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 1.146 by root, Wed Jul 8 13:46:46 2009 UTC

1package AnyEvent::Handle; 1package AnyEvent::Handle;
2 2
3use warnings; 3no warnings;
4use strict; 4use strict qw(subs vars);
5 5
6use AnyEvent; 6use AnyEvent ();
7use IO::Handle; 7use AnyEvent::Util qw(WSAEWOULDBLOCK);
8use Scalar::Util ();
9use Carp ();
10use Fcntl ();
8use Errno qw/EAGAIN EINTR/; 11use Errno qw(EAGAIN EINTR);
9 12
10=head1 NAME 13=head1 NAME
11 14
12AnyEvent::Handle - non-blocking I/O on filehandles via AnyEvent 15AnyEvent::Handle - non-blocking I/O on file handles via AnyEvent
13 16
14=head1 VERSION
15
16Version 0.01
17
18=cut 17=cut
19 18
20our $VERSION = '0.01'; 19our $VERSION = 4.8;
21 20
22=head1 SYNOPSIS 21=head1 SYNOPSIS
23 22
24 use AnyEvent; 23 use AnyEvent;
25 use AnyEvent::Handle; 24 use AnyEvent::Handle;
26 25
27 my $cv = AnyEvent->condvar; 26 my $cv = AnyEvent->condvar;
28 27
29 my $ae_fh = AnyEvent::Handle->new (fh => \*STDIN); 28 my $handle =
30
31 $ae_fh->on_eof (sub { $cv->broadcast });
32
33 $ae_fh->readlines (sub {
34 my ($ae_fh, @lines) = @_;
35 for (@lines) {
36 chomp;
37 print "Line: $_";
38 }
39 });
40
41 # or use the constructor to pass the callback:
42
43 my $ae_fh2 =
44 AnyEvent::Handle->new ( 29 AnyEvent::Handle->new (
45 fh => \*STDIN, 30 fh => \*STDIN,
46 on_eof => sub { 31 on_eof => sub {
47 $cv->broadcast; 32 $cv->send;
48 }, 33 },
34 );
35
36 # send some request line
37 $handle->push_write ("getinfo\015\012");
38
39 # read the response line
40 $handle->push_read (line => sub {
41 my ($handle, $line) = @_;
42 warn "read line <$line>\n";
43 $cv->send;
44 });
45
46 $cv->recv;
47
48=head1 DESCRIPTION
49
50This module is a helper module to make it easier to do event-based I/O on
51filehandles. For utility functions for doing non-blocking connects and accepts
52on sockets see L<AnyEvent::Util>.
53
54The L<AnyEvent::Intro> tutorial contains some well-documented
55AnyEvent::Handle examples.
56
57In the following, when the documentation refers to of "bytes" then this
58means characters. As sysread and syswrite are used for all I/O, their
59treatment of characters applies to this module as well.
60
61All callbacks will be invoked with the handle object as their first
62argument.
63
64=head1 METHODS
65
66=over 4
67
68=item $handle = B<new> AnyEvent::TLS fh => $filehandle, key => value...
69
70The constructor supports these arguments (all as C<< key => value >> pairs).
71
72=over 4
73
74=item fh => $filehandle [MANDATORY]
75
76The filehandle this L<AnyEvent::Handle> object will operate on.
77
78NOTE: The filehandle will be set to non-blocking mode (using
79C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
80that mode.
81
82=item on_eof => $cb->($handle)
83
84Set the callback to be called when an end-of-file condition is detected,
85i.e. in the case of a socket, when the other side has closed the
86connection cleanly.
87
88For sockets, this just means that the other side has stopped sending data,
89you can still try to write data, and, in fact, one can return from the EOF
90callback and continue writing data, as only the read part has been shut
91down.
92
93While not mandatory, it is I<highly> recommended to set an EOF callback,
94otherwise you might end up with a closed socket while you are still
95waiting for data.
96
97If an EOF condition has been detected but no C<on_eof> callback has been
98set, then a fatal error will be raised with C<$!> set to <0>.
99
100=item on_error => $cb->($handle, $fatal, $message)
101
102This is the error callback, which is called when, well, some error
103occured, such as not being able to resolve the hostname, failure to
104connect or a read error.
105
106Some errors are fatal (which is indicated by C<$fatal> being true). On
107fatal errors the handle object will be shut down and will not be usable
108(but you are free to look at the current C<< ->rbuf >>). Examples of fatal
109errors are an EOF condition with active (but unsatisifable) read watchers
110(C<EPIPE>) or I/O errors.
111
112AnyEvent::Handle tries to find an appropriate error code for you to check
113against, but in some cases (TLS errors), this does not work well. It is
114recommended to always output the C<$message> argument in human-readable
115error messages (it's usually the same as C<"$!">).
116
117Non-fatal errors can be retried by simply returning, but it is recommended
118to simply ignore this parameter and instead abondon the handle object
119when this callback is invoked. Examples of non-fatal errors are timeouts
120C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
121
122On callback entrance, the value of C<$!> contains the operating system
123error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
124C<EPROTO>).
125
126While not mandatory, it is I<highly> recommended to set this callback, as
127you will not be notified of errors otherwise. The default simply calls
128C<croak>.
129
130=item on_read => $cb->($handle)
131
132This sets the default read callback, which is called when data arrives
133and no read request is in the queue (unlike read queue callbacks, this
134callback will only be called when at least one octet of data is in the
135read buffer).
136
137To access (and remove data from) the read buffer, use the C<< ->rbuf >>
138method or access the C<< $handle->{rbuf} >> member directly. Note that you
139must not enlarge or modify the read buffer, you can only remove data at
140the beginning from it.
141
142When an EOF condition is detected then AnyEvent::Handle will first try to
143feed all the remaining data to the queued callbacks and C<on_read> before
144calling the C<on_eof> callback. If no progress can be made, then a fatal
145error will be raised (with C<$!> set to C<EPIPE>).
146
147=item on_drain => $cb->($handle)
148
149This sets the callback that is called when the write buffer becomes empty
150(or when the callback is set and the buffer is empty already).
151
152To append to the write buffer, use the C<< ->push_write >> method.
153
154This callback is useful when you don't want to put all of your write data
155into the queue at once, for example, when you want to write the contents
156of some file to the socket you might not want to read the whole file into
157memory and push it into the queue, but instead only read more data from
158the file when the write queue becomes empty.
159
160=item timeout => $fractional_seconds
161
162If non-zero, then this enables an "inactivity" timeout: whenever this many
163seconds pass without a successful read or write on the underlying file
164handle, the C<on_timeout> callback will be invoked (and if that one is
165missing, a non-fatal C<ETIMEDOUT> error will be raised).
166
167Note that timeout processing is also active when you currently do not have
168any outstanding read or write requests: If you plan to keep the connection
169idle then you should disable the timout temporarily or ignore the timeout
170in the C<on_timeout> callback, in which case AnyEvent::Handle will simply
171restart the timeout.
172
173Zero (the default) disables this timeout.
174
175=item on_timeout => $cb->($handle)
176
177Called whenever the inactivity timeout passes. If you return from this
178callback, then the timeout will be reset as if some activity had happened,
179so this condition is not fatal in any way.
180
181=item rbuf_max => <bytes>
182
183If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
184when the read buffer ever (strictly) exceeds this size. This is useful to
185avoid some forms of denial-of-service attacks.
186
187For example, a server accepting connections from untrusted sources should
188be configured to accept only so-and-so much data that it cannot act on
189(for example, when expecting a line, an attacker could send an unlimited
190amount of data without a callback ever being called as long as the line
191isn't finished).
192
193=item autocork => <boolean>
194
195When disabled (the default), then C<push_write> will try to immediately
196write the data to the handle, if possible. This avoids having to register
197a write watcher and wait for the next event loop iteration, but can
198be inefficient if you write multiple small chunks (on the wire, this
199disadvantage is usually avoided by your kernel's nagle algorithm, see
200C<no_delay>, but this option can save costly syscalls).
201
202When enabled, then writes will always be queued till the next event loop
203iteration. This is efficient when you do many small writes per iteration,
204but less efficient when you do a single write only per iteration (or when
205the write buffer often is full). It also increases write latency.
206
207=item no_delay => <boolean>
208
209When doing small writes on sockets, your operating system kernel might
210wait a bit for more data before actually sending it out. This is called
211the Nagle algorithm, and usually it is beneficial.
212
213In some situations you want as low a delay as possible, which can be
214accomplishd by setting this option to a true value.
215
216The default is your opertaing system's default behaviour (most likely
217enabled), this option explicitly enables or disables it, if possible.
218
219=item read_size => <bytes>
220
221The default read block size (the amount of bytes this module will
222try to read during each loop iteration, which affects memory
223requirements). Default: C<8192>.
224
225=item low_water_mark => <bytes>
226
227Sets the amount of bytes (default: C<0>) that make up an "empty" write
228buffer: If the write reaches this size or gets even samller it is
229considered empty.
230
231Sometimes it can be beneficial (for performance reasons) to add data to
232the write buffer before it is fully drained, but this is a rare case, as
233the operating system kernel usually buffers data as well, so the default
234is good in almost all cases.
235
236=item linger => <seconds>
237
238If non-zero (default: C<3600>), then the destructor of the
239AnyEvent::Handle object will check whether there is still outstanding
240write data and will install a watcher that will write this data to the
241socket. No errors will be reported (this mostly matches how the operating
242system treats outstanding data at socket close time).
243
244This will not work for partial TLS data that could not be encoded
245yet. This data will be lost. Calling the C<stoptls> method in time might
246help.
247
248=item peername => $string
249
250A string used to identify the remote site - usually the DNS hostname
251(I<not> IDN!) used to create the connection, rarely the IP address.
252
253Apart from being useful in error messages, this string is also used in TLS
254peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
255verification will be skipped when C<peername> is not specified or
256C<undef>.
257
258=item tls => "accept" | "connect" | Net::SSLeay::SSL object
259
260When this parameter is given, it enables TLS (SSL) mode, that means
261AnyEvent will start a TLS handshake as soon as the conenction has been
262established and will transparently encrypt/decrypt data afterwards.
263
264All TLS protocol errors will be signalled as C<EPROTO>, with an
265appropriate error message.
266
267TLS mode requires Net::SSLeay to be installed (it will be loaded
268automatically when you try to create a TLS handle): this module doesn't
269have a dependency on that module, so if your module requires it, you have
270to add the dependency yourself.
271
272Unlike TCP, TLS has a server and client side: for the TLS server side, use
273C<accept>, and for the TLS client side of a connection, use C<connect>
274mode.
275
276You can also provide your own TLS connection object, but you have
277to make sure that you call either C<Net::SSLeay::set_connect_state>
278or C<Net::SSLeay::set_accept_state> on it before you pass it to
279AnyEvent::Handle. Also, this module will take ownership of this connection
280object.
281
282At some future point, AnyEvent::Handle might switch to another TLS
283implementation, then the option to use your own session object will go
284away.
285
286B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
287passing in the wrong integer will lead to certain crash. This most often
288happens when one uses a stylish C<< tls => 1 >> and is surprised about the
289segmentation fault.
290
291See the C<< ->starttls >> method for when need to start TLS negotiation later.
292
293=item tls_ctx => $anyevent_tls
294
295Use the given C<AnyEvent::TLS> object to create the new TLS connection
296(unless a connection object was specified directly). If this parameter is
297missing, then AnyEvent::Handle will use C<AnyEvent::Handle::TLS_CTX>.
298
299Instead of an object, you can also specify a hash reference with C<< key
300=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
301new TLS context object.
302
303=item on_starttls => $cb->($handle, $success[, $error_message])
304
305This callback will be invoked when the TLS/SSL handshake has finished. If
306C<$success> is true, then the TLS handshake succeeded, otherwise it failed
307(C<on_stoptls> will not be called in this case).
308
309The session in C<< $handle->{tls} >> can still be examined in this
310callback, even when the handshake was not successful.
311
312TLS handshake failures will not cause C<on_error> to be invoked when this
313callback is in effect, instead, the error message will be passed to C<on_starttls>.
314
315Without this callback, handshake failures lead to C<on_error> being
316called, as normal.
317
318Note that you cannot call C<starttls> right again in this callback. If you
319need to do that, start an zero-second timer instead whose callback can
320then call C<< ->starttls >> again.
321
322=item on_stoptls => $cb->($handle)
323
324When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
325set, then it will be invoked after freeing the TLS session. If it is not,
326then a TLS shutdown condition will be treated like a normal EOF condition
327on the handle.
328
329The session in C<< $handle->{tls} >> can still be examined in this
330callback.
331
332This callback will only be called on TLS shutdowns, not when the
333underlying handle signals EOF.
334
335=item json => JSON or JSON::XS object
336
337This is the json coder object used by the C<json> read and write types.
338
339If you don't supply it, then AnyEvent::Handle will create and use a
340suitable one (on demand), which will write and expect UTF-8 encoded JSON
341texts.
342
343Note that you are responsible to depend on the JSON module if you want to
344use this functionality, as AnyEvent does not have a dependency itself.
345
346=back
347
348=cut
349
350sub new {
351 my $class = shift;
352 my $self = bless { @_ }, $class;
353
354 $self->{fh} or Carp::croak "mandatory argument fh is missing";
355
356 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
357
358 $self->{_activity} = AnyEvent->now;
359 $self->_timeout;
360
361 $self->no_delay (delete $self->{no_delay}) if exists $self->{no_delay};
362
363 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
364 if $self->{tls};
365
366 $self->on_drain (delete $self->{on_drain}) if $self->{on_drain};
367
368 $self->start_read
369 if $self->{on_read};
370
371 $self->{fh} && $self
372}
373
374sub _shutdown {
375 my ($self) = @_;
376
377 delete @$self{qw(_tw _rw _ww fh wbuf on_read _queue)};
378 $self->{_eof} = 1; # tell starttls et. al to stop trying
379
380 &_freetls;
381}
382
383sub _error {
384 my ($self, $errno, $fatal, $message) = @_;
385
386 $self->_shutdown
387 if $fatal;
388
389 $! = $errno;
390 $message ||= "$!";
391
392 if ($self->{on_error}) {
393 $self->{on_error}($self, $fatal, $message);
394 } elsif ($self->{fh}) {
395 Carp::croak "AnyEvent::Handle uncaught error: $message";
396 }
397}
398
399=item $fh = $handle->fh
400
401This method returns the file handle used to create the L<AnyEvent::Handle> object.
402
403=cut
404
405sub fh { $_[0]{fh} }
406
407=item $handle->on_error ($cb)
408
409Replace the current C<on_error> callback (see the C<on_error> constructor argument).
410
411=cut
412
413sub on_error {
414 $_[0]{on_error} = $_[1];
415}
416
417=item $handle->on_eof ($cb)
418
419Replace the current C<on_eof> callback (see the C<on_eof> constructor argument).
420
421=cut
422
423sub on_eof {
424 $_[0]{on_eof} = $_[1];
425}
426
427=item $handle->on_timeout ($cb)
428
429Replace the current C<on_timeout> callback, or disables the callback (but
430not the timeout) if C<$cb> = C<undef>. See the C<timeout> constructor
431argument and method.
432
433=cut
434
435sub on_timeout {
436 $_[0]{on_timeout} = $_[1];
437}
438
439=item $handle->autocork ($boolean)
440
441Enables or disables the current autocork behaviour (see C<autocork>
442constructor argument). Changes will only take effect on the next write.
443
444=cut
445
446sub autocork {
447 $_[0]{autocork} = $_[1];
448}
449
450=item $handle->no_delay ($boolean)
451
452Enables or disables the C<no_delay> setting (see constructor argument of
453the same name for details).
454
455=cut
456
457sub no_delay {
458 $_[0]{no_delay} = $_[1];
459
460 eval {
461 local $SIG{__DIE__};
462 setsockopt $_[0]{fh}, &Socket::IPPROTO_TCP, &Socket::TCP_NODELAY, int $_[1];
463 };
464}
465
466=item $handle->on_starttls ($cb)
467
468Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
469
470=cut
471
472sub on_starttls {
473 $_[0]{on_starttls} = $_[1];
474}
475
476=item $handle->on_stoptls ($cb)
477
478Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
479
480=cut
481
482sub on_starttls {
483 $_[0]{on_stoptls} = $_[1];
484}
485
486#############################################################################
487
488=item $handle->timeout ($seconds)
489
490Configures (or disables) the inactivity timeout.
491
492=cut
493
494sub timeout {
495 my ($self, $timeout) = @_;
496
497 $self->{timeout} = $timeout;
498 $self->_timeout;
499}
500
501# reset the timeout watcher, as neccessary
502# also check for time-outs
503sub _timeout {
504 my ($self) = @_;
505
506 if ($self->{timeout}) {
507 my $NOW = AnyEvent->now;
508
509 # when would the timeout trigger?
510 my $after = $self->{_activity} + $self->{timeout} - $NOW;
511
512 # now or in the past already?
513 if ($after <= 0) {
514 $self->{_activity} = $NOW;
515
516 if ($self->{on_timeout}) {
517 $self->{on_timeout}($self);
518 } else {
519 $self->_error (&Errno::ETIMEDOUT);
520 }
521
522 # callback could have changed timeout value, optimise
523 return unless $self->{timeout};
524
525 # calculate new after
526 $after = $self->{timeout};
527 }
528
529 Scalar::Util::weaken $self;
530 return unless $self; # ->error could have destroyed $self
531
532 $self->{_tw} ||= AnyEvent->timer (after => $after, cb => sub {
533 delete $self->{_tw};
534 $self->_timeout;
535 });
536 } else {
537 delete $self->{_tw};
538 }
539}
540
541#############################################################################
542
543=back
544
545=head2 WRITE QUEUE
546
547AnyEvent::Handle manages two queues per handle, one for writing and one
548for reading.
549
550The write queue is very simple: you can add data to its end, and
551AnyEvent::Handle will automatically try to get rid of it for you.
552
553When data could be written and the write buffer is shorter then the low
554water mark, the C<on_drain> callback will be invoked.
555
556=over 4
557
558=item $handle->on_drain ($cb)
559
560Sets the C<on_drain> callback or clears it (see the description of
561C<on_drain> in the constructor).
562
563=cut
564
565sub on_drain {
566 my ($self, $cb) = @_;
567
568 $self->{on_drain} = $cb;
569
570 $cb->($self)
571 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
572}
573
574=item $handle->push_write ($data)
575
576Queues the given scalar to be written. You can push as much data as you
577want (only limited by the available memory), as C<AnyEvent::Handle>
578buffers it independently of the kernel.
579
580=cut
581
582sub _drain_wbuf {
583 my ($self) = @_;
584
585 if (!$self->{_ww} && length $self->{wbuf}) {
586
587 Scalar::Util::weaken $self;
588
589 my $cb = sub {
590 my $len = syswrite $self->{fh}, $self->{wbuf};
591
592 if (defined $len) {
593 substr $self->{wbuf}, 0, $len, "";
594
595 $self->{_activity} = AnyEvent->now;
596
597 $self->{on_drain}($self)
598 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
599 && $self->{on_drain};
600
601 delete $self->{_ww} unless length $self->{wbuf};
602 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
603 $self->_error ($!, 1);
604 }
605 };
606
607 # try to write data immediately
608 $cb->() unless $self->{autocork};
609
610 # if still data left in wbuf, we need to poll
611 $self->{_ww} = AnyEvent->io (fh => $self->{fh}, poll => "w", cb => $cb)
612 if length $self->{wbuf};
613 };
614}
615
616our %WH;
617
618sub register_write_type($$) {
619 $WH{$_[0]} = $_[1];
620}
621
622sub push_write {
623 my $self = shift;
624
625 if (@_ > 1) {
626 my $type = shift;
627
628 @_ = ($WH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::push_write")
629 ->($self, @_);
630 }
631
632 if ($self->{tls}) {
633 $self->{_tls_wbuf} .= $_[0];
634
635 &_dotls ($self);
636 } else {
637 $self->{wbuf} .= $_[0];
638 $self->_drain_wbuf;
639 }
640}
641
642=item $handle->push_write (type => @args)
643
644Instead of formatting your data yourself, you can also let this module do
645the job by specifying a type and type-specific arguments.
646
647Predefined types are (if you have ideas for additional types, feel free to
648drop by and tell us):
649
650=over 4
651
652=item netstring => $string
653
654Formats the given value as netstring
655(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
656
657=cut
658
659register_write_type netstring => sub {
660 my ($self, $string) = @_;
661
662 (length $string) . ":$string,"
663};
664
665=item packstring => $format, $data
666
667An octet string prefixed with an encoded length. The encoding C<$format>
668uses the same format as a Perl C<pack> format, but must specify a single
669integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
670optional C<!>, C<< < >> or C<< > >> modifier).
671
672=cut
673
674register_write_type packstring => sub {
675 my ($self, $format, $string) = @_;
676
677 pack "$format/a*", $string
678};
679
680=item json => $array_or_hashref
681
682Encodes the given hash or array reference into a JSON object. Unless you
683provide your own JSON object, this means it will be encoded to JSON text
684in UTF-8.
685
686JSON objects (and arrays) are self-delimiting, so you can write JSON at
687one end of a handle and read them at the other end without using any
688additional framing.
689
690The generated JSON text is guaranteed not to contain any newlines: While
691this module doesn't need delimiters after or between JSON texts to be
692able to read them, many other languages depend on that.
693
694A simple RPC protocol that interoperates easily with others is to send
695JSON arrays (or objects, although arrays are usually the better choice as
696they mimic how function argument passing works) and a newline after each
697JSON text:
698
699 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
700 $handle->push_write ("\012");
701
702An AnyEvent::Handle receiver would simply use the C<json> read type and
703rely on the fact that the newline will be skipped as leading whitespace:
704
705 $handle->push_read (json => sub { my $array = $_[1]; ... });
706
707Other languages could read single lines terminated by a newline and pass
708this line into their JSON decoder of choice.
709
710=cut
711
712register_write_type json => sub {
713 my ($self, $ref) = @_;
714
715 require JSON;
716
717 $self->{json} ? $self->{json}->encode ($ref)
718 : JSON::encode_json ($ref)
719};
720
721=item storable => $reference
722
723Freezes the given reference using L<Storable> and writes it to the
724handle. Uses the C<nfreeze> format.
725
726=cut
727
728register_write_type storable => sub {
729 my ($self, $ref) = @_;
730
731 require Storable;
732
733 pack "w/a*", Storable::nfreeze ($ref)
734};
735
736=back
737
738=item $handle->push_shutdown
739
740Sometimes you know you want to close the socket after writing your data
741before it was actually written. One way to do that is to replace your
742C<on_drain> handler by a callback that shuts down the socket (and set
743C<low_water_mark> to C<0>). This method is a shorthand for just that, and
744replaces the C<on_drain> callback with:
745
746 sub { shutdown $_[0]{fh}, 1 } # for push_shutdown
747
748This simply shuts down the write side and signals an EOF condition to the
749the peer.
750
751You can rely on the normal read queue and C<on_eof> handling
752afterwards. This is the cleanest way to close a connection.
753
754=cut
755
756sub push_shutdown {
757 my ($self) = @_;
758
759 delete $self->{low_water_mark};
760 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
761}
762
763=item AnyEvent::Handle::register_write_type type => $coderef->($handle, @args)
764
765This function (not method) lets you add your own types to C<push_write>.
766Whenever the given C<type> is used, C<push_write> will invoke the code
767reference with the handle object and the remaining arguments.
768
769The code reference is supposed to return a single octet string that will
770be appended to the write buffer.
771
772Note that this is a function, and all types registered this way will be
773global, so try to use unique names.
774
775=cut
776
777#############################################################################
778
779=back
780
781=head2 READ QUEUE
782
783AnyEvent::Handle manages two queues per handle, one for writing and one
784for reading.
785
786The read queue is more complex than the write queue. It can be used in two
787ways, the "simple" way, using only C<on_read> and the "complex" way, using
788a queue.
789
790In the simple case, you just install an C<on_read> callback and whenever
791new data arrives, it will be called. You can then remove some data (if
792enough is there) from the read buffer (C<< $handle->rbuf >>). Or you cna
793leave the data there if you want to accumulate more (e.g. when only a
794partial message has been received so far).
795
796In the more complex case, you want to queue multiple callbacks. In this
797case, AnyEvent::Handle will call the first queued callback each time new
798data arrives (also the first time it is queued) and removes it when it has
799done its job (see C<push_read>, below).
800
801This way you can, for example, push three line-reads, followed by reading
802a chunk of data, and AnyEvent::Handle will execute them in order.
803
804Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
805the specified number of bytes which give an XML datagram.
806
807 # in the default state, expect some header bytes
808 $handle->on_read (sub {
809 # some data is here, now queue the length-header-read (4 octets)
810 shift->unshift_read (chunk => 4, sub {
811 # header arrived, decode
812 my $len = unpack "N", $_[1];
813
814 # now read the payload
815 shift->unshift_read (chunk => $len, sub {
816 my $xml = $_[1];
817 # handle xml
818 });
819 });
820 });
821
822Example 2: Implement a client for a protocol that replies either with "OK"
823and another line or "ERROR" for the first request that is sent, and 64
824bytes for the second request. Due to the availability of a queue, we can
825just pipeline sending both requests and manipulate the queue as necessary
826in the callbacks.
827
828When the first callback is called and sees an "OK" response, it will
829C<unshift> another line-read. This line-read will be queued I<before> the
83064-byte chunk callback.
831
832 # request one, returns either "OK + extra line" or "ERROR"
833 $handle->push_write ("request 1\015\012");
834
835 # we expect "ERROR" or "OK" as response, so push a line read
836 $handle->push_read (line => sub {
837 # if we got an "OK", we have to _prepend_ another line,
838 # so it will be read before the second request reads its 64 bytes
839 # which are already in the queue when this callback is called
840 # we don't do this in case we got an error
841 if ($_[1] eq "OK") {
49 on_readline => sub { 842 $_[0]->unshift_read (line => sub {
50 my ($ae_fh, @lines) = @_; 843 my $response = $_[1];
51 for (@lines) { 844 ...
52 chomp; 845 });
53 print "Line: $_"; 846 }
847 });
848
849 # request two, simply returns 64 octets
850 $handle->push_write ("request 2\015\012");
851
852 # simply read 64 bytes, always
853 $handle->push_read (chunk => 64, sub {
854 my $response = $_[1];
855 ...
856 });
857
858=over 4
859
860=cut
861
862sub _drain_rbuf {
863 my ($self) = @_;
864
865 local $self->{_in_drain} = 1;
866
867 if (
868 defined $self->{rbuf_max}
869 && $self->{rbuf_max} < length $self->{rbuf}
870 ) {
871 $self->_error (&Errno::ENOSPC, 1), return;
872 }
873
874 while () {
875 # we need to use a separate tls read buffer, as we must not receive data while
876 # we are draining the buffer, and this can only happen with TLS.
877 $self->{rbuf} .= delete $self->{_tls_rbuf} if exists $self->{_tls_rbuf};
878
879 my $len = length $self->{rbuf};
880
881 if (my $cb = shift @{ $self->{_queue} }) {
882 unless ($cb->($self)) {
883 if ($self->{_eof}) {
884 # no progress can be made (not enough data and no data forthcoming)
885 $self->_error (&Errno::EPIPE, 1), return;
54 } 886 }
887
888 unshift @{ $self->{_queue} }, $cb;
889 last;
55 } 890 }
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}) { 891 } elsif ($self->{on_read}) {
121 $self->readlines ($self->{on_readline}); 892 last unless $len;
122 893
894 $self->{on_read}($self);
895
896 if (
897 $len == length $self->{rbuf} # if no data has been consumed
898 && !@{ $self->{_queue} } # and the queue is still empty
899 && $self->{on_read} # but we still have on_read
900 ) {
901 # no further data will arrive
902 # so no progress can be made
903 $self->_error (&Errno::EPIPE, 1), return
904 if $self->{_eof};
905
906 last; # more data might arrive
907 }
908 } else {
909 # read side becomes idle
910 delete $self->{_rw} unless $self->{tls};
911 last;
912 }
913 }
914
123 } elsif ($self->{on_eof}) { 915 if ($self->{_eof}) {
124 $self->on_eof ($self->{on_eof});
125
126 } elsif ($self->{on_error}) { 916 if ($self->{on_eof}) {
127 $self->on_eof ($self->{on_error}); 917 $self->{on_eof}($self)
918 } else {
919 $self->_error (0, 1, "Unexpected end-of-file");
920 }
128 } 921 }
129 922
130 return $self 923 # may need to restart read watcher
924 unless ($self->{_rw}) {
925 $self->start_read
926 if $self->{on_read} || @{ $self->{_queue} };
927 }
131} 928}
132 929
133=item B<fh> 930=item $handle->on_read ($cb)
134 931
135This method returns the filehandle of the L<AnyEvent::Handle> object. 932This replaces the currently set C<on_read> callback, or clears it (when
136 933the new callback is C<undef>). See the description of C<on_read> in the
137=cut 934constructor.
138
139sub fh { $_[0]->{fh} }
140
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.
148 935
149=cut 936=cut
150 937
151sub on_read { 938sub on_read {
152 my ($self, $cb) = @_; 939 my ($self, $cb) = @_;
940
153 $self->{on_read} = $cb; 941 $self->{on_read} = $cb;
942 $self->_drain_rbuf if $cb && !$self->{_in_drain};
943}
154 944
155 unless (defined $self->{on_read}) { 945=item $handle->rbuf
156 delete $self->{on_read_w}; 946
947Returns the read buffer (as a modifiable lvalue).
948
949You can access the read buffer directly as the C<< ->{rbuf} >>
950member, if you want. However, the only operation allowed on the
951read buffer (apart from looking at it) is removing data from its
952beginning. Otherwise modifying or appending to it is not allowed and will
953lead to hard-to-track-down bugs.
954
955NOTE: The read buffer should only be used or modified if the C<on_read>,
956C<push_read> or C<unshift_read> methods are used. The other read methods
957automatically manage the read buffer.
958
959=cut
960
961sub rbuf : lvalue {
962 $_[0]{rbuf}
963}
964
965=item $handle->push_read ($cb)
966
967=item $handle->unshift_read ($cb)
968
969Append the given callback to the end of the queue (C<push_read>) or
970prepend it (C<unshift_read>).
971
972The callback is called each time some additional read data arrives.
973
974It must check whether enough data is in the read buffer already.
975
976If not enough data is available, it must return the empty list or a false
977value, in which case it will be called repeatedly until enough data is
978available (or an error condition is detected).
979
980If enough data was available, then the callback must remove all data it is
981interested in (which can be none at all) and return a true value. After returning
982true, it will be removed from the queue.
983
984=cut
985
986our %RH;
987
988sub register_read_type($$) {
989 $RH{$_[0]} = $_[1];
990}
991
992sub push_read {
993 my $self = shift;
994 my $cb = pop;
995
996 if (@_) {
997 my $type = shift;
998
999 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::push_read")
1000 ->($self, $cb, @_);
1001 }
1002
1003 push @{ $self->{_queue} }, $cb;
1004 $self->_drain_rbuf unless $self->{_in_drain};
1005}
1006
1007sub unshift_read {
1008 my $self = shift;
1009 my $cb = pop;
1010
1011 if (@_) {
1012 my $type = shift;
1013
1014 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::unshift_read")
1015 ->($self, $cb, @_);
1016 }
1017
1018
1019 unshift @{ $self->{_queue} }, $cb;
1020 $self->_drain_rbuf unless $self->{_in_drain};
1021}
1022
1023=item $handle->push_read (type => @args, $cb)
1024
1025=item $handle->unshift_read (type => @args, $cb)
1026
1027Instead of providing a callback that parses the data itself you can chose
1028between a number of predefined parsing formats, for chunks of data, lines
1029etc.
1030
1031Predefined types are (if you have ideas for additional types, feel free to
1032drop by and tell us):
1033
1034=over 4
1035
1036=item chunk => $octets, $cb->($handle, $data)
1037
1038Invoke the callback only once C<$octets> bytes have been read. Pass the
1039data read to the callback. The callback will never be called with less
1040data.
1041
1042Example: read 2 bytes.
1043
1044 $handle->push_read (chunk => 2, sub {
1045 warn "yay ", unpack "H*", $_[1];
1046 });
1047
1048=cut
1049
1050register_read_type chunk => sub {
1051 my ($self, $cb, $len) = @_;
1052
1053 sub {
1054 $len <= length $_[0]{rbuf} or return;
1055 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
1056 1
1057 }
1058};
1059
1060=item line => [$eol, ]$cb->($handle, $line, $eol)
1061
1062The callback will be called only once a full line (including the end of
1063line marker, C<$eol>) has been read. This line (excluding the end of line
1064marker) will be passed to the callback as second argument (C<$line>), and
1065the end of line marker as the third argument (C<$eol>).
1066
1067The end of line marker, C<$eol>, can be either a string, in which case it
1068will be interpreted as a fixed record end marker, or it can be a regex
1069object (e.g. created by C<qr>), in which case it is interpreted as a
1070regular expression.
1071
1072The end of line marker argument C<$eol> is optional, if it is missing (NOT
1073undef), then C<qr|\015?\012|> is used (which is good for most internet
1074protocols).
1075
1076Partial lines at the end of the stream will never be returned, as they are
1077not marked by the end of line marker.
1078
1079=cut
1080
1081register_read_type line => sub {
1082 my ($self, $cb, $eol) = @_;
1083
1084 if (@_ < 3) {
1085 # this is more than twice as fast as the generic code below
1086 sub {
1087 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1088
1089 $cb->($_[0], $1, $2);
1090 1
1091 }
1092 } else {
1093 $eol = quotemeta $eol unless ref $eol;
1094 $eol = qr|^(.*?)($eol)|s;
1095
1096 sub {
1097 $_[0]{rbuf} =~ s/$eol// or return;
1098
1099 $cb->($_[0], $1, $2);
1100 1
1101 }
1102 }
1103};
1104
1105=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1106
1107Makes a regex match against the regex object C<$accept> and returns
1108everything up to and including the match.
1109
1110Example: read a single line terminated by '\n'.
1111
1112 $handle->push_read (regex => qr<\n>, sub { ... });
1113
1114If C<$reject> is given and not undef, then it determines when the data is
1115to be rejected: it is matched against the data when the C<$accept> regex
1116does not match and generates an C<EBADMSG> error when it matches. This is
1117useful to quickly reject wrong data (to avoid waiting for a timeout or a
1118receive buffer overflow).
1119
1120Example: expect a single decimal number followed by whitespace, reject
1121anything else (not the use of an anchor).
1122
1123 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1124
1125If C<$skip> is given and not C<undef>, then it will be matched against
1126the receive buffer when neither C<$accept> nor C<$reject> match,
1127and everything preceding and including the match will be accepted
1128unconditionally. This is useful to skip large amounts of data that you
1129know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1130have to start matching from the beginning. This is purely an optimisation
1131and is usually worth only when you expect more than a few kilobytes.
1132
1133Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1134expect the header to be very large (it isn't in practise, but...), we use
1135a skip regex to skip initial portions. The skip regex is tricky in that
1136it only accepts something not ending in either \015 or \012, as these are
1137required for the accept regex.
1138
1139 $handle->push_read (regex =>
1140 qr<\015\012\015\012>,
1141 undef, # no reject
1142 qr<^.*[^\015\012]>,
1143 sub { ... });
1144
1145=cut
1146
1147register_read_type regex => sub {
1148 my ($self, $cb, $accept, $reject, $skip) = @_;
1149
1150 my $data;
1151 my $rbuf = \$self->{rbuf};
1152
1153 sub {
1154 # accept
1155 if ($$rbuf =~ $accept) {
1156 $data .= substr $$rbuf, 0, $+[0], "";
1157 $cb->($self, $data);
157 return; 1158 return 1;
1159 }
1160
1161 # reject
1162 if ($reject && $$rbuf =~ $reject) {
1163 $self->_error (&Errno::EBADMSG);
1164 }
1165
1166 # skip
1167 if ($skip && $$rbuf =~ $skip) {
1168 $data .= substr $$rbuf, 0, $+[0], "";
1169 }
1170
1171 ()
158 } 1172 }
159 1173};
160 $self->{on_read_w} = 1174
161 AnyEvent->io (poll => 'r', fh => $self->{fh}, cb => sub { 1175=item netstring => $cb->($handle, $string)
162 #d# warn "READ:[$self->{read_size}] $self->{read_block_size} : ".length ($self->{rbuf})."\n"; 1176
163 my $rbuf_len = length $self->{rbuf}; 1177A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
164 my $l; 1178
165 if (defined $self->{read_size}) { 1179Throws an error with C<$!> set to EBADMSG on format violations.
166 $l = sysread $self->{fh}, $self->{rbuf}, 1180
167 ($self->{read_size} - $rbuf_len), $rbuf_len; 1181=cut
168 } else { 1182
169 $l = sysread $self->{fh}, $self->{rbuf}, $self->{read_block_size}, $rbuf_len; 1183register_read_type netstring => sub {
1184 my ($self, $cb) = @_;
1185
1186 sub {
1187 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1188 if ($_[0]{rbuf} =~ /[^0-9]/) {
1189 $self->_error (&Errno::EBADMSG);
170 } 1190 }
171 #d# warn "READL $l [$self->{rbuf}]\n"; 1191 return;
1192 }
172 1193
173 if (not defined $l) { 1194 my $len = $1;
174 return if $! == EAGAIN || $! == EINTR;
175 $self->{on_error}->($self) if $self->{on_error};
176 delete $self->{on_read_w};
177 1195
178 } elsif ($l == 0) { 1196 $self->unshift_read (chunk => $len, sub {
179 $self->{on_eof}->($self) if $self->{on_eof}; 1197 my $string = $_[1];
180 delete $self->{on_read_w}; 1198 $_[0]->unshift_read (chunk => 1, sub {
181 1199 if ($_[1] eq ",") {
1200 $cb->($_[0], $string);
182 } else { 1201 } else {
183 $self->{on_read}->($self); 1202 $self->_error (&Errno::EBADMSG);
1203 }
1204 });
1205 });
1206
1207 1
1208 }
1209};
1210
1211=item packstring => $format, $cb->($handle, $string)
1212
1213An octet string prefixed with an encoded length. The encoding C<$format>
1214uses the same format as a Perl C<pack> format, but must specify a single
1215integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1216optional C<!>, C<< < >> or C<< > >> modifier).
1217
1218For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1219EPP uses a prefix of C<N> (4 octtes).
1220
1221Example: read a block of data prefixed by its length in BER-encoded
1222format (very efficient).
1223
1224 $handle->push_read (packstring => "w", sub {
1225 my ($handle, $data) = @_;
1226 });
1227
1228=cut
1229
1230register_read_type packstring => sub {
1231 my ($self, $cb, $format) = @_;
1232
1233 sub {
1234 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1235 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1236 or return;
1237
1238 $format = length pack $format, $len;
1239
1240 # bypass unshift if we already have the remaining chunk
1241 if ($format + $len <= length $_[0]{rbuf}) {
1242 my $data = substr $_[0]{rbuf}, $format, $len;
1243 substr $_[0]{rbuf}, 0, $format + $len, "";
1244 $cb->($_[0], $data);
1245 } else {
1246 # remove prefix
1247 substr $_[0]{rbuf}, 0, $format, "";
1248
1249 # read remaining chunk
1250 $_[0]->unshift_read (chunk => $len, $cb);
1251 }
1252
1253 1
1254 }
1255};
1256
1257=item json => $cb->($handle, $hash_or_arrayref)
1258
1259Reads a JSON object or array, decodes it and passes it to the
1260callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1261
1262If a C<json> object was passed to the constructor, then that will be used
1263for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1264
1265This read type uses the incremental parser available with JSON version
12662.09 (and JSON::XS version 2.2) and above. You have to provide a
1267dependency on your own: this module will load the JSON module, but
1268AnyEvent does not depend on it itself.
1269
1270Since JSON texts are fully self-delimiting, the C<json> read and write
1271types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1272the C<json> write type description, above, for an actual example.
1273
1274=cut
1275
1276register_read_type json => sub {
1277 my ($self, $cb) = @_;
1278
1279 my $json = $self->{json} ||=
1280 eval { require JSON::XS; JSON::XS->new->utf8 }
1281 || do { require JSON; JSON->new->utf8 };
1282
1283 my $data;
1284 my $rbuf = \$self->{rbuf};
1285
1286 sub {
1287 my $ref = eval { $json->incr_parse ($self->{rbuf}) };
1288
1289 if ($ref) {
1290 $self->{rbuf} = $json->incr_text;
1291 $json->incr_text = "";
1292 $cb->($self, $ref);
1293
1294 1
1295 } elsif ($@) {
1296 # error case
1297 $json->incr_skip;
1298
1299 $self->{rbuf} = $json->incr_text;
1300 $json->incr_text = "";
1301
1302 $self->_error (&Errno::EBADMSG);
1303
1304 ()
1305 } else {
1306 $self->{rbuf} = "";
1307
1308 ()
1309 }
1310 }
1311};
1312
1313=item storable => $cb->($handle, $ref)
1314
1315Deserialises a L<Storable> frozen representation as written by the
1316C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1317data).
1318
1319Raises C<EBADMSG> error if the data could not be decoded.
1320
1321=cut
1322
1323register_read_type storable => sub {
1324 my ($self, $cb) = @_;
1325
1326 require Storable;
1327
1328 sub {
1329 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1330 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1331 or return;
1332
1333 my $format = length pack "w", $len;
1334
1335 # bypass unshift if we already have the remaining chunk
1336 if ($format + $len <= length $_[0]{rbuf}) {
1337 my $data = substr $_[0]{rbuf}, $format, $len;
1338 substr $_[0]{rbuf}, 0, $format + $len, "";
1339 $cb->($_[0], Storable::thaw ($data));
1340 } else {
1341 # remove prefix
1342 substr $_[0]{rbuf}, 0, $format, "";
1343
1344 # read remaining chunk
1345 $_[0]->unshift_read (chunk => $len, sub {
1346 if (my $ref = eval { Storable::thaw ($_[1]) }) {
1347 $cb->($_[0], $ref);
1348 } else {
1349 $self->_error (&Errno::EBADMSG);
1350 }
1351 });
1352 }
1353
1354 1
1355 }
1356};
1357
1358=back
1359
1360=item AnyEvent::Handle::register_read_type type => $coderef->($handle, $cb, @args)
1361
1362This function (not method) lets you add your own types to C<push_read>.
1363
1364Whenever the given C<type> is used, C<push_read> will invoke the code
1365reference with the handle object, the callback and the remaining
1366arguments.
1367
1368The code reference is supposed to return a callback (usually a closure)
1369that works as a plain read callback (see C<< ->push_read ($cb) >>).
1370
1371It should invoke the passed callback when it is done reading (remember to
1372pass C<$handle> as first argument as all other callbacks do that).
1373
1374Note that this is a function, and all types registered this way will be
1375global, so try to use unique names.
1376
1377For examples, see the source of this module (F<perldoc -m AnyEvent::Handle>,
1378search for C<register_read_type>)).
1379
1380=item $handle->stop_read
1381
1382=item $handle->start_read
1383
1384In rare cases you actually do not want to read anything from the
1385socket. In this case you can call C<stop_read>. Neither C<on_read> nor
1386any queued callbacks will be executed then. To start reading again, call
1387C<start_read>.
1388
1389Note that AnyEvent::Handle will automatically C<start_read> for you when
1390you change the C<on_read> callback or push/unshift a read callback, and it
1391will automatically C<stop_read> for you when neither C<on_read> is set nor
1392there are any read requests in the queue.
1393
1394These methods will have no effect when in TLS mode (as TLS doesn't support
1395half-duplex connections).
1396
1397=cut
1398
1399sub stop_read {
1400 my ($self) = @_;
1401
1402 delete $self->{_rw} unless $self->{tls};
1403}
1404
1405sub start_read {
1406 my ($self) = @_;
1407
1408 unless ($self->{_rw} || $self->{_eof}) {
1409 Scalar::Util::weaken $self;
1410
1411 $self->{_rw} = AnyEvent->io (fh => $self->{fh}, poll => "r", cb => sub {
1412 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
1413 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size} || 8192, length $$rbuf;
1414
1415 if ($len > 0) {
1416 $self->{_activity} = AnyEvent->now;
1417
1418 if ($self->{tls}) {
1419 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1420
1421 &_dotls ($self);
1422 } else {
1423 $self->_drain_rbuf unless $self->{_in_drain};
1424 }
1425
1426 } elsif (defined $len) {
1427 delete $self->{_rw};
1428 $self->{_eof} = 1;
1429 $self->_drain_rbuf unless $self->{_in_drain};
1430
1431 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
1432 return $self->_error ($!, 1);
184 } 1433 }
185 }); 1434 });
1435 }
186} 1436}
187 1437
188=item B<on_error ($callback)> 1438our $ERROR_SYSCALL;
1439our $ERROR_WANT_READ;
189 1440
190Whenever a read or write operation resulted in an error the C<$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 { 1441sub _tls_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, $cb) = @_; 1442 my ($self, $err) = @_;
240 1443
241 $self->{read_cb} = $cb; 1444 return $self->_error ($!, 1)
242 my $old_blk_size = $self->{read_block_size}; 1445 if $err == Net::SSLeay::ERROR_SYSCALL ();
243 $self->{read_block_size} = $len;
244 1446
245 $self->on_read (sub { 1447 my $err =Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
246 #d# warn "OFOFO $len || ".length($_[0]->{rbuf})."||\n";
247 1448
248 if ($len == length $_[0]->{rbuf}) { 1449 # reduce error string to look less scary
249 $_[0]->{read_block_size} = $old_blk_size; 1450 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
250 $_[0]->on_read (undef); 1451
251 $_[0]->{read_cb}->($_[0], (substr $self->{rbuf}, 0, $len, '')); 1452 if ($self->{_on_starttls}) {
1453 (delete $self->{_on_starttls})->($self, undef, $err);
1454 &_freetls;
1455 } else {
1456 &_freetls;
1457 $self->_error (&Errno::EPROTO, 1, $err);
1458 }
1459}
1460
1461# poll the write BIO and send the data if applicable
1462# also decode read data if possible
1463# this is basiclaly our TLS state machine
1464# more efficient implementations are possible with openssl,
1465# but not with the buggy and incomplete Net::SSLeay.
1466sub _dotls {
1467 my ($self) = @_;
1468
1469 my $tmp;
1470
1471 if (length $self->{_tls_wbuf}) {
1472 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
1473 substr $self->{_tls_wbuf}, 0, $tmp, "";
252 } 1474 }
1475
1476 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1477 return $self->_tls_error ($tmp)
1478 if $tmp != $ERROR_WANT_READ
1479 && ($tmp != $ERROR_SYSCALL || $!);
253 }); 1480 }
254}
255 1481
256=item B<readlines ($callback)> 1482 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1483 unless (length $tmp) {
1484 $self->{_on_starttls}
1485 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1486 &_freetls;
257 1487
258=item B<readlines ($sep, $callback)> 1488 if ($self->{on_stoptls}) {
259 1489 $self->{on_stoptls}($self);
260This method will read lines from the filehandle, seperated by C<$sep> or C<"\n"> 1490 return;
261if C<$sep> is not provided. C<$sep> will be used as part of a regex, so it can be 1491 } else {
262a regex itself and won't be quoted! 1492 # let's treat SSL-eof as we treat normal EOF
263 1493 delete $self->{_rw};
264The C<$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]; 1494 $self->{_eof} = 1;
316 } 1495 }
317 } 1496 }
318 return; 1497
1498 $self->{_tls_rbuf} .= $tmp;
1499 $self->_drain_rbuf unless $self->{_in_drain};
1500 $self->{tls} or return; # tls session might have gone away in callback
1501 }
1502
1503 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
1504 return $self->_tls_error ($tmp)
1505 if $tmp != $ERROR_WANT_READ
1506 && ($tmp != $ERROR_SYSCALL || $!);
1507
1508 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
1509 $self->{wbuf} .= $tmp;
1510 $self->_drain_wbuf;
1511 }
1512
1513 $self->{_on_starttls}
1514 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
1515 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
1516}
1517
1518=item $handle->starttls ($tls[, $tls_ctx])
1519
1520Instead of starting TLS negotiation immediately when the AnyEvent::Handle
1521object is created, you can also do that at a later time by calling
1522C<starttls>.
1523
1524The first argument is the same as the C<tls> constructor argument (either
1525C<"connect">, C<"accept"> or an existing Net::SSLeay object).
1526
1527The second argument is the optional C<AnyEvent::TLS> object that is used
1528when AnyEvent::Handle has to create its own TLS connection object, or
1529a hash reference with C<< key => value >> pairs that will be used to
1530construct a new context.
1531
1532The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1533context in C<< $handle->{tls_ctx} >> after this call and can be used or
1534changed to your liking. Note that the handshake might have already started
1535when this function returns.
1536
1537If it an error to start a TLS handshake more than once per
1538AnyEvent::Handle object (this is due to bugs in OpenSSL).
1539
1540=cut
1541
1542our %TLS_CACHE; #TODO not yet documented, should we?
1543
1544sub starttls {
1545 my ($self, $ssl, $ctx) = @_;
1546
1547 require Net::SSLeay;
1548
1549 Carp::croak "it is an error to call starttls more than once on an AnyEvent::Handle object"
1550 if $self->{tls};
1551
1552 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
1553 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
1554
1555 $ctx ||= $self->{tls_ctx};
1556
1557 if ("HASH" eq ref $ctx) {
1558 require AnyEvent::TLS;
1559
1560 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context
1561
1562 if ($ctx->{cache}) {
1563 my $key = $ctx+0;
1564 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
1565 } else {
1566 $ctx = new AnyEvent::TLS %$ctx;
1567 }
1568 }
319 } 1569
1570 $self->{tls_ctx} = $ctx || TLS_CTX ();
1571 $self->{tls} = $ssl = $self->{tls_ctx}->_get_session ($ssl, $self, $self->{peername});
320 1572
321 my $wba = shift @{$self->{write_bufs}} 1573 # basically, this is deep magic (because SSL_read should have the same issues)
322 or return; 1574 # but the openssl maintainers basically said: "trust us, it just works".
1575 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
1576 # and mismaintained ssleay-module doesn't even offer them).
1577 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
1578 #
1579 # in short: this is a mess.
1580 #
1581 # note that we do not try to keep the length constant between writes as we are required to do.
1582 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
1583 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
1584 # have identity issues in that area.
1585# Net::SSLeay::CTX_set_mode ($ssl,
1586# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
1587# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
1588 Net::SSLeay::CTX_set_mode ($ssl, 1|2);
323 1589
324 unless (defined $wba->[0]) { 1590 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
325 $wba->[1]->($self) if $wba->[1]; 1591 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
326 $self->_check_writer; 1592
327 return; 1593 Net::SSLeay::set_bio ($ssl, $self->{_rbio}, $self->{_wbio});
1594
1595 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
1596 if $self->{on_starttls};
1597
1598 &_dotls; # need to trigger the initial handshake
1599 $self->start_read; # make sure we actually do read
1600}
1601
1602=item $handle->stoptls
1603
1604Shuts down the SSL connection - this makes a proper EOF handshake by
1605sending a close notify to the other side, but since OpenSSL doesn't
1606support non-blocking shut downs, it is not possible to re-use the stream
1607afterwards.
1608
1609=cut
1610
1611sub stoptls {
1612 my ($self) = @_;
1613
1614 if ($self->{tls}) {
1615 Net::SSLeay::shutdown ($self->{tls});
1616
1617 &_dotls;
1618
1619# # we don't give a shit. no, we do, but we can't. no...#d#
1620# # we, we... have to use openssl :/#d#
1621# &_freetls;#d#
1622 }
1623}
1624
1625sub _freetls {
1626 my ($self) = @_;
1627
1628 return unless $self->{tls};
1629
1630 $self->{tls_ctx}->_put_session (delete $self->{tls});
328 } 1631
1632 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
1633}
329 1634
330 $self->{wbuf} = $wba->[0]; 1635sub DESTROY {
331 $self->{write_cb} = $wba->[1]; 1636 my ($self) = @_;
332 1637
333 $self->{write_w} = 1638 &_freetls;
334 AnyEvent->io (poll => 'w', fh => $self->{fh}, cb => sub { 1639
1640 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
1641
1642 if ($linger && length $self->{wbuf}) {
1643 my $fh = delete $self->{fh};
1644 my $wbuf = delete $self->{wbuf};
1645
1646 my @linger;
1647
1648 push @linger, AnyEvent->io (fh => $fh, poll => "w", cb => sub {
335 my $l = syswrite $self->{fh}, $self->{wbuf}, length $self->{wbuf}; 1649 my $len = syswrite $fh, $wbuf, length $wbuf;
336 1650
337 if (not defined $l) { 1651 if ($len > 0) {
338 return if $! == EAGAIN || $! == EINTR; 1652 substr $wbuf, 0, $len, "";
339 delete $self->{write_w};
340 $self->{on_error}->($self) if $self->{on_error};
341
342 } else { 1653 } else {
343 substr $self->{wbuf}, 0, $l, ''; 1654 @linger = (); # end
344
345 if (length ($self->{wbuf}) == 0) {
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 } 1655 }
355 }); 1656 });
1657 push @linger, AnyEvent->timer (after => $linger, cb => sub {
1658 @linger = ();
1659 });
1660 }
1661}
1662
1663=item $handle->destroy
1664
1665Shuts down the handle object as much as possible - this call ensures that
1666no further callbacks will be invoked and as many resources as possible
1667will be freed. You must not call any methods on the object afterwards.
1668
1669Normally, you can just "forget" any references to an AnyEvent::Handle
1670object and it will simply shut down. This works in fatal error and EOF
1671callbacks, as well as code outside. It does I<NOT> work in a read or write
1672callback, so when you want to destroy the AnyEvent::Handle object from
1673within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
1674that case.
1675
1676The handle might still linger in the background and write out remaining
1677data, as specified by the C<linger> option, however.
1678
1679=cut
1680
1681sub destroy {
1682 my ($self) = @_;
1683
1684 $self->DESTROY;
1685 %$self = ();
1686}
1687
1688=item AnyEvent::Handle::TLS_CTX
1689
1690This function creates and returns the AnyEvent::TLS object used by default
1691for TLS mode.
1692
1693The context is created by calling L<AnyEvent::TLS> without any arguments.
1694
1695=cut
1696
1697our $TLS_CTX;
1698
1699sub TLS_CTX() {
1700 $TLS_CTX ||= do {
1701 require AnyEvent::TLS;
1702
1703 new AnyEvent::TLS
1704 }
356} 1705}
357 1706
358=back 1707=back
359 1708
1709
1710=head1 NONFREQUENTLY ASKED QUESTIONS
1711
1712=over 4
1713
1714=item I C<undef> the AnyEvent::Handle reference inside my callback and
1715still get further invocations!
1716
1717That's because AnyEvent::Handle keeps a reference to itself when handling
1718read or write callbacks.
1719
1720It is only safe to "forget" the reference inside EOF or error callbacks,
1721from within all other callbacks, you need to explicitly call the C<<
1722->destroy >> method.
1723
1724=item I get different callback invocations in TLS mode/Why can't I pause
1725reading?
1726
1727Unlike, say, TCP, TLS connections do not consist of two independent
1728communication channels, one for each direction. Or put differently. The
1729read and write directions are not independent of each other: you cannot
1730write data unless you are also prepared to read, and vice versa.
1731
1732This can mean than, in TLS mode, you might get C<on_error> or C<on_eof>
1733callback invocations when you are not expecting any read data - the reason
1734is that AnyEvent::Handle always reads in TLS mode.
1735
1736During the connection, you have to make sure that you always have a
1737non-empty read-queue, or an C<on_read> watcher. At the end of the
1738connection (or when you no longer want to use it) you can call the
1739C<destroy> method.
1740
1741=item How do I read data until the other side closes the connection?
1742
1743If you just want to read your data into a perl scalar, the easiest way
1744to achieve this is by setting an C<on_read> callback that does nothing,
1745clearing the C<on_eof> callback and in the C<on_error> callback, the data
1746will be in C<$_[0]{rbuf}>:
1747
1748 $handle->on_read (sub { });
1749 $handle->on_eof (undef);
1750 $handle->on_error (sub {
1751 my $data = delete $_[0]{rbuf};
1752 undef $handle;
1753 });
1754
1755The reason to use C<on_error> is that TCP connections, due to latencies
1756and packets loss, might get closed quite violently with an error, when in
1757fact, all data has been received.
1758
1759It is usually better to use acknowledgements when transferring data,
1760to make sure the other side hasn't just died and you got the data
1761intact. This is also one reason why so many internet protocols have an
1762explicit QUIT command.
1763
1764=item I don't want to destroy the handle too early - how do I wait until
1765all data has been written?
1766
1767After writing your last bits of data, set the C<on_drain> callback
1768and destroy the handle in there - with the default setting of
1769C<low_water_mark> this will be called precisely when all data has been
1770written to the socket:
1771
1772 $handle->push_write (...);
1773 $handle->on_drain (sub {
1774 warn "all data submitted to the kernel\n";
1775 undef $handle;
1776 });
1777
1778If you just want to queue some data and then signal EOF to the other side,
1779consider using C<< ->push_shutdown >> instead.
1780
1781=item I want to contact a TLS/SSL server, I don't care about security.
1782
1783If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
1784simply connect to it and then create the AnyEvent::Handle with the C<tls>
1785parameter:
1786
1787 tcp_connect $host, $port, sub {
1788 my ($fh) = @_;
1789
1790 my $handle = new AnyEvent::Handle
1791 fh => $fh,
1792 tls => "connect",
1793 on_error => sub { ... };
1794
1795 $handle->push_write (...);
1796 };
1797
1798=item I want to contact a TLS/SSL server, I do care about security.
1799
1800Then you should additionally enable certificate verification, including
1801peername verification, if the protocol you use supports it (see
1802L<AnyEvent::TLS>, C<verify_peername>).
1803
1804E.g. for HTTPS:
1805
1806 tcp_connect $host, $port, sub {
1807 my ($fh) = @_;
1808
1809 my $handle = new AnyEvent::Handle
1810 fh => $fh,
1811 peername => $host,
1812 tls => "connect",
1813 tls_ctx => { verify => 1, verify_peername => "https" },
1814 ...
1815
1816Note that you must specify the hostname you connected to (or whatever
1817"peername" the protocol needs) as the C<peername> argument, otherwise no
1818peername verification will be done.
1819
1820The above will use the system-dependent default set of trusted CA
1821certificates. If you want to check against a specific CA, add the
1822C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
1823
1824 tls_ctx => {
1825 verify => 1,
1826 verify_peername => "https",
1827 ca_file => "my-ca-cert.pem",
1828 },
1829
1830=item I want to create a TLS/SSL server, how do I do that?
1831
1832Well, you first need to get a server certificate and key. You have
1833three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
1834self-signed certificate (cheap. check the search engine of your choice,
1835there are many tutorials on the net) or c) make your own CA (tinyca2 is a
1836nice program for that purpose).
1837
1838Then create a file with your private key (in PEM format, see
1839L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
1840file should then look like this:
1841
1842 -----BEGIN RSA PRIVATE KEY-----
1843 ...header data
1844 ... lots of base64'y-stuff
1845 -----END RSA PRIVATE KEY-----
1846
1847 -----BEGIN CERTIFICATE-----
1848 ... lots of base64'y-stuff
1849 -----END CERTIFICATE-----
1850
1851The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
1852specify this file as C<cert_file>:
1853
1854 tcp_server undef, $port, sub {
1855 my ($fh) = @_;
1856
1857 my $handle = new AnyEvent::Handle
1858 fh => $fh,
1859 tls => "accept",
1860 tls_ctx => { cert_file => "my-server-keycert.pem" },
1861 ...
1862
1863When you have intermediate CA certificates that your clients might not
1864know about, just append them to the C<cert_file>.
1865
1866=back
1867
1868
1869=head1 SUBCLASSING AnyEvent::Handle
1870
1871In many cases, you might want to subclass AnyEvent::Handle.
1872
1873To make this easier, a given version of AnyEvent::Handle uses these
1874conventions:
1875
1876=over 4
1877
1878=item * all constructor arguments become object members.
1879
1880At least initially, when you pass a C<tls>-argument to the constructor it
1881will end up in C<< $handle->{tls} >>. Those members might be changed or
1882mutated later on (for example C<tls> will hold the TLS connection object).
1883
1884=item * other object member names are prefixed with an C<_>.
1885
1886All object members not explicitly documented (internal use) are prefixed
1887with an underscore character, so the remaining non-C<_>-namespace is free
1888for use for subclasses.
1889
1890=item * all members not documented here and not prefixed with an underscore
1891are free to use in subclasses.
1892
1893Of course, new versions of AnyEvent::Handle may introduce more "public"
1894member variables, but thats just life, at least it is documented.
1895
1896=back
1897
360=head1 AUTHOR 1898=head1 AUTHOR
361 1899
362Robin Redeker, C<< <elmex at ta-sa.org> >> 1900Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
363 1901
364=cut 1902=cut
365 1903
3661; # End of AnyEvent::Handle 19041; # End of AnyEvent::Handle

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines