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 1.9 by root, Fri May 2 16:07:46 2008 UTC vs.
Revision 1.193 by root, Mon Mar 15 18:51:30 2010 UTC

1package AnyEvent::Handle;
2
3no warnings;
4use strict;
5
6use AnyEvent ();
7use AnyEvent::Util ();
8use Scalar::Util ();
9use Carp ();
10use Fcntl ();
11use Errno qw/EAGAIN EINTR/;
12
13=head1 NAME 1=head1 NAME
14 2
15AnyEvent::Handle - non-blocking I/O on filehandles via AnyEvent 3AnyEvent::Handle - non-blocking I/O on file handles via AnyEvent
16
17=cut
18
19our $VERSION = '0.02';
20 4
21=head1 SYNOPSIS 5=head1 SYNOPSIS
22 6
23 use AnyEvent; 7 use AnyEvent;
24 use AnyEvent::Handle; 8 use AnyEvent::Handle;
25 9
26 my $cv = AnyEvent->condvar; 10 my $cv = AnyEvent->condvar;
27 11
28 my $ae_fh = AnyEvent::Handle->new (fh => \*STDIN); 12 my $hdl; $hdl = new AnyEvent::Handle
29
30 #TODO
31
32 # or use the constructor to pass the callback:
33
34 my $ae_fh2 =
35 AnyEvent::Handle->new (
36 fh => \*STDIN, 13 fh => \*STDIN,
37 on_eof => sub { 14 on_error => sub {
38 $cv->broadcast; 15 my ($hdl, $fatal, $msg) = @_;
39 }, 16 warn "got error $msg\n";
40 #TODO 17 $hdl->destroy;
18 $cv->send;
41 ); 19 };
42 20
43 $cv->wait; 21 # send some request line
22 $hdl->push_write ("getinfo\015\012");
23
24 # read the response line
25 $hdl->push_read (line => sub {
26 my ($hdl, $line) = @_;
27 warn "got line <$line>\n";
28 $cv->send;
29 });
30
31 $cv->recv;
44 32
45=head1 DESCRIPTION 33=head1 DESCRIPTION
46 34
47This module is a helper module to make it easier to do event-based I/O on 35This module is a helper module to make it easier to do event-based I/O on
48filehandles (and sockets, see L<AnyEvent::Socket> for an easy way to make 36filehandles.
49non-blocking resolves and connects). 37
38The L<AnyEvent::Intro> tutorial contains some well-documented
39AnyEvent::Handle examples.
50 40
51In the following, when the documentation refers to of "bytes" then this 41In the following, when the documentation refers to of "bytes" then this
52means characters. As sysread and syswrite are used for all I/O, their 42means characters. As sysread and syswrite are used for all I/O, their
53treatment of characters applies to this module as well. 43treatment of characters applies to this module as well.
54 44
45At the very minimum, you should specify C<fh> or C<connect>, and the
46C<on_error> callback.
47
55All callbacks will be invoked with the handle object as their first 48All callbacks will be invoked with the handle object as their first
56argument. 49argument.
57 50
51=cut
52
53package AnyEvent::Handle;
54
55use Scalar::Util ();
56use List::Util ();
57use Carp ();
58use Errno qw(EAGAIN EINTR);
59
60use AnyEvent (); BEGIN { AnyEvent::common_sense }
61use AnyEvent::Util qw(WSAEWOULDBLOCK);
62
63our $VERSION = $AnyEvent::VERSION;
64
65sub _load_func($) {
66 my $func = $_[0];
67
68 unless (defined &$func) {
69 my $pkg = $func;
70 do {
71 $pkg =~ s/::[^:]+$//
72 or return;
73 eval "require $pkg";
74 } until defined &$func;
75 }
76
77 \&$func
78}
79
58=head1 METHODS 80=head1 METHODS
59 81
60=over 4 82=over 4
61 83
62=item B<new (%args)> 84=item $handle = B<new> AnyEvent::Handle fh => $filehandle, key => value...
63 85
64The constructor supports these arguments (all as key => value pairs). 86The constructor supports these arguments (all as C<< key => value >> pairs).
65 87
66=over 4 88=over 4
67 89
68=item fh => $filehandle [MANDATORY] 90=item fh => $filehandle [C<fh> or C<connect> MANDATORY]
69 91
70The filehandle this L<AnyEvent::Handle> object will operate on. 92The filehandle this L<AnyEvent::Handle> object will operate on.
71
72NOTE: The filehandle will be set to non-blocking (using 93NOTE: The filehandle will be set to non-blocking mode (using
73AnyEvent::Util::fh_nonblocking). 94C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
95that mode.
74 96
75=item on_error => $cb->($self) [MANDATORY] 97=item connect => [$host, $service] [C<fh> or C<connect> MANDATORY]
76 98
77This is the fatal error callback, that is called when a fatal error ocurs, 99Try to connect to the specified host and service (port), using
100C<AnyEvent::Socket::tcp_connect>. The C<$host> additionally becomes the
101default C<peername>.
102
103You have to specify either this parameter, or C<fh>, above.
104
105It is possible to push requests on the read and write queues, and modify
106properties of the stream, even while AnyEvent::Handle is connecting.
107
108When this parameter is specified, then the C<on_prepare>,
109C<on_connect_error> and C<on_connect> callbacks will be called under the
110appropriate circumstances:
111
112=over 4
113
114=item on_prepare => $cb->($handle)
115
116This (rarely used) callback is called before a new connection is
117attempted, but after the file handle has been created. It could be used to
118prepare the file handle with parameters required for the actual connect
119(as opposed to settings that can be changed when the connection is already
120established).
121
122The return value of this callback should be the connect timeout value in
123seconds (or C<0>, or C<undef>, or the empty list, to indicate the default
124timeout is to be used).
125
126=item on_connect => $cb->($handle, $host, $port, $retry->())
127
128This callback is called when a connection has been successfully established.
129
130The actual numeric host and port (the socket peername) are passed as
131parameters, together with a retry callback.
132
133When, for some reason, the handle is not acceptable, then calling
134C<$retry> will continue with the next connection target (in case of
135multi-homed hosts or SRV records there can be multiple connection
136endpoints). At the time it is called the read and write queues, eof
137status, tls status and similar properties of the handle will have been
138reset.
139
140In most cases, ignoring the C<$retry> parameter is the way to go.
141
142=item on_connect_error => $cb->($handle, $message)
143
144This callback is called when the connection could not be
145established. C<$!> will contain the relevant error code, and C<$message> a
146message describing it (usually the same as C<"$!">).
147
148If this callback isn't specified, then C<on_error> will be called with a
149fatal error instead.
150
151=back
152
153=item on_error => $cb->($handle, $fatal, $message)
154
155This is the error callback, which is called when, well, some error
78such as not being able to resolve the hostname, failure to connect or a 156occured, such as not being able to resolve the hostname, failure to
79read error. 157connect or a read error.
80 158
81The object will not be in a usable state when this callback has been 159Some errors are fatal (which is indicated by C<$fatal> being true). On
82called. 160fatal errors the handle object will be destroyed (by a call to C<< ->
161destroy >>) after invoking the error callback (which means you are free to
162examine the handle object). Examples of fatal errors are an EOF condition
163with active (but unsatisifable) read watchers (C<EPIPE>) or I/O errors. In
164cases where the other side can close the connection at their will it is
165often easiest to not report C<EPIPE> errors in this callback.
83 166
167AnyEvent::Handle tries to find an appropriate error code for you to check
168against, but in some cases (TLS errors), this does not work well. It is
169recommended to always output the C<$message> argument in human-readable
170error messages (it's usually the same as C<"$!">).
171
172Non-fatal errors can be retried by simply returning, but it is recommended
173to simply ignore this parameter and instead abondon the handle object
174when this callback is invoked. Examples of non-fatal errors are timeouts
175C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
176
84On callback entrance, the value of C<$!> contains the opertaing system 177On callback entrance, the value of C<$!> contains the operating system
85error (or C<ENOSPC> or C<EPIPE>). 178error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
179C<EPROTO>).
86 180
87=item on_eof => $cb->($self) [MANDATORY] 181While not mandatory, it is I<highly> recommended to set this callback, as
182you will not be notified of errors otherwise. The default simply calls
183C<croak>.
88 184
89Set the callback to be called on EOF.
90
91=item on_read => $cb->($self) 185=item on_read => $cb->($handle)
92 186
93This sets the default read callback, which is called when data arrives 187This sets the default read callback, which is called when data arrives
94and no read request is in the queue. If the read callback is C<undef> 188and no read request is in the queue (unlike read queue callbacks, this
95or has never been set, than AnyEvent::Handle will cease reading from the 189callback will only be called when at least one octet of data is in the
96filehandle. 190read buffer).
97 191
98To access (and remove data from) the read buffer, use the C<< ->rbuf >> 192To access (and remove data from) the read buffer, use the C<< ->rbuf >>
99method or acces sthe C<$self->{rbuf}> member directly. 193method or access the C<< $handle->{rbuf} >> member directly. Note that you
194must not enlarge or modify the read buffer, you can only remove data at
195the beginning from it.
100 196
101When an EOF condition is detected then AnyEvent::Handle will first try to 197When an EOF condition is detected then AnyEvent::Handle will first try to
102feed all the remaining data to the queued callbacks and C<on_read> before 198feed all the remaining data to the queued callbacks and C<on_read> before
103calling the C<on_eof> callback. If no progress can be made, then a fatal 199calling the C<on_eof> callback. If no progress can be made, then a fatal
104error will be raised (with C<$!> set to C<EPIPE>). 200error will be raised (with C<$!> set to C<EPIPE>).
105 201
202Note that, unlike requests in the read queue, an C<on_read> callback
203doesn't mean you I<require> some data: if there is an EOF and there
204are outstanding read requests then an error will be flagged. With an
205C<on_read> callback, the C<on_eof> callback will be invoked.
206
207=item on_eof => $cb->($handle)
208
209Set the callback to be called when an end-of-file condition is detected,
210i.e. in the case of a socket, when the other side has closed the
211connection cleanly, and there are no outstanding read requests in the
212queue (if there are read requests, then an EOF counts as an unexpected
213connection close and will be flagged as an error).
214
215For sockets, this just means that the other side has stopped sending data,
216you can still try to write data, and, in fact, one can return from the EOF
217callback and continue writing data, as only the read part has been shut
218down.
219
220If an EOF condition has been detected but no C<on_eof> callback has been
221set, then a fatal error will be raised with C<$!> set to <0>.
222
106=item on_drain => $cb->() 223=item on_drain => $cb->($handle)
107 224
108This sets the callback that is called when the write buffer becomes empty 225This sets the callback that is called when the write buffer becomes empty
109(or when the callback is set and the buffer is empty already). 226(or when the callback is set and the buffer is empty already).
110 227
111To append to the write buffer, use the C<< ->push_write >> method. 228To append to the write buffer, use the C<< ->push_write >> method.
112 229
230This callback is useful when you don't want to put all of your write data
231into the queue at once, for example, when you want to write the contents
232of some file to the socket you might not want to read the whole file into
233memory and push it into the queue, but instead only read more data from
234the file when the write queue becomes empty.
235
236=item timeout => $fractional_seconds
237
238=item rtimeout => $fractional_seconds
239
240=item wtimeout => $fractional_seconds
241
242If non-zero, then these enables an "inactivity" timeout: whenever this
243many seconds pass without a successful read or write on the underlying
244file handle (or a call to C<timeout_reset>), the C<on_timeout> callback
245will be invoked (and if that one is missing, a non-fatal C<ETIMEDOUT>
246error will be raised).
247
248There are three variants of the timeouts that work fully independent
249of each other, for both read and write, just read, and just write:
250C<timeout>, C<rtimeout> and C<wtimeout>, with corresponding callbacks
251C<on_timeout>, C<on_rtimeout> and C<on_wtimeout>, and reset functions
252C<timeout_reset>, C<rtimeout_reset>, and C<wtimeout_reset>.
253
254Note that timeout processing is also active when you currently do not have
255any outstanding read or write requests: If you plan to keep the connection
256idle then you should disable the timout temporarily or ignore the timeout
257in the C<on_timeout> callback, in which case AnyEvent::Handle will simply
258restart the timeout.
259
260Zero (the default) disables this timeout.
261
262=item on_timeout => $cb->($handle)
263
264Called whenever the inactivity timeout passes. If you return from this
265callback, then the timeout will be reset as if some activity had happened,
266so this condition is not fatal in any way.
267
113=item rbuf_max => <bytes> 268=item rbuf_max => <bytes>
114 269
115If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>) 270If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
116when the read buffer ever (strictly) exceeds this size. This is useful to 271when the read buffer ever (strictly) exceeds this size. This is useful to
117avoid denial-of-service attacks. 272avoid some forms of denial-of-service attacks.
118 273
119For example, a server accepting connections from untrusted sources should 274For example, a server accepting connections from untrusted sources should
120be configured to accept only so-and-so much data that it cannot act on 275be configured to accept only so-and-so much data that it cannot act on
121(for example, when expecting a line, an attacker could send an unlimited 276(for example, when expecting a line, an attacker could send an unlimited
122amount of data without a callback ever being called as long as the line 277amount of data without a callback ever being called as long as the line
123isn't finished). 278isn't finished).
124 279
280=item autocork => <boolean>
281
282When disabled (the default), then C<push_write> will try to immediately
283write the data to the handle, if possible. This avoids having to register
284a write watcher and wait for the next event loop iteration, but can
285be inefficient if you write multiple small chunks (on the wire, this
286disadvantage is usually avoided by your kernel's nagle algorithm, see
287C<no_delay>, but this option can save costly syscalls).
288
289When enabled, then writes will always be queued till the next event loop
290iteration. This is efficient when you do many small writes per iteration,
291but less efficient when you do a single write only per iteration (or when
292the write buffer often is full). It also increases write latency.
293
294=item no_delay => <boolean>
295
296When doing small writes on sockets, your operating system kernel might
297wait a bit for more data before actually sending it out. This is called
298the Nagle algorithm, and usually it is beneficial.
299
300In some situations you want as low a delay as possible, which can be
301accomplishd by setting this option to a true value.
302
303The default is your opertaing system's default behaviour (most likely
304enabled), this option explicitly enables or disables it, if possible.
305
306=item keepalive => <boolean>
307
308Enables (default disable) the SO_KEEPALIVE option on the stream socket:
309normally, TCP connections have no time-out once established, so TCP
310connections, once established, can stay alive forever even when the other
311side has long gone. TCP keepalives are a cheap way to take down long-lived
312TCP connections whent he other side becomes unreachable. While the default
313is OS-dependent, TCP keepalives usually kick in after around two hours,
314and, if the other side doesn't reply, take down the TCP connection some 10
315to 15 minutes later.
316
317It is harmless to specify this option for file handles that do not support
318keepalives, and enabling it on connections that are potentially long-lived
319is usually a good idea.
320
321=item oobinline => <boolean>
322
323BSD majorly fucked up the implementation of TCP urgent data. The result
324is that almost no OS implements TCP according to the specs, and every OS
325implements it slightly differently.
326
327If you want to handle TCP urgent data, then setting this flag (the default
328is enabled) gives you the most portable way of getting urgent data, by
329putting it into the stream.
330
331Since BSD emulation of OOB data on top of TCP's urgent data can have
332security implications, AnyEvent::Handle sets this flag automatically
333unless explicitly specified. Note that setting this flag after
334establishing a connection I<may> be a bit too late (data loss could
335already have occured on BSD systems), but at least it will protect you
336from most attacks.
337
125=item read_size => <bytes> 338=item read_size => <bytes>
126 339
127The default read block size (the amount of bytes this module will try to read 340The default read block size (the amount of bytes this module will
128on each [loop iteration). Default: C<4096>. 341try to read during each loop iteration, which affects memory
342requirements). Default: C<8192>.
129 343
130=item low_water_mark => <bytes> 344=item low_water_mark => <bytes>
131 345
132Sets the amount of bytes (default: C<0>) that make up an "empty" write 346Sets the amount of bytes (default: C<0>) that make up an "empty" write
133buffer: If the write reaches this size or gets even samller it is 347buffer: If the write reaches this size or gets even samller it is
134considered empty. 348considered empty.
135 349
350Sometimes it can be beneficial (for performance reasons) to add data to
351the write buffer before it is fully drained, but this is a rare case, as
352the operating system kernel usually buffers data as well, so the default
353is good in almost all cases.
354
355=item linger => <seconds>
356
357If non-zero (default: C<3600>), then the destructor of the
358AnyEvent::Handle object will check whether there is still outstanding
359write data and will install a watcher that will write this data to the
360socket. No errors will be reported (this mostly matches how the operating
361system treats outstanding data at socket close time).
362
363This will not work for partial TLS data that could not be encoded
364yet. This data will be lost. Calling the C<stoptls> method in time might
365help.
366
367=item peername => $string
368
369A string used to identify the remote site - usually the DNS hostname
370(I<not> IDN!) used to create the connection, rarely the IP address.
371
372Apart from being useful in error messages, this string is also used in TLS
373peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
374verification will be skipped when C<peername> is not specified or
375C<undef>.
376
377=item tls => "accept" | "connect" | Net::SSLeay::SSL object
378
379When this parameter is given, it enables TLS (SSL) mode, that means
380AnyEvent will start a TLS handshake as soon as the connection has been
381established and will transparently encrypt/decrypt data afterwards.
382
383All TLS protocol errors will be signalled as C<EPROTO>, with an
384appropriate error message.
385
386TLS mode requires Net::SSLeay to be installed (it will be loaded
387automatically when you try to create a TLS handle): this module doesn't
388have a dependency on that module, so if your module requires it, you have
389to add the dependency yourself.
390
391Unlike TCP, TLS has a server and client side: for the TLS server side, use
392C<accept>, and for the TLS client side of a connection, use C<connect>
393mode.
394
395You can also provide your own TLS connection object, but you have
396to make sure that you call either C<Net::SSLeay::set_connect_state>
397or C<Net::SSLeay::set_accept_state> on it before you pass it to
398AnyEvent::Handle. Also, this module will take ownership of this connection
399object.
400
401At some future point, AnyEvent::Handle might switch to another TLS
402implementation, then the option to use your own session object will go
403away.
404
405B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
406passing in the wrong integer will lead to certain crash. This most often
407happens when one uses a stylish C<< tls => 1 >> and is surprised about the
408segmentation fault.
409
410See the C<< ->starttls >> method for when need to start TLS negotiation later.
411
412=item tls_ctx => $anyevent_tls
413
414Use the given C<AnyEvent::TLS> object to create the new TLS connection
415(unless a connection object was specified directly). If this parameter is
416missing, then AnyEvent::Handle will use C<AnyEvent::Handle::TLS_CTX>.
417
418Instead of an object, you can also specify a hash reference with C<< key
419=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
420new TLS context object.
421
422=item on_starttls => $cb->($handle, $success[, $error_message])
423
424This callback will be invoked when the TLS/SSL handshake has finished. If
425C<$success> is true, then the TLS handshake succeeded, otherwise it failed
426(C<on_stoptls> will not be called in this case).
427
428The session in C<< $handle->{tls} >> can still be examined in this
429callback, even when the handshake was not successful.
430
431TLS handshake failures will not cause C<on_error> to be invoked when this
432callback is in effect, instead, the error message will be passed to C<on_starttls>.
433
434Without this callback, handshake failures lead to C<on_error> being
435called, as normal.
436
437Note that you cannot call C<starttls> right again in this callback. If you
438need to do that, start an zero-second timer instead whose callback can
439then call C<< ->starttls >> again.
440
441=item on_stoptls => $cb->($handle)
442
443When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
444set, then it will be invoked after freeing the TLS session. If it is not,
445then a TLS shutdown condition will be treated like a normal EOF condition
446on the handle.
447
448The session in C<< $handle->{tls} >> can still be examined in this
449callback.
450
451This callback will only be called on TLS shutdowns, not when the
452underlying handle signals EOF.
453
454=item json => JSON or JSON::XS object
455
456This is the json coder object used by the C<json> read and write types.
457
458If you don't supply it, then AnyEvent::Handle will create and use a
459suitable one (on demand), which will write and expect UTF-8 encoded JSON
460texts.
461
462Note that you are responsible to depend on the JSON module if you want to
463use this functionality, as AnyEvent does not have a dependency itself.
464
136=back 465=back
137 466
138=cut 467=cut
139 468
140sub new { 469sub new {
141 my $class = shift; 470 my $class = shift;
142
143 my $self = bless { @_ }, $class; 471 my $self = bless { @_ }, $class;
144 472
145 $self->{fh} or Carp::croak "mandatory argument fh is missing"; 473 if ($self->{fh}) {
474 $self->_start;
475 return unless $self->{fh}; # could be gone by now
476
477 } elsif ($self->{connect}) {
478 require AnyEvent::Socket;
479
480 $self->{peername} = $self->{connect}[0]
481 unless exists $self->{peername};
482
483 $self->{_skip_drain_rbuf} = 1;
484
485 {
486 Scalar::Util::weaken (my $self = $self);
487
488 $self->{_connect} =
489 AnyEvent::Socket::tcp_connect (
490 $self->{connect}[0],
491 $self->{connect}[1],
492 sub {
493 my ($fh, $host, $port, $retry) = @_;
494
495 if ($fh) {
496 $self->{fh} = $fh;
497
498 delete $self->{_skip_drain_rbuf};
499 $self->_start;
500
501 $self->{on_connect}
502 and $self->{on_connect}($self, $host, $port, sub {
503 delete @$self{qw(fh _tw _rtw _wtw _ww _rw _eof _queue rbuf _wbuf tls _tls_rbuf _tls_wbuf)};
504 $self->{_skip_drain_rbuf} = 1;
505 &$retry;
506 });
507
508 } else {
509 if ($self->{on_connect_error}) {
510 $self->{on_connect_error}($self, "$!");
511 $self->destroy;
512 } else {
513 $self->_error ($!, 1);
514 }
515 }
516 },
517 sub {
518 local $self->{fh} = $_[0];
519
520 $self->{on_prepare}
521 ? $self->{on_prepare}->($self)
522 : ()
523 }
524 );
525 }
526
527 } else {
528 Carp::croak "AnyEvent::Handle: either an existing fh or the connect parameter must be specified";
529 }
530
531 $self
532}
533
534sub _start {
535 my ($self) = @_;
146 536
147 AnyEvent::Util::fh_nonblocking $self->{fh}, 1; 537 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
148 538
149 $self->on_error ((delete $self->{on_error}) or Carp::croak "mandatory argument on_error is missing"); 539 $self->{_activity} =
150 $self->on_eof ((delete $self->{on_eof} ) or Carp::croak "mandatory argument on_eof is missing"); 540 $self->{_ractivity} =
541 $self->{_wactivity} = AE::now;
151 542
543 $self->timeout (delete $self->{timeout} ) if $self->{timeout};
544 $self->rtimeout (delete $self->{rtimeout} ) if $self->{rtimeout};
545 $self->wtimeout (delete $self->{wtimeout} ) if $self->{wtimeout};
546
547 $self->no_delay (delete $self->{no_delay} ) if exists $self->{no_delay} && $self->{no_delay};
548 $self->keepalive (delete $self->{keepalive}) if exists $self->{keepalive} && $self->{keepalive};
549
550 $self->oobinline (exists $self->{oobinline} ? delete $self->{oobinline} : 1);
551
552 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
553 if $self->{tls};
554
152 $self->on_drain (delete $self->{on_drain}) if $self->{on_drain}; 555 $self->on_drain (delete $self->{on_drain}) if $self->{on_drain};
153 $self->on_read (delete $self->{on_read} ) if $self->{on_read};
154 556
155 $self 557 $self->start_read
156} 558 if $self->{on_read} || @{ $self->{_queue} };
157 559
158sub _shutdown { 560 $self->_drain_wbuf;
159 my ($self) = @_;
160
161 delete $self->{rw};
162 delete $self->{ww};
163 delete $self->{fh};
164} 561}
165 562
166sub error { 563sub _error {
167 my ($self) = @_; 564 my ($self, $errno, $fatal, $message) = @_;
168 565
169 { 566 $! = $errno;
170 local $!; 567 $message ||= "$!";
568
569 if ($self->{on_error}) {
570 $self->{on_error}($self, $fatal, $message);
571 $self->destroy if $fatal;
572 } elsif ($self->{fh} || $self->{connect}) {
171 $self->_shutdown; 573 $self->destroy;
574 Carp::croak "AnyEvent::Handle uncaught error: $message";
172 } 575 }
173
174 $self->{on_error}($self);
175} 576}
176 577
177=item $fh = $handle->fh 578=item $fh = $handle->fh
178 579
179This method returns the filehandle of the L<AnyEvent::Handle> object. 580This method returns the file handle used to create the L<AnyEvent::Handle> object.
180 581
181=cut 582=cut
182 583
183sub fh { $_[0]->{fh} } 584sub fh { $_[0]{fh} }
184 585
185=item $handle->on_error ($cb) 586=item $handle->on_error ($cb)
186 587
187Replace the current C<on_error> callback (see the C<on_error> constructor argument). 588Replace the current C<on_error> callback (see the C<on_error> constructor argument).
188 589
200 601
201sub on_eof { 602sub on_eof {
202 $_[0]{on_eof} = $_[1]; 603 $_[0]{on_eof} = $_[1];
203} 604}
204 605
606=item $handle->on_timeout ($cb)
607
608=item $handle->on_rtimeout ($cb)
609
610=item $handle->on_wtimeout ($cb)
611
612Replace the current C<on_timeout>, C<on_rtimeout> or C<on_wtimeout>
613callback, or disables the callback (but not the timeout) if C<$cb> =
614C<undef>. See the C<timeout> constructor argument and method.
615
616=cut
617
618# see below
619
620=item $handle->autocork ($boolean)
621
622Enables or disables the current autocork behaviour (see C<autocork>
623constructor argument). Changes will only take effect on the next write.
624
625=cut
626
627sub autocork {
628 $_[0]{autocork} = $_[1];
629}
630
631=item $handle->no_delay ($boolean)
632
633Enables or disables the C<no_delay> setting (see constructor argument of
634the same name for details).
635
636=cut
637
638sub no_delay {
639 $_[0]{no_delay} = $_[1];
640
641 eval {
642 local $SIG{__DIE__};
643 setsockopt $_[0]{fh}, Socket::IPPROTO_TCP (), Socket::TCP_NODELAY (), int $_[1]
644 if $_[0]{fh};
645 };
646}
647
648=item $handle->keepalive ($boolean)
649
650Enables or disables the C<keepalive> setting (see constructor argument of
651the same name for details).
652
653=cut
654
655sub keepalive {
656 $_[0]{keepalive} = $_[1];
657
658 eval {
659 local $SIG{__DIE__};
660 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
661 if $_[0]{fh};
662 };
663}
664
665=item $handle->oobinline ($boolean)
666
667Enables or disables the C<oobinline> setting (see constructor argument of
668the same name for details).
669
670=cut
671
672sub oobinline {
673 $_[0]{oobinline} = $_[1];
674
675 eval {
676 local $SIG{__DIE__};
677 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_OOBINLINE (), int $_[1]
678 if $_[0]{fh};
679 };
680}
681
682=item $handle->keepalive ($boolean)
683
684Enables or disables the C<keepalive> setting (see constructor argument of
685the same name for details).
686
687=cut
688
689sub keepalive {
690 $_[0]{keepalive} = $_[1];
691
692 eval {
693 local $SIG{__DIE__};
694 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
695 if $_[0]{fh};
696 };
697}
698
699=item $handle->on_starttls ($cb)
700
701Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
702
703=cut
704
705sub on_starttls {
706 $_[0]{on_starttls} = $_[1];
707}
708
709=item $handle->on_stoptls ($cb)
710
711Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
712
713=cut
714
715sub on_stoptls {
716 $_[0]{on_stoptls} = $_[1];
717}
718
719=item $handle->rbuf_max ($max_octets)
720
721Configures the C<rbuf_max> setting (C<undef> disables it).
722
723=cut
724
725sub rbuf_max {
726 $_[0]{rbuf_max} = $_[1];
727}
728
729#############################################################################
730
731=item $handle->timeout ($seconds)
732
733=item $handle->rtimeout ($seconds)
734
735=item $handle->wtimeout ($seconds)
736
737Configures (or disables) the inactivity timeout.
738
739=item $handle->timeout_reset
740
741=item $handle->rtimeout_reset
742
743=item $handle->wtimeout_reset
744
745Reset the activity timeout, as if data was received or sent.
746
747These methods are cheap to call.
748
749=cut
750
751for my $dir ("", "r", "w") {
752 my $timeout = "${dir}timeout";
753 my $tw = "_${dir}tw";
754 my $on_timeout = "on_${dir}timeout";
755 my $activity = "_${dir}activity";
756 my $cb;
757
758 *$on_timeout = sub {
759 $_[0]{$on_timeout} = $_[1];
760 };
761
762 *$timeout = sub {
763 my ($self, $new_value) = @_;
764
765 $self->{$timeout} = $new_value;
766 delete $self->{$tw}; &$cb;
767 };
768
769 *{"${dir}timeout_reset"} = sub {
770 $_[0]{$activity} = AE::now;
771 };
772
773 # main workhorse:
774 # reset the timeout watcher, as neccessary
775 # also check for time-outs
776 $cb = sub {
777 my ($self) = @_;
778
779 if ($self->{$timeout} && $self->{fh}) {
780 my $NOW = AE::now;
781
782 # when would the timeout trigger?
783 my $after = $self->{$activity} + $self->{$timeout} - $NOW;
784
785 # now or in the past already?
786 if ($after <= 0) {
787 $self->{$activity} = $NOW;
788
789 if ($self->{$on_timeout}) {
790 $self->{$on_timeout}($self);
791 } else {
792 $self->_error (Errno::ETIMEDOUT);
793 }
794
795 # callback could have changed timeout value, optimise
796 return unless $self->{$timeout};
797
798 # calculate new after
799 $after = $self->{$timeout};
800 }
801
802 Scalar::Util::weaken $self;
803 return unless $self; # ->error could have destroyed $self
804
805 $self->{$tw} ||= AE::timer $after, 0, sub {
806 delete $self->{$tw};
807 $cb->($self);
808 };
809 } else {
810 delete $self->{$tw};
811 }
812 }
813}
814
205############################################################################# 815#############################################################################
206 816
207=back 817=back
208 818
209=head2 WRITE QUEUE 819=head2 WRITE QUEUE
212for reading. 822for reading.
213 823
214The write queue is very simple: you can add data to its end, and 824The write queue is very simple: you can add data to its end, and
215AnyEvent::Handle will automatically try to get rid of it for you. 825AnyEvent::Handle will automatically try to get rid of it for you.
216 826
217When data could be writtena nd the write buffer is shorter then the low 827When data could be written and the write buffer is shorter then the low
218water mark, the C<on_drain> callback will be invoked. 828water mark, the C<on_drain> callback will be invoked.
219 829
220=over 4 830=over 4
221 831
222=item $handle->on_drain ($cb) 832=item $handle->on_drain ($cb)
223 833
224Sets the C<on_drain> callback or clears it (see the description of 834Sets the C<on_drain> callback or clears it (see the description of
225C<on_drain> in the constructor). 835C<on_drain> in the constructor).
226 836
837This method may invoke callbacks (and therefore the handle might be
838destroyed after it returns).
839
227=cut 840=cut
228 841
229sub on_drain { 842sub on_drain {
230 my ($self, $cb) = @_; 843 my ($self, $cb) = @_;
231 844
232 $self->{on_drain} = $cb; 845 $self->{on_drain} = $cb;
233 846
234 $cb->($self) 847 $cb->($self)
235 if $cb && $self->{low_water_mark} >= length $self->{wbuf}; 848 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
236} 849}
237 850
238=item $handle->push_write ($data) 851=item $handle->push_write ($data)
239 852
240Queues the given scalar to be written. You can push as much data as you 853Queues the given scalar to be written. You can push as much data as you
241want (only limited by the available memory), as C<AnyEvent::Handle> 854want (only limited by the available memory), as C<AnyEvent::Handle>
242buffers it independently of the kernel. 855buffers it independently of the kernel.
243 856
244=cut 857This method may invoke callbacks (and therefore the handle might be
858destroyed after it returns).
245 859
246sub push_write { 860=cut
861
862sub _drain_wbuf {
247 my ($self, $data) = @_; 863 my ($self) = @_;
248 864
249 $self->{wbuf} .= $data; 865 if (!$self->{_ww} && length $self->{wbuf}) {
250 866
251 unless ($self->{ww}) {
252 Scalar::Util::weaken $self; 867 Scalar::Util::weaken $self;
868
253 my $cb = sub { 869 my $cb = sub {
254 my $len = syswrite $self->{fh}, $self->{wbuf}; 870 my $len = syswrite $self->{fh}, $self->{wbuf};
255 871
256 if ($len > 0) { 872 if (defined $len) {
257 substr $self->{wbuf}, 0, $len, ""; 873 substr $self->{wbuf}, 0, $len, "";
258 874
875 $self->{_activity} = $self->{_wactivity} = AE::now;
259 876
260 $self->{on_drain}($self) 877 $self->{on_drain}($self)
261 if $self->{low_water_mark} >= length $self->{wbuf} 878 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
262 && $self->{on_drain}; 879 && $self->{on_drain};
263 880
264 delete $self->{ww} unless length $self->{wbuf}; 881 delete $self->{_ww} unless length $self->{wbuf};
265 } elsif ($! != EAGAIN && $! != EINTR) { 882 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
266 $self->error; 883 $self->_error ($!, 1);
267 } 884 }
268 }; 885 };
269 886
270 $self->{ww} = AnyEvent->io (fh => $self->{fh}, poll => "w", cb => $cb); 887 # try to write data immediately
888 $cb->() unless $self->{autocork};
271 889
272 $cb->($self); 890 # if still data left in wbuf, we need to poll
891 $self->{_ww} = AE::io $self->{fh}, 1, $cb
892 if length $self->{wbuf};
273 }; 893 };
274} 894}
895
896our %WH;
897
898# deprecated
899sub register_write_type($$) {
900 $WH{$_[0]} = $_[1];
901}
902
903sub push_write {
904 my $self = shift;
905
906 if (@_ > 1) {
907 my $type = shift;
908
909 @_ = ($WH{$type} ||= _load_func "$type\::anyevent_write_type"
910 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_write")
911 ->($self, @_);
912 }
913
914 # we downgrade here to avoid hard-to-track-down bugs,
915 # and diagnose the problem earlier and better.
916
917 if ($self->{tls}) {
918 utf8::downgrade $self->{_tls_wbuf} .= $_[0];
919 &_dotls ($self) if $self->{fh};
920 } else {
921 utf8::downgrade $self->{wbuf} .= $_[0];
922 $self->_drain_wbuf if $self->{fh};
923 }
924}
925
926=item $handle->push_write (type => @args)
927
928Instead of formatting your data yourself, you can also let this module
929do the job by specifying a type and type-specific arguments. You
930can also specify the (fully qualified) name of a package, in which
931case AnyEvent tries to load the package and then expects to find the
932C<anyevent_read_type> function inside (see "custom write types", below).
933
934Predefined types are (if you have ideas for additional types, feel free to
935drop by and tell us):
936
937=over 4
938
939=item netstring => $string
940
941Formats the given value as netstring
942(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
943
944=cut
945
946register_write_type netstring => sub {
947 my ($self, $string) = @_;
948
949 (length $string) . ":$string,"
950};
951
952=item packstring => $format, $data
953
954An octet string prefixed with an encoded length. The encoding C<$format>
955uses the same format as a Perl C<pack> format, but must specify a single
956integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
957optional C<!>, C<< < >> or C<< > >> modifier).
958
959=cut
960
961register_write_type packstring => sub {
962 my ($self, $format, $string) = @_;
963
964 pack "$format/a*", $string
965};
966
967=item json => $array_or_hashref
968
969Encodes the given hash or array reference into a JSON object. Unless you
970provide your own JSON object, this means it will be encoded to JSON text
971in UTF-8.
972
973JSON objects (and arrays) are self-delimiting, so you can write JSON at
974one end of a handle and read them at the other end without using any
975additional framing.
976
977The generated JSON text is guaranteed not to contain any newlines: While
978this module doesn't need delimiters after or between JSON texts to be
979able to read them, many other languages depend on that.
980
981A simple RPC protocol that interoperates easily with others is to send
982JSON arrays (or objects, although arrays are usually the better choice as
983they mimic how function argument passing works) and a newline after each
984JSON text:
985
986 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
987 $handle->push_write ("\012");
988
989An AnyEvent::Handle receiver would simply use the C<json> read type and
990rely on the fact that the newline will be skipped as leading whitespace:
991
992 $handle->push_read (json => sub { my $array = $_[1]; ... });
993
994Other languages could read single lines terminated by a newline and pass
995this line into their JSON decoder of choice.
996
997=cut
998
999sub json_coder() {
1000 eval { require JSON::XS; JSON::XS->new->utf8 }
1001 || do { require JSON; JSON->new->utf8 }
1002}
1003
1004register_write_type json => sub {
1005 my ($self, $ref) = @_;
1006
1007 my $json = $self->{json} ||= json_coder;
1008
1009 $json->encode ($ref)
1010};
1011
1012=item storable => $reference
1013
1014Freezes the given reference using L<Storable> and writes it to the
1015handle. Uses the C<nfreeze> format.
1016
1017=cut
1018
1019register_write_type storable => sub {
1020 my ($self, $ref) = @_;
1021
1022 require Storable;
1023
1024 pack "w/a*", Storable::nfreeze ($ref)
1025};
1026
1027=back
1028
1029=item $handle->push_shutdown
1030
1031Sometimes you know you want to close the socket after writing your data
1032before it was actually written. One way to do that is to replace your
1033C<on_drain> handler by a callback that shuts down the socket (and set
1034C<low_water_mark> to C<0>). This method is a shorthand for just that, and
1035replaces the C<on_drain> callback with:
1036
1037 sub { shutdown $_[0]{fh}, 1 } # for push_shutdown
1038
1039This simply shuts down the write side and signals an EOF condition to the
1040the peer.
1041
1042You can rely on the normal read queue and C<on_eof> handling
1043afterwards. This is the cleanest way to close a connection.
1044
1045This method may invoke callbacks (and therefore the handle might be
1046destroyed after it returns).
1047
1048=cut
1049
1050sub push_shutdown {
1051 my ($self) = @_;
1052
1053 delete $self->{low_water_mark};
1054 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
1055}
1056
1057=item custom write types - Package::anyevent_write_type $handle, @args
1058
1059Instead of one of the predefined types, you can also specify the name of
1060a package. AnyEvent will try to load the package and then expects to find
1061a function named C<anyevent_write_type> inside. If it isn't found, it
1062progressively tries to load the parent package until it either finds the
1063function (good) or runs out of packages (bad).
1064
1065Whenever the given C<type> is used, C<push_write> will the function with
1066the handle object and the remaining arguments.
1067
1068The function is supposed to return a single octet string that will be
1069appended to the write buffer, so you cna mentally treat this function as a
1070"arguments to on-the-wire-format" converter.
1071
1072Example: implement a custom write type C<join> that joins the remaining
1073arguments using the first one.
1074
1075 $handle->push_write (My::Type => " ", 1,2,3);
1076
1077 # uses the following package, which can be defined in the "My::Type" or in
1078 # the "My" modules to be auto-loaded, or just about anywhere when the
1079 # My::Type::anyevent_write_type is defined before invoking it.
1080
1081 package My::Type;
1082
1083 sub anyevent_write_type {
1084 my ($handle, $delim, @args) = @_;
1085
1086 join $delim, @args
1087 }
1088
1089=cut
275 1090
276############################################################################# 1091#############################################################################
277 1092
278=back 1093=back
279 1094
286ways, the "simple" way, using only C<on_read> and the "complex" way, using 1101ways, the "simple" way, using only C<on_read> and the "complex" way, using
287a queue. 1102a queue.
288 1103
289In the simple case, you just install an C<on_read> callback and whenever 1104In the simple case, you just install an C<on_read> callback and whenever
290new data arrives, it will be called. You can then remove some data (if 1105new data arrives, it will be called. You can then remove some data (if
291enough is there) from the read buffer (C<< $handle->rbuf >>) if you want 1106enough is there) from the read buffer (C<< $handle->rbuf >>). Or you cna
292or not. 1107leave the data there if you want to accumulate more (e.g. when only a
1108partial message has been received so far).
293 1109
294In the more complex case, you want to queue multiple callbacks. In this 1110In the more complex case, you want to queue multiple callbacks. In this
295case, AnyEvent::Handle will call the first queued callback each time new 1111case, AnyEvent::Handle will call the first queued callback each time new
296data arrives and removes it when it has done its job (see C<push_read>, 1112data arrives (also the first time it is queued) and removes it when it has
297below). 1113done its job (see C<push_read>, below).
298 1114
299This way you can, for example, push three line-reads, followed by reading 1115This way you can, for example, push three line-reads, followed by reading
300a chunk of data, and AnyEvent::Handle will execute them in order. 1116a chunk of data, and AnyEvent::Handle will execute them in order.
301 1117
302Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by 1118Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
303the specified number of bytes which give an XML datagram. 1119the specified number of bytes which give an XML datagram.
304 1120
305 # in the default state, expect some header bytes 1121 # in the default state, expect some header bytes
306 $handle->on_read (sub { 1122 $handle->on_read (sub {
307 # some data is here, now queue the length-header-read (4 octets) 1123 # some data is here, now queue the length-header-read (4 octets)
308 shift->unshift_read_chunk (4, sub { 1124 shift->unshift_read (chunk => 4, sub {
309 # header arrived, decode 1125 # header arrived, decode
310 my $len = unpack "N", $_[1]; 1126 my $len = unpack "N", $_[1];
311 1127
312 # now read the payload 1128 # now read the payload
313 shift->unshift_read_chunk ($len, sub { 1129 shift->unshift_read (chunk => $len, sub {
314 my $xml = $_[1]; 1130 my $xml = $_[1];
315 # handle xml 1131 # handle xml
316 }); 1132 });
317 }); 1133 });
318 }); 1134 });
319 1135
320Example 2: Implement a client for a protocol that replies either with 1136Example 2: Implement a client for a protocol that replies either with "OK"
321"OK" and another line or "ERROR" for one request, and 64 bytes for the 1137and another line or "ERROR" for the first request that is sent, and 64
322second request. Due tot he availability of a full queue, we can just 1138bytes for the second request. Due to the availability of a queue, we can
323pipeline sending both requests and manipulate the queue as necessary in 1139just pipeline sending both requests and manipulate the queue as necessary
324the callbacks: 1140in the callbacks.
325 1141
326 # request one 1142When the first callback is called and sees an "OK" response, it will
1143C<unshift> another line-read. This line-read will be queued I<before> the
114464-byte chunk callback.
1145
1146 # request one, returns either "OK + extra line" or "ERROR"
327 $handle->push_write ("request 1\015\012"); 1147 $handle->push_write ("request 1\015\012");
328 1148
329 # we expect "ERROR" or "OK" as response, so push a line read 1149 # we expect "ERROR" or "OK" as response, so push a line read
330 $handle->push_read_line (sub { 1150 $handle->push_read (line => sub {
331 # if we got an "OK", we have to _prepend_ another line, 1151 # if we got an "OK", we have to _prepend_ another line,
332 # so it will be read before the second request reads its 64 bytes 1152 # so it will be read before the second request reads its 64 bytes
333 # which are already in the queue when this callback is called 1153 # which are already in the queue when this callback is called
334 # we don't do this in case we got an error 1154 # we don't do this in case we got an error
335 if ($_[1] eq "OK") { 1155 if ($_[1] eq "OK") {
336 $_[0]->unshift_read_line (sub { 1156 $_[0]->unshift_read (line => sub {
337 my $response = $_[1]; 1157 my $response = $_[1];
338 ... 1158 ...
339 }); 1159 });
340 } 1160 }
341 }); 1161 });
342 1162
343 # request two 1163 # request two, simply returns 64 octets
344 $handle->push_write ("request 2\015\012"); 1164 $handle->push_write ("request 2\015\012");
345 1165
346 # simply read 64 bytes, always 1166 # simply read 64 bytes, always
347 $handle->push_read_chunk (64, sub { 1167 $handle->push_read (chunk => 64, sub {
348 my $response = $_[1]; 1168 my $response = $_[1];
349 ... 1169 ...
350 }); 1170 });
351 1171
352=over 4 1172=over 4
353 1173
1174=cut
1175
354sub _drain_rbuf { 1176sub _drain_rbuf {
355 my ($self) = @_; 1177 my ($self) = @_;
356 1178
1179 # avoid recursion
357 return if exists $self->{in_drain}; 1180 return if $self->{_skip_drain_rbuf};
358 local $self->{in_drain} = 1; 1181 local $self->{_skip_drain_rbuf} = 1;
359 1182
1183 while () {
1184 # we need to use a separate tls read buffer, as we must not receive data while
1185 # we are draining the buffer, and this can only happen with TLS.
1186 $self->{rbuf} .= delete $self->{_tls_rbuf}
1187 if exists $self->{_tls_rbuf};
1188
360 while (my $len = length $self->{rbuf}) { 1189 my $len = length $self->{rbuf};
361 no strict 'refs'; 1190
362 if (@{ $self->{queue} }) { 1191 if (my $cb = shift @{ $self->{_queue} }) {
363 if ($self->{queue}[0]($self)) { 1192 unless ($cb->($self)) {
1193 # no progress can be made
1194 # (not enough data and no data forthcoming)
1195 $self->_error (Errno::EPIPE, 1), return
1196 if $self->{_eof};
1197
364 shift @{ $self->{queue} }; 1198 unshift @{ $self->{_queue} }, $cb;
365 } elsif ($self->{eof}) {
366 # no progress can be made (not enough data and no data forthcoming)
367 $! = &Errno::EPIPE; return $self->error;
368 } else {
369 return; 1199 last;
370 } 1200 }
371 } elsif ($self->{on_read}) { 1201 } elsif ($self->{on_read}) {
1202 last unless $len;
1203
372 $self->{on_read}($self); 1204 $self->{on_read}($self);
373 1205
374 if ( 1206 if (
375 $self->{eof} # if no further data will arrive
376 && $len == length $self->{rbuf} # and no data has been consumed 1207 $len == length $self->{rbuf} # if no data has been consumed
377 && !@{ $self->{queue} } # and the queue is still empty 1208 && !@{ $self->{_queue} } # and the queue is still empty
378 && $self->{on_read} # and we still want to read data 1209 && $self->{on_read} # but we still have on_read
379 ) { 1210 ) {
1211 # no further data will arrive
380 # then no progress can be made 1212 # so no progress can be made
381 $! = &Errno::EPIPE; return $self->error; 1213 $self->_error (Errno::EPIPE, 1), return
1214 if $self->{_eof};
1215
1216 last; # more data might arrive
382 } 1217 }
383 } else { 1218 } else {
384 # read side becomes idle 1219 # read side becomes idle
385 delete $self->{rw}; 1220 delete $self->{_rw} unless $self->{tls};
386 return; 1221 last;
387 } 1222 }
388 } 1223 }
389 1224
390 if ($self->{eof}) { 1225 if ($self->{_eof}) {
391 $self->_shutdown; 1226 $self->{on_eof}
392 $self->{on_eof}($self); 1227 ? $self->{on_eof}($self)
1228 : $self->_error (0, 1, "Unexpected end-of-file");
1229
1230 return;
1231 }
1232
1233 if (
1234 defined $self->{rbuf_max}
1235 && $self->{rbuf_max} < length $self->{rbuf}
1236 ) {
1237 $self->_error (Errno::ENOSPC, 1), return;
1238 }
1239
1240 # may need to restart read watcher
1241 unless ($self->{_rw}) {
1242 $self->start_read
1243 if $self->{on_read} || @{ $self->{_queue} };
393 } 1244 }
394} 1245}
395 1246
396=item $handle->on_read ($cb) 1247=item $handle->on_read ($cb)
397 1248
398This replaces the currently set C<on_read> callback, or clears it (when 1249This replaces the currently set C<on_read> callback, or clears it (when
399the new callback is C<undef>). See the description of C<on_read> in the 1250the new callback is C<undef>). See the description of C<on_read> in the
400constructor. 1251constructor.
401 1252
1253This method may invoke callbacks (and therefore the handle might be
1254destroyed after it returns).
1255
402=cut 1256=cut
403 1257
404sub on_read { 1258sub on_read {
405 my ($self, $cb) = @_; 1259 my ($self, $cb) = @_;
406 1260
407 $self->{on_read} = $cb; 1261 $self->{on_read} = $cb;
408
409 unless ($self->{rw} || $self->{eof}) {
410 Scalar::Util::weaken $self;
411
412 $self->{rw} = AnyEvent->io (fh => $self->{fh}, poll => "r", cb => sub {
413 my $len = sysread $self->{fh}, $self->{rbuf}, $self->{read_size} || 8192, length $self->{rbuf};
414
415 if ($len > 0) {
416 if (exists $self->{rbuf_max}) {
417 if ($self->{rbuf_max} < length $self->{rbuf}) {
418 $! = &Errno::ENOSPC; return $self->error;
419 }
420 }
421
422 } elsif (defined $len) {
423 $self->{eof} = 1;
424 delete $self->{rw};
425
426 } elsif ($! != EAGAIN && $! != EINTR) {
427 return $self->error;
428 }
429
430 $self->_drain_rbuf; 1262 $self->_drain_rbuf if $cb;
431 });
432 }
433} 1263}
434 1264
435=item $handle->rbuf 1265=item $handle->rbuf
436 1266
437Returns the read buffer (as a modifiable lvalue). 1267Returns the read buffer (as a modifiable lvalue).
438 1268
439You can access the read buffer directly as the C<< ->{rbuf} >> member, if 1269You can access the read buffer directly as the C<< ->{rbuf} >>
440you want. 1270member, if you want. However, the only operation allowed on the
1271read buffer (apart from looking at it) is removing data from its
1272beginning. Otherwise modifying or appending to it is not allowed and will
1273lead to hard-to-track-down bugs.
441 1274
442NOTE: The read buffer should only be used or modified if the C<on_read>, 1275NOTE: The read buffer should only be used or modified if the C<on_read>,
443C<push_read> or C<unshift_read> methods are used. The other read methods 1276C<push_read> or C<unshift_read> methods are used. The other read methods
444automatically manage the read buffer. 1277automatically manage the read buffer.
445 1278
456Append the given callback to the end of the queue (C<push_read>) or 1289Append the given callback to the end of the queue (C<push_read>) or
457prepend it (C<unshift_read>). 1290prepend it (C<unshift_read>).
458 1291
459The callback is called each time some additional read data arrives. 1292The callback is called each time some additional read data arrives.
460 1293
461It must check wether enough data is in the read buffer already. 1294It must check whether enough data is in the read buffer already.
462 1295
463If not enough data is available, it must return the empty list or a false 1296If not enough data is available, it must return the empty list or a false
464value, in which case it will be called repeatedly until enough data is 1297value, in which case it will be called repeatedly until enough data is
465available (or an error condition is detected). 1298available (or an error condition is detected).
466 1299
467If enough data was available, then the callback must remove all data it is 1300If enough data was available, then the callback must remove all data it is
468interested in (which can be none at all) and return a true value. After returning 1301interested in (which can be none at all) and return a true value. After returning
469true, it will be removed from the queue. 1302true, it will be removed from the queue.
470 1303
1304These methods may invoke callbacks (and therefore the handle might be
1305destroyed after it returns).
1306
471=cut 1307=cut
1308
1309our %RH;
1310
1311sub register_read_type($$) {
1312 $RH{$_[0]} = $_[1];
1313}
472 1314
473sub push_read { 1315sub push_read {
474 my ($self, $cb) = @_; 1316 my $self = shift;
1317 my $cb = pop;
475 1318
1319 if (@_) {
1320 my $type = shift;
1321
1322 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1323 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_read")
1324 ->($self, $cb, @_);
1325 }
1326
476 push @{ $self->{queue} }, $cb; 1327 push @{ $self->{_queue} }, $cb;
477 $self->_drain_rbuf; 1328 $self->_drain_rbuf;
478} 1329}
479 1330
480sub unshift_read { 1331sub unshift_read {
481 my ($self, $cb) = @_; 1332 my $self = shift;
1333 my $cb = pop;
482 1334
1335 if (@_) {
1336 my $type = shift;
1337
1338 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::unshift_read")
1339 ->($self, $cb, @_);
1340 }
1341
483 push @{ $self->{queue} }, $cb; 1342 unshift @{ $self->{_queue} }, $cb;
484 $self->_drain_rbuf; 1343 $self->_drain_rbuf;
485} 1344}
486 1345
487=item $handle->push_read_chunk ($len, $cb->($self, $data)) 1346=item $handle->push_read (type => @args, $cb)
488 1347
489=item $handle->unshift_read_chunk ($len, $cb->($self, $data)) 1348=item $handle->unshift_read (type => @args, $cb)
490 1349
491Append the given callback to the end of the queue (C<push_read_chunk>) or 1350Instead of providing a callback that parses the data itself you can chose
492prepend it (C<unshift_read_chunk>). 1351between a number of predefined parsing formats, for chunks of data, lines
1352etc. You can also specify the (fully qualified) name of a package, in
1353which case AnyEvent tries to load the package and then expects to find the
1354C<anyevent_read_type> function inside (see "custom read types", below).
493 1355
494The callback will be called only once C<$len> bytes have been read, and 1356Predefined types are (if you have ideas for additional types, feel free to
495these C<$len> bytes will be passed to the callback. 1357drop by and tell us):
496 1358
497=cut 1359=over 4
498 1360
499sub _read_chunk($$) { 1361=item chunk => $octets, $cb->($handle, $data)
500 my ($len, $cb) = @_; 1362
1363Invoke the callback only once C<$octets> bytes have been read. Pass the
1364data read to the callback. The callback will never be called with less
1365data.
1366
1367Example: read 2 bytes.
1368
1369 $handle->push_read (chunk => 2, sub {
1370 warn "yay ", unpack "H*", $_[1];
1371 });
1372
1373=cut
1374
1375register_read_type chunk => sub {
1376 my ($self, $cb, $len) = @_;
501 1377
502 sub { 1378 sub {
503 $len <= length $_[0]{rbuf} or return; 1379 $len <= length $_[0]{rbuf} or return;
504 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, ""); 1380 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
505 1 1381 1
506 } 1382 }
507} 1383};
508 1384
509sub push_read_chunk { 1385=item line => [$eol, ]$cb->($handle, $line, $eol)
510 my ($self, $len, $cb) = @_;
511
512 $self->push_read (_read_chunk $len, $cb);
513}
514
515
516sub unshift_read_chunk {
517 my ($self, $len, $cb) = @_;
518
519 $self->unshift_read (_read_chunk $len, $cb);
520}
521
522=item $handle->push_read_line ([$eol, ]$cb->($self, $line, $eol))
523
524=item $handle->unshift_read_line ([$eol, ]$cb->($self, $line, $eol))
525
526Append the given callback to the end of the queue (C<push_read_line>) or
527prepend it (C<unshift_read_line>).
528 1386
529The callback will be called only once a full line (including the end of 1387The callback will be called only once a full line (including the end of
530line marker, C<$eol>) has been read. This line (excluding the end of line 1388line marker, C<$eol>) has been read. This line (excluding the end of line
531marker) will be passed to the callback as second argument (C<$line>), and 1389marker) will be passed to the callback as second argument (C<$line>), and
532the end of line marker as the third argument (C<$eol>). 1390the end of line marker as the third argument (C<$eol>).
543Partial lines at the end of the stream will never be returned, as they are 1401Partial lines at the end of the stream will never be returned, as they are
544not marked by the end of line marker. 1402not marked by the end of line marker.
545 1403
546=cut 1404=cut
547 1405
548sub _read_line($$) { 1406register_read_type line => sub {
549 my $cb = pop; 1407 my ($self, $cb, $eol) = @_;
550 my $eol = @_ ? shift : qr|(\015?\012)|;
551 my $pos;
552 1408
1409 if (@_ < 3) {
1410 # this is more than twice as fast as the generic code below
1411 sub {
1412 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1413
1414 $cb->($_[0], $1, $2);
1415 1
1416 }
1417 } else {
553 $eol = qr|(\Q$eol\E)| unless ref $eol; 1418 $eol = quotemeta $eol unless ref $eol;
554 $eol = qr|^(.*?)($eol)|; 1419 $eol = qr|^(.*?)($eol)|s;
1420
1421 sub {
1422 $_[0]{rbuf} =~ s/$eol// or return;
1423
1424 $cb->($_[0], $1, $2);
1425 1
1426 }
1427 }
1428};
1429
1430=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1431
1432Makes a regex match against the regex object C<$accept> and returns
1433everything up to and including the match.
1434
1435Example: read a single line terminated by '\n'.
1436
1437 $handle->push_read (regex => qr<\n>, sub { ... });
1438
1439If C<$reject> is given and not undef, then it determines when the data is
1440to be rejected: it is matched against the data when the C<$accept> regex
1441does not match and generates an C<EBADMSG> error when it matches. This is
1442useful to quickly reject wrong data (to avoid waiting for a timeout or a
1443receive buffer overflow).
1444
1445Example: expect a single decimal number followed by whitespace, reject
1446anything else (not the use of an anchor).
1447
1448 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1449
1450If C<$skip> is given and not C<undef>, then it will be matched against
1451the receive buffer when neither C<$accept> nor C<$reject> match,
1452and everything preceding and including the match will be accepted
1453unconditionally. This is useful to skip large amounts of data that you
1454know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1455have to start matching from the beginning. This is purely an optimisation
1456and is usually worth only when you expect more than a few kilobytes.
1457
1458Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1459expect the header to be very large (it isn't in practise, but...), we use
1460a skip regex to skip initial portions. The skip regex is tricky in that
1461it only accepts something not ending in either \015 or \012, as these are
1462required for the accept regex.
1463
1464 $handle->push_read (regex =>
1465 qr<\015\012\015\012>,
1466 undef, # no reject
1467 qr<^.*[^\015\012]>,
1468 sub { ... });
1469
1470=cut
1471
1472register_read_type regex => sub {
1473 my ($self, $cb, $accept, $reject, $skip) = @_;
1474
1475 my $data;
1476 my $rbuf = \$self->{rbuf};
555 1477
556 sub { 1478 sub {
557 $_[0]{rbuf} =~ s/$eol// or return; 1479 # accept
1480 if ($$rbuf =~ $accept) {
1481 $data .= substr $$rbuf, 0, $+[0], "";
1482 $cb->($self, $data);
1483 return 1;
1484 }
1485
1486 # reject
1487 if ($reject && $$rbuf =~ $reject) {
1488 $self->_error (Errno::EBADMSG);
1489 }
558 1490
559 $cb->($1, $2); 1491 # skip
1492 if ($skip && $$rbuf =~ $skip) {
1493 $data .= substr $$rbuf, 0, $+[0], "";
1494 }
1495
1496 ()
1497 }
1498};
1499
1500=item netstring => $cb->($handle, $string)
1501
1502A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
1503
1504Throws an error with C<$!> set to EBADMSG on format violations.
1505
1506=cut
1507
1508register_read_type netstring => sub {
1509 my ($self, $cb) = @_;
1510
1511 sub {
1512 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1513 if ($_[0]{rbuf} =~ /[^0-9]/) {
1514 $self->_error (Errno::EBADMSG);
1515 }
1516 return;
1517 }
1518
1519 my $len = $1;
1520
1521 $self->unshift_read (chunk => $len, sub {
1522 my $string = $_[1];
1523 $_[0]->unshift_read (chunk => 1, sub {
1524 if ($_[1] eq ",") {
1525 $cb->($_[0], $string);
1526 } else {
1527 $self->_error (Errno::EBADMSG);
1528 }
1529 });
1530 });
1531
560 1 1532 1
561 } 1533 }
562} 1534};
563 1535
564sub push_read_line { 1536=item packstring => $format, $cb->($handle, $string)
565 my $self = shift;
566 1537
567 $self->push_read (&_read_line); 1538An octet string prefixed with an encoded length. The encoding C<$format>
568} 1539uses the same format as a Perl C<pack> format, but must specify a single
1540integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1541optional C<!>, C<< < >> or C<< > >> modifier).
569 1542
570sub unshift_read_line { 1543For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
571 my $self = shift; 1544EPP uses a prefix of C<N> (4 octtes).
572 1545
573 $self->unshift_read (&_read_line); 1546Example: read a block of data prefixed by its length in BER-encoded
574} 1547format (very efficient).
1548
1549 $handle->push_read (packstring => "w", sub {
1550 my ($handle, $data) = @_;
1551 });
1552
1553=cut
1554
1555register_read_type packstring => sub {
1556 my ($self, $cb, $format) = @_;
1557
1558 sub {
1559 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1560 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1561 or return;
1562
1563 $format = length pack $format, $len;
1564
1565 # bypass unshift if we already have the remaining chunk
1566 if ($format + $len <= length $_[0]{rbuf}) {
1567 my $data = substr $_[0]{rbuf}, $format, $len;
1568 substr $_[0]{rbuf}, 0, $format + $len, "";
1569 $cb->($_[0], $data);
1570 } else {
1571 # remove prefix
1572 substr $_[0]{rbuf}, 0, $format, "";
1573
1574 # read remaining chunk
1575 $_[0]->unshift_read (chunk => $len, $cb);
1576 }
1577
1578 1
1579 }
1580};
1581
1582=item json => $cb->($handle, $hash_or_arrayref)
1583
1584Reads a JSON object or array, decodes it and passes it to the
1585callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1586
1587If a C<json> object was passed to the constructor, then that will be used
1588for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1589
1590This read type uses the incremental parser available with JSON version
15912.09 (and JSON::XS version 2.2) and above. You have to provide a
1592dependency on your own: this module will load the JSON module, but
1593AnyEvent does not depend on it itself.
1594
1595Since JSON texts are fully self-delimiting, the C<json> read and write
1596types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1597the C<json> write type description, above, for an actual example.
1598
1599=cut
1600
1601register_read_type json => sub {
1602 my ($self, $cb) = @_;
1603
1604 my $json = $self->{json} ||= json_coder;
1605
1606 my $data;
1607 my $rbuf = \$self->{rbuf};
1608
1609 sub {
1610 my $ref = eval { $json->incr_parse ($self->{rbuf}) };
1611
1612 if ($ref) {
1613 $self->{rbuf} = $json->incr_text;
1614 $json->incr_text = "";
1615 $cb->($self, $ref);
1616
1617 1
1618 } elsif ($@) {
1619 # error case
1620 $json->incr_skip;
1621
1622 $self->{rbuf} = $json->incr_text;
1623 $json->incr_text = "";
1624
1625 $self->_error (Errno::EBADMSG);
1626
1627 ()
1628 } else {
1629 $self->{rbuf} = "";
1630
1631 ()
1632 }
1633 }
1634};
1635
1636=item storable => $cb->($handle, $ref)
1637
1638Deserialises a L<Storable> frozen representation as written by the
1639C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1640data).
1641
1642Raises C<EBADMSG> error if the data could not be decoded.
1643
1644=cut
1645
1646register_read_type storable => sub {
1647 my ($self, $cb) = @_;
1648
1649 require Storable;
1650
1651 sub {
1652 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1653 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1654 or return;
1655
1656 my $format = length pack "w", $len;
1657
1658 # bypass unshift if we already have the remaining chunk
1659 if ($format + $len <= length $_[0]{rbuf}) {
1660 my $data = substr $_[0]{rbuf}, $format, $len;
1661 substr $_[0]{rbuf}, 0, $format + $len, "";
1662 $cb->($_[0], Storable::thaw ($data));
1663 } else {
1664 # remove prefix
1665 substr $_[0]{rbuf}, 0, $format, "";
1666
1667 # read remaining chunk
1668 $_[0]->unshift_read (chunk => $len, sub {
1669 if (my $ref = eval { Storable::thaw ($_[1]) }) {
1670 $cb->($_[0], $ref);
1671 } else {
1672 $self->_error (Errno::EBADMSG);
1673 }
1674 });
1675 }
1676
1677 1
1678 }
1679};
575 1680
576=back 1681=back
577 1682
1683=item custom read types - Package::anyevent_read_type $handle, $cb, @args
1684
1685Instead of one of the predefined types, you can also specify the name
1686of a package. AnyEvent will try to load the package and then expects to
1687find a function named C<anyevent_read_type> inside. If it isn't found, it
1688progressively tries to load the parent package until it either finds the
1689function (good) or runs out of packages (bad).
1690
1691Whenever this type is used, C<push_read> will invoke the function with the
1692handle object, the original callback and the remaining arguments.
1693
1694The function is supposed to return a callback (usually a closure) that
1695works as a plain read callback (see C<< ->push_read ($cb) >>), so you can
1696mentally treat the function as a "configurable read type to read callback"
1697converter.
1698
1699It should invoke the original callback when it is done reading (remember
1700to pass C<$handle> as first argument as all other callbacks do that,
1701although there is no strict requirement on this).
1702
1703For examples, see the source of this module (F<perldoc -m
1704AnyEvent::Handle>, search for C<register_read_type>)).
1705
1706=item $handle->stop_read
1707
1708=item $handle->start_read
1709
1710In rare cases you actually do not want to read anything from the
1711socket. In this case you can call C<stop_read>. Neither C<on_read> nor
1712any queued callbacks will be executed then. To start reading again, call
1713C<start_read>.
1714
1715Note that AnyEvent::Handle will automatically C<start_read> for you when
1716you change the C<on_read> callback or push/unshift a read callback, and it
1717will automatically C<stop_read> for you when neither C<on_read> is set nor
1718there are any read requests in the queue.
1719
1720These methods will have no effect when in TLS mode (as TLS doesn't support
1721half-duplex connections).
1722
1723=cut
1724
1725sub stop_read {
1726 my ($self) = @_;
1727
1728 delete $self->{_rw} unless $self->{tls};
1729}
1730
1731sub start_read {
1732 my ($self) = @_;
1733
1734 unless ($self->{_rw} || $self->{_eof} || !$self->{fh}) {
1735 Scalar::Util::weaken $self;
1736
1737 $self->{_rw} = AE::io $self->{fh}, 0, sub {
1738 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
1739 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size} || 8192, length $$rbuf;
1740
1741 if ($len > 0) {
1742 $self->{_activity} = $self->{_ractivity} = AE::now;
1743
1744 if ($self->{tls}) {
1745 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1746
1747 &_dotls ($self);
1748 } else {
1749 $self->_drain_rbuf;
1750 }
1751
1752 } elsif (defined $len) {
1753 delete $self->{_rw};
1754 $self->{_eof} = 1;
1755 $self->_drain_rbuf;
1756
1757 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
1758 return $self->_error ($!, 1);
1759 }
1760 };
1761 }
1762}
1763
1764our $ERROR_SYSCALL;
1765our $ERROR_WANT_READ;
1766
1767sub _tls_error {
1768 my ($self, $err) = @_;
1769
1770 return $self->_error ($!, 1)
1771 if $err == Net::SSLeay::ERROR_SYSCALL ();
1772
1773 my $err =Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
1774
1775 # reduce error string to look less scary
1776 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
1777
1778 if ($self->{_on_starttls}) {
1779 (delete $self->{_on_starttls})->($self, undef, $err);
1780 &_freetls;
1781 } else {
1782 &_freetls;
1783 $self->_error (Errno::EPROTO, 1, $err);
1784 }
1785}
1786
1787# poll the write BIO and send the data if applicable
1788# also decode read data if possible
1789# this is basiclaly our TLS state machine
1790# more efficient implementations are possible with openssl,
1791# but not with the buggy and incomplete Net::SSLeay.
1792sub _dotls {
1793 my ($self) = @_;
1794
1795 my $tmp;
1796
1797 if (length $self->{_tls_wbuf}) {
1798 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
1799 substr $self->{_tls_wbuf}, 0, $tmp, "";
1800 }
1801
1802 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1803 return $self->_tls_error ($tmp)
1804 if $tmp != $ERROR_WANT_READ
1805 && ($tmp != $ERROR_SYSCALL || $!);
1806 }
1807
1808 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1809 unless (length $tmp) {
1810 $self->{_on_starttls}
1811 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1812 &_freetls;
1813
1814 if ($self->{on_stoptls}) {
1815 $self->{on_stoptls}($self);
1816 return;
1817 } else {
1818 # let's treat SSL-eof as we treat normal EOF
1819 delete $self->{_rw};
1820 $self->{_eof} = 1;
1821 }
1822 }
1823
1824 $self->{_tls_rbuf} .= $tmp;
1825 $self->_drain_rbuf;
1826 $self->{tls} or return; # tls session might have gone away in callback
1827 }
1828
1829 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
1830 return $self->_tls_error ($tmp)
1831 if $tmp != $ERROR_WANT_READ
1832 && ($tmp != $ERROR_SYSCALL || $!);
1833
1834 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
1835 $self->{wbuf} .= $tmp;
1836 $self->_drain_wbuf;
1837 $self->{tls} or return; # tls session might have gone away in callback
1838 }
1839
1840 $self->{_on_starttls}
1841 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
1842 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
1843}
1844
1845=item $handle->starttls ($tls[, $tls_ctx])
1846
1847Instead of starting TLS negotiation immediately when the AnyEvent::Handle
1848object is created, you can also do that at a later time by calling
1849C<starttls>.
1850
1851Starting TLS is currently an asynchronous operation - when you push some
1852write data and then call C<< ->starttls >> then TLS negotiation will start
1853immediately, after which the queued write data is then sent.
1854
1855The first argument is the same as the C<tls> constructor argument (either
1856C<"connect">, C<"accept"> or an existing Net::SSLeay object).
1857
1858The second argument is the optional C<AnyEvent::TLS> object that is used
1859when AnyEvent::Handle has to create its own TLS connection object, or
1860a hash reference with C<< key => value >> pairs that will be used to
1861construct a new context.
1862
1863The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1864context in C<< $handle->{tls_ctx} >> after this call and can be used or
1865changed to your liking. Note that the handshake might have already started
1866when this function returns.
1867
1868Due to bugs in OpenSSL, it might or might not be possible to do multiple
1869handshakes on the same stream. Best do not attempt to use the stream after
1870stopping TLS.
1871
1872This method may invoke callbacks (and therefore the handle might be
1873destroyed after it returns).
1874
1875=cut
1876
1877our %TLS_CACHE; #TODO not yet documented, should we?
1878
1879sub starttls {
1880 my ($self, $tls, $ctx) = @_;
1881
1882 Carp::croak "It is an error to call starttls on an AnyEvent::Handle object while TLS is already active, caught"
1883 if $self->{tls};
1884
1885 $self->{tls} = $tls;
1886 $self->{tls_ctx} = $ctx if @_ > 2;
1887
1888 return unless $self->{fh};
1889
1890 require Net::SSLeay;
1891
1892 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
1893 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
1894
1895 $tls = delete $self->{tls};
1896 $ctx = $self->{tls_ctx};
1897
1898 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context or session
1899
1900 if ("HASH" eq ref $ctx) {
1901 require AnyEvent::TLS;
1902
1903 if ($ctx->{cache}) {
1904 my $key = $ctx+0;
1905 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
1906 } else {
1907 $ctx = new AnyEvent::TLS %$ctx;
1908 }
1909 }
1910
1911 $self->{tls_ctx} = $ctx || TLS_CTX ();
1912 $self->{tls} = $tls = $self->{tls_ctx}->_get_session ($tls, $self, $self->{peername});
1913
1914 # basically, this is deep magic (because SSL_read should have the same issues)
1915 # but the openssl maintainers basically said: "trust us, it just works".
1916 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
1917 # and mismaintained ssleay-module doesn't even offer them).
1918 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
1919 #
1920 # in short: this is a mess.
1921 #
1922 # note that we do not try to keep the length constant between writes as we are required to do.
1923 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
1924 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
1925 # have identity issues in that area.
1926# Net::SSLeay::CTX_set_mode ($ssl,
1927# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
1928# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
1929 Net::SSLeay::CTX_set_mode ($tls, 1|2);
1930
1931 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
1932 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
1933
1934 Net::SSLeay::BIO_write ($self->{_rbio}, delete $self->{rbuf});
1935
1936 Net::SSLeay::set_bio ($tls, $self->{_rbio}, $self->{_wbio});
1937
1938 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
1939 if $self->{on_starttls};
1940
1941 &_dotls; # need to trigger the initial handshake
1942 $self->start_read; # make sure we actually do read
1943}
1944
1945=item $handle->stoptls
1946
1947Shuts down the SSL connection - this makes a proper EOF handshake by
1948sending a close notify to the other side, but since OpenSSL doesn't
1949support non-blocking shut downs, it is not guaranteed that you can re-use
1950the stream afterwards.
1951
1952This method may invoke callbacks (and therefore the handle might be
1953destroyed after it returns).
1954
1955=cut
1956
1957sub stoptls {
1958 my ($self) = @_;
1959
1960 if ($self->{tls} && $self->{fh}) {
1961 Net::SSLeay::shutdown ($self->{tls});
1962
1963 &_dotls;
1964
1965# # we don't give a shit. no, we do, but we can't. no...#d#
1966# # we, we... have to use openssl :/#d#
1967# &_freetls;#d#
1968 }
1969}
1970
1971sub _freetls {
1972 my ($self) = @_;
1973
1974 return unless $self->{tls};
1975
1976 $self->{tls_ctx}->_put_session (delete $self->{tls})
1977 if $self->{tls} > 0;
1978
1979 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
1980}
1981
1982sub DESTROY {
1983 my ($self) = @_;
1984
1985 &_freetls;
1986
1987 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
1988
1989 if ($linger && length $self->{wbuf} && $self->{fh}) {
1990 my $fh = delete $self->{fh};
1991 my $wbuf = delete $self->{wbuf};
1992
1993 my @linger;
1994
1995 push @linger, AE::io $fh, 1, sub {
1996 my $len = syswrite $fh, $wbuf, length $wbuf;
1997
1998 if ($len > 0) {
1999 substr $wbuf, 0, $len, "";
2000 } else {
2001 @linger = (); # end
2002 }
2003 };
2004 push @linger, AE::timer $linger, 0, sub {
2005 @linger = ();
2006 };
2007 }
2008}
2009
2010=item $handle->destroy
2011
2012Shuts down the handle object as much as possible - this call ensures that
2013no further callbacks will be invoked and as many resources as possible
2014will be freed. Any method you will call on the handle object after
2015destroying it in this way will be silently ignored (and it will return the
2016empty list).
2017
2018Normally, you can just "forget" any references to an AnyEvent::Handle
2019object and it will simply shut down. This works in fatal error and EOF
2020callbacks, as well as code outside. It does I<NOT> work in a read or write
2021callback, so when you want to destroy the AnyEvent::Handle object from
2022within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
2023that case.
2024
2025Destroying the handle object in this way has the advantage that callbacks
2026will be removed as well, so if those are the only reference holders (as
2027is common), then one doesn't need to do anything special to break any
2028reference cycles.
2029
2030The handle might still linger in the background and write out remaining
2031data, as specified by the C<linger> option, however.
2032
2033=cut
2034
2035sub destroy {
2036 my ($self) = @_;
2037
2038 $self->DESTROY;
2039 %$self = ();
2040 bless $self, "AnyEvent::Handle::destroyed";
2041}
2042
2043sub AnyEvent::Handle::destroyed::AUTOLOAD {
2044 #nop
2045}
2046
2047=item $handle->destroyed
2048
2049Returns false as long as the handle hasn't been destroyed by a call to C<<
2050->destroy >>, true otherwise.
2051
2052Can be useful to decide whether the handle is still valid after some
2053callback possibly destroyed the handle. For example, C<< ->push_write >>,
2054C<< ->starttls >> and other methods can call user callbacks, which in turn
2055can destroy the handle, so work can be avoided by checking sometimes:
2056
2057 $hdl->starttls ("accept");
2058 return if $hdl->destroyed;
2059 $hdl->push_write (...
2060
2061Note that the call to C<push_write> will silently be ignored if the handle
2062has been destroyed, so often you can just ignore the possibility of the
2063handle being destroyed.
2064
2065=cut
2066
2067sub destroyed { 0 }
2068sub AnyEvent::Handle::destroyed::destroyed { 1 }
2069
2070=item AnyEvent::Handle::TLS_CTX
2071
2072This function creates and returns the AnyEvent::TLS object used by default
2073for TLS mode.
2074
2075The context is created by calling L<AnyEvent::TLS> without any arguments.
2076
2077=cut
2078
2079our $TLS_CTX;
2080
2081sub TLS_CTX() {
2082 $TLS_CTX ||= do {
2083 require AnyEvent::TLS;
2084
2085 new AnyEvent::TLS
2086 }
2087}
2088
2089=back
2090
2091
2092=head1 NONFREQUENTLY ASKED QUESTIONS
2093
2094=over 4
2095
2096=item I C<undef> the AnyEvent::Handle reference inside my callback and
2097still get further invocations!
2098
2099That's because AnyEvent::Handle keeps a reference to itself when handling
2100read or write callbacks.
2101
2102It is only safe to "forget" the reference inside EOF or error callbacks,
2103from within all other callbacks, you need to explicitly call the C<<
2104->destroy >> method.
2105
2106=item I get different callback invocations in TLS mode/Why can't I pause
2107reading?
2108
2109Unlike, say, TCP, TLS connections do not consist of two independent
2110communication channels, one for each direction. Or put differently. The
2111read and write directions are not independent of each other: you cannot
2112write data unless you are also prepared to read, and vice versa.
2113
2114This can mean than, in TLS mode, you might get C<on_error> or C<on_eof>
2115callback invocations when you are not expecting any read data - the reason
2116is that AnyEvent::Handle always reads in TLS mode.
2117
2118During the connection, you have to make sure that you always have a
2119non-empty read-queue, or an C<on_read> watcher. At the end of the
2120connection (or when you no longer want to use it) you can call the
2121C<destroy> method.
2122
2123=item How do I read data until the other side closes the connection?
2124
2125If you just want to read your data into a perl scalar, the easiest way
2126to achieve this is by setting an C<on_read> callback that does nothing,
2127clearing the C<on_eof> callback and in the C<on_error> callback, the data
2128will be in C<$_[0]{rbuf}>:
2129
2130 $handle->on_read (sub { });
2131 $handle->on_eof (undef);
2132 $handle->on_error (sub {
2133 my $data = delete $_[0]{rbuf};
2134 });
2135
2136The reason to use C<on_error> is that TCP connections, due to latencies
2137and packets loss, might get closed quite violently with an error, when in
2138fact, all data has been received.
2139
2140It is usually better to use acknowledgements when transferring data,
2141to make sure the other side hasn't just died and you got the data
2142intact. This is also one reason why so many internet protocols have an
2143explicit QUIT command.
2144
2145=item I don't want to destroy the handle too early - how do I wait until
2146all data has been written?
2147
2148After writing your last bits of data, set the C<on_drain> callback
2149and destroy the handle in there - with the default setting of
2150C<low_water_mark> this will be called precisely when all data has been
2151written to the socket:
2152
2153 $handle->push_write (...);
2154 $handle->on_drain (sub {
2155 warn "all data submitted to the kernel\n";
2156 undef $handle;
2157 });
2158
2159If you just want to queue some data and then signal EOF to the other side,
2160consider using C<< ->push_shutdown >> instead.
2161
2162=item I want to contact a TLS/SSL server, I don't care about security.
2163
2164If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
2165simply connect to it and then create the AnyEvent::Handle with the C<tls>
2166parameter:
2167
2168 tcp_connect $host, $port, sub {
2169 my ($fh) = @_;
2170
2171 my $handle = new AnyEvent::Handle
2172 fh => $fh,
2173 tls => "connect",
2174 on_error => sub { ... };
2175
2176 $handle->push_write (...);
2177 };
2178
2179=item I want to contact a TLS/SSL server, I do care about security.
2180
2181Then you should additionally enable certificate verification, including
2182peername verification, if the protocol you use supports it (see
2183L<AnyEvent::TLS>, C<verify_peername>).
2184
2185E.g. for HTTPS:
2186
2187 tcp_connect $host, $port, sub {
2188 my ($fh) = @_;
2189
2190 my $handle = new AnyEvent::Handle
2191 fh => $fh,
2192 peername => $host,
2193 tls => "connect",
2194 tls_ctx => { verify => 1, verify_peername => "https" },
2195 ...
2196
2197Note that you must specify the hostname you connected to (or whatever
2198"peername" the protocol needs) as the C<peername> argument, otherwise no
2199peername verification will be done.
2200
2201The above will use the system-dependent default set of trusted CA
2202certificates. If you want to check against a specific CA, add the
2203C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
2204
2205 tls_ctx => {
2206 verify => 1,
2207 verify_peername => "https",
2208 ca_file => "my-ca-cert.pem",
2209 },
2210
2211=item I want to create a TLS/SSL server, how do I do that?
2212
2213Well, you first need to get a server certificate and key. You have
2214three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
2215self-signed certificate (cheap. check the search engine of your choice,
2216there are many tutorials on the net) or c) make your own CA (tinyca2 is a
2217nice program for that purpose).
2218
2219Then create a file with your private key (in PEM format, see
2220L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
2221file should then look like this:
2222
2223 -----BEGIN RSA PRIVATE KEY-----
2224 ...header data
2225 ... lots of base64'y-stuff
2226 -----END RSA PRIVATE KEY-----
2227
2228 -----BEGIN CERTIFICATE-----
2229 ... lots of base64'y-stuff
2230 -----END CERTIFICATE-----
2231
2232The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
2233specify this file as C<cert_file>:
2234
2235 tcp_server undef, $port, sub {
2236 my ($fh) = @_;
2237
2238 my $handle = new AnyEvent::Handle
2239 fh => $fh,
2240 tls => "accept",
2241 tls_ctx => { cert_file => "my-server-keycert.pem" },
2242 ...
2243
2244When you have intermediate CA certificates that your clients might not
2245know about, just append them to the C<cert_file>.
2246
2247=back
2248
2249
2250=head1 SUBCLASSING AnyEvent::Handle
2251
2252In many cases, you might want to subclass AnyEvent::Handle.
2253
2254To make this easier, a given version of AnyEvent::Handle uses these
2255conventions:
2256
2257=over 4
2258
2259=item * all constructor arguments become object members.
2260
2261At least initially, when you pass a C<tls>-argument to the constructor it
2262will end up in C<< $handle->{tls} >>. Those members might be changed or
2263mutated later on (for example C<tls> will hold the TLS connection object).
2264
2265=item * other object member names are prefixed with an C<_>.
2266
2267All object members not explicitly documented (internal use) are prefixed
2268with an underscore character, so the remaining non-C<_>-namespace is free
2269for use for subclasses.
2270
2271=item * all members not documented here and not prefixed with an underscore
2272are free to use in subclasses.
2273
2274Of course, new versions of AnyEvent::Handle may introduce more "public"
2275member variables, but thats just life, at least it is documented.
2276
2277=back
2278
578=head1 AUTHOR 2279=head1 AUTHOR
579 2280
580Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>. 2281Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
581 2282
582=cut 2283=cut

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines