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 1.151 by root, Thu Jul 16 04:20:23 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 = 4.82;
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 $hdl; $hdl = new AnyEvent::Handle
29 fh => \*STDIN,
30 on_error => sub {
31 my ($hdl, $fatal, $msg) = @_;
32 warn "got error $msg\n";
33 $hdl->destroy;
34 $cv->send;
35 );
30 36
31 $ae_fh->on_eof (sub { $cv->broadcast }); 37 # send some request line
38 $hdl->push_write ("getinfo\015\012");
32 39
33 $ae_fh->readlines (sub { 40 # read the response line
41 $hdl->push_read (line => sub {
34 my ($ae_fh, @lines) = @_; 42 my ($hdl, $line) = @_;
35 for (@lines) { 43 warn "got line <$line>\n";
36 chomp; 44 $cv->send;
37 print "Line: $_"; 45 });
46
47 $cv->recv;
48
49=head1 DESCRIPTION
50
51This module is a helper module to make it easier to do event-based I/O on
52filehandles. For utility functions for doing non-blocking connects and accepts
53on sockets see L<AnyEvent::Util>.
54
55The L<AnyEvent::Intro> tutorial contains some well-documented
56AnyEvent::Handle examples.
57
58In the following, when the documentation refers to of "bytes" then this
59means characters. As sysread and syswrite are used for all I/O, their
60treatment of characters applies to this module as well.
61
62All callbacks will be invoked with the handle object as their first
63argument.
64
65=head1 METHODS
66
67=over 4
68
69=item $handle = B<new> AnyEvent::TLS fh => $filehandle, key => value...
70
71The constructor supports these arguments (all as C<< key => value >> pairs).
72
73=over 4
74
75=item fh => $filehandle [MANDATORY]
76
77The filehandle this L<AnyEvent::Handle> object will operate on.
78
79NOTE: The filehandle will be set to non-blocking mode (using
80C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
81that mode.
82
83=item on_eof => $cb->($handle)
84
85Set the callback to be called when an end-of-file condition is detected,
86i.e. in the case of a socket, when the other side has closed the
87connection cleanly, and there are no outstanding read requests in the
88queue (if there are read requests, then an EOF counts as an unexpected
89connection close and will be flagged as an error).
90
91For sockets, this just means that the other side has stopped sending data,
92you can still try to write data, and, in fact, one can return from the EOF
93callback and continue writing data, as only the read part has been shut
94down.
95
96If an EOF condition has been detected but no C<on_eof> callback has been
97set, then a fatal error will be raised with C<$!> set to <0>.
98
99=item on_error => $cb->($handle, $fatal, $message)
100
101This is the error callback, which is called when, well, some error
102occured, such as not being able to resolve the hostname, failure to
103connect or a read error.
104
105Some errors are fatal (which is indicated by C<$fatal> being true). On
106fatal errors the handle object will be destroyed (by a call to C<< ->
107destroy >>) after invoking the error callback (which means you are free to
108examine the handle object). Examples of fatal errors are an EOF condition
109with active (but unsatisifable) read watchers (C<EPIPE>) or I/O errors.
110
111AnyEvent::Handle tries to find an appropriate error code for you to check
112against, but in some cases (TLS errors), this does not work well. It is
113recommended to always output the C<$message> argument in human-readable
114error messages (it's usually the same as C<"$!">).
115
116Non-fatal errors can be retried by simply returning, but it is recommended
117to simply ignore this parameter and instead abondon the handle object
118when this callback is invoked. Examples of non-fatal errors are timeouts
119C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
120
121On callback entrance, the value of C<$!> contains the operating system
122error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
123C<EPROTO>).
124
125While not mandatory, it is I<highly> recommended to set this callback, as
126you will not be notified of errors otherwise. The default simply calls
127C<croak>.
128
129=item on_read => $cb->($handle)
130
131This sets the default read callback, which is called when data arrives
132and no read request is in the queue (unlike read queue callbacks, this
133callback will only be called when at least one octet of data is in the
134read buffer).
135
136To access (and remove data from) the read buffer, use the C<< ->rbuf >>
137method or access the C<< $handle->{rbuf} >> member directly. Note that you
138must not enlarge or modify the read buffer, you can only remove data at
139the beginning from it.
140
141When an EOF condition is detected then AnyEvent::Handle will first try to
142feed all the remaining data to the queued callbacks and C<on_read> before
143calling the C<on_eof> callback. If no progress can be made, then a fatal
144error will be raised (with C<$!> set to C<EPIPE>).
145
146Note that, unlike requests in the read queue, an C<on_read> callback
147doesn't mean you I<require> some data: if there is an EOF and there
148are outstanding read requests then an error will be flagged. With an
149C<on_read> callback, the C<on_eof> callback will be invoked.
150
151=item on_drain => $cb->($handle)
152
153This sets the callback that is called when the write buffer becomes empty
154(or when the callback is set and the buffer is empty already).
155
156To append to the write buffer, use the C<< ->push_write >> method.
157
158This callback is useful when you don't want to put all of your write data
159into the queue at once, for example, when you want to write the contents
160of some file to the socket you might not want to read the whole file into
161memory and push it into the queue, but instead only read more data from
162the file when the write queue becomes empty.
163
164=item timeout => $fractional_seconds
165
166If non-zero, then this enables an "inactivity" timeout: whenever this many
167seconds pass without a successful read or write on the underlying file
168handle, the C<on_timeout> callback will be invoked (and if that one is
169missing, a non-fatal C<ETIMEDOUT> error will be raised).
170
171Note that timeout processing is also active when you currently do not have
172any outstanding read or write requests: If you plan to keep the connection
173idle then you should disable the timout temporarily or ignore the timeout
174in the C<on_timeout> callback, in which case AnyEvent::Handle will simply
175restart the timeout.
176
177Zero (the default) disables this timeout.
178
179=item on_timeout => $cb->($handle)
180
181Called whenever the inactivity timeout passes. If you return from this
182callback, then the timeout will be reset as if some activity had happened,
183so this condition is not fatal in any way.
184
185=item rbuf_max => <bytes>
186
187If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
188when the read buffer ever (strictly) exceeds this size. This is useful to
189avoid some forms of denial-of-service attacks.
190
191For example, a server accepting connections from untrusted sources should
192be configured to accept only so-and-so much data that it cannot act on
193(for example, when expecting a line, an attacker could send an unlimited
194amount of data without a callback ever being called as long as the line
195isn't finished).
196
197=item autocork => <boolean>
198
199When disabled (the default), then C<push_write> will try to immediately
200write the data to the handle, if possible. This avoids having to register
201a write watcher and wait for the next event loop iteration, but can
202be inefficient if you write multiple small chunks (on the wire, this
203disadvantage is usually avoided by your kernel's nagle algorithm, see
204C<no_delay>, but this option can save costly syscalls).
205
206When enabled, then writes will always be queued till the next event loop
207iteration. This is efficient when you do many small writes per iteration,
208but less efficient when you do a single write only per iteration (or when
209the write buffer often is full). It also increases write latency.
210
211=item no_delay => <boolean>
212
213When doing small writes on sockets, your operating system kernel might
214wait a bit for more data before actually sending it out. This is called
215the Nagle algorithm, and usually it is beneficial.
216
217In some situations you want as low a delay as possible, which can be
218accomplishd by setting this option to a true value.
219
220The default is your opertaing system's default behaviour (most likely
221enabled), this option explicitly enables or disables it, if possible.
222
223=item read_size => <bytes>
224
225The default read block size (the amount of bytes this module will
226try to read during each loop iteration, which affects memory
227requirements). Default: C<8192>.
228
229=item low_water_mark => <bytes>
230
231Sets the amount of bytes (default: C<0>) that make up an "empty" write
232buffer: If the write reaches this size or gets even samller it is
233considered empty.
234
235Sometimes it can be beneficial (for performance reasons) to add data to
236the write buffer before it is fully drained, but this is a rare case, as
237the operating system kernel usually buffers data as well, so the default
238is good in almost all cases.
239
240=item linger => <seconds>
241
242If non-zero (default: C<3600>), then the destructor of the
243AnyEvent::Handle object will check whether there is still outstanding
244write data and will install a watcher that will write this data to the
245socket. No errors will be reported (this mostly matches how the operating
246system treats outstanding data at socket close time).
247
248This will not work for partial TLS data that could not be encoded
249yet. This data will be lost. Calling the C<stoptls> method in time might
250help.
251
252=item peername => $string
253
254A string used to identify the remote site - usually the DNS hostname
255(I<not> IDN!) used to create the connection, rarely the IP address.
256
257Apart from being useful in error messages, this string is also used in TLS
258peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
259verification will be skipped when C<peername> is not specified or
260C<undef>.
261
262=item tls => "accept" | "connect" | Net::SSLeay::SSL object
263
264When this parameter is given, it enables TLS (SSL) mode, that means
265AnyEvent will start a TLS handshake as soon as the conenction has been
266established and will transparently encrypt/decrypt data afterwards.
267
268All TLS protocol errors will be signalled as C<EPROTO>, with an
269appropriate error message.
270
271TLS mode requires Net::SSLeay to be installed (it will be loaded
272automatically when you try to create a TLS handle): this module doesn't
273have a dependency on that module, so if your module requires it, you have
274to add the dependency yourself.
275
276Unlike TCP, TLS has a server and client side: for the TLS server side, use
277C<accept>, and for the TLS client side of a connection, use C<connect>
278mode.
279
280You can also provide your own TLS connection object, but you have
281to make sure that you call either C<Net::SSLeay::set_connect_state>
282or C<Net::SSLeay::set_accept_state> on it before you pass it to
283AnyEvent::Handle. Also, this module will take ownership of this connection
284object.
285
286At some future point, AnyEvent::Handle might switch to another TLS
287implementation, then the option to use your own session object will go
288away.
289
290B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
291passing in the wrong integer will lead to certain crash. This most often
292happens when one uses a stylish C<< tls => 1 >> and is surprised about the
293segmentation fault.
294
295See the C<< ->starttls >> method for when need to start TLS negotiation later.
296
297=item tls_ctx => $anyevent_tls
298
299Use the given C<AnyEvent::TLS> object to create the new TLS connection
300(unless a connection object was specified directly). If this parameter is
301missing, then AnyEvent::Handle will use C<AnyEvent::Handle::TLS_CTX>.
302
303Instead of an object, you can also specify a hash reference with C<< key
304=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
305new TLS context object.
306
307=item on_starttls => $cb->($handle, $success[, $error_message])
308
309This callback will be invoked when the TLS/SSL handshake has finished. If
310C<$success> is true, then the TLS handshake succeeded, otherwise it failed
311(C<on_stoptls> will not be called in this case).
312
313The session in C<< $handle->{tls} >> can still be examined in this
314callback, even when the handshake was not successful.
315
316TLS handshake failures will not cause C<on_error> to be invoked when this
317callback is in effect, instead, the error message will be passed to C<on_starttls>.
318
319Without this callback, handshake failures lead to C<on_error> being
320called, as normal.
321
322Note that you cannot call C<starttls> right again in this callback. If you
323need to do that, start an zero-second timer instead whose callback can
324then call C<< ->starttls >> again.
325
326=item on_stoptls => $cb->($handle)
327
328When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
329set, then it will be invoked after freeing the TLS session. If it is not,
330then a TLS shutdown condition will be treated like a normal EOF condition
331on the handle.
332
333The session in C<< $handle->{tls} >> can still be examined in this
334callback.
335
336This callback will only be called on TLS shutdowns, not when the
337underlying handle signals EOF.
338
339=item json => JSON or JSON::XS object
340
341This is the json coder object used by the C<json> read and write types.
342
343If you don't supply it, then AnyEvent::Handle will create and use a
344suitable one (on demand), which will write and expect UTF-8 encoded JSON
345texts.
346
347Note that you are responsible to depend on the JSON module if you want to
348use this functionality, as AnyEvent does not have a dependency itself.
349
350=back
351
352=cut
353
354sub new {
355 my $class = shift;
356 my $self = bless { @_ }, $class;
357
358 $self->{fh} or Carp::croak "mandatory argument fh is missing";
359
360 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
361
362 $self->{_activity} = AnyEvent->now;
363 $self->_timeout;
364
365 $self->no_delay (delete $self->{no_delay}) if exists $self->{no_delay};
366
367 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
368 if $self->{tls};
369
370 $self->on_drain (delete $self->{on_drain}) if $self->{on_drain};
371
372 $self->start_read
373 if $self->{on_read};
374
375 $self->{fh} && $self
376}
377
378#sub _shutdown {
379# my ($self) = @_;
380#
381# delete @$self{qw(_tw _rw _ww fh wbuf on_read _queue)};
382# $self->{_eof} = 1; # tell starttls et. al to stop trying
383#
384# &_freetls;
385#}
386
387sub _error {
388 my ($self, $errno, $fatal, $message) = @_;
389
390 $! = $errno;
391 $message ||= "$!";
392
393 if ($self->{on_error}) {
394 $self->{on_error}($self, $fatal, $message);
395 $self->destroy if $fatal;
396 } elsif ($self->{fh}) {
397 $self->destroy;
398 Carp::croak "AnyEvent::Handle uncaught error: $message";
399 }
400}
401
402=item $fh = $handle->fh
403
404This method returns the file handle used to create the L<AnyEvent::Handle> object.
405
406=cut
407
408sub fh { $_[0]{fh} }
409
410=item $handle->on_error ($cb)
411
412Replace the current C<on_error> callback (see the C<on_error> constructor argument).
413
414=cut
415
416sub on_error {
417 $_[0]{on_error} = $_[1];
418}
419
420=item $handle->on_eof ($cb)
421
422Replace the current C<on_eof> callback (see the C<on_eof> constructor argument).
423
424=cut
425
426sub on_eof {
427 $_[0]{on_eof} = $_[1];
428}
429
430=item $handle->on_timeout ($cb)
431
432Replace the current C<on_timeout> callback, or disables the callback (but
433not the timeout) if C<$cb> = C<undef>. See the C<timeout> constructor
434argument and method.
435
436=cut
437
438sub on_timeout {
439 $_[0]{on_timeout} = $_[1];
440}
441
442=item $handle->autocork ($boolean)
443
444Enables or disables the current autocork behaviour (see C<autocork>
445constructor argument). Changes will only take effect on the next write.
446
447=cut
448
449sub autocork {
450 $_[0]{autocork} = $_[1];
451}
452
453=item $handle->no_delay ($boolean)
454
455Enables or disables the C<no_delay> setting (see constructor argument of
456the same name for details).
457
458=cut
459
460sub no_delay {
461 $_[0]{no_delay} = $_[1];
462
463 eval {
464 local $SIG{__DIE__};
465 setsockopt $_[0]{fh}, &Socket::IPPROTO_TCP, &Socket::TCP_NODELAY, int $_[1];
466 };
467}
468
469=item $handle->on_starttls ($cb)
470
471Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
472
473=cut
474
475sub on_starttls {
476 $_[0]{on_starttls} = $_[1];
477}
478
479=item $handle->on_stoptls ($cb)
480
481Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
482
483=cut
484
485sub on_starttls {
486 $_[0]{on_stoptls} = $_[1];
487}
488
489#############################################################################
490
491=item $handle->timeout ($seconds)
492
493Configures (or disables) the inactivity timeout.
494
495=cut
496
497sub timeout {
498 my ($self, $timeout) = @_;
499
500 $self->{timeout} = $timeout;
501 $self->_timeout;
502}
503
504# reset the timeout watcher, as neccessary
505# also check for time-outs
506sub _timeout {
507 my ($self) = @_;
508
509 if ($self->{timeout}) {
510 my $NOW = AnyEvent->now;
511
512 # when would the timeout trigger?
513 my $after = $self->{_activity} + $self->{timeout} - $NOW;
514
515 # now or in the past already?
516 if ($after <= 0) {
517 $self->{_activity} = $NOW;
518
519 if ($self->{on_timeout}) {
520 $self->{on_timeout}($self);
521 } else {
522 $self->_error (Errno::ETIMEDOUT);
523 }
524
525 # callback could have changed timeout value, optimise
526 return unless $self->{timeout};
527
528 # calculate new after
529 $after = $self->{timeout};
530 }
531
532 Scalar::Util::weaken $self;
533 return unless $self; # ->error could have destroyed $self
534
535 $self->{_tw} ||= AnyEvent->timer (after => $after, cb => sub {
536 delete $self->{_tw};
537 $self->_timeout;
538 });
539 } else {
540 delete $self->{_tw};
541 }
542}
543
544#############################################################################
545
546=back
547
548=head2 WRITE QUEUE
549
550AnyEvent::Handle manages two queues per handle, one for writing and one
551for reading.
552
553The write queue is very simple: you can add data to its end, and
554AnyEvent::Handle will automatically try to get rid of it for you.
555
556When data could be written and the write buffer is shorter then the low
557water mark, the C<on_drain> callback will be invoked.
558
559=over 4
560
561=item $handle->on_drain ($cb)
562
563Sets the C<on_drain> callback or clears it (see the description of
564C<on_drain> in the constructor).
565
566=cut
567
568sub on_drain {
569 my ($self, $cb) = @_;
570
571 $self->{on_drain} = $cb;
572
573 $cb->($self)
574 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
575}
576
577=item $handle->push_write ($data)
578
579Queues the given scalar to be written. You can push as much data as you
580want (only limited by the available memory), as C<AnyEvent::Handle>
581buffers it independently of the kernel.
582
583=cut
584
585sub _drain_wbuf {
586 my ($self) = @_;
587
588 if (!$self->{_ww} && length $self->{wbuf}) {
589
590 Scalar::Util::weaken $self;
591
592 my $cb = sub {
593 my $len = syswrite $self->{fh}, $self->{wbuf};
594
595 if (defined $len) {
596 substr $self->{wbuf}, 0, $len, "";
597
598 $self->{_activity} = AnyEvent->now;
599
600 $self->{on_drain}($self)
601 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
602 && $self->{on_drain};
603
604 delete $self->{_ww} unless length $self->{wbuf};
605 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
606 $self->_error ($!, 1);
607 }
608 };
609
610 # try to write data immediately
611 $cb->() unless $self->{autocork};
612
613 # if still data left in wbuf, we need to poll
614 $self->{_ww} = AnyEvent->io (fh => $self->{fh}, poll => "w", cb => $cb)
615 if length $self->{wbuf};
616 };
617}
618
619our %WH;
620
621sub register_write_type($$) {
622 $WH{$_[0]} = $_[1];
623}
624
625sub push_write {
626 my $self = shift;
627
628 if (@_ > 1) {
629 my $type = shift;
630
631 @_ = ($WH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::push_write")
632 ->($self, @_);
633 }
634
635 if ($self->{tls}) {
636 $self->{_tls_wbuf} .= $_[0];
637
638 &_dotls ($self);
639 } else {
640 $self->{wbuf} .= $_[0];
641 $self->_drain_wbuf;
642 }
643}
644
645=item $handle->push_write (type => @args)
646
647Instead of formatting your data yourself, you can also let this module do
648the job by specifying a type and type-specific arguments.
649
650Predefined types are (if you have ideas for additional types, feel free to
651drop by and tell us):
652
653=over 4
654
655=item netstring => $string
656
657Formats the given value as netstring
658(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
659
660=cut
661
662register_write_type netstring => sub {
663 my ($self, $string) = @_;
664
665 (length $string) . ":$string,"
666};
667
668=item packstring => $format, $data
669
670An octet string prefixed with an encoded length. The encoding C<$format>
671uses the same format as a Perl C<pack> format, but must specify a single
672integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
673optional C<!>, C<< < >> or C<< > >> modifier).
674
675=cut
676
677register_write_type packstring => sub {
678 my ($self, $format, $string) = @_;
679
680 pack "$format/a*", $string
681};
682
683=item json => $array_or_hashref
684
685Encodes the given hash or array reference into a JSON object. Unless you
686provide your own JSON object, this means it will be encoded to JSON text
687in UTF-8.
688
689JSON objects (and arrays) are self-delimiting, so you can write JSON at
690one end of a handle and read them at the other end without using any
691additional framing.
692
693The generated JSON text is guaranteed not to contain any newlines: While
694this module doesn't need delimiters after or between JSON texts to be
695able to read them, many other languages depend on that.
696
697A simple RPC protocol that interoperates easily with others is to send
698JSON arrays (or objects, although arrays are usually the better choice as
699they mimic how function argument passing works) and a newline after each
700JSON text:
701
702 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
703 $handle->push_write ("\012");
704
705An AnyEvent::Handle receiver would simply use the C<json> read type and
706rely on the fact that the newline will be skipped as leading whitespace:
707
708 $handle->push_read (json => sub { my $array = $_[1]; ... });
709
710Other languages could read single lines terminated by a newline and pass
711this line into their JSON decoder of choice.
712
713=cut
714
715register_write_type json => sub {
716 my ($self, $ref) = @_;
717
718 require JSON;
719
720 $self->{json} ? $self->{json}->encode ($ref)
721 : JSON::encode_json ($ref)
722};
723
724=item storable => $reference
725
726Freezes the given reference using L<Storable> and writes it to the
727handle. Uses the C<nfreeze> format.
728
729=cut
730
731register_write_type storable => sub {
732 my ($self, $ref) = @_;
733
734 require Storable;
735
736 pack "w/a*", Storable::nfreeze ($ref)
737};
738
739=back
740
741=item $handle->push_shutdown
742
743Sometimes you know you want to close the socket after writing your data
744before it was actually written. One way to do that is to replace your
745C<on_drain> handler by a callback that shuts down the socket (and set
746C<low_water_mark> to C<0>). This method is a shorthand for just that, and
747replaces the C<on_drain> callback with:
748
749 sub { shutdown $_[0]{fh}, 1 } # for push_shutdown
750
751This simply shuts down the write side and signals an EOF condition to the
752the peer.
753
754You can rely on the normal read queue and C<on_eof> handling
755afterwards. This is the cleanest way to close a connection.
756
757=cut
758
759sub push_shutdown {
760 my ($self) = @_;
761
762 delete $self->{low_water_mark};
763 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
764}
765
766=item AnyEvent::Handle::register_write_type type => $coderef->($handle, @args)
767
768This function (not method) lets you add your own types to C<push_write>.
769Whenever the given C<type> is used, C<push_write> will invoke the code
770reference with the handle object and the remaining arguments.
771
772The code reference is supposed to return a single octet string that will
773be appended to the write buffer.
774
775Note that this is a function, and all types registered this way will be
776global, so try to use unique names.
777
778=cut
779
780#############################################################################
781
782=back
783
784=head2 READ QUEUE
785
786AnyEvent::Handle manages two queues per handle, one for writing and one
787for reading.
788
789The read queue is more complex than the write queue. It can be used in two
790ways, the "simple" way, using only C<on_read> and the "complex" way, using
791a queue.
792
793In the simple case, you just install an C<on_read> callback and whenever
794new data arrives, it will be called. You can then remove some data (if
795enough is there) from the read buffer (C<< $handle->rbuf >>). Or you cna
796leave the data there if you want to accumulate more (e.g. when only a
797partial message has been received so far).
798
799In the more complex case, you want to queue multiple callbacks. In this
800case, AnyEvent::Handle will call the first queued callback each time new
801data arrives (also the first time it is queued) and removes it when it has
802done its job (see C<push_read>, below).
803
804This way you can, for example, push three line-reads, followed by reading
805a chunk of data, and AnyEvent::Handle will execute them in order.
806
807Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
808the specified number of bytes which give an XML datagram.
809
810 # in the default state, expect some header bytes
811 $handle->on_read (sub {
812 # some data is here, now queue the length-header-read (4 octets)
813 shift->unshift_read (chunk => 4, sub {
814 # header arrived, decode
815 my $len = unpack "N", $_[1];
816
817 # now read the payload
818 shift->unshift_read (chunk => $len, sub {
819 my $xml = $_[1];
820 # handle xml
821 });
822 });
823 });
824
825Example 2: Implement a client for a protocol that replies either with "OK"
826and another line or "ERROR" for the first request that is sent, and 64
827bytes for the second request. Due to the availability of a queue, we can
828just pipeline sending both requests and manipulate the queue as necessary
829in the callbacks.
830
831When the first callback is called and sees an "OK" response, it will
832C<unshift> another line-read. This line-read will be queued I<before> the
83364-byte chunk callback.
834
835 # request one, returns either "OK + extra line" or "ERROR"
836 $handle->push_write ("request 1\015\012");
837
838 # we expect "ERROR" or "OK" as response, so push a line read
839 $handle->push_read (line => sub {
840 # if we got an "OK", we have to _prepend_ another line,
841 # so it will be read before the second request reads its 64 bytes
842 # which are already in the queue when this callback is called
843 # we don't do this in case we got an error
844 if ($_[1] eq "OK") {
845 $_[0]->unshift_read (line => sub {
846 my $response = $_[1];
847 ...
848 });
38 } 849 }
39 }); 850 });
40 851
41 # or use the constructor to pass the callback: 852 # request two, simply returns 64 octets
853 $handle->push_write ("request 2\015\012");
42 854
43 my $ae_fh2 = 855 # simply read 64 bytes, always
44 AnyEvent::Handle->new ( 856 $handle->push_read (chunk => 64, sub {
45 fh => \*STDIN, 857 my $response = $_[1];
46 on_eof => sub { 858 ...
47 $cv->broadcast; 859 });
48 }, 860
49 on_readline => sub { 861=over 4
50 my ($ae_fh, @lines) = @_; 862
51 for (@lines) { 863=cut
52 chomp; 864
53 print "Line: $_"; 865sub _drain_rbuf {
866 my ($self) = @_;
867
868 local $self->{_in_drain} = 1;
869
870 if (
871 defined $self->{rbuf_max}
872 && $self->{rbuf_max} < length $self->{rbuf}
873 ) {
874 $self->_error (Errno::ENOSPC, 1), return;
875 }
876
877 while () {
878 # we need to use a separate tls read buffer, as we must not receive data while
879 # we are draining the buffer, and this can only happen with TLS.
880 $self->{rbuf} .= delete $self->{_tls_rbuf} if exists $self->{_tls_rbuf};
881
882 my $len = length $self->{rbuf};
883
884 if (my $cb = shift @{ $self->{_queue} }) {
885 unless ($cb->($self)) {
886 if ($self->{_eof}) {
887 # no progress can be made (not enough data and no data forthcoming)
888 $self->_error (Errno::EPIPE, 1), return;
54 } 889 }
890
891 unshift @{ $self->{_queue} }, $cb;
892 last;
55 } 893 }
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}) { 894 } elsif ($self->{on_read}) {
121 $self->readlines ($self->{on_readline}); 895 last unless $len;
122 896
897 $self->{on_read}($self);
898
899 if (
900 $len == length $self->{rbuf} # if no data has been consumed
901 && !@{ $self->{_queue} } # and the queue is still empty
902 && $self->{on_read} # but we still have on_read
903 ) {
904 # no further data will arrive
905 # so no progress can be made
906 $self->_error (Errno::EPIPE, 1), return
907 if $self->{_eof};
908
909 last; # more data might arrive
910 }
911 } else {
912 # read side becomes idle
913 delete $self->{_rw} unless $self->{tls};
914 last;
915 }
916 }
917
123 } elsif ($self->{on_eof}) { 918 if ($self->{_eof}) {
124 $self->on_eof ($self->{on_eof});
125
126 } elsif ($self->{on_error}) { 919 if ($self->{on_eof}) {
127 $self->on_eof ($self->{on_error}); 920 $self->{on_eof}($self)
921 } else {
922 $self->_error (0, 1, "Unexpected end-of-file");
923 }
128 } 924 }
129 925
130 return $self 926 # may need to restart read watcher
927 unless ($self->{_rw}) {
928 $self->start_read
929 if $self->{on_read} || @{ $self->{_queue} };
930 }
131} 931}
132 932
133=item B<fh> 933=item $handle->on_read ($cb)
134 934
135This method returns the filehandle of the L<AnyEvent::Handle> object. 935This replaces the currently set C<on_read> callback, or clears it (when
136 936the new callback is C<undef>). See the description of C<on_read> in the
137=cut 937constructor.
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.
148 938
149=cut 939=cut
150 940
151sub on_read { 941sub on_read {
152 my ($self, $cb) = @_; 942 my ($self, $cb) = @_;
943
153 $self->{on_read} = $cb; 944 $self->{on_read} = $cb;
945 $self->_drain_rbuf if $cb && !$self->{_in_drain};
946}
154 947
155 unless (defined $self->{on_read}) { 948=item $handle->rbuf
156 delete $self->{on_read_w}; 949
950Returns the read buffer (as a modifiable lvalue).
951
952You can access the read buffer directly as the C<< ->{rbuf} >>
953member, if you want. However, the only operation allowed on the
954read buffer (apart from looking at it) is removing data from its
955beginning. Otherwise modifying or appending to it is not allowed and will
956lead to hard-to-track-down bugs.
957
958NOTE: The read buffer should only be used or modified if the C<on_read>,
959C<push_read> or C<unshift_read> methods are used. The other read methods
960automatically manage the read buffer.
961
962=cut
963
964sub rbuf : lvalue {
965 $_[0]{rbuf}
966}
967
968=item $handle->push_read ($cb)
969
970=item $handle->unshift_read ($cb)
971
972Append the given callback to the end of the queue (C<push_read>) or
973prepend it (C<unshift_read>).
974
975The callback is called each time some additional read data arrives.
976
977It must check whether enough data is in the read buffer already.
978
979If not enough data is available, it must return the empty list or a false
980value, in which case it will be called repeatedly until enough data is
981available (or an error condition is detected).
982
983If enough data was available, then the callback must remove all data it is
984interested in (which can be none at all) and return a true value. After returning
985true, it will be removed from the queue.
986
987=cut
988
989our %RH;
990
991sub register_read_type($$) {
992 $RH{$_[0]} = $_[1];
993}
994
995sub push_read {
996 my $self = shift;
997 my $cb = pop;
998
999 if (@_) {
1000 my $type = shift;
1001
1002 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::push_read")
1003 ->($self, $cb, @_);
1004 }
1005
1006 push @{ $self->{_queue} }, $cb;
1007 $self->_drain_rbuf unless $self->{_in_drain};
1008}
1009
1010sub unshift_read {
1011 my $self = shift;
1012 my $cb = pop;
1013
1014 if (@_) {
1015 my $type = shift;
1016
1017 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::unshift_read")
1018 ->($self, $cb, @_);
1019 }
1020
1021
1022 unshift @{ $self->{_queue} }, $cb;
1023 $self->_drain_rbuf unless $self->{_in_drain};
1024}
1025
1026=item $handle->push_read (type => @args, $cb)
1027
1028=item $handle->unshift_read (type => @args, $cb)
1029
1030Instead of providing a callback that parses the data itself you can chose
1031between a number of predefined parsing formats, for chunks of data, lines
1032etc.
1033
1034Predefined types are (if you have ideas for additional types, feel free to
1035drop by and tell us):
1036
1037=over 4
1038
1039=item chunk => $octets, $cb->($handle, $data)
1040
1041Invoke the callback only once C<$octets> bytes have been read. Pass the
1042data read to the callback. The callback will never be called with less
1043data.
1044
1045Example: read 2 bytes.
1046
1047 $handle->push_read (chunk => 2, sub {
1048 warn "yay ", unpack "H*", $_[1];
1049 });
1050
1051=cut
1052
1053register_read_type chunk => sub {
1054 my ($self, $cb, $len) = @_;
1055
1056 sub {
1057 $len <= length $_[0]{rbuf} or return;
1058 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
1059 1
1060 }
1061};
1062
1063=item line => [$eol, ]$cb->($handle, $line, $eol)
1064
1065The callback will be called only once a full line (including the end of
1066line marker, C<$eol>) has been read. This line (excluding the end of line
1067marker) will be passed to the callback as second argument (C<$line>), and
1068the end of line marker as the third argument (C<$eol>).
1069
1070The end of line marker, C<$eol>, can be either a string, in which case it
1071will be interpreted as a fixed record end marker, or it can be a regex
1072object (e.g. created by C<qr>), in which case it is interpreted as a
1073regular expression.
1074
1075The end of line marker argument C<$eol> is optional, if it is missing (NOT
1076undef), then C<qr|\015?\012|> is used (which is good for most internet
1077protocols).
1078
1079Partial lines at the end of the stream will never be returned, as they are
1080not marked by the end of line marker.
1081
1082=cut
1083
1084register_read_type line => sub {
1085 my ($self, $cb, $eol) = @_;
1086
1087 if (@_ < 3) {
1088 # this is more than twice as fast as the generic code below
1089 sub {
1090 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1091
1092 $cb->($_[0], $1, $2);
1093 1
1094 }
1095 } else {
1096 $eol = quotemeta $eol unless ref $eol;
1097 $eol = qr|^(.*?)($eol)|s;
1098
1099 sub {
1100 $_[0]{rbuf} =~ s/$eol// or return;
1101
1102 $cb->($_[0], $1, $2);
1103 1
1104 }
1105 }
1106};
1107
1108=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1109
1110Makes a regex match against the regex object C<$accept> and returns
1111everything up to and including the match.
1112
1113Example: read a single line terminated by '\n'.
1114
1115 $handle->push_read (regex => qr<\n>, sub { ... });
1116
1117If C<$reject> is given and not undef, then it determines when the data is
1118to be rejected: it is matched against the data when the C<$accept> regex
1119does not match and generates an C<EBADMSG> error when it matches. This is
1120useful to quickly reject wrong data (to avoid waiting for a timeout or a
1121receive buffer overflow).
1122
1123Example: expect a single decimal number followed by whitespace, reject
1124anything else (not the use of an anchor).
1125
1126 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1127
1128If C<$skip> is given and not C<undef>, then it will be matched against
1129the receive buffer when neither C<$accept> nor C<$reject> match,
1130and everything preceding and including the match will be accepted
1131unconditionally. This is useful to skip large amounts of data that you
1132know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1133have to start matching from the beginning. This is purely an optimisation
1134and is usually worth only when you expect more than a few kilobytes.
1135
1136Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1137expect the header to be very large (it isn't in practise, but...), we use
1138a skip regex to skip initial portions. The skip regex is tricky in that
1139it only accepts something not ending in either \015 or \012, as these are
1140required for the accept regex.
1141
1142 $handle->push_read (regex =>
1143 qr<\015\012\015\012>,
1144 undef, # no reject
1145 qr<^.*[^\015\012]>,
1146 sub { ... });
1147
1148=cut
1149
1150register_read_type regex => sub {
1151 my ($self, $cb, $accept, $reject, $skip) = @_;
1152
1153 my $data;
1154 my $rbuf = \$self->{rbuf};
1155
1156 sub {
1157 # accept
1158 if ($$rbuf =~ $accept) {
1159 $data .= substr $$rbuf, 0, $+[0], "";
1160 $cb->($self, $data);
157 return; 1161 return 1;
1162 }
1163
1164 # reject
1165 if ($reject && $$rbuf =~ $reject) {
1166 $self->_error (Errno::EBADMSG);
1167 }
1168
1169 # skip
1170 if ($skip && $$rbuf =~ $skip) {
1171 $data .= substr $$rbuf, 0, $+[0], "";
1172 }
1173
1174 ()
158 } 1175 }
159 1176};
160 $self->{on_read_w} = 1177
161 AnyEvent->io (poll => 'r', fh => $self->{fh}, cb => sub { 1178=item netstring => $cb->($handle, $string)
162 #d# warn "READ:[$self->{read_size}] $self->{read_block_size} : ".length ($self->{rbuf})."\n"; 1179
163 my $rbuf_len = length $self->{rbuf}; 1180A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
164 my $l; 1181
165 if (defined $self->{read_size}) { 1182Throws an error with C<$!> set to EBADMSG on format violations.
166 $l = sysread $self->{fh}, $self->{rbuf}, 1183
167 ($self->{read_size} - $rbuf_len), $rbuf_len; 1184=cut
168 } else { 1185
169 $l = sysread $self->{fh}, $self->{rbuf}, $self->{read_block_size}, $rbuf_len; 1186register_read_type netstring => sub {
1187 my ($self, $cb) = @_;
1188
1189 sub {
1190 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1191 if ($_[0]{rbuf} =~ /[^0-9]/) {
1192 $self->_error (Errno::EBADMSG);
170 } 1193 }
171 #d# warn "READL $l [$self->{rbuf}]\n"; 1194 return;
1195 }
172 1196
173 if (not defined $l) { 1197 my $len = $1;
174 return if $! == EAGAIN || $! == EINTR;
175 $self->{on_error}->($self) if $self->{on_error};
176 delete $self->{on_read_w};
177 1198
178 } elsif ($l == 0) { 1199 $self->unshift_read (chunk => $len, sub {
179 $self->{on_eof}->($self) if $self->{on_eof}; 1200 my $string = $_[1];
180 delete $self->{on_read_w}; 1201 $_[0]->unshift_read (chunk => 1, sub {
181 1202 if ($_[1] eq ",") {
1203 $cb->($_[0], $string);
182 } else { 1204 } else {
183 $self->{on_read}->($self); 1205 $self->_error (Errno::EBADMSG);
1206 }
1207 });
1208 });
1209
1210 1
1211 }
1212};
1213
1214=item packstring => $format, $cb->($handle, $string)
1215
1216An octet string prefixed with an encoded length. The encoding C<$format>
1217uses the same format as a Perl C<pack> format, but must specify a single
1218integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1219optional C<!>, C<< < >> or C<< > >> modifier).
1220
1221For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1222EPP uses a prefix of C<N> (4 octtes).
1223
1224Example: read a block of data prefixed by its length in BER-encoded
1225format (very efficient).
1226
1227 $handle->push_read (packstring => "w", sub {
1228 my ($handle, $data) = @_;
1229 });
1230
1231=cut
1232
1233register_read_type packstring => sub {
1234 my ($self, $cb, $format) = @_;
1235
1236 sub {
1237 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1238 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1239 or return;
1240
1241 $format = length pack $format, $len;
1242
1243 # bypass unshift if we already have the remaining chunk
1244 if ($format + $len <= length $_[0]{rbuf}) {
1245 my $data = substr $_[0]{rbuf}, $format, $len;
1246 substr $_[0]{rbuf}, 0, $format + $len, "";
1247 $cb->($_[0], $data);
1248 } else {
1249 # remove prefix
1250 substr $_[0]{rbuf}, 0, $format, "";
1251
1252 # read remaining chunk
1253 $_[0]->unshift_read (chunk => $len, $cb);
1254 }
1255
1256 1
1257 }
1258};
1259
1260=item json => $cb->($handle, $hash_or_arrayref)
1261
1262Reads a JSON object or array, decodes it and passes it to the
1263callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1264
1265If a C<json> object was passed to the constructor, then that will be used
1266for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1267
1268This read type uses the incremental parser available with JSON version
12692.09 (and JSON::XS version 2.2) and above. You have to provide a
1270dependency on your own: this module will load the JSON module, but
1271AnyEvent does not depend on it itself.
1272
1273Since JSON texts are fully self-delimiting, the C<json> read and write
1274types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1275the C<json> write type description, above, for an actual example.
1276
1277=cut
1278
1279register_read_type json => sub {
1280 my ($self, $cb) = @_;
1281
1282 my $json = $self->{json} ||=
1283 eval { require JSON::XS; JSON::XS->new->utf8 }
1284 || do { require JSON; JSON->new->utf8 };
1285
1286 my $data;
1287 my $rbuf = \$self->{rbuf};
1288
1289 sub {
1290 my $ref = eval { $json->incr_parse ($self->{rbuf}) };
1291
1292 if ($ref) {
1293 $self->{rbuf} = $json->incr_text;
1294 $json->incr_text = "";
1295 $cb->($self, $ref);
1296
1297 1
1298 } elsif ($@) {
1299 # error case
1300 $json->incr_skip;
1301
1302 $self->{rbuf} = $json->incr_text;
1303 $json->incr_text = "";
1304
1305 $self->_error (Errno::EBADMSG);
1306
1307 ()
1308 } else {
1309 $self->{rbuf} = "";
1310
1311 ()
1312 }
1313 }
1314};
1315
1316=item storable => $cb->($handle, $ref)
1317
1318Deserialises a L<Storable> frozen representation as written by the
1319C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1320data).
1321
1322Raises C<EBADMSG> error if the data could not be decoded.
1323
1324=cut
1325
1326register_read_type storable => sub {
1327 my ($self, $cb) = @_;
1328
1329 require Storable;
1330
1331 sub {
1332 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1333 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1334 or return;
1335
1336 my $format = length pack "w", $len;
1337
1338 # bypass unshift if we already have the remaining chunk
1339 if ($format + $len <= length $_[0]{rbuf}) {
1340 my $data = substr $_[0]{rbuf}, $format, $len;
1341 substr $_[0]{rbuf}, 0, $format + $len, "";
1342 $cb->($_[0], Storable::thaw ($data));
1343 } else {
1344 # remove prefix
1345 substr $_[0]{rbuf}, 0, $format, "";
1346
1347 # read remaining chunk
1348 $_[0]->unshift_read (chunk => $len, sub {
1349 if (my $ref = eval { Storable::thaw ($_[1]) }) {
1350 $cb->($_[0], $ref);
1351 } else {
1352 $self->_error (Errno::EBADMSG);
1353 }
1354 });
1355 }
1356
1357 1
1358 }
1359};
1360
1361=back
1362
1363=item AnyEvent::Handle::register_read_type type => $coderef->($handle, $cb, @args)
1364
1365This function (not method) lets you add your own types to C<push_read>.
1366
1367Whenever the given C<type> is used, C<push_read> will invoke the code
1368reference with the handle object, the callback and the remaining
1369arguments.
1370
1371The code reference is supposed to return a callback (usually a closure)
1372that works as a plain read callback (see C<< ->push_read ($cb) >>).
1373
1374It should invoke the passed callback when it is done reading (remember to
1375pass C<$handle> as first argument as all other callbacks do that).
1376
1377Note that this is a function, and all types registered this way will be
1378global, so try to use unique names.
1379
1380For examples, see the source of this module (F<perldoc -m AnyEvent::Handle>,
1381search for C<register_read_type>)).
1382
1383=item $handle->stop_read
1384
1385=item $handle->start_read
1386
1387In rare cases you actually do not want to read anything from the
1388socket. In this case you can call C<stop_read>. Neither C<on_read> nor
1389any queued callbacks will be executed then. To start reading again, call
1390C<start_read>.
1391
1392Note that AnyEvent::Handle will automatically C<start_read> for you when
1393you change the C<on_read> callback or push/unshift a read callback, and it
1394will automatically C<stop_read> for you when neither C<on_read> is set nor
1395there are any read requests in the queue.
1396
1397These methods will have no effect when in TLS mode (as TLS doesn't support
1398half-duplex connections).
1399
1400=cut
1401
1402sub stop_read {
1403 my ($self) = @_;
1404
1405 delete $self->{_rw} unless $self->{tls};
1406}
1407
1408sub start_read {
1409 my ($self) = @_;
1410
1411 unless ($self->{_rw} || $self->{_eof}) {
1412 Scalar::Util::weaken $self;
1413
1414 $self->{_rw} = AnyEvent->io (fh => $self->{fh}, poll => "r", cb => sub {
1415 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
1416 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size} || 8192, length $$rbuf;
1417
1418 if ($len > 0) {
1419 $self->{_activity} = AnyEvent->now;
1420
1421 if ($self->{tls}) {
1422 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1423
1424 &_dotls ($self);
1425 } else {
1426 $self->_drain_rbuf unless $self->{_in_drain};
1427 }
1428
1429 } elsif (defined $len) {
1430 delete $self->{_rw};
1431 $self->{_eof} = 1;
1432 $self->_drain_rbuf unless $self->{_in_drain};
1433
1434 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
1435 return $self->_error ($!, 1);
184 } 1436 }
185 }); 1437 });
1438 }
186} 1439}
187 1440
188=item B<on_error ($callback)> 1441our $ERROR_SYSCALL;
1442our $ERROR_WANT_READ;
189 1443
190Whenever a read or write operation resulted in an error the C<$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 { 1444sub _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, $cb) = @_; 1445 my ($self, $err) = @_;
240 1446
241 $self->{read_cb} = $cb; 1447 return $self->_error ($!, 1)
242 my $old_blk_size = $self->{read_block_size}; 1448 if $err == Net::SSLeay::ERROR_SYSCALL ();
243 $self->{read_block_size} = $len;
244 1449
245 $self->on_read (sub { 1450 my $err =Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
246 #d# warn "OFOFO $len || ".length($_[0]->{rbuf})."||\n";
247 1451
248 if ($len == length $_[0]->{rbuf}) { 1452 # reduce error string to look less scary
249 $_[0]->{read_block_size} = $old_blk_size; 1453 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
250 $_[0]->on_read (undef); 1454
251 $_[0]->{read_cb}->($_[0], (substr $self->{rbuf}, 0, $len, '')); 1455 if ($self->{_on_starttls}) {
1456 (delete $self->{_on_starttls})->($self, undef, $err);
1457 &_freetls;
1458 } else {
1459 &_freetls;
1460 $self->_error (Errno::EPROTO, 1, $err);
1461 }
1462}
1463
1464# poll the write BIO and send the data if applicable
1465# also decode read data if possible
1466# this is basiclaly our TLS state machine
1467# more efficient implementations are possible with openssl,
1468# but not with the buggy and incomplete Net::SSLeay.
1469sub _dotls {
1470 my ($self) = @_;
1471
1472 my $tmp;
1473
1474 if (length $self->{_tls_wbuf}) {
1475 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
1476 substr $self->{_tls_wbuf}, 0, $tmp, "";
252 } 1477 }
1478
1479 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1480 return $self->_tls_error ($tmp)
1481 if $tmp != $ERROR_WANT_READ
1482 && ($tmp != $ERROR_SYSCALL || $!);
253 }); 1483 }
254}
255 1484
256=item B<readlines ($callback)> 1485 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1486 unless (length $tmp) {
1487 $self->{_on_starttls}
1488 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1489 &_freetls;
257 1490
258=item B<readlines ($sep, $callback)> 1491 if ($self->{on_stoptls}) {
259 1492 $self->{on_stoptls}($self);
260This method will read lines from the filehandle, seperated by C<$sep> or C<"\n"> 1493 return;
261if C<$sep> is not provided. C<$sep> will be used as part of a regex, so it can be 1494 } else {
262a regex itself and won't be quoted! 1495 # let's treat SSL-eof as we treat normal EOF
263 1496 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]; 1497 $self->{_eof} = 1;
316 } 1498 }
317 } 1499 }
318 return; 1500
1501 $self->{_tls_rbuf} .= $tmp;
1502 $self->_drain_rbuf unless $self->{_in_drain};
1503 $self->{tls} or return; # tls session might have gone away in callback
1504 }
1505
1506 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
1507 return $self->_tls_error ($tmp)
1508 if $tmp != $ERROR_WANT_READ
1509 && ($tmp != $ERROR_SYSCALL || $!);
1510
1511 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
1512 $self->{wbuf} .= $tmp;
1513 $self->_drain_wbuf;
1514 }
1515
1516 $self->{_on_starttls}
1517 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
1518 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
1519}
1520
1521=item $handle->starttls ($tls[, $tls_ctx])
1522
1523Instead of starting TLS negotiation immediately when the AnyEvent::Handle
1524object is created, you can also do that at a later time by calling
1525C<starttls>.
1526
1527The first argument is the same as the C<tls> constructor argument (either
1528C<"connect">, C<"accept"> or an existing Net::SSLeay object).
1529
1530The second argument is the optional C<AnyEvent::TLS> object that is used
1531when AnyEvent::Handle has to create its own TLS connection object, or
1532a hash reference with C<< key => value >> pairs that will be used to
1533construct a new context.
1534
1535The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1536context in C<< $handle->{tls_ctx} >> after this call and can be used or
1537changed to your liking. Note that the handshake might have already started
1538when this function returns.
1539
1540If it an error to start a TLS handshake more than once per
1541AnyEvent::Handle object (this is due to bugs in OpenSSL).
1542
1543=cut
1544
1545our %TLS_CACHE; #TODO not yet documented, should we?
1546
1547sub starttls {
1548 my ($self, $ssl, $ctx) = @_;
1549
1550 require Net::SSLeay;
1551
1552 Carp::croak "it is an error to call starttls more than once on an AnyEvent::Handle object"
1553 if $self->{tls};
1554
1555 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
1556 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
1557
1558 $ctx ||= $self->{tls_ctx};
1559
1560 if ("HASH" eq ref $ctx) {
1561 require AnyEvent::TLS;
1562
1563 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context
1564
1565 if ($ctx->{cache}) {
1566 my $key = $ctx+0;
1567 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
1568 } else {
1569 $ctx = new AnyEvent::TLS %$ctx;
1570 }
1571 }
319 } 1572
1573 $self->{tls_ctx} = $ctx || TLS_CTX ();
1574 $self->{tls} = $ssl = $self->{tls_ctx}->_get_session ($ssl, $self, $self->{peername});
320 1575
321 my $wba = shift @{$self->{write_bufs}} 1576 # basically, this is deep magic (because SSL_read should have the same issues)
322 or return; 1577 # but the openssl maintainers basically said: "trust us, it just works".
1578 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
1579 # and mismaintained ssleay-module doesn't even offer them).
1580 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
1581 #
1582 # in short: this is a mess.
1583 #
1584 # note that we do not try to keep the length constant between writes as we are required to do.
1585 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
1586 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
1587 # have identity issues in that area.
1588# Net::SSLeay::CTX_set_mode ($ssl,
1589# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
1590# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
1591 Net::SSLeay::CTX_set_mode ($ssl, 1|2);
323 1592
324 unless (defined $wba->[0]) { 1593 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
325 $wba->[1]->($self) if $wba->[1]; 1594 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
326 $self->_check_writer; 1595
327 return; 1596 Net::SSLeay::set_bio ($ssl, $self->{_rbio}, $self->{_wbio});
1597
1598 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
1599 if $self->{on_starttls};
1600
1601 &_dotls; # need to trigger the initial handshake
1602 $self->start_read; # make sure we actually do read
1603}
1604
1605=item $handle->stoptls
1606
1607Shuts down the SSL connection - this makes a proper EOF handshake by
1608sending a close notify to the other side, but since OpenSSL doesn't
1609support non-blocking shut downs, it is not possible to re-use the stream
1610afterwards.
1611
1612=cut
1613
1614sub stoptls {
1615 my ($self) = @_;
1616
1617 if ($self->{tls}) {
1618 Net::SSLeay::shutdown ($self->{tls});
1619
1620 &_dotls;
1621
1622# # we don't give a shit. no, we do, but we can't. no...#d#
1623# # we, we... have to use openssl :/#d#
1624# &_freetls;#d#
1625 }
1626}
1627
1628sub _freetls {
1629 my ($self) = @_;
1630
1631 return unless $self->{tls};
1632
1633 $self->{tls_ctx}->_put_session (delete $self->{tls});
328 } 1634
1635 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
1636}
329 1637
330 $self->{wbuf} = $wba->[0]; 1638sub DESTROY {
331 $self->{write_cb} = $wba->[1]; 1639 my ($self) = @_;
332 1640
333 $self->{write_w} = 1641 &_freetls;
334 AnyEvent->io (poll => 'w', fh => $self->{fh}, cb => sub { 1642
1643 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
1644
1645 if ($linger && length $self->{wbuf}) {
1646 my $fh = delete $self->{fh};
1647 my $wbuf = delete $self->{wbuf};
1648
1649 my @linger;
1650
1651 push @linger, AnyEvent->io (fh => $fh, poll => "w", cb => sub {
335 my $l = syswrite $self->{fh}, $self->{wbuf}, length $self->{wbuf}; 1652 my $len = syswrite $fh, $wbuf, length $wbuf;
336 1653
337 if (not defined $l) { 1654 if ($len > 0) {
338 return if $! == EAGAIN || $! == EINTR; 1655 substr $wbuf, 0, $len, "";
339 delete $self->{write_w};
340 $self->{on_error}->($self) if $self->{on_error};
341
342 } else { 1656 } else {
343 substr $self->{wbuf}, 0, $l, ''; 1657 @linger = (); # end
344
345 if (length ($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 } 1658 }
355 }); 1659 });
1660 push @linger, AnyEvent->timer (after => $linger, cb => sub {
1661 @linger = ();
1662 });
1663 }
1664}
1665
1666=item $handle->destroy
1667
1668Shuts down the handle object as much as possible - this call ensures that
1669no further callbacks will be invoked and as many resources as possible
1670will be freed. You must not call any methods on the object afterwards.
1671
1672Normally, you can just "forget" any references to an AnyEvent::Handle
1673object and it will simply shut down. This works in fatal error and EOF
1674callbacks, as well as code outside. It does I<NOT> work in a read or write
1675callback, so when you want to destroy the AnyEvent::Handle object from
1676within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
1677that case.
1678
1679Destroying the handle object in this way has the advantage that callbacks
1680will be removed as well, so if those are the only reference holders (as
1681is common), then one doesn't need to do anything special to break any
1682reference cycles.
1683
1684The handle might still linger in the background and write out remaining
1685data, as specified by the C<linger> option, however.
1686
1687=cut
1688
1689sub destroy {
1690 my ($self) = @_;
1691
1692 $self->DESTROY;
1693 %$self = ();
1694}
1695
1696=item AnyEvent::Handle::TLS_CTX
1697
1698This function creates and returns the AnyEvent::TLS object used by default
1699for TLS mode.
1700
1701The context is created by calling L<AnyEvent::TLS> without any arguments.
1702
1703=cut
1704
1705our $TLS_CTX;
1706
1707sub TLS_CTX() {
1708 $TLS_CTX ||= do {
1709 require AnyEvent::TLS;
1710
1711 new AnyEvent::TLS
1712 }
356} 1713}
357 1714
358=back 1715=back
359 1716
1717
1718=head1 NONFREQUENTLY ASKED QUESTIONS
1719
1720=over 4
1721
1722=item I C<undef> the AnyEvent::Handle reference inside my callback and
1723still get further invocations!
1724
1725That's because AnyEvent::Handle keeps a reference to itself when handling
1726read or write callbacks.
1727
1728It is only safe to "forget" the reference inside EOF or error callbacks,
1729from within all other callbacks, you need to explicitly call the C<<
1730->destroy >> method.
1731
1732=item I get different callback invocations in TLS mode/Why can't I pause
1733reading?
1734
1735Unlike, say, TCP, TLS connections do not consist of two independent
1736communication channels, one for each direction. Or put differently. The
1737read and write directions are not independent of each other: you cannot
1738write data unless you are also prepared to read, and vice versa.
1739
1740This can mean than, in TLS mode, you might get C<on_error> or C<on_eof>
1741callback invocations when you are not expecting any read data - the reason
1742is that AnyEvent::Handle always reads in TLS mode.
1743
1744During the connection, you have to make sure that you always have a
1745non-empty read-queue, or an C<on_read> watcher. At the end of the
1746connection (or when you no longer want to use it) you can call the
1747C<destroy> method.
1748
1749=item How do I read data until the other side closes the connection?
1750
1751If you just want to read your data into a perl scalar, the easiest way
1752to achieve this is by setting an C<on_read> callback that does nothing,
1753clearing the C<on_eof> callback and in the C<on_error> callback, the data
1754will be in C<$_[0]{rbuf}>:
1755
1756 $handle->on_read (sub { });
1757 $handle->on_eof (undef);
1758 $handle->on_error (sub {
1759 my $data = delete $_[0]{rbuf};
1760 });
1761
1762The reason to use C<on_error> is that TCP connections, due to latencies
1763and packets loss, might get closed quite violently with an error, when in
1764fact, all data has been received.
1765
1766It is usually better to use acknowledgements when transferring data,
1767to make sure the other side hasn't just died and you got the data
1768intact. This is also one reason why so many internet protocols have an
1769explicit QUIT command.
1770
1771=item I don't want to destroy the handle too early - how do I wait until
1772all data has been written?
1773
1774After writing your last bits of data, set the C<on_drain> callback
1775and destroy the handle in there - with the default setting of
1776C<low_water_mark> this will be called precisely when all data has been
1777written to the socket:
1778
1779 $handle->push_write (...);
1780 $handle->on_drain (sub {
1781 warn "all data submitted to the kernel\n";
1782 undef $handle;
1783 });
1784
1785If you just want to queue some data and then signal EOF to the other side,
1786consider using C<< ->push_shutdown >> instead.
1787
1788=item I want to contact a TLS/SSL server, I don't care about security.
1789
1790If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
1791simply connect to it and then create the AnyEvent::Handle with the C<tls>
1792parameter:
1793
1794 tcp_connect $host, $port, sub {
1795 my ($fh) = @_;
1796
1797 my $handle = new AnyEvent::Handle
1798 fh => $fh,
1799 tls => "connect",
1800 on_error => sub { ... };
1801
1802 $handle->push_write (...);
1803 };
1804
1805=item I want to contact a TLS/SSL server, I do care about security.
1806
1807Then you should additionally enable certificate verification, including
1808peername verification, if the protocol you use supports it (see
1809L<AnyEvent::TLS>, C<verify_peername>).
1810
1811E.g. for HTTPS:
1812
1813 tcp_connect $host, $port, sub {
1814 my ($fh) = @_;
1815
1816 my $handle = new AnyEvent::Handle
1817 fh => $fh,
1818 peername => $host,
1819 tls => "connect",
1820 tls_ctx => { verify => 1, verify_peername => "https" },
1821 ...
1822
1823Note that you must specify the hostname you connected to (or whatever
1824"peername" the protocol needs) as the C<peername> argument, otherwise no
1825peername verification will be done.
1826
1827The above will use the system-dependent default set of trusted CA
1828certificates. If you want to check against a specific CA, add the
1829C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
1830
1831 tls_ctx => {
1832 verify => 1,
1833 verify_peername => "https",
1834 ca_file => "my-ca-cert.pem",
1835 },
1836
1837=item I want to create a TLS/SSL server, how do I do that?
1838
1839Well, you first need to get a server certificate and key. You have
1840three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
1841self-signed certificate (cheap. check the search engine of your choice,
1842there are many tutorials on the net) or c) make your own CA (tinyca2 is a
1843nice program for that purpose).
1844
1845Then create a file with your private key (in PEM format, see
1846L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
1847file should then look like this:
1848
1849 -----BEGIN RSA PRIVATE KEY-----
1850 ...header data
1851 ... lots of base64'y-stuff
1852 -----END RSA PRIVATE KEY-----
1853
1854 -----BEGIN CERTIFICATE-----
1855 ... lots of base64'y-stuff
1856 -----END CERTIFICATE-----
1857
1858The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
1859specify this file as C<cert_file>:
1860
1861 tcp_server undef, $port, sub {
1862 my ($fh) = @_;
1863
1864 my $handle = new AnyEvent::Handle
1865 fh => $fh,
1866 tls => "accept",
1867 tls_ctx => { cert_file => "my-server-keycert.pem" },
1868 ...
1869
1870When you have intermediate CA certificates that your clients might not
1871know about, just append them to the C<cert_file>.
1872
1873=back
1874
1875
1876=head1 SUBCLASSING AnyEvent::Handle
1877
1878In many cases, you might want to subclass AnyEvent::Handle.
1879
1880To make this easier, a given version of AnyEvent::Handle uses these
1881conventions:
1882
1883=over 4
1884
1885=item * all constructor arguments become object members.
1886
1887At least initially, when you pass a C<tls>-argument to the constructor it
1888will end up in C<< $handle->{tls} >>. Those members might be changed or
1889mutated later on (for example C<tls> will hold the TLS connection object).
1890
1891=item * other object member names are prefixed with an C<_>.
1892
1893All object members not explicitly documented (internal use) are prefixed
1894with an underscore character, so the remaining non-C<_>-namespace is free
1895for use for subclasses.
1896
1897=item * all members not documented here and not prefixed with an underscore
1898are free to use in subclasses.
1899
1900Of course, new versions of AnyEvent::Handle may introduce more "public"
1901member variables, but thats just life, at least it is documented.
1902
1903=back
1904
360=head1 AUTHOR 1905=head1 AUTHOR
361 1906
362Robin Redeker, C<< <elmex at ta-sa.org> >> 1907Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
363 1908
364=cut 1909=cut
365 1910
3661; # End of AnyEvent::Handle 19111; # End of AnyEvent::Handle

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines