[0]->{rbuf}
226}
227
228=item B<read ($len, $callback)>
229
230Will read exactly C<$len> bytes from the filehandle and call the C<$callback>
231if done so. The first argument to the C<$callback> will be the L<AnyEvent::Handle>
232object itself and the second argument the read data.
233
234NOTE: This method will override any callbacks installed via the C<on_read> method.
235
236=cut
237
238sub read {
239 my ($self, $len, $cb) = @_;
240
241 $self->{read_cb} = $cb;
242 my $old_blk_size = $self->{read_block_size};
243 $self->{read_block_size} = $len;
244
245 $self->on_read (sub {
246 #d# warn "OFOFO $len || ".length($_[0]->{rbuf})."||\n";
247
248 if ($len == length $_[0]->{rbuf}) {
249 $_[0]->{read_block_size} = $old_blk_size;
250 $_[0]->on_read (undef);
251 $_[0]->{read_cb}->($_[0], (substr $self->{rbuf}, 0, $len, ''));
252 }
253 });
254}
255
256=item B<readlines ($callback)>
257
258=item B<readlines ($sep, $callback)>
259
260This method will read lines from the filehandle, seperated by C<$sep> or C<"\n">
261if C<$sep> is not provided. C<$sep> will be used as part of a regex, so it can be
262a regex itself and won't be quoted!
263
264The C<$callback> will be called when at least one
265line could be read. The first argument to the C<$callback> will be the L<AnyEvent::Handle>
266object itself and the rest of the arguments will be the read lines.
267
268NOTE: This method will override any callbacks installed via the C<on_read> method.
269
270=cut
271
272sub readlines {
273 my ($self, $NL, $cb) = @_;
274
275 if (ref $NL) {
276 $cb = $NL;
277 $NL = "\n";
278 }
279
280 $self->{on_readline} = $cb;
281
282 $self->on_read (sub {
283 my @lines;
284 push @lines, $1 while $_[0]->{rbuf} =~ s/(.*)$NL//;
285 $self->{on_readline}->($_[0], @lines);
286 });
287}
288
289=item B<write ($data)>
290
291=item B<write ($callback)>
292
293=item B<write ($data, $callback)>
294
295This method will write C<$data> to the filehandle and call the C<$callback>
296afterwards. If only C<$callback> is provided it will be called when the
297write buffer becomes empty the next time (or immediately if it already is empty).
298
299=cut
300
301sub write {
302 my ($self, $data, $cb) = @_;
303 if (ref $data) { $cb = $data; undef $data }
304 push @{$self->{write_bufs}}, [$data, $cb];
305 $self->_check_writer;
306}
307
308sub _check_writer {
309 my ($self) = @_;
310
311 if ($self->{write_w}) {
312 unless ($self->{write_cb}) {
313 while (@{$self->{write_bufs}} && not defined $self->{write_bufs}->[0]->[1]) {
314 my $wba = shift @{$self->{write_bufs}};
315 $self->{wbuf} .= $wba->[0]; 1901 $self->{_eof} = 1;
316 } 1902 }
317 } 1903 }
318 return;
319 }
320 1904
321 my $wba = shift @{$self->{write_bufs}} 1905 $self->{_tls_rbuf} .= $tmp;
322 or return; 1906 $self->_drain_rbuf;
323 1907 $self->{tls} or return; # tls session might have gone away in callback
324 unless (defined $wba->[0]) {
325 $wba->[1]->($self) if $wba->[1];
326 $self->_check_writer;
327 return;
328 } 1908 }
329 1909
330 $self->{wbuf} = $wba->[0]; 1910 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
331 $self->{write_cb} = $wba->[1]; 1911 return $self->_tls_error ($tmp)
1912 if $tmp != $ERROR_WANT_READ
1913 && ($tmp != $ERROR_SYSCALL || $!);
332 1914
333 $self->{write_w} = 1915 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
334 AnyEvent->io (poll => 'w', fh => $self->{fh}, cb => sub { 1916 $self->{wbuf} .= $tmp;
335 my $l = syswrite $self->{fh}, $self->{wbuf}, length $self->{wbuf}; 1917 $self->_drain_wbuf;
1918 $self->{tls} or return; # tls session might have gone away in callback
1919 }
336 1920
337 if (not defined $l) { 1921 $self->{_on_starttls}
338 return if $! == EAGAIN || $! == EINTR; 1922 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
339 delete $self->{write_w}; 1923 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
340 $self->{on_error}->($self) if $self->{on_error}; 1924}
341 1925
1926=item $handle->starttls ($tls[, $tls_ctx])
1927
1928Instead of starting TLS negotiation immediately when the AnyEvent::Handle
1929object is created, you can also do that at a later time by calling
1930C<starttls>.
1931
1932Starting TLS is currently an asynchronous operation - when you push some
1933write data and then call C<< ->starttls >> then TLS negotiation will start
1934immediately, after which the queued write data is then sent.
1935
1936The first argument is the same as the C<tls> constructor argument (either
1937C<"connect">, C<"accept"> or an existing Net::SSLeay object).
1938
1939The second argument is the optional C<AnyEvent::TLS> object that is used
1940when AnyEvent::Handle has to create its own TLS connection object, or
1941a hash reference with C<< key => value >> pairs that will be used to
1942construct a new context.
1943
1944The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1945context in C<< $handle->{tls_ctx} >> after this call and can be used or
1946changed to your liking. Note that the handshake might have already started
1947when this function returns.
1948
1949Due to bugs in OpenSSL, it might or might not be possible to do multiple
1950handshakes on the same stream. It is best to not attempt to use the
1951stream after stopping TLS.
1952
1953This method may invoke callbacks (and therefore the handle might be
1954destroyed after it returns).
1955
1956=cut
1957
1958our %TLS_CACHE; #TODO not yet documented, should we?
1959
1960sub starttls {
1961 my ($self, $tls, $ctx) = @_;
1962
1963 Carp::croak "It is an error to call starttls on an AnyEvent::Handle object while TLS is already active, caught"
1964 if $self->{tls};
1965
1966 $self->{tls} = $tls;
1967 $self->{tls_ctx} = $ctx if @_ > 2;
1968
1969 return unless $self->{fh};
1970
1971 require Net::SSLeay;
1972
1973 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
1974 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
1975
1976 $tls = delete $self->{tls};
1977 $ctx = $self->{tls_ctx};
1978
1979 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context or session
1980
1981 if ("HASH" eq ref $ctx) {
1982 require AnyEvent::TLS;
1983
1984 if ($ctx->{cache}) {
1985 my $key = $ctx+0;
1986 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
342 } else { 1987 } else {
1988 $ctx = new AnyEvent::TLS %$ctx;
1989 }
1990 }
1991
1992 $self->{tls_ctx} = $ctx || TLS_CTX ();
1993 $self->{tls} = $tls = $self->{tls_ctx}->_get_session ($tls, $self, $self->{peername});
1994
1995 # basically, this is deep magic (because SSL_read should have the same issues)
1996 # but the openssl maintainers basically said: "trust us, it just works".
1997 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
1998 # and mismaintained ssleay-module doesn't even offer them).
1999 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
2000 #
2001 # in short: this is a mess.
2002 #
2003 # note that we do not try to keep the length constant between writes as we are required to do.
2004 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
2005 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
2006 # have identity issues in that area.
2007# Net::SSLeay::CTX_set_mode ($ssl,
2008# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
2009# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
2010 Net::SSLeay::CTX_set_mode ($tls, 1|2);
2011
2012 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
2013 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
2014
2015 Net::SSLeay::BIO_write ($self->{_rbio}, $self->{rbuf});
2016 $self->{rbuf} = "";
2017
2018 Net::SSLeay::set_bio ($tls, $self->{_rbio}, $self->{_wbio});
2019
2020 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
2021 if $self->{on_starttls};
2022
2023 &_dotls; # need to trigger the initial handshake
2024 $self->start_read; # make sure we actually do read
2025}
2026
2027=item $handle->stoptls
2028
2029Shuts down the SSL connection - this makes a proper EOF handshake by
2030sending a close notify to the other side, but since OpenSSL doesn't
2031support non-blocking shut downs, it is not guaranteed that you can re-use
2032the stream afterwards.
2033
2034This method may invoke callbacks (and therefore the handle might be
2035destroyed after it returns).
2036
2037=cut
2038
2039sub stoptls {
2040 my ($self) = @_;
2041
2042 if ($self->{tls} && $self->{fh}) {
2043 Net::SSLeay::shutdown ($self->{tls});
2044
2045 &_dotls;
2046
2047# # we don't give a shit. no, we do, but we can't. no...#d#
2048# # we, we... have to use openssl :/#d#
2049# &_freetls;#d#
2050 }
2051}
2052
2053sub _freetls {
2054 my ($self) = @_;
2055
2056 return unless $self->{tls};
2057
2058 $self->{tls_ctx}->_put_session (delete $self->{tls})
2059 if $self->{tls} > 0;
2060
2061 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
2062}
2063
2064=item $handle->resettls
2065
2066This rarely-used method simply resets and TLS state on the handle, usually
2067causing data loss.
2068
2069One case where it may be useful is when you want to skip over the data in
2070the stream but you are not interested in interpreting it, so data loss is
2071no concern.
2072
2073=cut
2074
2075*resettls = \&_freetls;
2076
2077sub DESTROY {
2078 my ($self) = @_;
2079
2080 &_freetls;
2081
2082 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
2083
2084 if ($linger && length $self->{wbuf} && $self->{fh}) {
2085 my $fh = delete $self->{fh};
2086 my $wbuf = delete $self->{wbuf};
2087
2088 my @linger;
2089
2090 push @linger, AE::io $fh, 1, sub {
2091 my $len = syswrite $fh, $wbuf, length $wbuf;
2092
2093 if ($len > 0) {
343 substr $self->{wbuf}, 0, $l, ''; 2094 substr $wbuf, 0, $len, "";
344 2095 } elsif (defined $len || ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK)) {
345 if (length ($self->{wbuf}) == 0) { 2096 @linger = (); # end
346 $self->{write_cb}->($self) if $self->{write_cb};
347
348 delete $self->{write_w};
349 delete $self->{wbuf};
350 delete $self->{write_cb};
351
352 $self->_check_writer;
353 }
354 } 2097 }
2098 };
2099 push @linger, AE::timer $linger, 0, sub {
2100 @linger = ();
2101 };
2102 }
2103}
2104
2105=item $handle->destroy
2106
2107Shuts down the handle object as much as possible - this call ensures that
2108no further callbacks will be invoked and as many resources as possible
2109will be freed. Any method you will call on the handle object after
2110destroying it in this way will be silently ignored (and it will return the
2111empty list).
2112
2113Normally, you can just "forget" any references to an AnyEvent::Handle
2114object and it will simply shut down. This works in fatal error and EOF
2115callbacks, as well as code outside. It does I<NOT> work in a read or write
2116callback, so when you want to destroy the AnyEvent::Handle object from
2117within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
2118that case.
2119
2120Destroying the handle object in this way has the advantage that callbacks
2121will be removed as well, so if those are the only reference holders (as
2122is common), then one doesn't need to do anything special to break any
2123reference cycles.
2124
2125The handle might still linger in the background and write out remaining
2126data, as specified by the C<linger> option, however.
2127
2128=cut
2129
2130sub destroy {
2131 my ($self) = @_;
2132
2133 $self->DESTROY;
2134 %$self = ();
2135 bless $self, "AnyEvent::Handle::destroyed";
2136}
2137
2138sub AnyEvent::Handle::destroyed::AUTOLOAD {
2139 #nop
2140}
2141
2142=item $handle->destroyed
2143
2144Returns false as long as the handle hasn't been destroyed by a call to C<<
2145->destroy >>, true otherwise.
2146
2147Can be useful to decide whether the handle is still valid after some
2148callback possibly destroyed the handle. For example, C<< ->push_write >>,
2149C<< ->starttls >> and other methods can call user callbacks, which in turn
2150can destroy the handle, so work can be avoided by checking sometimes:
2151
2152 $hdl->starttls ("accept");
2153 return if $hdl->destroyed;
2154 $hdl->push_write (...
2155
2156Note that the call to C<push_write> will silently be ignored if the handle
2157has been destroyed, so often you can just ignore the possibility of the
2158handle being destroyed.
2159
2160=cut
2161
2162sub destroyed { 0 }
2163sub AnyEvent::Handle::destroyed::destroyed { 1 }
2164
2165=item AnyEvent::Handle::TLS_CTX
2166
2167This function creates and returns the AnyEvent::TLS object used by default
2168for TLS mode.
2169
2170The context is created by calling L<AnyEvent::TLS> without any arguments.
2171
2172=cut
2173
2174our $TLS_CTX;
2175
2176sub TLS_CTX() {
2177 $TLS_CTX ||= do {
2178 require AnyEvent::TLS;
2179
2180 new AnyEvent::TLS
2181 }
2182}
2183
2184=back
2185
2186
2187=head1 NONFREQUENTLY ASKED QUESTIONS
2188
2189=over 4
2190
2191=item I C<undef> the AnyEvent::Handle reference inside my callback and
2192still get further invocations!
2193
2194That's because AnyEvent::Handle keeps a reference to itself when handling
2195read or write callbacks.
2196
2197It is only safe to "forget" the reference inside EOF or error callbacks,
2198from within all other callbacks, you need to explicitly call the C<<
2199->destroy >> method.
2200
2201=item Why is my C<on_eof> callback never called?
2202
2203Probably because your C<on_error> callback is being called instead: When
2204you have outstanding requests in your read queue, then an EOF is
2205considered an error as you clearly expected some data.
2206
2207To avoid this, make sure you have an empty read queue whenever your handle
2208is supposed to be "idle" (i.e. connection closes are O.K.). You cna set
2209an C<on_read> handler that simply pushes the first read requests in the
2210queue.
2211
2212See also the next question, which explains this in a bit more detail.
2213
2214=item How can I serve requests in a loop?
2215
2216Most protocols consist of some setup phase (authentication for example)
2217followed by a request handling phase, where the server waits for requests
2218and handles them, in a loop.
2219
2220There are two important variants: The first (traditional, better) variant
2221handles requests until the server gets some QUIT command, causing it to
2222close the connection first (highly desirable for a busy TCP server). A
2223client dropping the connection is an error, which means this variant can
2224detect an unexpected detection close.
2225
2226To handle this case, always make sure you have a on-empty read queue, by
2227pushing the "read request start" handler on it:
2228
2229 # we assume a request starts with a single line
2230 my @start_request; @start_request = (line => sub {
2231 my ($hdl, $line) = @_;
2232
2233 ... handle request
2234
2235 # push next request read, possibly from a nested callback
2236 $hdl->push_read (@start_request);
2237 });
2238
2239 # auth done, now go into request handling loop
2240 # now push the first @start_request
2241 $hdl->push_read (@start_request);
2242
2243By always having an outstanding C<push_read>, the handle always expects
2244some data and raises the C<EPIPE> error when the connction is dropped
2245unexpectedly.
2246
2247The second variant is a protocol where the client can drop the connection
2248at any time. For TCP, this means that the server machine may run out of
2249sockets easier, and in general, it means you cnanot distinguish a protocl
2250failure/client crash from a normal connection close. Nevertheless, these
2251kinds of protocols are common (and sometimes even the best solution to the
2252problem).
2253
2254Having an outstanding read request at all times is possible if you ignore
2255C<EPIPE> errors, but this doesn't help with when the client drops the
2256connection during a request, which would still be an error.
2257
2258A better solution is to push the initial request read in an C<on_read>
2259callback. This avoids an error, as when the server doesn't expect data
2260(i.e. is idly waiting for the next request, an EOF will not raise an
2261error, but simply result in an C<on_eof> callback. It is also a bit slower
2262and simpler:
2263
2264 # auth done, now go into request handling loop
2265 $hdl->on_read (sub {
2266 my ($hdl) = @_;
2267
2268 # called each time we receive data but the read queue is empty
2269 # simply start read the request
2270
2271 $hdl->push_read (line => sub {
2272 my ($hdl, $line) = @_;
2273
2274 ... handle request
2275
2276 # do nothing special when the request has been handled, just
2277 # let the request queue go empty.
355 }); 2278 });
356} 2279 });
2280
2281=item I get different callback invocations in TLS mode/Why can't I pause
2282reading?
2283
2284Unlike, say, TCP, TLS connections do not consist of two independent
2285communication channels, one for each direction. Or put differently, the
2286read and write directions are not independent of each other: you cannot
2287write data unless you are also prepared to read, and vice versa.
2288
2289This means that, in TLS mode, you might get C<on_error> or C<on_eof>
2290callback invocations when you are not expecting any read data - the reason
2291is that AnyEvent::Handle always reads in TLS mode.
2292
2293During the connection, you have to make sure that you always have a
2294non-empty read-queue, or an C<on_read> watcher. At the end of the
2295connection (or when you no longer want to use it) you can call the
2296C<destroy> method.
2297
2298=item How do I read data until the other side closes the connection?
2299
2300If you just want to read your data into a perl scalar, the easiest way
2301to achieve this is by setting an C<on_read> callback that does nothing,
2302clearing the C<on_eof> callback and in the C<on_error> callback, the data
2303will be in C<$_[0]{rbuf}>:
2304
2305 $handle->on_read (sub { });
2306 $handle->on_eof (undef);
2307 $handle->on_error (sub {
2308 my $data = delete $_[0]{rbuf};
2309 });
2310
2311Note that this example removes the C<rbuf> member from the handle object,
2312which is not normally allowed by the API. It is expressly permitted in
2313this case only, as the handle object needs to be destroyed afterwards.
2314
2315The reason to use C<on_error> is that TCP connections, due to latencies
2316and packets loss, might get closed quite violently with an error, when in
2317fact all data has been received.
2318
2319It is usually better to use acknowledgements when transferring data,
2320to make sure the other side hasn't just died and you got the data
2321intact. This is also one reason why so many internet protocols have an
2322explicit QUIT command.
2323
2324=item I don't want to destroy the handle too early - how do I wait until
2325all data has been written?
2326
2327After writing your last bits of data, set the C<on_drain> callback
2328and destroy the handle in there - with the default setting of
2329C<low_water_mark> this will be called precisely when all data has been
2330written to the socket:
2331
2332 $handle->push_write (...);
2333 $handle->on_drain (sub {
2334 warn "all data submitted to the kernel\n";
2335 undef $handle;
2336 });
2337
2338If you just want to queue some data and then signal EOF to the other side,
2339consider using C<< ->push_shutdown >> instead.
2340
2341=item I want to contact a TLS/SSL server, I don't care about security.
2342
2343If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
2344connect to it and then create the AnyEvent::Handle with the C<tls>
2345parameter:
2346
2347 tcp_connect $host, $port, sub {
2348 my ($fh) = @_;
2349
2350 my $handle = new AnyEvent::Handle
2351 fh => $fh,
2352 tls => "connect",
2353 on_error => sub { ... };
2354
2355 $handle->push_write (...);
2356 };
2357
2358=item I want to contact a TLS/SSL server, I do care about security.
2359
2360Then you should additionally enable certificate verification, including
2361peername verification, if the protocol you use supports it (see
2362L<AnyEvent::TLS>, C<verify_peername>).
2363
2364E.g. for HTTPS:
2365
2366 tcp_connect $host, $port, sub {
2367 my ($fh) = @_;
2368
2369 my $handle = new AnyEvent::Handle
2370 fh => $fh,
2371 peername => $host,
2372 tls => "connect",
2373 tls_ctx => { verify => 1, verify_peername => "https" },
2374 ...
2375
2376Note that you must specify the hostname you connected to (or whatever
2377"peername" the protocol needs) as the C<peername> argument, otherwise no
2378peername verification will be done.
2379
2380The above will use the system-dependent default set of trusted CA
2381certificates. If you want to check against a specific CA, add the
2382C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
2383
2384 tls_ctx => {
2385 verify => 1,
2386 verify_peername => "https",
2387 ca_file => "my-ca-cert.pem",
2388 },
2389
2390=item I want to create a TLS/SSL server, how do I do that?
2391
2392Well, you first need to get a server certificate and key. You have
2393three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
2394self-signed certificate (cheap. check the search engine of your choice,
2395there are many tutorials on the net) or c) make your own CA (tinyca2 is a
2396nice program for that purpose).
2397
2398Then create a file with your private key (in PEM format, see
2399L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
2400file should then look like this:
2401
2402 -----BEGIN RSA PRIVATE KEY-----
2403 ...header data
2404 ... lots of base64'y-stuff
2405 -----END RSA PRIVATE KEY-----
2406
2407 -----BEGIN CERTIFICATE-----
2408 ... lots of base64'y-stuff
2409 -----END CERTIFICATE-----
2410
2411The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
2412specify this file as C<cert_file>:
2413
2414 tcp_server undef, $port, sub {
2415 my ($fh) = @_;
2416
2417 my $handle = new AnyEvent::Handle
2418 fh => $fh,
2419 tls => "accept",
2420 tls_ctx => { cert_file => "my-server-keycert.pem" },
2421 ...
2422
2423When you have intermediate CA certificates that your clients might not
2424know about, just append them to the C<cert_file>.
357 2425
358=back 2426=back
359 2427
2428
2429=head1 SUBCLASSING AnyEvent::Handle
2430
2431In many cases, you might want to subclass AnyEvent::Handle.
2432
2433To make this easier, a given version of AnyEvent::Handle uses these
2434conventions:
2435
2436=over 4
2437
2438=item * all constructor arguments become object members.
2439
2440At least initially, when you pass a C<tls>-argument to the constructor it
2441will end up in C<< $handle->{tls} >>. Those members might be changed or
2442mutated later on (for example C<tls> will hold the TLS connection object).
2443
2444=item * other object member names are prefixed with an C<_>.
2445
2446All object members not explicitly documented (internal use) are prefixed
2447with an underscore character, so the remaining non-C<_>-namespace is free
2448for use for subclasses.
2449
2450=item * all members not documented here and not prefixed with an underscore
2451are free to use in subclasses.
2452
2453Of course, new versions of AnyEvent::Handle may introduce more "public"
2454member variables, but that's just life. At least it is documented.
2455
2456=back
2457
360=head1 AUTHOR 2458=head1 AUTHOR
361 2459
362Robin Redeker, C<< <elmex at ta-sa.org> >> 2460Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
363 2461
364=cut 2462=cut
365 2463
3661; # End of AnyEvent::Handle 24641; # End of AnyEvent::Handle

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines