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 1.6 by elmex, Mon Apr 28 09:27:47 2008 UTC vs.
Revision 1.229 by root, Mon Feb 27 17:14:02 2012 UTC

1package AnyEvent::Handle;
2
3no 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 AE::log error => "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, $line) = @_;
35 for (@lines) { 27 say "got line <$line>";
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, $retry->())
129
130This callback is called when a connection has been successfully established.
131
132The peer's numeric host and port (the socket peername) are passed as
133parameters, together with a retry callback. At the time it is called the
134read and write queues, EOF status, TLS status and similar properties of
135the handle will have been reset.
136
137It is not allowed to use the read or write queues while the handle object
138is connecting.
139
140If, for some reason, the handle is not acceptable, calling C<$retry> will
141continue with the next connection target (in case of multi-homed hosts or
142SRV records there can be multiple connection endpoints). The C<$retry>
143callback can be invoked after the connect callback returns, i.e. one can
144start a handshake and then decide to retry with the next host if the
145handshake fails.
146
147In most cases, you should ignore the C<$retry> parameter.
148
149=item on_connect_error => $cb->($handle, $message)
150
151This callback is called when the connection could not be
152established. C<$!> will contain the relevant error code, and C<$message> a
153message describing it (usually the same as C<"$!">).
154
155If this callback isn't specified, then C<on_error> will be called with a
156fatal error instead.
157
158=back
159
160=item on_error => $cb->($handle, $fatal, $message)
161
162This is the error callback, which is called when, well, some error
163occured, such as not being able to resolve the hostname, failure to
164connect, or a read error.
165
166Some errors are fatal (which is indicated by C<$fatal> being true). On
167fatal errors the handle object will be destroyed (by a call to C<< ->
168destroy >>) after invoking the error callback (which means you are free to
169examine the handle object). Examples of fatal errors are an EOF condition
170with active (but unsatisfiable) read watchers (C<EPIPE>) or I/O errors. In
171cases where the other side can close the connection at will, it is
172often easiest to not report C<EPIPE> errors in this callback.
173
174AnyEvent::Handle tries to find an appropriate error code for you to check
175against, but in some cases (TLS errors), this does not work well. It is
176recommended to always output the C<$message> argument in human-readable
177error messages (it's usually the same as C<"$!">).
178
179Non-fatal errors can be retried by returning, but it is recommended
180to simply ignore this parameter and instead abondon the handle object
181when this callback is invoked. Examples of non-fatal errors are timeouts
182C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
183
184On entry to the callback, the value of C<$!> contains the operating
185system error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
186C<EPROTO>).
187
188While not mandatory, it is I<highly> recommended to set this callback, as
189you will not be notified of errors otherwise. The default just calls
190C<croak>.
191
192=item on_read => $cb->($handle)
193
194This sets the default read callback, which is called when data arrives
195and no read request is in the queue (unlike read queue callbacks, this
196callback will only be called when at least one octet of data is in the
197read buffer).
198
199To access (and remove data from) the read buffer, use the C<< ->rbuf >>
200method or access the C<< $handle->{rbuf} >> member directly. Note that you
201must not enlarge or modify the read buffer, you can only remove data at
202the beginning from it.
203
204You can also call C<< ->push_read (...) >> or any other function that
205modifies the read queue. Or do both. Or ...
206
207When an EOF condition is detected, AnyEvent::Handle will first try to
208feed all the remaining data to the queued callbacks and C<on_read> before
209calling the C<on_eof> callback. If no progress can be made, then a fatal
210error will be raised (with C<$!> set to C<EPIPE>).
211
212Note that, unlike requests in the read queue, an C<on_read> callback
213doesn't mean you I<require> some data: if there is an EOF and there
214are outstanding read requests then an error will be flagged. With an
215C<on_read> callback, the C<on_eof> callback will be invoked.
216
217=item on_eof => $cb->($handle)
218
219Set the callback to be called when an end-of-file condition is detected,
220i.e. in the case of a socket, when the other side has closed the
221connection cleanly, and there are no outstanding read requests in the
222queue (if there are read requests, then an EOF counts as an unexpected
223connection close and will be flagged as an error).
224
225For sockets, this just means that the other side has stopped sending data,
226you can still try to write data, and, in fact, one can return from the EOF
227callback and continue writing data, as only the read part has been shut
228down.
229
230If an EOF condition has been detected but no C<on_eof> callback has been
231set, then a fatal error will be raised with C<$!> set to <0>.
232
233=item on_drain => $cb->($handle)
234
235This sets the callback that is called once when the write buffer becomes
236empty (and immediately when the handle object is created).
237
238To append to the write buffer, use the C<< ->push_write >> method.
239
240This callback is useful when you don't want to put all of your write data
241into the queue at once, for example, when you want to write the contents
242of some file to the socket you might not want to read the whole file into
243memory and push it into the queue, but instead only read more data from
244the file when the write queue becomes empty.
245
246=item timeout => $fractional_seconds
247
248=item rtimeout => $fractional_seconds
249
250=item wtimeout => $fractional_seconds
251
252If non-zero, then these enables an "inactivity" timeout: whenever this
253many seconds pass without a successful read or write on the underlying
254file handle (or a call to C<timeout_reset>), the C<on_timeout> callback
255will be invoked (and if that one is missing, a non-fatal C<ETIMEDOUT>
256error will be raised).
257
258There are three variants of the timeouts that work independently of each
259other, for both read and write (triggered when nothing was read I<OR>
260written), just read (triggered when nothing was read), and just write:
261C<timeout>, C<rtimeout> and C<wtimeout>, with corresponding callbacks
262C<on_timeout>, C<on_rtimeout> and C<on_wtimeout>, and reset functions
263C<timeout_reset>, C<rtimeout_reset>, and C<wtimeout_reset>.
264
265Note that timeout processing is active even when you do not have any
266outstanding read or write requests: If you plan to keep the connection
267idle then you should disable the timeout temporarily or ignore the
268timeout in the corresponding C<on_timeout> callback, in which case
269AnyEvent::Handle will simply restart the timeout.
270
271Zero (the default) disables the corresponding timeout.
272
273=item on_timeout => $cb->($handle)
274
275=item on_rtimeout => $cb->($handle)
276
277=item on_wtimeout => $cb->($handle)
278
279Called whenever the inactivity timeout passes. If you return from this
280callback, then the timeout will be reset as if some activity had happened,
281so this condition is not fatal in any way.
282
283=item rbuf_max => <bytes>
284
285If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
286when the read buffer ever (strictly) exceeds this size. This is useful to
287avoid some forms of denial-of-service attacks.
288
289For example, a server accepting connections from untrusted sources should
290be configured to accept only so-and-so much data that it cannot act on
291(for example, when expecting a line, an attacker could send an unlimited
292amount of data without a callback ever being called as long as the line
293isn't finished).
294
295=item wbuf_max => <bytes>
296
297If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
298when the write buffer ever (strictly) exceeds this size. This is useful to
299avoid some forms of denial-of-service attacks.
300
301Although the units of this parameter is bytes, this is the I<raw> number
302of bytes not yet accepted by the kernel. This can make a difference when
303you e.g. use TLS, as TLS typically makes your write data larger (but it
304can also make it smaller due to compression).
305
306As an example of when this limit is useful, take a chat server that sends
307chat messages to a client. If the client does not read those in a timely
308manner then the send buffer in the server would grow unbounded.
309
310=item autocork => <boolean>
311
312When disabled (the default), C<push_write> will try to immediately
313write the data to the handle if possible. This avoids having to register
314a write watcher and wait for the next event loop iteration, but can
315be inefficient if you write multiple small chunks (on the wire, this
316disadvantage is usually avoided by your kernel's nagle algorithm, see
317C<no_delay>, but this option can save costly syscalls).
318
319When enabled, writes will always be queued till the next event loop
320iteration. This is efficient when you do many small writes per iteration,
321but less efficient when you do a single write only per iteration (or when
322the write buffer often is full). It also increases write latency.
323
324=item no_delay => <boolean>
325
326When doing small writes on sockets, your operating system kernel might
327wait a bit for more data before actually sending it out. This is called
328the Nagle algorithm, and usually it is beneficial.
329
330In some situations you want as low a delay as possible, which can be
331accomplishd by setting this option to a true value.
332
333The default is your operating system's default behaviour (most likely
334enabled). This option explicitly enables or disables it, if possible.
335
336=item keepalive => <boolean>
337
338Enables (default disable) the SO_KEEPALIVE option on the stream socket:
339normally, TCP connections have no time-out once established, so TCP
340connections, once established, can stay alive forever even when the other
341side has long gone. TCP keepalives are a cheap way to take down long-lived
342TCP connections when the other side becomes unreachable. While the default
343is OS-dependent, TCP keepalives usually kick in after around two hours,
344and, if the other side doesn't reply, take down the TCP connection some 10
345to 15 minutes later.
346
347It is harmless to specify this option for file handles that do not support
348keepalives, and enabling it on connections that are potentially long-lived
349is usually a good idea.
350
351=item oobinline => <boolean>
352
353BSD majorly fucked up the implementation of TCP urgent data. The result
354is that almost no OS implements TCP according to the specs, and every OS
355implements it slightly differently.
356
357If you want to handle TCP urgent data, then setting this flag (the default
358is enabled) gives you the most portable way of getting urgent data, by
359putting it into the stream.
360
361Since BSD emulation of OOB data on top of TCP's urgent data can have
362security implications, AnyEvent::Handle sets this flag automatically
363unless explicitly specified. Note that setting this flag after
364establishing a connection I<may> be a bit too late (data loss could
365already have occured on BSD systems), but at least it will protect you
366from most attacks.
367
368=item read_size => <bytes>
369
370The initial read block size, the number of bytes this module will try
371to read during each loop iteration. Each handle object will consume
372at least this amount of memory for the read buffer as well, so when
373handling many connections watch out for memory requirements). See also
374C<max_read_size>. Default: C<2048>.
375
376=item max_read_size => <bytes>
377
378The maximum read buffer size used by the dynamic adjustment
379algorithm: Each time AnyEvent::Handle can read C<read_size> bytes in
380one go it will double C<read_size> up to the maximum given by this
381option. Default: C<131072> or C<read_size>, whichever is higher.
382
383=item low_water_mark => <bytes>
384
385Sets the number of bytes (default: C<0>) that make up an "empty" write
386buffer: If the buffer reaches this size or gets even samller it is
387considered empty.
388
389Sometimes it can be beneficial (for performance reasons) to add data to
390the write buffer before it is fully drained, but this is a rare case, as
391the operating system kernel usually buffers data as well, so the default
392is good in almost all cases.
393
394=item linger => <seconds>
395
396If this is non-zero (default: C<3600>), the destructor of the
397AnyEvent::Handle object will check whether there is still outstanding
398write data and will install a watcher that will write this data to the
399socket. No errors will be reported (this mostly matches how the operating
400system treats outstanding data at socket close time).
401
402This will not work for partial TLS data that could not be encoded
403yet. This data will be lost. Calling the C<stoptls> method in time might
404help.
405
406=item peername => $string
407
408A string used to identify the remote site - usually the DNS hostname
409(I<not> IDN!) used to create the connection, rarely the IP address.
410
411Apart from being useful in error messages, this string is also used in TLS
412peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
413verification will be skipped when C<peername> is not specified or is
414C<undef>.
415
416=item tls => "accept" | "connect" | Net::SSLeay::SSL object
417
418When this parameter is given, it enables TLS (SSL) mode, that means
419AnyEvent will start a TLS handshake as soon as the connection has been
420established and will transparently encrypt/decrypt data afterwards.
421
422All TLS protocol errors will be signalled as C<EPROTO>, with an
423appropriate error message.
424
425TLS mode requires Net::SSLeay to be installed (it will be loaded
426automatically when you try to create a TLS handle): this module doesn't
427have a dependency on that module, so if your module requires it, you have
428to add the dependency yourself.
429
430Unlike TCP, TLS has a server and client side: for the TLS server side, use
431C<accept>, and for the TLS client side of a connection, use C<connect>
432mode.
433
434You can also provide your own TLS connection object, but you have
435to make sure that you call either C<Net::SSLeay::set_connect_state>
436or C<Net::SSLeay::set_accept_state> on it before you pass it to
437AnyEvent::Handle. Also, this module will take ownership of this connection
438object.
439
440At some future point, AnyEvent::Handle might switch to another TLS
441implementation, then the option to use your own session object will go
442away.
443
444B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
445passing in the wrong integer will lead to certain crash. This most often
446happens when one uses a stylish C<< tls => 1 >> and is surprised about the
447segmentation fault.
448
449Use the C<< ->starttls >> method if you need to start TLS negotiation later.
450
451=item tls_ctx => $anyevent_tls
452
453Use the given C<AnyEvent::TLS> object to create the new TLS connection
454(unless a connection object was specified directly). If this
455parameter is missing (or C<undef>), then AnyEvent::Handle will use
456C<AnyEvent::Handle::TLS_CTX>.
457
458Instead of an object, you can also specify a hash reference with C<< key
459=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
460new TLS context object.
461
462=item on_starttls => $cb->($handle, $success[, $error_message])
463
464This callback will be invoked when the TLS/SSL handshake has finished. If
465C<$success> is true, then the TLS handshake succeeded, otherwise it failed
466(C<on_stoptls> will not be called in this case).
467
468The session in C<< $handle->{tls} >> can still be examined in this
469callback, even when the handshake was not successful.
470
471TLS handshake failures will not cause C<on_error> to be invoked when this
472callback is in effect, instead, the error message will be passed to C<on_starttls>.
473
474Without this callback, handshake failures lead to C<on_error> being
475called as usual.
476
477Note that you cannot just call C<starttls> again in this callback. If you
478need to do that, start an zero-second timer instead whose callback can
479then call C<< ->starttls >> again.
480
481=item on_stoptls => $cb->($handle)
482
483When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
484set, then it will be invoked after freeing the TLS session. If it is not,
485then a TLS shutdown condition will be treated like a normal EOF condition
486on the handle.
487
488The session in C<< $handle->{tls} >> can still be examined in this
489callback.
490
491This callback will only be called on TLS shutdowns, not when the
492underlying handle signals EOF.
493
494=item json => JSON or JSON::XS object
495
496This is the json coder object used by the C<json> read and write types.
497
498If you don't supply it, then AnyEvent::Handle will create and use a
499suitable one (on demand), which will write and expect UTF-8 encoded JSON
500texts.
501
502Note that you are responsible to depend on the JSON module if you want to
503use this functionality, as AnyEvent does not have a dependency itself.
504
505=back
506
507=cut
508
509sub new {
510 my $class = shift;
511 my $self = bless { @_ }, $class;
512
513 if ($self->{fh}) {
514 $self->_start;
515 return unless $self->{fh}; # could be gone by now
516
517 } elsif ($self->{connect}) {
518 require AnyEvent::Socket;
519
520 $self->{peername} = $self->{connect}[0]
521 unless exists $self->{peername};
522
523 $self->{_skip_drain_rbuf} = 1;
524
525 {
526 Scalar::Util::weaken (my $self = $self);
527
528 $self->{_connect} =
529 AnyEvent::Socket::tcp_connect (
530 $self->{connect}[0],
531 $self->{connect}[1],
532 sub {
533 my ($fh, $host, $port, $retry) = @_;
534
535 delete $self->{_connect}; # no longer needed
536
537 if ($fh) {
538 $self->{fh} = $fh;
539
540 delete $self->{_skip_drain_rbuf};
541 $self->_start;
542
543 $self->{on_connect}
544 and $self->{on_connect}($self, $host, $port, sub {
545 delete @$self{qw(fh _tw _rtw _wtw _ww _rw _eof _queue rbuf _wbuf tls _tls_rbuf _tls_wbuf)};
546 $self->{_skip_drain_rbuf} = 1;
547 &$retry;
548 });
549
550 } else {
551 if ($self->{on_connect_error}) {
552 $self->{on_connect_error}($self, "$!");
553 $self->destroy if $self;
554 } else {
555 $self->_error ($!, 1);
556 }
557 }
558 },
559 sub {
560 local $self->{fh} = $_[0];
561
562 $self->{on_prepare}
563 ? $self->{on_prepare}->($self)
564 : ()
565 }
566 );
567 }
568
569 } else {
570 Carp::croak "AnyEvent::Handle: either an existing fh or the connect parameter must be specified";
571 }
572
573 $self
574}
575
576sub _start {
577 my ($self) = @_;
578
579 # too many clueless people try to use udp and similar sockets
580 # with AnyEvent::Handle, do them a favour.
581 my $type = getsockopt $self->{fh}, Socket::SOL_SOCKET (), Socket::SO_TYPE ();
582 Carp::croak "AnyEvent::Handle: only stream sockets supported, anything else will NOT work!"
583 if Socket::SOCK_STREAM () != (unpack "I", $type) && defined $type;
584
585 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
586
587 $self->{_activity} =
588 $self->{_ractivity} =
589 $self->{_wactivity} = AE::now;
590
591 $self->{read_size} ||= 2048;
592 $self->{max_read_size} = $self->{read_size}
593 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
594
595 $self->timeout (delete $self->{timeout} ) if $self->{timeout};
596 $self->rtimeout (delete $self->{rtimeout} ) if $self->{rtimeout};
597 $self->wtimeout (delete $self->{wtimeout} ) if $self->{wtimeout};
598
599 $self->no_delay (delete $self->{no_delay} ) if exists $self->{no_delay} && $self->{no_delay};
600 $self->keepalive (delete $self->{keepalive}) if exists $self->{keepalive} && $self->{keepalive};
601
602 $self->oobinline (exists $self->{oobinline} ? delete $self->{oobinline} : 1);
603
604 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
605 if $self->{tls};
606
607 $self->on_drain (delete $self->{on_drain} ) if $self->{on_drain};
608
609 $self->start_read
610 if $self->{on_read} || @{ $self->{_queue} };
611
612 $self->_drain_wbuf;
613}
614
615sub _error {
616 my ($self, $errno, $fatal, $message) = @_;
617
618 $! = $errno;
619 $message ||= "$!";
620
621 if ($self->{on_error}) {
622 $self->{on_error}($self, $fatal, $message);
623 $self->destroy if $fatal;
624 } elsif ($self->{fh} || $self->{connect}) {
625 $self->destroy;
626 Carp::croak "AnyEvent::Handle uncaught error: $message";
627 }
628}
629
630=item $fh = $handle->fh
631
632This method returns the file handle used to create the L<AnyEvent::Handle> object.
633
634=cut
635
636sub fh { $_[0]{fh} }
637
638=item $handle->on_error ($cb)
639
640Replace the current C<on_error> callback (see the C<on_error> constructor argument).
641
642=cut
643
644sub on_error {
645 $_[0]{on_error} = $_[1];
646}
647
648=item $handle->on_eof ($cb)
649
650Replace the current C<on_eof> callback (see the C<on_eof> constructor argument).
651
652=cut
653
654sub on_eof {
655 $_[0]{on_eof} = $_[1];
656}
657
658=item $handle->on_timeout ($cb)
659
660=item $handle->on_rtimeout ($cb)
661
662=item $handle->on_wtimeout ($cb)
663
664Replace the current C<on_timeout>, C<on_rtimeout> or C<on_wtimeout>
665callback, or disables the callback (but not the timeout) if C<$cb> =
666C<undef>. See the C<timeout> constructor argument and method.
667
668=cut
669
670# see below
671
672=item $handle->autocork ($boolean)
673
674Enables or disables the current autocork behaviour (see C<autocork>
675constructor argument). Changes will only take effect on the next write.
676
677=cut
678
679sub autocork {
680 $_[0]{autocork} = $_[1];
681}
682
683=item $handle->no_delay ($boolean)
684
685Enables or disables the C<no_delay> setting (see constructor argument of
686the same name for details).
687
688=cut
689
690sub no_delay {
691 $_[0]{no_delay} = $_[1];
692
693 setsockopt $_[0]{fh}, Socket::IPPROTO_TCP (), Socket::TCP_NODELAY (), int $_[1]
694 if $_[0]{fh};
695}
696
697=item $handle->keepalive ($boolean)
698
699Enables or disables the C<keepalive> setting (see constructor argument of
700the same name for details).
701
702=cut
703
704sub keepalive {
705 $_[0]{keepalive} = $_[1];
706
707 eval {
708 local $SIG{__DIE__};
709 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
710 if $_[0]{fh};
711 };
712}
713
714=item $handle->oobinline ($boolean)
715
716Enables or disables the C<oobinline> setting (see constructor argument of
717the same name for details).
718
719=cut
720
721sub oobinline {
722 $_[0]{oobinline} = $_[1];
723
724 eval {
725 local $SIG{__DIE__};
726 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_OOBINLINE (), int $_[1]
727 if $_[0]{fh};
728 };
729}
730
731=item $handle->keepalive ($boolean)
732
733Enables or disables the C<keepalive> setting (see constructor argument of
734the same name for details).
735
736=cut
737
738sub keepalive {
739 $_[0]{keepalive} = $_[1];
740
741 eval {
742 local $SIG{__DIE__};
743 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
744 if $_[0]{fh};
745 };
746}
747
748=item $handle->on_starttls ($cb)
749
750Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
751
752=cut
753
754sub on_starttls {
755 $_[0]{on_starttls} = $_[1];
756}
757
758=item $handle->on_stoptls ($cb)
759
760Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
761
762=cut
763
764sub on_stoptls {
765 $_[0]{on_stoptls} = $_[1];
766}
767
768=item $handle->rbuf_max ($max_octets)
769
770Configures the C<rbuf_max> setting (C<undef> disables it).
771
772=item $handle->wbuf_max ($max_octets)
773
774Configures the C<wbuf_max> setting (C<undef> disables it).
775
776=cut
777
778sub rbuf_max {
779 $_[0]{rbuf_max} = $_[1];
780}
781
782sub wbuf_max {
783 $_[0]{wbuf_max} = $_[1];
784}
785
786#############################################################################
787
788=item $handle->timeout ($seconds)
789
790=item $handle->rtimeout ($seconds)
791
792=item $handle->wtimeout ($seconds)
793
794Configures (or disables) the inactivity timeout.
795
796The timeout will be checked instantly, so this method might destroy the
797handle before it returns.
798
799=item $handle->timeout_reset
800
801=item $handle->rtimeout_reset
802
803=item $handle->wtimeout_reset
804
805Reset the activity timeout, as if data was received or sent.
806
807These methods are cheap to call.
808
809=cut
810
811for my $dir ("", "r", "w") {
812 my $timeout = "${dir}timeout";
813 my $tw = "_${dir}tw";
814 my $on_timeout = "on_${dir}timeout";
815 my $activity = "_${dir}activity";
816 my $cb;
817
818 *$on_timeout = sub {
819 $_[0]{$on_timeout} = $_[1];
820 };
821
822 *$timeout = sub {
823 my ($self, $new_value) = @_;
824
825 $new_value >= 0
826 or Carp::croak "AnyEvent::Handle->$timeout called with negative timeout ($new_value), caught";
827
828 $self->{$timeout} = $new_value;
829 delete $self->{$tw}; &$cb;
830 };
831
832 *{"${dir}timeout_reset"} = sub {
833 $_[0]{$activity} = AE::now;
834 };
835
836 # main workhorse:
837 # reset the timeout watcher, as neccessary
838 # also check for time-outs
839 $cb = sub {
840 my ($self) = @_;
841
842 if ($self->{$timeout} && $self->{fh}) {
843 my $NOW = AE::now;
844
845 # when would the timeout trigger?
846 my $after = $self->{$activity} + $self->{$timeout} - $NOW;
847
848 # now or in the past already?
849 if ($after <= 0) {
850 $self->{$activity} = $NOW;
851
852 if ($self->{$on_timeout}) {
853 $self->{$on_timeout}($self);
854 } else {
855 $self->_error (Errno::ETIMEDOUT);
856 }
857
858 # callback could have changed timeout value, optimise
859 return unless $self->{$timeout};
860
861 # calculate new after
862 $after = $self->{$timeout};
863 }
864
865 Scalar::Util::weaken $self;
866 return unless $self; # ->error could have destroyed $self
867
868 $self->{$tw} ||= AE::timer $after, 0, sub {
869 delete $self->{$tw};
870 $cb->($self);
36 chomp; 871 };
37 print "Line: $_"; 872 } else {
873 delete $self->{$tw};
874 }
875 }
876}
877
878#############################################################################
879
880=back
881
882=head2 WRITE QUEUE
883
884AnyEvent::Handle manages two queues per handle, one for writing and one
885for reading.
886
887The write queue is very simple: you can add data to its end, and
888AnyEvent::Handle will automatically try to get rid of it for you.
889
890When data could be written and the write buffer is shorter then the low
891water mark, the C<on_drain> callback will be invoked once.
892
893=over 4
894
895=item $handle->on_drain ($cb)
896
897Sets the C<on_drain> callback or clears it (see the description of
898C<on_drain> in the constructor).
899
900This method may invoke callbacks (and therefore the handle might be
901destroyed after it returns).
902
903=cut
904
905sub on_drain {
906 my ($self, $cb) = @_;
907
908 $self->{on_drain} = $cb;
909
910 $cb->($self)
911 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
912}
913
914=item $handle->push_write ($data)
915
916Queues the given scalar to be written. You can push as much data as
917you want (only limited by the available memory and C<wbuf_max>), as
918C<AnyEvent::Handle> buffers it independently of the kernel.
919
920This method may invoke callbacks (and therefore the handle might be
921destroyed after it returns).
922
923=cut
924
925sub _drain_wbuf {
926 my ($self) = @_;
927
928 if (!$self->{_ww} && length $self->{wbuf}) {
929
930 Scalar::Util::weaken $self;
931
932 my $cb = sub {
933 my $len = syswrite $self->{fh}, $self->{wbuf};
934
935 if (defined $len) {
936 substr $self->{wbuf}, 0, $len, "";
937
938 $self->{_activity} = $self->{_wactivity} = AE::now;
939
940 $self->{on_drain}($self)
941 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
942 && $self->{on_drain};
943
944 delete $self->{_ww} unless length $self->{wbuf};
945 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
946 $self->_error ($!, 1);
947 }
948 };
949
950 # try to write data immediately
951 $cb->() unless $self->{autocork};
952
953 # if still data left in wbuf, we need to poll
954 $self->{_ww} = AE::io $self->{fh}, 1, $cb
955 if length $self->{wbuf};
956
957 if (
958 defined $self->{wbuf_max}
959 && $self->{wbuf_max} < length $self->{wbuf}
960 ) {
961 $self->_error (Errno::ENOSPC, 1), return;
962 }
963 };
964}
965
966our %WH;
967
968# deprecated
969sub register_write_type($$) {
970 $WH{$_[0]} = $_[1];
971}
972
973sub push_write {
974 my $self = shift;
975
976 if (@_ > 1) {
977 my $type = shift;
978
979 @_ = ($WH{$type} ||= _load_func "$type\::anyevent_write_type"
980 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_write")
981 ->($self, @_);
982 }
983
984 # we downgrade here to avoid hard-to-track-down bugs,
985 # and diagnose the problem earlier and better.
986
987 if ($self->{tls}) {
988 utf8::downgrade $self->{_tls_wbuf} .= $_[0];
989 &_dotls ($self) if $self->{fh};
990 } else {
991 utf8::downgrade $self->{wbuf} .= $_[0];
992 $self->_drain_wbuf if $self->{fh};
993 }
994}
995
996=item $handle->push_write (type => @args)
997
998Instead of formatting your data yourself, you can also let this module
999do the job by specifying a type and type-specific arguments. You
1000can also specify the (fully qualified) name of a package, in which
1001case AnyEvent tries to load the package and then expects to find the
1002C<anyevent_write_type> function inside (see "custom write types", below).
1003
1004Predefined types are (if you have ideas for additional types, feel free to
1005drop by and tell us):
1006
1007=over 4
1008
1009=item netstring => $string
1010
1011Formats the given value as netstring
1012(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
1013
1014=cut
1015
1016register_write_type netstring => sub {
1017 my ($self, $string) = @_;
1018
1019 (length $string) . ":$string,"
1020};
1021
1022=item packstring => $format, $data
1023
1024An octet string prefixed with an encoded length. The encoding C<$format>
1025uses the same format as a Perl C<pack> format, but must specify a single
1026integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1027optional C<!>, C<< < >> or C<< > >> modifier).
1028
1029=cut
1030
1031register_write_type packstring => sub {
1032 my ($self, $format, $string) = @_;
1033
1034 pack "$format/a*", $string
1035};
1036
1037=item json => $array_or_hashref
1038
1039Encodes the given hash or array reference into a JSON object. Unless you
1040provide your own JSON object, this means it will be encoded to JSON text
1041in UTF-8.
1042
1043JSON objects (and arrays) are self-delimiting, so you can write JSON at
1044one end of a handle and read them at the other end without using any
1045additional framing.
1046
1047The generated JSON text is guaranteed not to contain any newlines: While
1048this module doesn't need delimiters after or between JSON texts to be
1049able to read them, many other languages depend on that.
1050
1051A simple RPC protocol that interoperates easily with others is to send
1052JSON arrays (or objects, although arrays are usually the better choice as
1053they mimic how function argument passing works) and a newline after each
1054JSON text:
1055
1056 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
1057 $handle->push_write ("\012");
1058
1059An AnyEvent::Handle receiver would simply use the C<json> read type and
1060rely on the fact that the newline will be skipped as leading whitespace:
1061
1062 $handle->push_read (json => sub { my $array = $_[1]; ... });
1063
1064Other languages could read single lines terminated by a newline and pass
1065this line into their JSON decoder of choice.
1066
1067=cut
1068
1069sub json_coder() {
1070 eval { require JSON::XS; JSON::XS->new->utf8 }
1071 || do { require JSON; JSON->new->utf8 }
1072}
1073
1074register_write_type json => sub {
1075 my ($self, $ref) = @_;
1076
1077 my $json = $self->{json} ||= json_coder;
1078
1079 $json->encode ($ref)
1080};
1081
1082=item storable => $reference
1083
1084Freezes the given reference using L<Storable> and writes it to the
1085handle. Uses the C<nfreeze> format.
1086
1087=cut
1088
1089register_write_type storable => sub {
1090 my ($self, $ref) = @_;
1091
1092 require Storable unless $Storable::VERSION;
1093
1094 pack "w/a*", Storable::nfreeze ($ref)
1095};
1096
1097=back
1098
1099=item $handle->push_shutdown
1100
1101Sometimes you know you want to close the socket after writing your data
1102before it was actually written. One way to do that is to replace your
1103C<on_drain> handler by a callback that shuts down the socket (and set
1104C<low_water_mark> to C<0>). This method is a shorthand for just that, and
1105replaces the C<on_drain> callback with:
1106
1107 sub { shutdown $_[0]{fh}, 1 }
1108
1109This simply shuts down the write side and signals an EOF condition to the
1110the peer.
1111
1112You can rely on the normal read queue and C<on_eof> handling
1113afterwards. This is the cleanest way to close a connection.
1114
1115This method may invoke callbacks (and therefore the handle might be
1116destroyed after it returns).
1117
1118=cut
1119
1120sub push_shutdown {
1121 my ($self) = @_;
1122
1123 delete $self->{low_water_mark};
1124 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
1125}
1126
1127=item custom write types - Package::anyevent_write_type $handle, @args
1128
1129Instead of one of the predefined types, you can also specify the name of
1130a package. AnyEvent will try to load the package and then expects to find
1131a function named C<anyevent_write_type> inside. If it isn't found, it
1132progressively tries to load the parent package until it either finds the
1133function (good) or runs out of packages (bad).
1134
1135Whenever the given C<type> is used, C<push_write> will the function with
1136the handle object and the remaining arguments.
1137
1138The function is supposed to return a single octet string that will be
1139appended to the write buffer, so you can mentally treat this function as a
1140"arguments to on-the-wire-format" converter.
1141
1142Example: implement a custom write type C<join> that joins the remaining
1143arguments using the first one.
1144
1145 $handle->push_write (My::Type => " ", 1,2,3);
1146
1147 # uses the following package, which can be defined in the "My::Type" or in
1148 # the "My" modules to be auto-loaded, or just about anywhere when the
1149 # My::Type::anyevent_write_type is defined before invoking it.
1150
1151 package My::Type;
1152
1153 sub anyevent_write_type {
1154 my ($handle, $delim, @args) = @_;
1155
1156 join $delim, @args
1157 }
1158
1159=cut
1160
1161#############################################################################
1162
1163=back
1164
1165=head2 READ QUEUE
1166
1167AnyEvent::Handle manages two queues per handle, one for writing and one
1168for reading.
1169
1170The read queue is more complex than the write queue. It can be used in two
1171ways, the "simple" way, using only C<on_read> and the "complex" way, using
1172a queue.
1173
1174In the simple case, you just install an C<on_read> callback and whenever
1175new data arrives, it will be called. You can then remove some data (if
1176enough is there) from the read buffer (C<< $handle->rbuf >>). Or you can
1177leave the data there if you want to accumulate more (e.g. when only a
1178partial message has been received so far), or change the read queue with
1179e.g. C<push_read>.
1180
1181In the more complex case, you want to queue multiple callbacks. In this
1182case, AnyEvent::Handle will call the first queued callback each time new
1183data arrives (also the first time it is queued) and remove it when it has
1184done its job (see C<push_read>, below).
1185
1186This way you can, for example, push three line-reads, followed by reading
1187a chunk of data, and AnyEvent::Handle will execute them in order.
1188
1189Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
1190the specified number of bytes which give an XML datagram.
1191
1192 # in the default state, expect some header bytes
1193 $handle->on_read (sub {
1194 # some data is here, now queue the length-header-read (4 octets)
1195 shift->unshift_read (chunk => 4, sub {
1196 # header arrived, decode
1197 my $len = unpack "N", $_[1];
1198
1199 # now read the payload
1200 shift->unshift_read (chunk => $len, sub {
1201 my $xml = $_[1];
1202 # handle xml
1203 });
1204 });
1205 });
1206
1207Example 2: Implement a client for a protocol that replies either with "OK"
1208and another line or "ERROR" for the first request that is sent, and 64
1209bytes for the second request. Due to the availability of a queue, we can
1210just pipeline sending both requests and manipulate the queue as necessary
1211in the callbacks.
1212
1213When the first callback is called and sees an "OK" response, it will
1214C<unshift> another line-read. This line-read will be queued I<before> the
121564-byte chunk callback.
1216
1217 # request one, returns either "OK + extra line" or "ERROR"
1218 $handle->push_write ("request 1\015\012");
1219
1220 # we expect "ERROR" or "OK" as response, so push a line read
1221 $handle->push_read (line => sub {
1222 # if we got an "OK", we have to _prepend_ another line,
1223 # so it will be read before the second request reads its 64 bytes
1224 # which are already in the queue when this callback is called
1225 # we don't do this in case we got an error
1226 if ($_[1] eq "OK") {
1227 $_[0]->unshift_read (line => sub {
1228 my $response = $_[1];
1229 ...
1230 });
38 } 1231 }
39 }); 1232 });
40 1233
41 # or use the constructor to pass the callback: 1234 # request two, simply returns 64 octets
1235 $handle->push_write ("request 2\015\012");
42 1236
43 my $ae_fh2 = 1237 # simply read 64 bytes, always
44 AnyEvent::Handle->new ( 1238 $handle->push_read (chunk => 64, sub {
45 fh => \*STDIN, 1239 my $response = $_[1];
46 on_eof => sub { 1240 ...
47 $cv->broadcast; 1241 });
48 }, 1242
49 on_readline => sub { 1243=over 4
50 my ($ae_fh, @lines) = @_; 1244
51 for (@lines) { 1245=cut
52 chomp; 1246
53 print "Line: $_"; 1247sub _drain_rbuf {
1248 my ($self) = @_;
1249
1250 # avoid recursion
1251 return if $self->{_skip_drain_rbuf};
1252 local $self->{_skip_drain_rbuf} = 1;
1253
1254 while () {
1255 # we need to use a separate tls read buffer, as we must not receive data while
1256 # we are draining the buffer, and this can only happen with TLS.
1257 $self->{rbuf} .= delete $self->{_tls_rbuf}
1258 if exists $self->{_tls_rbuf};
1259
1260 my $len = length $self->{rbuf};
1261
1262 if (my $cb = shift @{ $self->{_queue} }) {
1263 unless ($cb->($self)) {
1264 # no progress can be made
1265 # (not enough data and no data forthcoming)
1266 $self->_error (Errno::EPIPE, 1), return
1267 if $self->{_eof};
1268
1269 unshift @{ $self->{_queue} }, $cb;
54 } 1270 last;
55 } 1271 }
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}) { 1272 } elsif ($self->{on_read}) {
121 $self->readlines ($self->{on_readline}); 1273 last unless $len;
122 1274
1275 $self->{on_read}($self);
1276
1277 if (
1278 $len == length $self->{rbuf} # if no data has been consumed
1279 && !@{ $self->{_queue} } # and the queue is still empty
1280 && $self->{on_read} # but we still have on_read
1281 ) {
1282 # no further data will arrive
1283 # so no progress can be made
1284 $self->_error (Errno::EPIPE, 1), return
1285 if $self->{_eof};
1286
1287 last; # more data might arrive
1288 }
1289 } else {
1290 # read side becomes idle
1291 delete $self->{_rw} unless $self->{tls};
1292 last;
1293 }
1294 }
1295
123 } elsif ($self->{on_eof}) { 1296 if ($self->{_eof}) {
124 $self->on_eof ($self->{on_eof}); 1297 $self->{on_eof}
1298 ? $self->{on_eof}($self)
1299 : $self->_error (0, 1, "Unexpected end-of-file");
125 1300
126 } elsif ($self->{on_error}) { 1301 return;
127 $self->on_eof ($self->{on_error});
128 } 1302 }
129 1303
130 return $self 1304 if (
131} 1305 defined $self->{rbuf_max}
1306 && $self->{rbuf_max} < length $self->{rbuf}
1307 ) {
1308 $self->_error (Errno::ENOSPC, 1), return;
1309 }
132 1310
133=item B<fh> 1311 # may need to restart read watcher
1312 unless ($self->{_rw}) {
1313 $self->start_read
1314 if $self->{on_read} || @{ $self->{_queue} };
1315 }
1316}
134 1317
135This method returns the filehandle of the L<AnyEvent::Handle> object. 1318=item $handle->on_read ($cb)
136 1319
137=cut 1320This replaces the currently set C<on_read> callback, or clears it (when
1321the new callback is C<undef>). See the description of C<on_read> in the
1322constructor.
138 1323
139sub fh { $_[0]->{fh} } 1324This method may invoke callbacks (and therefore the handle might be
140 1325destroyed 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 1326
149=cut 1327=cut
150 1328
151sub on_read { 1329sub on_read {
152 my ($self, $cb) = @_; 1330 my ($self, $cb) = @_;
1331
153 $self->{on_read} = $cb; 1332 $self->{on_read} = $cb;
1333 $self->_drain_rbuf if $cb;
1334}
154 1335
155 unless (defined $self->{on_read}) { 1336=item $handle->rbuf
156 delete $self->{on_read_w}; 1337
1338Returns the read buffer (as a modifiable lvalue). You can also access the
1339read buffer directly as the C<< ->{rbuf} >> member, if you want (this is
1340much faster, and no less clean).
1341
1342The only operation allowed on the read buffer (apart from looking at it)
1343is removing data from its beginning. Otherwise modifying or appending to
1344it is not allowed and will lead to hard-to-track-down bugs.
1345
1346NOTE: The read buffer should only be used or modified in the C<on_read>
1347callback or when C<push_read> or C<unshift_read> are used with a single
1348callback (i.e. untyped). Typed C<push_read> and C<unshift_read> methods
1349will manage the read buffer on their own.
1350
1351=cut
1352
1353sub rbuf : lvalue {
1354 $_[0]{rbuf}
1355}
1356
1357=item $handle->push_read ($cb)
1358
1359=item $handle->unshift_read ($cb)
1360
1361Append the given callback to the end of the queue (C<push_read>) or
1362prepend it (C<unshift_read>).
1363
1364The callback is called each time some additional read data arrives.
1365
1366It must check whether enough data is in the read buffer already.
1367
1368If not enough data is available, it must return the empty list or a false
1369value, in which case it will be called repeatedly until enough data is
1370available (or an error condition is detected).
1371
1372If enough data was available, then the callback must remove all data it is
1373interested in (which can be none at all) and return a true value. After returning
1374true, it will be removed from the queue.
1375
1376These methods may invoke callbacks (and therefore the handle might be
1377destroyed after it returns).
1378
1379=cut
1380
1381our %RH;
1382
1383sub register_read_type($$) {
1384 $RH{$_[0]} = $_[1];
1385}
1386
1387sub push_read {
1388 my $self = shift;
1389 my $cb = pop;
1390
1391 if (@_) {
1392 my $type = shift;
1393
1394 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1395 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_read")
1396 ->($self, $cb, @_);
1397 }
1398
1399 push @{ $self->{_queue} }, $cb;
1400 $self->_drain_rbuf;
1401}
1402
1403sub unshift_read {
1404 my $self = shift;
1405 my $cb = pop;
1406
1407 if (@_) {
1408 my $type = shift;
1409
1410 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1411 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::unshift_read")
1412 ->($self, $cb, @_);
1413 }
1414
1415 unshift @{ $self->{_queue} }, $cb;
1416 $self->_drain_rbuf;
1417}
1418
1419=item $handle->push_read (type => @args, $cb)
1420
1421=item $handle->unshift_read (type => @args, $cb)
1422
1423Instead of providing a callback that parses the data itself you can chose
1424between a number of predefined parsing formats, for chunks of data, lines
1425etc. You can also specify the (fully qualified) name of a package, in
1426which case AnyEvent tries to load the package and then expects to find the
1427C<anyevent_read_type> function inside (see "custom read types", below).
1428
1429Predefined types are (if you have ideas for additional types, feel free to
1430drop by and tell us):
1431
1432=over 4
1433
1434=item chunk => $octets, $cb->($handle, $data)
1435
1436Invoke the callback only once C<$octets> bytes have been read. Pass the
1437data read to the callback. The callback will never be called with less
1438data.
1439
1440Example: read 2 bytes.
1441
1442 $handle->push_read (chunk => 2, sub {
1443 say "yay " . unpack "H*", $_[1];
1444 });
1445
1446=cut
1447
1448register_read_type chunk => sub {
1449 my ($self, $cb, $len) = @_;
1450
1451 sub {
1452 $len <= length $_[0]{rbuf} or return;
1453 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
1454 1
1455 }
1456};
1457
1458=item line => [$eol, ]$cb->($handle, $line, $eol)
1459
1460The callback will be called only once a full line (including the end of
1461line marker, C<$eol>) has been read. This line (excluding the end of line
1462marker) will be passed to the callback as second argument (C<$line>), and
1463the end of line marker as the third argument (C<$eol>).
1464
1465The end of line marker, C<$eol>, can be either a string, in which case it
1466will be interpreted as a fixed record end marker, or it can be a regex
1467object (e.g. created by C<qr>), in which case it is interpreted as a
1468regular expression.
1469
1470The end of line marker argument C<$eol> is optional, if it is missing (NOT
1471undef), then C<qr|\015?\012|> is used (which is good for most internet
1472protocols).
1473
1474Partial lines at the end of the stream will never be returned, as they are
1475not marked by the end of line marker.
1476
1477=cut
1478
1479register_read_type line => sub {
1480 my ($self, $cb, $eol) = @_;
1481
1482 if (@_ < 3) {
1483 # this is more than twice as fast as the generic code below
1484 sub {
1485 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1486
1487 $cb->($_[0], "$1", "$2");
1488 1
1489 }
1490 } else {
1491 $eol = quotemeta $eol unless ref $eol;
1492 $eol = qr|^(.*?)($eol)|s;
1493
1494 sub {
1495 $_[0]{rbuf} =~ s/$eol// or return;
1496
1497 $cb->($_[0], "$1", "$2");
1498 1
1499 }
1500 }
1501};
1502
1503=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1504
1505Makes a regex match against the regex object C<$accept> and returns
1506everything up to and including the match.
1507
1508Example: read a single line terminated by '\n'.
1509
1510 $handle->push_read (regex => qr<\n>, sub { ... });
1511
1512If C<$reject> is given and not undef, then it determines when the data is
1513to be rejected: it is matched against the data when the C<$accept> regex
1514does not match and generates an C<EBADMSG> error when it matches. This is
1515useful to quickly reject wrong data (to avoid waiting for a timeout or a
1516receive buffer overflow).
1517
1518Example: expect a single decimal number followed by whitespace, reject
1519anything else (not the use of an anchor).
1520
1521 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1522
1523If C<$skip> is given and not C<undef>, then it will be matched against
1524the receive buffer when neither C<$accept> nor C<$reject> match,
1525and everything preceding and including the match will be accepted
1526unconditionally. This is useful to skip large amounts of data that you
1527know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1528have to start matching from the beginning. This is purely an optimisation
1529and is usually worth it only when you expect more than a few kilobytes.
1530
1531Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1532expect the header to be very large (it isn't in practice, but...), we use
1533a skip regex to skip initial portions. The skip regex is tricky in that
1534it only accepts something not ending in either \015 or \012, as these are
1535required for the accept regex.
1536
1537 $handle->push_read (regex =>
1538 qr<\015\012\015\012>,
1539 undef, # no reject
1540 qr<^.*[^\015\012]>,
1541 sub { ... });
1542
1543=cut
1544
1545register_read_type regex => sub {
1546 my ($self, $cb, $accept, $reject, $skip) = @_;
1547
1548 my $data;
1549 my $rbuf = \$self->{rbuf};
1550
1551 sub {
1552 # accept
1553 if ($$rbuf =~ $accept) {
1554 $data .= substr $$rbuf, 0, $+[0], "";
1555 $cb->($_[0], $data);
1556 return 1;
1557 }
1558
1559 # reject
1560 if ($reject && $$rbuf =~ $reject) {
1561 $_[0]->_error (Errno::EBADMSG);
1562 }
1563
1564 # skip
1565 if ($skip && $$rbuf =~ $skip) {
1566 $data .= substr $$rbuf, 0, $+[0], "";
1567 }
1568
1569 ()
1570 }
1571};
1572
1573=item netstring => $cb->($handle, $string)
1574
1575A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
1576
1577Throws an error with C<$!> set to EBADMSG on format violations.
1578
1579=cut
1580
1581register_read_type netstring => sub {
1582 my ($self, $cb) = @_;
1583
1584 sub {
1585 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1586 if ($_[0]{rbuf} =~ /[^0-9]/) {
1587 $_[0]->_error (Errno::EBADMSG);
1588 }
157 return; 1589 return;
1590 }
1591
1592 my $len = $1;
1593
1594 $_[0]->unshift_read (chunk => $len, sub {
1595 my $string = $_[1];
1596 $_[0]->unshift_read (chunk => 1, sub {
1597 if ($_[1] eq ",") {
1598 $cb->($_[0], $string);
1599 } else {
1600 $_[0]->_error (Errno::EBADMSG);
1601 }
1602 });
1603 });
1604
1605 1
1606 }
1607};
1608
1609=item packstring => $format, $cb->($handle, $string)
1610
1611An octet string prefixed with an encoded length. The encoding C<$format>
1612uses the same format as a Perl C<pack> format, but must specify a single
1613integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1614optional C<!>, C<< < >> or C<< > >> modifier).
1615
1616For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1617EPP uses a prefix of C<N> (4 octtes).
1618
1619Example: read a block of data prefixed by its length in BER-encoded
1620format (very efficient).
1621
1622 $handle->push_read (packstring => "w", sub {
1623 my ($handle, $data) = @_;
158 } 1624 });
159 1625
160 $self->{on_read_w} = 1626=cut
161 AnyEvent->io (poll => 'r', fh => $self->{fh}, cb => sub { 1627
162 #d# warn "READ:[$self->{read_size}] $self->{read_block_size} : ".length ($self->{rbuf})."\n"; 1628register_read_type packstring => sub {
1629 my ($self, $cb, $format) = @_;
1630
1631 sub {
1632 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1633 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1634 or return;
1635
1636 $format = length pack $format, $len;
1637
1638 # bypass unshift if we already have the remaining chunk
1639 if ($format + $len <= length $_[0]{rbuf}) {
1640 my $data = substr $_[0]{rbuf}, $format, $len;
1641 substr $_[0]{rbuf}, 0, $format + $len, "";
1642 $cb->($_[0], $data);
1643 } else {
1644 # remove prefix
1645 substr $_[0]{rbuf}, 0, $format, "";
1646
1647 # read remaining chunk
1648 $_[0]->unshift_read (chunk => $len, $cb);
1649 }
1650
1651 1
1652 }
1653};
1654
1655=item json => $cb->($handle, $hash_or_arrayref)
1656
1657Reads a JSON object or array, decodes it and passes it to the
1658callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1659
1660If a C<json> object was passed to the constructor, then that will be used
1661for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1662
1663This read type uses the incremental parser available with JSON version
16642.09 (and JSON::XS version 2.2) and above. You have to provide a
1665dependency on your own: this module will load the JSON module, but
1666AnyEvent does not depend on it itself.
1667
1668Since JSON texts are fully self-delimiting, the C<json> read and write
1669types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1670the C<json> write type description, above, for an actual example.
1671
1672=cut
1673
1674register_read_type json => sub {
1675 my ($self, $cb) = @_;
1676
1677 my $json = $self->{json} ||= json_coder;
1678
1679 my $data;
163 my $rbuf_len = length $self->{rbuf}; 1680 my $rbuf = \$self->{rbuf};
164 my $l; 1681
1682 sub {
1683 my $ref = eval { $json->incr_parse ($_[0]{rbuf}) };
1684
1685 if ($ref) {
1686 $_[0]{rbuf} = $json->incr_text;
1687 $json->incr_text = "";
1688 $cb->($_[0], $ref);
1689
1690 1
1691 } elsif ($@) {
1692 # error case
1693 $json->incr_skip;
1694
1695 $_[0]{rbuf} = $json->incr_text;
1696 $json->incr_text = "";
1697
1698 $_[0]->_error (Errno::EBADMSG);
1699
1700 ()
1701 } else {
1702 $_[0]{rbuf} = "";
1703
1704 ()
1705 }
1706 }
1707};
1708
1709=item storable => $cb->($handle, $ref)
1710
1711Deserialises a L<Storable> frozen representation as written by the
1712C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1713data).
1714
1715Raises C<EBADMSG> error if the data could not be decoded.
1716
1717=cut
1718
1719register_read_type storable => sub {
1720 my ($self, $cb) = @_;
1721
1722 require Storable unless $Storable::VERSION;
1723
1724 sub {
1725 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1726 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1727 or return;
1728
1729 my $format = length pack "w", $len;
1730
1731 # bypass unshift if we already have the remaining chunk
1732 if ($format + $len <= length $_[0]{rbuf}) {
1733 my $data = substr $_[0]{rbuf}, $format, $len;
1734 substr $_[0]{rbuf}, 0, $format + $len, "";
1735 $cb->($_[0], Storable::thaw ($data));
1736 } else {
1737 # remove prefix
1738 substr $_[0]{rbuf}, 0, $format, "";
1739
1740 # read remaining chunk
1741 $_[0]->unshift_read (chunk => $len, sub {
1742 if (my $ref = eval { Storable::thaw ($_[1]) }) {
1743 $cb->($_[0], $ref);
1744 } else {
1745 $_[0]->_error (Errno::EBADMSG);
1746 }
1747 });
1748 }
1749
1750 1
1751 }
1752};
1753
1754=back
1755
1756=item custom read types - Package::anyevent_read_type $handle, $cb, @args
1757
1758Instead of one of the predefined types, you can also specify the name
1759of a package. AnyEvent will try to load the package and then expects to
1760find a function named C<anyevent_read_type> inside. If it isn't found, it
1761progressively tries to load the parent package until it either finds the
1762function (good) or runs out of packages (bad).
1763
1764Whenever this type is used, C<push_read> will invoke the function with the
1765handle object, the original callback and the remaining arguments.
1766
1767The function is supposed to return a callback (usually a closure) that
1768works as a plain read callback (see C<< ->push_read ($cb) >>), so you can
1769mentally treat the function as a "configurable read type to read callback"
1770converter.
1771
1772It should invoke the original callback when it is done reading (remember
1773to pass C<$handle> as first argument as all other callbacks do that,
1774although there is no strict requirement on this).
1775
1776For examples, see the source of this module (F<perldoc -m
1777AnyEvent::Handle>, search for C<register_read_type>)).
1778
1779=item $handle->stop_read
1780
1781=item $handle->start_read
1782
1783In rare cases you actually do not want to read anything from the
1784socket. In this case you can call C<stop_read>. Neither C<on_read> nor
1785any queued callbacks will be executed then. To start reading again, call
1786C<start_read>.
1787
1788Note that AnyEvent::Handle will automatically C<start_read> for you when
1789you change the C<on_read> callback or push/unshift a read callback, and it
1790will automatically C<stop_read> for you when neither C<on_read> is set nor
1791there are any read requests in the queue.
1792
1793In older versions of this module (<= 5.3), these methods had no effect,
1794as TLS does not support half-duplex connections. In current versions they
1795work as expected, as this behaviour is required to avoid certain resource
1796attacks, where the program would be forced to read (and buffer) arbitrary
1797amounts of data before being able to send some data. The drawback is that
1798some readings of the the SSL/TLS specifications basically require this
1799attack to be working, as SSL/TLS implementations might stall sending data
1800during a rehandshake.
1801
1802As a guideline, during the initial handshake, you should not stop reading,
1803and as a client, it might cause problems, depending on your application.
1804
1805=cut
1806
1807sub stop_read {
1808 my ($self) = @_;
1809
1810 delete $self->{_rw};
1811}
1812
1813sub start_read {
1814 my ($self) = @_;
1815
1816 unless ($self->{_rw} || $self->{_eof} || !$self->{fh}) {
1817 Scalar::Util::weaken $self;
1818
1819 $self->{_rw} = AE::io $self->{fh}, 0, sub {
1820 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
1821 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size}, length $$rbuf;
1822
1823 if ($len > 0) {
1824 $self->{_activity} = $self->{_ractivity} = AE::now;
1825
1826 if ($self->{tls}) {
1827 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1828
1829 &_dotls ($self);
1830 } else {
1831 $self->_drain_rbuf;
1832 }
1833
165 if (defined $self->{read_size}) { 1834 if ($len == $self->{read_size}) {
166 $l = sysread $self->{fh}, $self->{rbuf}, 1835 $self->{read_size} *= 2;
167 ($self->{read_size} - $rbuf_len), $rbuf_len; 1836 $self->{read_size} = $self->{max_read_size} || MAX_READ_SIZE
1837 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
1838 }
1839
1840 } elsif (defined $len) {
1841 delete $self->{_rw};
1842 $self->{_eof} = 1;
1843 $self->_drain_rbuf;
1844
1845 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
1846 return $self->_error ($!, 1);
1847 }
1848 };
1849 }
1850}
1851
1852our $ERROR_SYSCALL;
1853our $ERROR_WANT_READ;
1854
1855sub _tls_error {
1856 my ($self, $err) = @_;
1857
1858 return $self->_error ($!, 1)
1859 if $err == Net::SSLeay::ERROR_SYSCALL ();
1860
1861 my $err =Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
1862
1863 # reduce error string to look less scary
1864 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
1865
1866 if ($self->{_on_starttls}) {
1867 (delete $self->{_on_starttls})->($self, undef, $err);
1868 &_freetls;
1869 } else {
1870 &_freetls;
1871 $self->_error (Errno::EPROTO, 1, $err);
1872 }
1873}
1874
1875# poll the write BIO and send the data if applicable
1876# also decode read data if possible
1877# this is basiclaly our TLS state machine
1878# more efficient implementations are possible with openssl,
1879# but not with the buggy and incomplete Net::SSLeay.
1880sub _dotls {
1881 my ($self) = @_;
1882
1883 my $tmp;
1884
1885 if (length $self->{_tls_wbuf}) {
1886 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
1887 substr $self->{_tls_wbuf}, 0, $tmp, "";
1888 }
1889
1890 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1891 return $self->_tls_error ($tmp)
1892 if $tmp != $ERROR_WANT_READ
1893 && ($tmp != $ERROR_SYSCALL || $!);
1894 }
1895
1896 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1897 unless (length $tmp) {
1898 $self->{_on_starttls}
1899 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1900 &_freetls;
1901
1902 if ($self->{on_stoptls}) {
1903 $self->{on_stoptls}($self);
1904 return;
168 } else { 1905 } else {
169 $l = sysread $self->{fh}, $self->{rbuf}, $self->{read_block_size}, $rbuf_len; 1906 # 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}; 1907 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 "line" seperator.
262
263The C<$callback> will be called when at least one
264line could be read. The first argument to the C<$callback> will be the L<AnyEvent::Handle>
265object itself and the rest of the arguments will be the read lines.
266
267NOTE: This method will override any callbacks installed via the C<on_read> method.
268
269=cut
270
271sub readlines {
272 my ($self, $sep, $cb) = @_;
273
274 if (ref $sep) {
275 $cb = $sep;
276 $sep = "\n";
277
278 } elsif (not defined $sep) {
279 $sep = "\n";
280 }
281
282 my $sep_len = length $sep;
283
284 $self->{on_readline} = $cb;
285
286 $self->on_read (sub {
287 my @lines;
288 my $rb = \$_[0]->{rbuf};
289 my $pos;
290 while (($pos = index ($$rb, $sep)) >= 0) {
291 push @lines, substr $$rb, 0, $pos + $sep_len, '';
292 }
293 $self->{on_readline}->($_[0], @lines);
294 });
295}
296
297=item B<write ($data)>
298
299=item B<write ($callback)>
300
301=item B<write ($data, $callback)>
302
303This method will write C<$data> to the filehandle and call the C<$callback>
304afterwards. If only C<$callback> is provided it will be called when the
305write buffer becomes empty the next time (or immediately if it already is empty).
306
307=cut
308
309sub write {
310 my ($self, $data, $cb) = @_;
311 if (ref $data) { $cb = $data; undef $data }
312 push @{$self->{write_bufs}}, [$data, $cb];
313 $self->_check_writer;
314}
315
316sub _check_writer {
317 my ($self) = @_;
318
319 if ($self->{write_w}) {
320 unless ($self->{write_cb}) {
321 while (@{$self->{write_bufs}} && not defined $self->{write_bufs}->[0]->[1]) {
322 my $wba = shift @{$self->{write_bufs}};
323 $self->{wbuf} .= $wba->[0]; 1908 $self->{_eof} = 1;
324 } 1909 }
325 } 1910 }
326 return;
327 }
328 1911
329 my $wba = shift @{$self->{write_bufs}} 1912 $self->{_tls_rbuf} .= $tmp;
330 or return; 1913 $self->_drain_rbuf;
331 1914 $self->{tls} or return; # tls session might have gone away in callback
332 unless (defined $wba->[0]) {
333 $wba->[1]->($self) if $wba->[1];
334 $self->_check_writer;
335 return;
336 } 1915 }
337 1916
338 $self->{wbuf} = $wba->[0]; 1917 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
339 $self->{write_cb} = $wba->[1]; 1918 return $self->_tls_error ($tmp)
1919 if $tmp != $ERROR_WANT_READ
1920 && ($tmp != $ERROR_SYSCALL || $!);
340 1921
341 $self->{write_w} = 1922 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
342 AnyEvent->io (poll => 'w', fh => $self->{fh}, cb => sub { 1923 $self->{wbuf} .= $tmp;
343 my $l = syswrite $self->{fh}, $self->{wbuf}, length $self->{wbuf}; 1924 $self->_drain_wbuf;
1925 $self->{tls} or return; # tls session might have gone away in callback
1926 }
344 1927
345 if (not defined $l) { 1928 $self->{_on_starttls}
346 return if $! == EAGAIN || $! == EINTR; 1929 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
347 delete $self->{write_w}; 1930 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
348 $self->{on_error}->($self) if $self->{on_error}; 1931}
349 1932
1933=item $handle->starttls ($tls[, $tls_ctx])
1934
1935Instead of starting TLS negotiation immediately when the AnyEvent::Handle
1936object is created, you can also do that at a later time by calling
1937C<starttls>.
1938
1939Starting TLS is currently an asynchronous operation - when you push some
1940write data and then call C<< ->starttls >> then TLS negotiation will start
1941immediately, after which the queued write data is then sent.
1942
1943The first argument is the same as the C<tls> constructor argument (either
1944C<"connect">, C<"accept"> or an existing Net::SSLeay object).
1945
1946The second argument is the optional C<AnyEvent::TLS> object that is used
1947when AnyEvent::Handle has to create its own TLS connection object, or
1948a hash reference with C<< key => value >> pairs that will be used to
1949construct a new context.
1950
1951The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1952context in C<< $handle->{tls_ctx} >> after this call and can be used or
1953changed to your liking. Note that the handshake might have already started
1954when this function returns.
1955
1956Due to bugs in OpenSSL, it might or might not be possible to do multiple
1957handshakes on the same stream. It is best to not attempt to use the
1958stream after stopping TLS.
1959
1960This method may invoke callbacks (and therefore the handle might be
1961destroyed after it returns).
1962
1963=cut
1964
1965our %TLS_CACHE; #TODO not yet documented, should we?
1966
1967sub starttls {
1968 my ($self, $tls, $ctx) = @_;
1969
1970 Carp::croak "It is an error to call starttls on an AnyEvent::Handle object while TLS is already active, caught"
1971 if $self->{tls};
1972
1973 $self->{tls} = $tls;
1974 $self->{tls_ctx} = $ctx if @_ > 2;
1975
1976 return unless $self->{fh};
1977
1978 require Net::SSLeay;
1979
1980 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
1981 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
1982
1983 $tls = delete $self->{tls};
1984 $ctx = $self->{tls_ctx};
1985
1986 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context or session
1987
1988 if ("HASH" eq ref $ctx) {
1989 require AnyEvent::TLS;
1990
1991 if ($ctx->{cache}) {
1992 my $key = $ctx+0;
1993 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
350 } else { 1994 } else {
1995 $ctx = new AnyEvent::TLS %$ctx;
1996 }
1997 }
1998
1999 $self->{tls_ctx} = $ctx || TLS_CTX ();
2000 $self->{tls} = $tls = $self->{tls_ctx}->_get_session ($tls, $self, $self->{peername});
2001
2002 # basically, this is deep magic (because SSL_read should have the same issues)
2003 # but the openssl maintainers basically said: "trust us, it just works".
2004 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
2005 # and mismaintained ssleay-module doesn't even offer them).
2006 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
2007 #
2008 # in short: this is a mess.
2009 #
2010 # note that we do not try to keep the length constant between writes as we are required to do.
2011 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
2012 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
2013 # have identity issues in that area.
2014# Net::SSLeay::CTX_set_mode ($ssl,
2015# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
2016# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
2017 Net::SSLeay::CTX_set_mode ($tls, 1|2);
2018
2019 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
2020 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
2021
2022 Net::SSLeay::BIO_write ($self->{_rbio}, $self->{rbuf});
2023 $self->{rbuf} = "";
2024
2025 Net::SSLeay::set_bio ($tls, $self->{_rbio}, $self->{_wbio});
2026
2027 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
2028 if $self->{on_starttls};
2029
2030 &_dotls; # need to trigger the initial handshake
2031 $self->start_read; # make sure we actually do read
2032}
2033
2034=item $handle->stoptls
2035
2036Shuts down the SSL connection - this makes a proper EOF handshake by
2037sending a close notify to the other side, but since OpenSSL doesn't
2038support non-blocking shut downs, it is not guaranteed that you can re-use
2039the stream afterwards.
2040
2041This method may invoke callbacks (and therefore the handle might be
2042destroyed after it returns).
2043
2044=cut
2045
2046sub stoptls {
2047 my ($self) = @_;
2048
2049 if ($self->{tls} && $self->{fh}) {
2050 Net::SSLeay::shutdown ($self->{tls});
2051
2052 &_dotls;
2053
2054# # we don't give a shit. no, we do, but we can't. no...#d#
2055# # we, we... have to use openssl :/#d#
2056# &_freetls;#d#
2057 }
2058}
2059
2060sub _freetls {
2061 my ($self) = @_;
2062
2063 return unless $self->{tls};
2064
2065 $self->{tls_ctx}->_put_session (delete $self->{tls})
2066 if $self->{tls} > 0;
2067
2068 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
2069}
2070
2071=item $handle->resettls
2072
2073This rarely-used method simply resets and TLS state on the handle, usually
2074causing data loss.
2075
2076One case where it may be useful is when you want to skip over the data in
2077the stream but you are not interested in interpreting it, so data loss is
2078no concern.
2079
2080=cut
2081
2082*resettls = \&_freetls;
2083
2084sub DESTROY {
2085 my ($self) = @_;
2086
2087 &_freetls;
2088
2089 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
2090
2091 if ($linger && length $self->{wbuf} && $self->{fh}) {
2092 my $fh = delete $self->{fh};
2093 my $wbuf = delete $self->{wbuf};
2094
2095 my @linger;
2096
2097 push @linger, AE::io $fh, 1, sub {
2098 my $len = syswrite $fh, $wbuf, length $wbuf;
2099
2100 if ($len > 0) {
351 substr $self->{wbuf}, 0, $l, ''; 2101 substr $wbuf, 0, $len, "";
352 2102 } elsif (defined $len || ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK)) {
353 if (length ($self->{wbuf}) == 0) { 2103 @linger = (); # end
354 $self->{write_cb}->($self) if $self->{write_cb};
355
356 delete $self->{write_w};
357 delete $self->{wbuf};
358 delete $self->{write_cb};
359
360 $self->_check_writer;
361 }
362 } 2104 }
2105 };
2106 push @linger, AE::timer $linger, 0, sub {
2107 @linger = ();
2108 };
2109 }
2110}
2111
2112=item $handle->destroy
2113
2114Shuts down the handle object as much as possible - this call ensures that
2115no further callbacks will be invoked and as many resources as possible
2116will be freed. Any method you will call on the handle object after
2117destroying it in this way will be silently ignored (and it will return the
2118empty list).
2119
2120Normally, you can just "forget" any references to an AnyEvent::Handle
2121object and it will simply shut down. This works in fatal error and EOF
2122callbacks, as well as code outside. It does I<NOT> work in a read or write
2123callback, so when you want to destroy the AnyEvent::Handle object from
2124within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
2125that case.
2126
2127Destroying the handle object in this way has the advantage that callbacks
2128will be removed as well, so if those are the only reference holders (as
2129is common), then one doesn't need to do anything special to break any
2130reference cycles.
2131
2132The handle might still linger in the background and write out remaining
2133data, as specified by the C<linger> option, however.
2134
2135=cut
2136
2137sub destroy {
2138 my ($self) = @_;
2139
2140 $self->DESTROY;
2141 %$self = ();
2142 bless $self, "AnyEvent::Handle::destroyed";
2143}
2144
2145sub AnyEvent::Handle::destroyed::AUTOLOAD {
2146 #nop
2147}
2148
2149=item $handle->destroyed
2150
2151Returns false as long as the handle hasn't been destroyed by a call to C<<
2152->destroy >>, true otherwise.
2153
2154Can be useful to decide whether the handle is still valid after some
2155callback possibly destroyed the handle. For example, C<< ->push_write >>,
2156C<< ->starttls >> and other methods can call user callbacks, which in turn
2157can destroy the handle, so work can be avoided by checking sometimes:
2158
2159 $hdl->starttls ("accept");
2160 return if $hdl->destroyed;
2161 $hdl->push_write (...
2162
2163Note that the call to C<push_write> will silently be ignored if the handle
2164has been destroyed, so often you can just ignore the possibility of the
2165handle being destroyed.
2166
2167=cut
2168
2169sub destroyed { 0 }
2170sub AnyEvent::Handle::destroyed::destroyed { 1 }
2171
2172=item AnyEvent::Handle::TLS_CTX
2173
2174This function creates and returns the AnyEvent::TLS object used by default
2175for TLS mode.
2176
2177The context is created by calling L<AnyEvent::TLS> without any arguments.
2178
2179=cut
2180
2181our $TLS_CTX;
2182
2183sub TLS_CTX() {
2184 $TLS_CTX ||= do {
2185 require AnyEvent::TLS;
2186
2187 new AnyEvent::TLS
2188 }
2189}
2190
2191=back
2192
2193
2194=head1 NONFREQUENTLY ASKED QUESTIONS
2195
2196=over 4
2197
2198=item I C<undef> the AnyEvent::Handle reference inside my callback and
2199still get further invocations!
2200
2201That's because AnyEvent::Handle keeps a reference to itself when handling
2202read or write callbacks.
2203
2204It is only safe to "forget" the reference inside EOF or error callbacks,
2205from within all other callbacks, you need to explicitly call the C<<
2206->destroy >> method.
2207
2208=item Why is my C<on_eof> callback never called?
2209
2210Probably because your C<on_error> callback is being called instead: When
2211you have outstanding requests in your read queue, then an EOF is
2212considered an error as you clearly expected some data.
2213
2214To avoid this, make sure you have an empty read queue whenever your handle
2215is supposed to be "idle" (i.e. connection closes are O.K.). You can set
2216an C<on_read> handler that simply pushes the first read requests in the
2217queue.
2218
2219See also the next question, which explains this in a bit more detail.
2220
2221=item How can I serve requests in a loop?
2222
2223Most protocols consist of some setup phase (authentication for example)
2224followed by a request handling phase, where the server waits for requests
2225and handles them, in a loop.
2226
2227There are two important variants: The first (traditional, better) variant
2228handles requests until the server gets some QUIT command, causing it to
2229close the connection first (highly desirable for a busy TCP server). A
2230client dropping the connection is an error, which means this variant can
2231detect an unexpected detection close.
2232
2233To handle this case, always make sure you have a on-empty read queue, by
2234pushing the "read request start" handler on it:
2235
2236 # we assume a request starts with a single line
2237 my @start_request; @start_request = (line => sub {
2238 my ($hdl, $line) = @_;
2239
2240 ... handle request
2241
2242 # push next request read, possibly from a nested callback
2243 $hdl->push_read (@start_request);
2244 });
2245
2246 # auth done, now go into request handling loop
2247 # now push the first @start_request
2248 $hdl->push_read (@start_request);
2249
2250By always having an outstanding C<push_read>, the handle always expects
2251some data and raises the C<EPIPE> error when the connction is dropped
2252unexpectedly.
2253
2254The second variant is a protocol where the client can drop the connection
2255at any time. For TCP, this means that the server machine may run out of
2256sockets easier, and in general, it means you cannot distinguish a protocl
2257failure/client crash from a normal connection close. Nevertheless, these
2258kinds of protocols are common (and sometimes even the best solution to the
2259problem).
2260
2261Having an outstanding read request at all times is possible if you ignore
2262C<EPIPE> errors, but this doesn't help with when the client drops the
2263connection during a request, which would still be an error.
2264
2265A better solution is to push the initial request read in an C<on_read>
2266callback. This avoids an error, as when the server doesn't expect data
2267(i.e. is idly waiting for the next request, an EOF will not raise an
2268error, but simply result in an C<on_eof> callback. It is also a bit slower
2269and simpler:
2270
2271 # auth done, now go into request handling loop
2272 $hdl->on_read (sub {
2273 my ($hdl) = @_;
2274
2275 # called each time we receive data but the read queue is empty
2276 # simply start read the request
2277
2278 $hdl->push_read (line => sub {
2279 my ($hdl, $line) = @_;
2280
2281 ... handle request
2282
2283 # do nothing special when the request has been handled, just
2284 # let the request queue go empty.
363 }); 2285 });
364} 2286 });
2287
2288=item I get different callback invocations in TLS mode/Why can't I pause
2289reading?
2290
2291Unlike, say, TCP, TLS connections do not consist of two independent
2292communication channels, one for each direction. Or put differently, the
2293read and write directions are not independent of each other: you cannot
2294write data unless you are also prepared to read, and vice versa.
2295
2296This means that, in TLS mode, you might get C<on_error> or C<on_eof>
2297callback invocations when you are not expecting any read data - the reason
2298is that AnyEvent::Handle always reads in TLS mode.
2299
2300During the connection, you have to make sure that you always have a
2301non-empty read-queue, or an C<on_read> watcher. At the end of the
2302connection (or when you no longer want to use it) you can call the
2303C<destroy> method.
2304
2305=item How do I read data until the other side closes the connection?
2306
2307If you just want to read your data into a perl scalar, the easiest way
2308to achieve this is by setting an C<on_read> callback that does nothing,
2309clearing the C<on_eof> callback and in the C<on_error> callback, the data
2310will be in C<$_[0]{rbuf}>:
2311
2312 $handle->on_read (sub { });
2313 $handle->on_eof (undef);
2314 $handle->on_error (sub {
2315 my $data = delete $_[0]{rbuf};
2316 });
2317
2318Note that this example removes the C<rbuf> member from the handle object,
2319which is not normally allowed by the API. It is expressly permitted in
2320this case only, as the handle object needs to be destroyed afterwards.
2321
2322The reason to use C<on_error> is that TCP connections, due to latencies
2323and packets loss, might get closed quite violently with an error, when in
2324fact all data has been received.
2325
2326It is usually better to use acknowledgements when transferring data,
2327to make sure the other side hasn't just died and you got the data
2328intact. This is also one reason why so many internet protocols have an
2329explicit QUIT command.
2330
2331=item I don't want to destroy the handle too early - how do I wait until
2332all data has been written?
2333
2334After writing your last bits of data, set the C<on_drain> callback
2335and destroy the handle in there - with the default setting of
2336C<low_water_mark> this will be called precisely when all data has been
2337written to the socket:
2338
2339 $handle->push_write (...);
2340 $handle->on_drain (sub {
2341 AE::log debug => "all data submitted to the kernel\n";
2342 undef $handle;
2343 });
2344
2345If you just want to queue some data and then signal EOF to the other side,
2346consider using C<< ->push_shutdown >> instead.
2347
2348=item I want to contact a TLS/SSL server, I don't care about security.
2349
2350If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
2351connect to it and then create the AnyEvent::Handle with the C<tls>
2352parameter:
2353
2354 tcp_connect $host, $port, sub {
2355 my ($fh) = @_;
2356
2357 my $handle = new AnyEvent::Handle
2358 fh => $fh,
2359 tls => "connect",
2360 on_error => sub { ... };
2361
2362 $handle->push_write (...);
2363 };
2364
2365=item I want to contact a TLS/SSL server, I do care about security.
2366
2367Then you should additionally enable certificate verification, including
2368peername verification, if the protocol you use supports it (see
2369L<AnyEvent::TLS>, C<verify_peername>).
2370
2371E.g. for HTTPS:
2372
2373 tcp_connect $host, $port, sub {
2374 my ($fh) = @_;
2375
2376 my $handle = new AnyEvent::Handle
2377 fh => $fh,
2378 peername => $host,
2379 tls => "connect",
2380 tls_ctx => { verify => 1, verify_peername => "https" },
2381 ...
2382
2383Note that you must specify the hostname you connected to (or whatever
2384"peername" the protocol needs) as the C<peername> argument, otherwise no
2385peername verification will be done.
2386
2387The above will use the system-dependent default set of trusted CA
2388certificates. If you want to check against a specific CA, add the
2389C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
2390
2391 tls_ctx => {
2392 verify => 1,
2393 verify_peername => "https",
2394 ca_file => "my-ca-cert.pem",
2395 },
2396
2397=item I want to create a TLS/SSL server, how do I do that?
2398
2399Well, you first need to get a server certificate and key. You have
2400three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
2401self-signed certificate (cheap. check the search engine of your choice,
2402there are many tutorials on the net) or c) make your own CA (tinyca2 is a
2403nice program for that purpose).
2404
2405Then create a file with your private key (in PEM format, see
2406L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
2407file should then look like this:
2408
2409 -----BEGIN RSA PRIVATE KEY-----
2410 ...header data
2411 ... lots of base64'y-stuff
2412 -----END RSA PRIVATE KEY-----
2413
2414 -----BEGIN CERTIFICATE-----
2415 ... lots of base64'y-stuff
2416 -----END CERTIFICATE-----
2417
2418The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
2419specify this file as C<cert_file>:
2420
2421 tcp_server undef, $port, sub {
2422 my ($fh) = @_;
2423
2424 my $handle = new AnyEvent::Handle
2425 fh => $fh,
2426 tls => "accept",
2427 tls_ctx => { cert_file => "my-server-keycert.pem" },
2428 ...
2429
2430When you have intermediate CA certificates that your clients might not
2431know about, just append them to the C<cert_file>.
365 2432
366=back 2433=back
367 2434
2435
2436=head1 SUBCLASSING AnyEvent::Handle
2437
2438In many cases, you might want to subclass AnyEvent::Handle.
2439
2440To make this easier, a given version of AnyEvent::Handle uses these
2441conventions:
2442
2443=over 4
2444
2445=item * all constructor arguments become object members.
2446
2447At least initially, when you pass a C<tls>-argument to the constructor it
2448will end up in C<< $handle->{tls} >>. Those members might be changed or
2449mutated later on (for example C<tls> will hold the TLS connection object).
2450
2451=item * other object member names are prefixed with an C<_>.
2452
2453All object members not explicitly documented (internal use) are prefixed
2454with an underscore character, so the remaining non-C<_>-namespace is free
2455for use for subclasses.
2456
2457=item * all members not documented here and not prefixed with an underscore
2458are free to use in subclasses.
2459
2460Of course, new versions of AnyEvent::Handle may introduce more "public"
2461member variables, but that's just life. At least it is documented.
2462
2463=back
2464
368=head1 AUTHOR 2465=head1 AUTHOR
369 2466
370Robin Redeker, C<< <elmex at ta-sa.org> >> 2467Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
371 2468
372=cut 2469=cut
373 2470
3741; # End of AnyEvent::Handle 24711; # End of AnyEvent::Handle

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines