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 1.20 by elmex, Sat May 24 08:16:50 2008 UTC vs.
Revision 1.204 by root, Mon Nov 15 03:29:17 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 streaming handles via AnyEvent
16
17This module is experimental.
18
19=cut
20
21our $VERSION = '0.04';
22 4
23=head1 SYNOPSIS 5=head1 SYNOPSIS
24 6
25 use AnyEvent; 7 use AnyEvent;
26 use AnyEvent::Handle; 8 use AnyEvent::Handle;
27 9
28 my $cv = AnyEvent->condvar; 10 my $cv = AnyEvent->condvar;
29 11
30 my $ae_fh = AnyEvent::Handle->new (fh => \*STDIN); 12 my $hdl; $hdl = new AnyEvent::Handle
31
32 #TODO
33
34 # or use the constructor to pass the callback:
35
36 my $ae_fh2 =
37 AnyEvent::Handle->new (
38 fh => \*STDIN, 13 fh => \*STDIN,
39 on_eof => sub { 14 on_error => sub {
40 $cv->broadcast; 15 my ($hdl, $fatal, $msg) = @_;
41 }, 16 warn "got error $msg\n";
42 #TODO 17 $hdl->destroy;
18 $cv->send;
43 ); 19 };
44 20
45 $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;
46 32
47=head1 DESCRIPTION 33=head1 DESCRIPTION
48 34
49This module is a helper module to make it easier to do event-based I/O on 35This is a helper module to make it easier to do event-based I/O on
50filehandles. For utility functions for doing non-blocking connects and accepts 36stream-based filehandles (sockets, pipes, and other stream things).
51on sockets see L<AnyEvent::Util>.
52 37
38The L<AnyEvent::Intro> tutorial contains some well-documented
39AnyEvent::Handle examples.
40
53In the following, when the documentation refers to of "bytes" then this 41In the following, where the documentation refers to "bytes", it means
54means characters. As sysread and syswrite are used for all I/O, their 42characters. As sysread and syswrite are used for all I/O, their
55treatment of characters applies to this module as well. 43treatment of characters applies to this module as well.
44
45At the very minimum, you should specify C<fh> or C<connect>, and the
46C<on_error> callback.
56 47
57All callbacks will be invoked with the handle object as their first 48All callbacks will be invoked with the handle object as their first
58argument. 49argument.
59 50
51=cut
52
53package AnyEvent::Handle;
54
55use Scalar::Util ();
56use List::Util ();
57use Carp ();
58use Errno qw(EAGAIN EINTR);
59
60use AnyEvent (); BEGIN { AnyEvent::common_sense }
61use AnyEvent::Util qw(WSAEWOULDBLOCK);
62
63our $VERSION = $AnyEvent::VERSION;
64
65sub _load_func($) {
66 my $func = $_[0];
67
68 unless (defined &$func) {
69 my $pkg = $func;
70 do {
71 $pkg =~ s/::[^:]+$//
72 or return;
73 eval "require $pkg";
74 } until defined &$func;
75 }
76
77 \&$func
78}
79
80sub MAX_READ_SIZE() { 131072 }
81
60=head1 METHODS 82=head1 METHODS
61 83
62=over 4 84=over 4
63 85
64=item B<new (%args)> 86=item $handle = B<new> AnyEvent::Handle fh => $filehandle, key => value...
65 87
66The constructor supports these arguments (all as key => value pairs). 88The constructor supports these arguments (all as C<< key => value >> pairs).
67 89
68=over 4 90=over 4
69 91
70=item fh => $filehandle [MANDATORY] 92=item fh => $filehandle [C<fh> or C<connect> MANDATORY]
71 93
72The filehandle this L<AnyEvent::Handle> object will operate on. 94The filehandle this L<AnyEvent::Handle> object will operate on.
73
74NOTE: The filehandle will be set to non-blocking (using 95NOTE: The filehandle will be set to non-blocking mode (using
75AnyEvent::Util::fh_nonblocking). 96C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
97that mode.
76 98
77=item on_eof => $cb->($self) 99=item connect => [$host, $service] [C<fh> or C<connect> MANDATORY]
78 100
79Set the callback to be called on EOF. 101Try to connect to the specified host and service (port), using
102C<AnyEvent::Socket::tcp_connect>. The C<$host> additionally becomes the
103default C<peername>.
80 104
81While not mandatory, it is highly recommended to set an eof callback, 105You have to specify either this parameter, or C<fh>, above.
82otherwise you might end up with a closed socket while you are still
83waiting for data.
84 106
107It is possible to push requests on the read and write queues, and modify
108properties of the stream, even while AnyEvent::Handle is connecting.
109
110When this parameter is specified, then the C<on_prepare>,
111C<on_connect_error> and C<on_connect> callbacks will be called under the
112appropriate circumstances:
113
114=over 4
115
85=item on_error => $cb->($self) 116=item on_prepare => $cb->($handle)
86 117
118This (rarely used) callback is called before a new connection is
119attempted, but after the file handle has been created. It could be used to
120prepare the file handle with parameters required for the actual connect
121(as opposed to settings that can be changed when the connection is already
122established).
123
124The return value of this callback should be the connect timeout value in
125seconds (or C<0>, or C<undef>, or the empty list, to indicate that the
126default timeout is to be used).
127
128=item on_connect => $cb->($handle, $host, $port, $retry->())
129
130This callback is called when a connection has been successfully established.
131
132The peer's numeric host and port (the socket peername) are passed as
133parameters, together with a retry callback.
134
135If, for some reason, the handle is not acceptable, calling C<$retry>
136will continue with the next connection target (in case of multi-homed
137hosts or SRV records there can be multiple connection endpoints). At the
138time it is called the read and write queues, eof status, tls status and
139similar properties of the handle will have been reset.
140
141In most cases, you should ignore the C<$retry> parameter.
142
143=item on_connect_error => $cb->($handle, $message)
144
145This callback is called when the connection could not be
146established. C<$!> will contain the relevant error code, and C<$message> a
147message describing it (usually the same as C<"$!">).
148
149If this callback isn't specified, then C<on_error> will be called with a
150fatal error instead.
151
152=back
153
154=item on_error => $cb->($handle, $fatal, $message)
155
87This is the fatal error callback, that is called when, well, a fatal error 156This is the error callback, which is called when, well, some error
88occurs, such as not being able to resolve the hostname, failure to connect 157occured, such as not being able to resolve the hostname, failure to
89or a read error. 158connect, or a read error.
90 159
91The object will not be in a usable state when this callback has been 160Some errors are fatal (which is indicated by C<$fatal> being true). On
92called. 161fatal errors the handle object will be destroyed (by a call to C<< ->
162destroy >>) after invoking the error callback (which means you are free to
163examine the handle object). Examples of fatal errors are an EOF condition
164with active (but unsatisfiable) read watchers (C<EPIPE>) or I/O errors. In
165cases where the other side can close the connection at will, it is
166often easiest to not report C<EPIPE> errors in this callback.
93 167
168AnyEvent::Handle tries to find an appropriate error code for you to check
169against, but in some cases (TLS errors), this does not work well. It is
170recommended to always output the C<$message> argument in human-readable
171error messages (it's usually the same as C<"$!">).
172
173Non-fatal errors can be retried by returning, but it is recommended
174to simply ignore this parameter and instead abondon the handle object
175when this callback is invoked. Examples of non-fatal errors are timeouts
176C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
177
94On callback entrance, the value of C<$!> contains the operating system 178On entry to the callback, the value of C<$!> contains the operating
95error (or C<ENOSPC> or C<EPIPE>). 179system error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
180C<EPROTO>).
96 181
97While not mandatory, it is I<highly> recommended to set this callback, as 182While not mandatory, it is I<highly> recommended to set this callback, as
98you will not be notified of errors otherwise. The default simply calls 183you will not be notified of errors otherwise. The default just calls
99die. 184C<croak>.
100 185
101=item on_read => $cb->($self) 186=item on_read => $cb->($handle)
102 187
103This sets the default read callback, which is called when data arrives 188This sets the default read callback, which is called when data arrives
104and no read request is in the queue. 189and no read request is in the queue (unlike read queue callbacks, this
190callback will only be called when at least one octet of data is in the
191read buffer).
105 192
106To access (and remove data from) the read buffer, use the C<< ->rbuf >> 193To access (and remove data from) the read buffer, use the C<< ->rbuf >>
107method or access the C<$self->{rbuf}> member directly. 194method or access the C<< $handle->{rbuf} >> member directly. Note that you
195must not enlarge or modify the read buffer, you can only remove data at
196the beginning from it.
108 197
198You can also call C<< ->push_read (...) >> or any other function that
199modifies the read queue. Or do both. Or ...
200
109When an EOF condition is detected then AnyEvent::Handle will first try to 201When an EOF condition is detected, AnyEvent::Handle will first try to
110feed all the remaining data to the queued callbacks and C<on_read> before 202feed all the remaining data to the queued callbacks and C<on_read> before
111calling the C<on_eof> callback. If no progress can be made, then a fatal 203calling the C<on_eof> callback. If no progress can be made, then a fatal
112error will be raised (with C<$!> set to C<EPIPE>). 204error will be raised (with C<$!> set to C<EPIPE>).
113 205
206Note that, unlike requests in the read queue, an C<on_read> callback
207doesn't mean you I<require> some data: if there is an EOF and there
208are outstanding read requests then an error will be flagged. With an
209C<on_read> callback, the C<on_eof> callback will be invoked.
210
211=item on_eof => $cb->($handle)
212
213Set the callback to be called when an end-of-file condition is detected,
214i.e. in the case of a socket, when the other side has closed the
215connection cleanly, and there are no outstanding read requests in the
216queue (if there are read requests, then an EOF counts as an unexpected
217connection close and will be flagged as an error).
218
219For sockets, this just means that the other side has stopped sending data,
220you can still try to write data, and, in fact, one can return from the EOF
221callback and continue writing data, as only the read part has been shut
222down.
223
224If an EOF condition has been detected but no C<on_eof> callback has been
225set, then a fatal error will be raised with C<$!> set to <0>.
226
114=item on_drain => $cb->() 227=item on_drain => $cb->($handle)
115 228
116This sets the callback that is called when the write buffer becomes empty 229This sets the callback that is called when the write buffer becomes empty
117(or when the callback is set and the buffer is empty already). 230(or immediately if the buffer is empty already).
118 231
119To append to the write buffer, use the C<< ->push_write >> method. 232To append to the write buffer, use the C<< ->push_write >> method.
233
234This callback is useful when you don't want to put all of your write data
235into the queue at once, for example, when you want to write the contents
236of some file to the socket you might not want to read the whole file into
237memory and push it into the queue, but instead only read more data from
238the file when the write queue becomes empty.
239
240=item timeout => $fractional_seconds
241
242=item rtimeout => $fractional_seconds
243
244=item wtimeout => $fractional_seconds
245
246If non-zero, then these enables an "inactivity" timeout: whenever this
247many seconds pass without a successful read or write on the underlying
248file handle (or a call to C<timeout_reset>), the C<on_timeout> callback
249will be invoked (and if that one is missing, a non-fatal C<ETIMEDOUT>
250error will be raised).
251
252There are three variants of the timeouts that work independently
253of each other, for both read and write, just read, and just write:
254C<timeout>, C<rtimeout> and C<wtimeout>, with corresponding callbacks
255C<on_timeout>, C<on_rtimeout> and C<on_wtimeout>, and reset functions
256C<timeout_reset>, C<rtimeout_reset>, and C<wtimeout_reset>.
257
258Note that timeout processing is active even when you do not have
259any outstanding read or write requests: If you plan to keep the connection
260idle then you should disable the timeout temporarily or ignore the timeout
261in the C<on_timeout> callback, in which case AnyEvent::Handle will simply
262restart the timeout.
263
264Zero (the default) disables this timeout.
265
266=item on_timeout => $cb->($handle)
267
268Called whenever the inactivity timeout passes. If you return from this
269callback, then the timeout will be reset as if some activity had happened,
270so this condition is not fatal in any way.
120 271
121=item rbuf_max => <bytes> 272=item rbuf_max => <bytes>
122 273
123If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>) 274If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
124when the read buffer ever (strictly) exceeds this size. This is useful to 275when the read buffer ever (strictly) exceeds this size. This is useful to
125avoid denial-of-service attacks. 276avoid some forms of denial-of-service attacks.
126 277
127For example, a server accepting connections from untrusted sources should 278For example, a server accepting connections from untrusted sources should
128be configured to accept only so-and-so much data that it cannot act on 279be configured to accept only so-and-so much data that it cannot act on
129(for example, when expecting a line, an attacker could send an unlimited 280(for example, when expecting a line, an attacker could send an unlimited
130amount of data without a callback ever being called as long as the line 281amount of data without a callback ever being called as long as the line
131isn't finished). 282isn't finished).
132 283
284=item autocork => <boolean>
285
286When disabled (the default), C<push_write> will try to immediately
287write the data to the handle if possible. This avoids having to register
288a write watcher and wait for the next event loop iteration, but can
289be inefficient if you write multiple small chunks (on the wire, this
290disadvantage is usually avoided by your kernel's nagle algorithm, see
291C<no_delay>, but this option can save costly syscalls).
292
293When enabled, writes will always be queued till the next event loop
294iteration. This is efficient when you do many small writes per iteration,
295but less efficient when you do a single write only per iteration (or when
296the write buffer often is full). It also increases write latency.
297
298=item no_delay => <boolean>
299
300When doing small writes on sockets, your operating system kernel might
301wait a bit for more data before actually sending it out. This is called
302the Nagle algorithm, and usually it is beneficial.
303
304In some situations you want as low a delay as possible, which can be
305accomplishd by setting this option to a true value.
306
307The default is your operating system's default behaviour (most likely
308enabled). This option explicitly enables or disables it, if possible.
309
310=item keepalive => <boolean>
311
312Enables (default disable) the SO_KEEPALIVE option on the stream socket:
313normally, TCP connections have no time-out once established, so TCP
314connections, once established, can stay alive forever even when the other
315side has long gone. TCP keepalives are a cheap way to take down long-lived
316TCP connections when the other side becomes unreachable. While the default
317is OS-dependent, TCP keepalives usually kick in after around two hours,
318and, if the other side doesn't reply, take down the TCP connection some 10
319to 15 minutes later.
320
321It is harmless to specify this option for file handles that do not support
322keepalives, and enabling it on connections that are potentially long-lived
323is usually a good idea.
324
325=item oobinline => <boolean>
326
327BSD majorly fucked up the implementation of TCP urgent data. The result
328is that almost no OS implements TCP according to the specs, and every OS
329implements it slightly differently.
330
331If you want to handle TCP urgent data, then setting this flag (the default
332is enabled) gives you the most portable way of getting urgent data, by
333putting it into the stream.
334
335Since BSD emulation of OOB data on top of TCP's urgent data can have
336security implications, AnyEvent::Handle sets this flag automatically
337unless explicitly specified. Note that setting this flag after
338establishing a connection I<may> be a bit too late (data loss could
339already have occured on BSD systems), but at least it will protect you
340from most attacks.
341
133=item read_size => <bytes> 342=item read_size => <bytes>
134 343
135The default read block size (the amount of bytes this module will try to read 344The initial read block size, the number of bytes this module will try to
136on each [loop iteration). Default: C<4096>. 345read during each loop iteration. Each handle object will consume at least
346this amount of memory for the read buffer as well, so when handling many
347connections requirements). See also C<max_read_size>. Default: C<2048>.
348
349=item max_read_size => <bytes>
350
351The maximum read buffer size used by the dynamic adjustment
352algorithm: Each time AnyEvent::Handle can read C<read_size> bytes in
353one go it will double C<read_size> up to the maximum given by this
354option. Default: C<131072> or C<read_size>, whichever is higher.
137 355
138=item low_water_mark => <bytes> 356=item low_water_mark => <bytes>
139 357
140Sets the amount of bytes (default: C<0>) that make up an "empty" write 358Sets the number of bytes (default: C<0>) that make up an "empty" write
141buffer: If the write reaches this size or gets even samller it is 359buffer: If the buffer reaches this size or gets even samller it is
142considered empty. 360considered empty.
143 361
362Sometimes it can be beneficial (for performance reasons) to add data to
363the write buffer before it is fully drained, but this is a rare case, as
364the operating system kernel usually buffers data as well, so the default
365is good in almost all cases.
366
367=item linger => <seconds>
368
369If this is non-zero (default: C<3600>), the destructor of the
370AnyEvent::Handle object will check whether there is still outstanding
371write data and will install a watcher that will write this data to the
372socket. No errors will be reported (this mostly matches how the operating
373system treats outstanding data at socket close time).
374
375This will not work for partial TLS data that could not be encoded
376yet. This data will be lost. Calling the C<stoptls> method in time might
377help.
378
379=item peername => $string
380
381A string used to identify the remote site - usually the DNS hostname
382(I<not> IDN!) used to create the connection, rarely the IP address.
383
384Apart from being useful in error messages, this string is also used in TLS
385peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
386verification will be skipped when C<peername> is not specified or is
387C<undef>.
388
144=item tls => "accept" | "connect" | Net::SSLeay::SSL object 389=item tls => "accept" | "connect" | Net::SSLeay::SSL object
145 390
146When this parameter is given, it enables TLS (SSL) mode, that means it 391When this parameter is given, it enables TLS (SSL) mode, that means
147will start making tls handshake and will transparently encrypt/decrypt 392AnyEvent will start a TLS handshake as soon as the connection has been
148data. 393established and will transparently encrypt/decrypt data afterwards.
149 394
150For the TLS server side, use C<accept>, and for the TLS client side of a 395All TLS protocol errors will be signalled as C<EPROTO>, with an
151connection, use C<connect> mode. 396appropriate error message.
397
398TLS mode requires Net::SSLeay to be installed (it will be loaded
399automatically when you try to create a TLS handle): this module doesn't
400have a dependency on that module, so if your module requires it, you have
401to add the dependency yourself.
402
403Unlike TCP, TLS has a server and client side: for the TLS server side, use
404C<accept>, and for the TLS client side of a connection, use C<connect>
405mode.
152 406
153You can also provide your own TLS connection object, but you have 407You can also provide your own TLS connection object, but you have
154to make sure that you call either C<Net::SSLeay::set_connect_state> 408to make sure that you call either C<Net::SSLeay::set_connect_state>
155or C<Net::SSLeay::set_accept_state> on it before you pass it to 409or C<Net::SSLeay::set_accept_state> on it before you pass it to
156AnyEvent::Handle. 410AnyEvent::Handle. Also, this module will take ownership of this connection
411object.
157 412
158=item tls_ctx => $ssl_ctx 413At some future point, AnyEvent::Handle might switch to another TLS
414implementation, then the option to use your own session object will go
415away.
159 416
417B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
418passing in the wrong integer will lead to certain crash. This most often
419happens when one uses a stylish C<< tls => 1 >> and is surprised about the
420segmentation fault.
421
422Use the C<< ->starttls >> method if you need to start TLS negotiation later.
423
424=item tls_ctx => $anyevent_tls
425
160Use the given Net::SSLeay::CTX object to create the new TLS connection 426Use the given C<AnyEvent::TLS> object to create the new TLS connection
161(unless a connection object was specified directly). If this parameter is 427(unless a connection object was specified directly). If this parameter is
162missing, then AnyEvent::Handle will use C<AnyEvent::Handle::TLS_CTX>. 428missing, then AnyEvent::Handle will use C<AnyEvent::Handle::TLS_CTX>.
163 429
430Instead of an object, you can also specify a hash reference with C<< key
431=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
432new TLS context object.
433
434=item on_starttls => $cb->($handle, $success[, $error_message])
435
436This callback will be invoked when the TLS/SSL handshake has finished. If
437C<$success> is true, then the TLS handshake succeeded, otherwise it failed
438(C<on_stoptls> will not be called in this case).
439
440The session in C<< $handle->{tls} >> can still be examined in this
441callback, even when the handshake was not successful.
442
443TLS handshake failures will not cause C<on_error> to be invoked when this
444callback is in effect, instead, the error message will be passed to C<on_starttls>.
445
446Without this callback, handshake failures lead to C<on_error> being
447called as usual.
448
449Note that you cannot just call C<starttls> again in this callback. If you
450need to do that, start an zero-second timer instead whose callback can
451then call C<< ->starttls >> again.
452
453=item on_stoptls => $cb->($handle)
454
455When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
456set, then it will be invoked after freeing the TLS session. If it is not,
457then a TLS shutdown condition will be treated like a normal EOF condition
458on the handle.
459
460The session in C<< $handle->{tls} >> can still be examined in this
461callback.
462
463This callback will only be called on TLS shutdowns, not when the
464underlying handle signals EOF.
465
466=item json => JSON or JSON::XS object
467
468This is the json coder object used by the C<json> read and write types.
469
470If you don't supply it, then AnyEvent::Handle will create and use a
471suitable one (on demand), which will write and expect UTF-8 encoded JSON
472texts.
473
474Note that you are responsible to depend on the JSON module if you want to
475use this functionality, as AnyEvent does not have a dependency itself.
476
164=back 477=back
165 478
166=cut 479=cut
167 480
168sub new { 481sub new {
169 my $class = shift; 482 my $class = shift;
170
171 my $self = bless { @_ }, $class; 483 my $self = bless { @_ }, $class;
172 484
173 $self->{fh} or Carp::croak "mandatory argument fh is missing"; 485 if ($self->{fh}) {
486 $self->_start;
487 return unless $self->{fh}; # could be gone by now
488
489 } elsif ($self->{connect}) {
490 require AnyEvent::Socket;
491
492 $self->{peername} = $self->{connect}[0]
493 unless exists $self->{peername};
494
495 $self->{_skip_drain_rbuf} = 1;
496
497 {
498 Scalar::Util::weaken (my $self = $self);
499
500 $self->{_connect} =
501 AnyEvent::Socket::tcp_connect (
502 $self->{connect}[0],
503 $self->{connect}[1],
504 sub {
505 my ($fh, $host, $port, $retry) = @_;
506
507 if ($fh) {
508 $self->{fh} = $fh;
509
510 delete $self->{_skip_drain_rbuf};
511 $self->_start;
512
513 $self->{on_connect}
514 and $self->{on_connect}($self, $host, $port, sub {
515 delete @$self{qw(fh _tw _rtw _wtw _ww _rw _eof _queue rbuf _wbuf tls _tls_rbuf _tls_wbuf)};
516 $self->{_skip_drain_rbuf} = 1;
517 &$retry;
518 });
519
520 } else {
521 if ($self->{on_connect_error}) {
522 $self->{on_connect_error}($self, "$!");
523 $self->destroy;
524 } else {
525 $self->_error ($!, 1);
526 }
527 }
528 },
529 sub {
530 local $self->{fh} = $_[0];
531
532 $self->{on_prepare}
533 ? $self->{on_prepare}->($self)
534 : ()
535 }
536 );
537 }
538
539 } else {
540 Carp::croak "AnyEvent::Handle: either an existing fh or the connect parameter must be specified";
541 }
542
543 $self
544}
545
546sub _start {
547 my ($self) = @_;
548
549 # too many clueless people try to use udp and similar sockets
550 # with AnyEvent::Handle, do them a favour.
551 my $type = getsockopt $self->{fh}, Socket::SOL_SOCKET (), Socket::SO_TYPE ();
552 Carp::croak "AnyEvent::Handle: only stream sockets supported, anything else will NOT work!"
553 if Socket::SOCK_STREAM () != (unpack "I", $type) && defined $type;
174 554
175 AnyEvent::Util::fh_nonblocking $self->{fh}, 1; 555 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
176 556
177 if ($self->{tls}) { 557 $self->{_activity} =
178 require Net::SSLeay; 558 $self->{_ractivity} =
559 $self->{_wactivity} = AE::now;
560
561 $self->{read_size} ||= 2048;
562 $self->{max_read_size} = $self->{read_size}
563 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
564
565 $self->timeout (delete $self->{timeout} ) if $self->{timeout};
566 $self->rtimeout (delete $self->{rtimeout} ) if $self->{rtimeout};
567 $self->wtimeout (delete $self->{wtimeout} ) if $self->{wtimeout};
568
569 $self->no_delay (delete $self->{no_delay} ) if exists $self->{no_delay} && $self->{no_delay};
570 $self->keepalive (delete $self->{keepalive}) if exists $self->{keepalive} && $self->{keepalive};
571
572 $self->oobinline (exists $self->{oobinline} ? delete $self->{oobinline} : 1);
573
179 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx}); 574 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
180 } 575 if $self->{tls};
181 576
182 $self->on_eof (delete $self->{on_eof} ) if $self->{on_eof};
183 $self->on_error (delete $self->{on_error}) if $self->{on_error};
184 $self->on_drain (delete $self->{on_drain}) if $self->{on_drain}; 577 $self->on_drain (delete $self->{on_drain} ) if $self->{on_drain};
185 $self->on_read (delete $self->{on_read} ) if $self->{on_read};
186 578
187 $self->start_read; 579 $self->start_read
580 if $self->{on_read} || @{ $self->{_queue} };
188 581
189 $self 582 $self->_drain_wbuf;
190} 583}
191 584
192sub _shutdown {
193 my ($self) = @_;
194
195 delete $self->{rw};
196 delete $self->{ww};
197 delete $self->{fh};
198}
199
200sub error { 585sub _error {
201 my ($self) = @_; 586 my ($self, $errno, $fatal, $message) = @_;
202 587
203 { 588 $! = $errno;
204 local $!; 589 $message ||= "$!";
205 $self->_shutdown;
206 }
207 590
208 if ($self->{on_error}) { 591 if ($self->{on_error}) {
209 $self->{on_error}($self); 592 $self->{on_error}($self, $fatal, $message);
210 } else { 593 $self->destroy if $fatal;
594 } elsif ($self->{fh} || $self->{connect}) {
595 $self->destroy;
211 die "AnyEvent::Handle uncaught fatal error: $!"; 596 Carp::croak "AnyEvent::Handle uncaught error: $message";
212 } 597 }
213} 598}
214 599
215=item $fh = $handle->fh 600=item $fh = $handle->fh
216 601
217This method returns the filehandle of the L<AnyEvent::Handle> object. 602This method returns the file handle used to create the L<AnyEvent::Handle> object.
218 603
219=cut 604=cut
220 605
221sub fh { $_[0]->{fh} } 606sub fh { $_[0]{fh} }
222 607
223=item $handle->on_error ($cb) 608=item $handle->on_error ($cb)
224 609
225Replace the current C<on_error> callback (see the C<on_error> constructor argument). 610Replace the current C<on_error> callback (see the C<on_error> constructor argument).
226 611
238 623
239sub on_eof { 624sub on_eof {
240 $_[0]{on_eof} = $_[1]; 625 $_[0]{on_eof} = $_[1];
241} 626}
242 627
628=item $handle->on_timeout ($cb)
629
630=item $handle->on_rtimeout ($cb)
631
632=item $handle->on_wtimeout ($cb)
633
634Replace the current C<on_timeout>, C<on_rtimeout> or C<on_wtimeout>
635callback, or disables the callback (but not the timeout) if C<$cb> =
636C<undef>. See the C<timeout> constructor argument and method.
637
638=cut
639
640# see below
641
642=item $handle->autocork ($boolean)
643
644Enables or disables the current autocork behaviour (see C<autocork>
645constructor argument). Changes will only take effect on the next write.
646
647=cut
648
649sub autocork {
650 $_[0]{autocork} = $_[1];
651}
652
653=item $handle->no_delay ($boolean)
654
655Enables or disables the C<no_delay> setting (see constructor argument of
656the same name for details).
657
658=cut
659
660sub no_delay {
661 $_[0]{no_delay} = $_[1];
662
663 setsockopt $_[0]{fh}, Socket::IPPROTO_TCP (), Socket::TCP_NODELAY (), int $_[1]
664 if $_[0]{fh};
665}
666
667=item $handle->keepalive ($boolean)
668
669Enables or disables the C<keepalive> setting (see constructor argument of
670the same name for details).
671
672=cut
673
674sub keepalive {
675 $_[0]{keepalive} = $_[1];
676
677 eval {
678 local $SIG{__DIE__};
679 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
680 if $_[0]{fh};
681 };
682}
683
684=item $handle->oobinline ($boolean)
685
686Enables or disables the C<oobinline> setting (see constructor argument of
687the same name for details).
688
689=cut
690
691sub oobinline {
692 $_[0]{oobinline} = $_[1];
693
694 eval {
695 local $SIG{__DIE__};
696 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_OOBINLINE (), int $_[1]
697 if $_[0]{fh};
698 };
699}
700
701=item $handle->keepalive ($boolean)
702
703Enables or disables the C<keepalive> setting (see constructor argument of
704the same name for details).
705
706=cut
707
708sub keepalive {
709 $_[0]{keepalive} = $_[1];
710
711 eval {
712 local $SIG{__DIE__};
713 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
714 if $_[0]{fh};
715 };
716}
717
718=item $handle->on_starttls ($cb)
719
720Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
721
722=cut
723
724sub on_starttls {
725 $_[0]{on_starttls} = $_[1];
726}
727
728=item $handle->on_stoptls ($cb)
729
730Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
731
732=cut
733
734sub on_stoptls {
735 $_[0]{on_stoptls} = $_[1];
736}
737
738=item $handle->rbuf_max ($max_octets)
739
740Configures the C<rbuf_max> setting (C<undef> disables it).
741
742=cut
743
744sub rbuf_max {
745 $_[0]{rbuf_max} = $_[1];
746}
747
748#############################################################################
749
750=item $handle->timeout ($seconds)
751
752=item $handle->rtimeout ($seconds)
753
754=item $handle->wtimeout ($seconds)
755
756Configures (or disables) the inactivity timeout.
757
758=item $handle->timeout_reset
759
760=item $handle->rtimeout_reset
761
762=item $handle->wtimeout_reset
763
764Reset the activity timeout, as if data was received or sent.
765
766These methods are cheap to call.
767
768=cut
769
770for my $dir ("", "r", "w") {
771 my $timeout = "${dir}timeout";
772 my $tw = "_${dir}tw";
773 my $on_timeout = "on_${dir}timeout";
774 my $activity = "_${dir}activity";
775 my $cb;
776
777 *$on_timeout = sub {
778 $_[0]{$on_timeout} = $_[1];
779 };
780
781 *$timeout = sub {
782 my ($self, $new_value) = @_;
783
784 $new_value >= 0
785 or Carp::croak "AnyEvent::Handle->$timeout called with negative timeout ($new_value), caught";
786
787 $self->{$timeout} = $new_value;
788 delete $self->{$tw}; &$cb;
789 };
790
791 *{"${dir}timeout_reset"} = sub {
792 $_[0]{$activity} = AE::now;
793 };
794
795 # main workhorse:
796 # reset the timeout watcher, as neccessary
797 # also check for time-outs
798 $cb = sub {
799 my ($self) = @_;
800
801 if ($self->{$timeout} && $self->{fh}) {
802 my $NOW = AE::now;
803
804 # when would the timeout trigger?
805 my $after = $self->{$activity} + $self->{$timeout} - $NOW;
806
807 # now or in the past already?
808 if ($after <= 0) {
809 $self->{$activity} = $NOW;
810
811 if ($self->{$on_timeout}) {
812 $self->{$on_timeout}($self);
813 } else {
814 $self->_error (Errno::ETIMEDOUT);
815 }
816
817 # callback could have changed timeout value, optimise
818 return unless $self->{$timeout};
819
820 # calculate new after
821 $after = $self->{$timeout};
822 }
823
824 Scalar::Util::weaken $self;
825 return unless $self; # ->error could have destroyed $self
826
827 $self->{$tw} ||= AE::timer $after, 0, sub {
828 delete $self->{$tw};
829 $cb->($self);
830 };
831 } else {
832 delete $self->{$tw};
833 }
834 }
835}
836
243############################################################################# 837#############################################################################
244 838
245=back 839=back
246 840
247=head2 WRITE QUEUE 841=head2 WRITE QUEUE
260=item $handle->on_drain ($cb) 854=item $handle->on_drain ($cb)
261 855
262Sets the C<on_drain> callback or clears it (see the description of 856Sets the C<on_drain> callback or clears it (see the description of
263C<on_drain> in the constructor). 857C<on_drain> in the constructor).
264 858
859This method may invoke callbacks (and therefore the handle might be
860destroyed after it returns).
861
265=cut 862=cut
266 863
267sub on_drain { 864sub on_drain {
268 my ($self, $cb) = @_; 865 my ($self, $cb) = @_;
269 866
270 $self->{on_drain} = $cb; 867 $self->{on_drain} = $cb;
271 868
272 $cb->($self) 869 $cb->($self)
273 if $cb && $self->{low_water_mark} >= length $self->{wbuf}; 870 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
274} 871}
275 872
276=item $handle->push_write ($data) 873=item $handle->push_write ($data)
277 874
278Queues the given scalar to be written. You can push as much data as you 875Queues the given scalar to be written. You can push as much data as you
279want (only limited by the available memory), as C<AnyEvent::Handle> 876want (only limited by the available memory), as C<AnyEvent::Handle>
280buffers it independently of the kernel. 877buffers it independently of the kernel.
281 878
879This method may invoke callbacks (and therefore the handle might be
880destroyed after it returns).
881
282=cut 882=cut
283 883
284sub _drain_wbuf { 884sub _drain_wbuf {
285 my ($self) = @_; 885 my ($self) = @_;
286 886
287 unless ($self->{ww}) { 887 if (!$self->{_ww} && length $self->{wbuf}) {
888
288 Scalar::Util::weaken $self; 889 Scalar::Util::weaken $self;
890
289 my $cb = sub { 891 my $cb = sub {
290 my $len = syswrite $self->{fh}, $self->{wbuf}; 892 my $len = syswrite $self->{fh}, $self->{wbuf};
291 893
292 if ($len > 0) { 894 if (defined $len) {
293 substr $self->{wbuf}, 0, $len, ""; 895 substr $self->{wbuf}, 0, $len, "";
294 896
897 $self->{_activity} = $self->{_wactivity} = AE::now;
898
295 $self->{on_drain}($self) 899 $self->{on_drain}($self)
296 if $self->{low_water_mark} >= length $self->{wbuf} 900 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
297 && $self->{on_drain}; 901 && $self->{on_drain};
298 902
299 delete $self->{ww} unless length $self->{wbuf}; 903 delete $self->{_ww} unless length $self->{wbuf};
300 } elsif ($! != EAGAIN && $! != EINTR) { 904 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
301 $self->error; 905 $self->_error ($!, 1);
302 } 906 }
303 }; 907 };
304 908
305 $self->{ww} = AnyEvent->io (fh => $self->{fh}, poll => "w", cb => $cb); 909 # try to write data immediately
910 $cb->() unless $self->{autocork};
306 911
307 $cb->($self); 912 # if still data left in wbuf, we need to poll
913 $self->{_ww} = AE::io $self->{fh}, 1, $cb
914 if length $self->{wbuf};
308 }; 915 };
916}
917
918our %WH;
919
920# deprecated
921sub register_write_type($$) {
922 $WH{$_[0]} = $_[1];
309} 923}
310 924
311sub push_write { 925sub push_write {
312 my $self = shift; 926 my $self = shift;
313 927
928 if (@_ > 1) {
929 my $type = shift;
930
931 @_ = ($WH{$type} ||= _load_func "$type\::anyevent_write_type"
932 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_write")
933 ->($self, @_);
934 }
935
936 # we downgrade here to avoid hard-to-track-down bugs,
937 # and diagnose the problem earlier and better.
938
314 if ($self->{filter_w}) { 939 if ($self->{tls}) {
315 $self->{filter_w}->($self, \$_[0]); 940 utf8::downgrade $self->{_tls_wbuf} .= $_[0];
941 &_dotls ($self) if $self->{fh};
316 } else { 942 } else {
317 $self->{wbuf} .= $_[0]; 943 utf8::downgrade $self->{wbuf} .= $_[0];
318 $self->_drain_wbuf; 944 $self->_drain_wbuf if $self->{fh};
319 } 945 }
320} 946}
947
948=item $handle->push_write (type => @args)
949
950Instead of formatting your data yourself, you can also let this module
951do the job by specifying a type and type-specific arguments. You
952can also specify the (fully qualified) name of a package, in which
953case AnyEvent tries to load the package and then expects to find the
954C<anyevent_write_type> function inside (see "custom write types", below).
955
956Predefined types are (if you have ideas for additional types, feel free to
957drop by and tell us):
958
959=over 4
960
961=item netstring => $string
962
963Formats the given value as netstring
964(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
965
966=cut
967
968register_write_type netstring => sub {
969 my ($self, $string) = @_;
970
971 (length $string) . ":$string,"
972};
973
974=item packstring => $format, $data
975
976An octet string prefixed with an encoded length. The encoding C<$format>
977uses the same format as a Perl C<pack> format, but must specify a single
978integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
979optional C<!>, C<< < >> or C<< > >> modifier).
980
981=cut
982
983register_write_type packstring => sub {
984 my ($self, $format, $string) = @_;
985
986 pack "$format/a*", $string
987};
988
989=item json => $array_or_hashref
990
991Encodes the given hash or array reference into a JSON object. Unless you
992provide your own JSON object, this means it will be encoded to JSON text
993in UTF-8.
994
995JSON objects (and arrays) are self-delimiting, so you can write JSON at
996one end of a handle and read them at the other end without using any
997additional framing.
998
999The generated JSON text is guaranteed not to contain any newlines: While
1000this module doesn't need delimiters after or between JSON texts to be
1001able to read them, many other languages depend on that.
1002
1003A simple RPC protocol that interoperates easily with others is to send
1004JSON arrays (or objects, although arrays are usually the better choice as
1005they mimic how function argument passing works) and a newline after each
1006JSON text:
1007
1008 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
1009 $handle->push_write ("\012");
1010
1011An AnyEvent::Handle receiver would simply use the C<json> read type and
1012rely on the fact that the newline will be skipped as leading whitespace:
1013
1014 $handle->push_read (json => sub { my $array = $_[1]; ... });
1015
1016Other languages could read single lines terminated by a newline and pass
1017this line into their JSON decoder of choice.
1018
1019=cut
1020
1021sub json_coder() {
1022 eval { require JSON::XS; JSON::XS->new->utf8 }
1023 || do { require JSON; JSON->new->utf8 }
1024}
1025
1026register_write_type json => sub {
1027 my ($self, $ref) = @_;
1028
1029 my $json = $self->{json} ||= json_coder;
1030
1031 $json->encode ($ref)
1032};
1033
1034=item storable => $reference
1035
1036Freezes the given reference using L<Storable> and writes it to the
1037handle. Uses the C<nfreeze> format.
1038
1039=cut
1040
1041register_write_type storable => sub {
1042 my ($self, $ref) = @_;
1043
1044 require Storable;
1045
1046 pack "w/a*", Storable::nfreeze ($ref)
1047};
1048
1049=back
1050
1051=item $handle->push_shutdown
1052
1053Sometimes you know you want to close the socket after writing your data
1054before it was actually written. One way to do that is to replace your
1055C<on_drain> handler by a callback that shuts down the socket (and set
1056C<low_water_mark> to C<0>). This method is a shorthand for just that, and
1057replaces the C<on_drain> callback with:
1058
1059 sub { shutdown $_[0]{fh}, 1 } # for push_shutdown
1060
1061This simply shuts down the write side and signals an EOF condition to the
1062the peer.
1063
1064You can rely on the normal read queue and C<on_eof> handling
1065afterwards. This is the cleanest way to close a connection.
1066
1067This method may invoke callbacks (and therefore the handle might be
1068destroyed after it returns).
1069
1070=cut
1071
1072sub push_shutdown {
1073 my ($self) = @_;
1074
1075 delete $self->{low_water_mark};
1076 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
1077}
1078
1079=item custom write types - Package::anyevent_write_type $handle, @args
1080
1081Instead of one of the predefined types, you can also specify the name of
1082a package. AnyEvent will try to load the package and then expects to find
1083a function named C<anyevent_write_type> inside. If it isn't found, it
1084progressively tries to load the parent package until it either finds the
1085function (good) or runs out of packages (bad).
1086
1087Whenever the given C<type> is used, C<push_write> will the function with
1088the handle object and the remaining arguments.
1089
1090The function is supposed to return a single octet string that will be
1091appended to the write buffer, so you cna mentally treat this function as a
1092"arguments to on-the-wire-format" converter.
1093
1094Example: implement a custom write type C<join> that joins the remaining
1095arguments using the first one.
1096
1097 $handle->push_write (My::Type => " ", 1,2,3);
1098
1099 # uses the following package, which can be defined in the "My::Type" or in
1100 # the "My" modules to be auto-loaded, or just about anywhere when the
1101 # My::Type::anyevent_write_type is defined before invoking it.
1102
1103 package My::Type;
1104
1105 sub anyevent_write_type {
1106 my ($handle, $delim, @args) = @_;
1107
1108 join $delim, @args
1109 }
1110
1111=cut
321 1112
322############################################################################# 1113#############################################################################
323 1114
324=back 1115=back
325 1116
332ways, the "simple" way, using only C<on_read> and the "complex" way, using 1123ways, the "simple" way, using only C<on_read> and the "complex" way, using
333a queue. 1124a queue.
334 1125
335In the simple case, you just install an C<on_read> callback and whenever 1126In the simple case, you just install an C<on_read> callback and whenever
336new data arrives, it will be called. You can then remove some data (if 1127new data arrives, it will be called. You can then remove some data (if
337enough is there) from the read buffer (C<< $handle->rbuf >>) if you want 1128enough is there) from the read buffer (C<< $handle->rbuf >>). Or you can
338or not. 1129leave the data there if you want to accumulate more (e.g. when only a
1130partial message has been received so far), or change the read queue with
1131e.g. C<push_read>.
339 1132
340In the more complex case, you want to queue multiple callbacks. In this 1133In the more complex case, you want to queue multiple callbacks. In this
341case, AnyEvent::Handle will call the first queued callback each time new 1134case, AnyEvent::Handle will call the first queued callback each time new
342data arrives and removes it when it has done its job (see C<push_read>, 1135data arrives (also the first time it is queued) and remove it when it has
343below). 1136done its job (see C<push_read>, below).
344 1137
345This way you can, for example, push three line-reads, followed by reading 1138This way you can, for example, push three line-reads, followed by reading
346a chunk of data, and AnyEvent::Handle will execute them in order. 1139a chunk of data, and AnyEvent::Handle will execute them in order.
347 1140
348Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by 1141Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
349the specified number of bytes which give an XML datagram. 1142the specified number of bytes which give an XML datagram.
350 1143
351 # in the default state, expect some header bytes 1144 # in the default state, expect some header bytes
352 $handle->on_read (sub { 1145 $handle->on_read (sub {
353 # some data is here, now queue the length-header-read (4 octets) 1146 # some data is here, now queue the length-header-read (4 octets)
354 shift->unshift_read_chunk (4, sub { 1147 shift->unshift_read (chunk => 4, sub {
355 # header arrived, decode 1148 # header arrived, decode
356 my $len = unpack "N", $_[1]; 1149 my $len = unpack "N", $_[1];
357 1150
358 # now read the payload 1151 # now read the payload
359 shift->unshift_read_chunk ($len, sub { 1152 shift->unshift_read (chunk => $len, sub {
360 my $xml = $_[1]; 1153 my $xml = $_[1];
361 # handle xml 1154 # handle xml
362 }); 1155 });
363 }); 1156 });
364 }); 1157 });
365 1158
366Example 2: Implement a client for a protocol that replies either with 1159Example 2: Implement a client for a protocol that replies either with "OK"
367"OK" and another line or "ERROR" for one request, and 64 bytes for the 1160and another line or "ERROR" for the first request that is sent, and 64
368second request. Due tot he availability of a full queue, we can just 1161bytes for the second request. Due to the availability of a queue, we can
369pipeline sending both requests and manipulate the queue as necessary in 1162just pipeline sending both requests and manipulate the queue as necessary
370the callbacks: 1163in the callbacks.
371 1164
372 # request one 1165When the first callback is called and sees an "OK" response, it will
1166C<unshift> another line-read. This line-read will be queued I<before> the
116764-byte chunk callback.
1168
1169 # request one, returns either "OK + extra line" or "ERROR"
373 $handle->push_write ("request 1\015\012"); 1170 $handle->push_write ("request 1\015\012");
374 1171
375 # we expect "ERROR" or "OK" as response, so push a line read 1172 # we expect "ERROR" or "OK" as response, so push a line read
376 $handle->push_read_line (sub { 1173 $handle->push_read (line => sub {
377 # if we got an "OK", we have to _prepend_ another line, 1174 # if we got an "OK", we have to _prepend_ another line,
378 # so it will be read before the second request reads its 64 bytes 1175 # so it will be read before the second request reads its 64 bytes
379 # which are already in the queue when this callback is called 1176 # which are already in the queue when this callback is called
380 # we don't do this in case we got an error 1177 # we don't do this in case we got an error
381 if ($_[1] eq "OK") { 1178 if ($_[1] eq "OK") {
382 $_[0]->unshift_read_line (sub { 1179 $_[0]->unshift_read (line => sub {
383 my $response = $_[1]; 1180 my $response = $_[1];
384 ... 1181 ...
385 }); 1182 });
386 } 1183 }
387 }); 1184 });
388 1185
389 # request two 1186 # request two, simply returns 64 octets
390 $handle->push_write ("request 2\015\012"); 1187 $handle->push_write ("request 2\015\012");
391 1188
392 # simply read 64 bytes, always 1189 # simply read 64 bytes, always
393 $handle->push_read_chunk (64, sub { 1190 $handle->push_read (chunk => 64, sub {
394 my $response = $_[1]; 1191 my $response = $_[1];
395 ... 1192 ...
396 }); 1193 });
397 1194
398=over 4 1195=over 4
399 1196
400=cut 1197=cut
401 1198
402sub _drain_rbuf { 1199sub _drain_rbuf {
403 my ($self) = @_; 1200 my ($self) = @_;
1201
1202 # avoid recursion
1203 return if $self->{_skip_drain_rbuf};
1204 local $self->{_skip_drain_rbuf} = 1;
1205
1206 while () {
1207 # we need to use a separate tls read buffer, as we must not receive data while
1208 # we are draining the buffer, and this can only happen with TLS.
1209 $self->{rbuf} .= delete $self->{_tls_rbuf}
1210 if exists $self->{_tls_rbuf};
1211
1212 my $len = length $self->{rbuf};
1213
1214 if (my $cb = shift @{ $self->{_queue} }) {
1215 unless ($cb->($self)) {
1216 # no progress can be made
1217 # (not enough data and no data forthcoming)
1218 $self->_error (Errno::EPIPE, 1), return
1219 if $self->{_eof};
1220
1221 unshift @{ $self->{_queue} }, $cb;
1222 last;
1223 }
1224 } elsif ($self->{on_read}) {
1225 last unless $len;
1226
1227 $self->{on_read}($self);
1228
1229 if (
1230 $len == length $self->{rbuf} # if no data has been consumed
1231 && !@{ $self->{_queue} } # and the queue is still empty
1232 && $self->{on_read} # but we still have on_read
1233 ) {
1234 # no further data will arrive
1235 # so no progress can be made
1236 $self->_error (Errno::EPIPE, 1), return
1237 if $self->{_eof};
1238
1239 last; # more data might arrive
1240 }
1241 } else {
1242 # read side becomes idle
1243 delete $self->{_rw} unless $self->{tls};
1244 last;
1245 }
1246 }
1247
1248 if ($self->{_eof}) {
1249 $self->{on_eof}
1250 ? $self->{on_eof}($self)
1251 : $self->_error (0, 1, "Unexpected end-of-file");
1252
1253 return;
1254 }
404 1255
405 if ( 1256 if (
406 defined $self->{rbuf_max} 1257 defined $self->{rbuf_max}
407 && $self->{rbuf_max} < length $self->{rbuf} 1258 && $self->{rbuf_max} < length $self->{rbuf}
408 ) { 1259 ) {
409 $! = &Errno::ENOSPC; return $self->error; 1260 $self->_error (Errno::ENOSPC, 1), return;
410 } 1261 }
411 1262
412 return if $self->{in_drain}; 1263 # may need to restart read watcher
413 local $self->{in_drain} = 1; 1264 unless ($self->{_rw}) {
414 1265 $self->start_read
415 while (my $len = length $self->{rbuf}) { 1266 if $self->{on_read} || @{ $self->{_queue} };
416 no strict 'refs';
417 if (my $cb = shift @{ $self->{queue} }) {
418 if (!$cb->($self)) {
419 if ($self->{eof}) {
420 # no progress can be made (not enough data and no data forthcoming)
421 $! = &Errno::EPIPE; return $self->error;
422 }
423
424 unshift @{ $self->{queue} }, $cb;
425 return;
426 }
427 } elsif ($self->{on_read}) {
428 $self->{on_read}($self);
429
430 if (
431 $self->{eof} # if no further data will arrive
432 && $len == length $self->{rbuf} # and no data has been consumed
433 && !@{ $self->{queue} } # and the queue is still empty
434 && $self->{on_read} # and we still want to read data
435 ) {
436 # then no progress can be made
437 $! = &Errno::EPIPE; return $self->error;
438 }
439 } else {
440 # read side becomes idle
441 delete $self->{rw};
442 return;
443 }
444 }
445
446 if ($self->{eof}) {
447 $self->_shutdown;
448 $self->{on_eof}($self)
449 if $self->{on_eof};
450 } 1267 }
451} 1268}
452 1269
453=item $handle->on_read ($cb) 1270=item $handle->on_read ($cb)
454 1271
455This replaces the currently set C<on_read> callback, or clears it (when 1272This replaces the currently set C<on_read> callback, or clears it (when
456the new callback is C<undef>). See the description of C<on_read> in the 1273the new callback is C<undef>). See the description of C<on_read> in the
457constructor. 1274constructor.
458 1275
1276This method may invoke callbacks (and therefore the handle might be
1277destroyed after it returns).
1278
459=cut 1279=cut
460 1280
461sub on_read { 1281sub on_read {
462 my ($self, $cb) = @_; 1282 my ($self, $cb) = @_;
463 1283
464 $self->{on_read} = $cb; 1284 $self->{on_read} = $cb;
1285 $self->_drain_rbuf if $cb;
465} 1286}
466 1287
467=item $handle->rbuf 1288=item $handle->rbuf
468 1289
469Returns the read buffer (as a modifiable lvalue). 1290Returns the read buffer (as a modifiable lvalue). You can also access the
1291read buffer directly as the C<< ->{rbuf} >> member, if you want (this is
1292much faster, and no less clean).
470 1293
471You can access the read buffer directly as the C<< ->{rbuf} >> member, if 1294The only operation allowed on the read buffer (apart from looking at it)
472you want. 1295is removing data from its beginning. Otherwise modifying or appending to
1296it is not allowed and will lead to hard-to-track-down bugs.
473 1297
474NOTE: The read buffer should only be used or modified if the C<on_read>, 1298NOTE: The read buffer should only be used or modified in the C<on_read>
475C<push_read> or C<unshift_read> methods are used. The other read methods 1299callback or when C<push_read> or C<unshift_read> are used with a single
476automatically manage the read buffer. 1300callback (i.e. untyped). Typed C<push_read> and C<unshift_read> methods
1301will manage the read buffer on their own.
477 1302
478=cut 1303=cut
479 1304
480sub rbuf : lvalue { 1305sub rbuf : lvalue {
481 $_[0]{rbuf} 1306 $_[0]{rbuf}
498 1323
499If enough data was available, then the callback must remove all data it is 1324If enough data was available, then the callback must remove all data it is
500interested in (which can be none at all) and return a true value. After returning 1325interested in (which can be none at all) and return a true value. After returning
501true, it will be removed from the queue. 1326true, it will be removed from the queue.
502 1327
1328These methods may invoke callbacks (and therefore the handle might be
1329destroyed after it returns).
1330
503=cut 1331=cut
1332
1333our %RH;
1334
1335sub register_read_type($$) {
1336 $RH{$_[0]} = $_[1];
1337}
504 1338
505sub push_read { 1339sub push_read {
506 my ($self, $cb) = @_; 1340 my $self = shift;
1341 my $cb = pop;
507 1342
1343 if (@_) {
1344 my $type = shift;
1345
1346 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1347 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_read")
1348 ->($self, $cb, @_);
1349 }
1350
508 push @{ $self->{queue} }, $cb; 1351 push @{ $self->{_queue} }, $cb;
509 $self->_drain_rbuf; 1352 $self->_drain_rbuf;
510} 1353}
511 1354
512sub unshift_read { 1355sub unshift_read {
513 my ($self, $cb) = @_; 1356 my $self = shift;
1357 my $cb = pop;
514 1358
1359 if (@_) {
1360 my $type = shift;
1361
1362 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
1363 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::unshift_read")
1364 ->($self, $cb, @_);
1365 }
1366
515 push @{ $self->{queue} }, $cb; 1367 unshift @{ $self->{_queue} }, $cb;
516 $self->_drain_rbuf; 1368 $self->_drain_rbuf;
517} 1369}
518 1370
519=item $handle->push_read_chunk ($len, $cb->($self, $data)) 1371=item $handle->push_read (type => @args, $cb)
520 1372
521=item $handle->unshift_read_chunk ($len, $cb->($self, $data)) 1373=item $handle->unshift_read (type => @args, $cb)
522 1374
523Append the given callback to the end of the queue (C<push_read_chunk>) or 1375Instead of providing a callback that parses the data itself you can chose
524prepend it (C<unshift_read_chunk>). 1376between a number of predefined parsing formats, for chunks of data, lines
1377etc. You can also specify the (fully qualified) name of a package, in
1378which case AnyEvent tries to load the package and then expects to find the
1379C<anyevent_read_type> function inside (see "custom read types", below).
525 1380
526The callback will be called only once C<$len> bytes have been read, and 1381Predefined types are (if you have ideas for additional types, feel free to
527these C<$len> bytes will be passed to the callback. 1382drop by and tell us):
528 1383
529=cut 1384=over 4
530 1385
531sub _read_chunk($$) { 1386=item chunk => $octets, $cb->($handle, $data)
1387
1388Invoke the callback only once C<$octets> bytes have been read. Pass the
1389data read to the callback. The callback will never be called with less
1390data.
1391
1392Example: read 2 bytes.
1393
1394 $handle->push_read (chunk => 2, sub {
1395 warn "yay ", unpack "H*", $_[1];
1396 });
1397
1398=cut
1399
1400register_read_type chunk => sub {
532 my ($self, $len, $cb) = @_; 1401 my ($self, $cb, $len) = @_;
533 1402
534 sub { 1403 sub {
535 $len <= length $_[0]{rbuf} or return; 1404 $len <= length $_[0]{rbuf} or return;
536 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, ""); 1405 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
537 1 1406 1
538 } 1407 }
539} 1408};
540 1409
541sub push_read_chunk { 1410=item line => [$eol, ]$cb->($handle, $line, $eol)
542 $_[0]->push_read (&_read_chunk);
543}
544
545
546sub unshift_read_chunk {
547 $_[0]->unshift_read (&_read_chunk);
548}
549
550=item $handle->push_read_line ([$eol, ]$cb->($self, $line, $eol))
551
552=item $handle->unshift_read_line ([$eol, ]$cb->($self, $line, $eol))
553
554Append the given callback to the end of the queue (C<push_read_line>) or
555prepend it (C<unshift_read_line>).
556 1411
557The callback will be called only once a full line (including the end of 1412The callback will be called only once a full line (including the end of
558line marker, C<$eol>) has been read. This line (excluding the end of line 1413line marker, C<$eol>) has been read. This line (excluding the end of line
559marker) will be passed to the callback as second argument (C<$line>), and 1414marker) will be passed to the callback as second argument (C<$line>), and
560the end of line marker as the third argument (C<$eol>). 1415the end of line marker as the third argument (C<$eol>).
571Partial lines at the end of the stream will never be returned, as they are 1426Partial lines at the end of the stream will never be returned, as they are
572not marked by the end of line marker. 1427not marked by the end of line marker.
573 1428
574=cut 1429=cut
575 1430
576sub _read_line($$) { 1431register_read_type line => sub {
577 my $self = shift; 1432 my ($self, $cb, $eol) = @_;
578 my $cb = pop;
579 my $eol = @_ ? shift : qr|(\015?\012)|;
580 my $pos;
581 1433
1434 if (@_ < 3) {
1435 # this is more than twice as fast as the generic code below
1436 sub {
1437 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1438
1439 $cb->($_[0], $1, $2);
1440 1
1441 }
1442 } else {
582 $eol = quotemeta $eol unless ref $eol; 1443 $eol = quotemeta $eol unless ref $eol;
583 $eol = qr|^(.*?)($eol)|s; 1444 $eol = qr|^(.*?)($eol)|s;
1445
1446 sub {
1447 $_[0]{rbuf} =~ s/$eol// or return;
1448
1449 $cb->($_[0], $1, $2);
1450 1
1451 }
1452 }
1453};
1454
1455=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1456
1457Makes a regex match against the regex object C<$accept> and returns
1458everything up to and including the match.
1459
1460Example: read a single line terminated by '\n'.
1461
1462 $handle->push_read (regex => qr<\n>, sub { ... });
1463
1464If C<$reject> is given and not undef, then it determines when the data is
1465to be rejected: it is matched against the data when the C<$accept> regex
1466does not match and generates an C<EBADMSG> error when it matches. This is
1467useful to quickly reject wrong data (to avoid waiting for a timeout or a
1468receive buffer overflow).
1469
1470Example: expect a single decimal number followed by whitespace, reject
1471anything else (not the use of an anchor).
1472
1473 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1474
1475If C<$skip> is given and not C<undef>, then it will be matched against
1476the receive buffer when neither C<$accept> nor C<$reject> match,
1477and everything preceding and including the match will be accepted
1478unconditionally. This is useful to skip large amounts of data that you
1479know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1480have to start matching from the beginning. This is purely an optimisation
1481and is usually worth it only when you expect more than a few kilobytes.
1482
1483Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1484expect the header to be very large (it isn't in practice, but...), we use
1485a skip regex to skip initial portions. The skip regex is tricky in that
1486it only accepts something not ending in either \015 or \012, as these are
1487required for the accept regex.
1488
1489 $handle->push_read (regex =>
1490 qr<\015\012\015\012>,
1491 undef, # no reject
1492 qr<^.*[^\015\012]>,
1493 sub { ... });
1494
1495=cut
1496
1497register_read_type regex => sub {
1498 my ($self, $cb, $accept, $reject, $skip) = @_;
1499
1500 my $data;
1501 my $rbuf = \$self->{rbuf};
584 1502
585 sub { 1503 sub {
586 $_[0]{rbuf} =~ s/$eol// or return; 1504 # accept
1505 if ($$rbuf =~ $accept) {
1506 $data .= substr $$rbuf, 0, $+[0], "";
1507 $cb->($self, $data);
1508 return 1;
1509 }
1510
1511 # reject
1512 if ($reject && $$rbuf =~ $reject) {
1513 $self->_error (Errno::EBADMSG);
1514 }
587 1515
588 $cb->($_[0], $1, $2); 1516 # skip
1517 if ($skip && $$rbuf =~ $skip) {
1518 $data .= substr $$rbuf, 0, $+[0], "";
1519 }
1520
1521 ()
1522 }
1523};
1524
1525=item netstring => $cb->($handle, $string)
1526
1527A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
1528
1529Throws an error with C<$!> set to EBADMSG on format violations.
1530
1531=cut
1532
1533register_read_type netstring => sub {
1534 my ($self, $cb) = @_;
1535
1536 sub {
1537 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
1538 if ($_[0]{rbuf} =~ /[^0-9]/) {
1539 $self->_error (Errno::EBADMSG);
1540 }
1541 return;
1542 }
1543
1544 my $len = $1;
1545
1546 $self->unshift_read (chunk => $len, sub {
1547 my $string = $_[1];
1548 $_[0]->unshift_read (chunk => 1, sub {
1549 if ($_[1] eq ",") {
1550 $cb->($_[0], $string);
1551 } else {
1552 $self->_error (Errno::EBADMSG);
1553 }
1554 });
1555 });
1556
589 1 1557 1
590 } 1558 }
591} 1559};
592 1560
593sub push_read_line { 1561=item packstring => $format, $cb->($handle, $string)
594 $_[0]->push_read (&_read_line);
595}
596 1562
597sub unshift_read_line { 1563An octet string prefixed with an encoded length. The encoding C<$format>
598 $_[0]->unshift_read (&_read_line); 1564uses the same format as a Perl C<pack> format, but must specify a single
599} 1565integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1566optional C<!>, C<< < >> or C<< > >> modifier).
1567
1568For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1569EPP uses a prefix of C<N> (4 octtes).
1570
1571Example: read a block of data prefixed by its length in BER-encoded
1572format (very efficient).
1573
1574 $handle->push_read (packstring => "w", sub {
1575 my ($handle, $data) = @_;
1576 });
1577
1578=cut
1579
1580register_read_type packstring => sub {
1581 my ($self, $cb, $format) = @_;
1582
1583 sub {
1584 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1585 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1586 or return;
1587
1588 $format = length pack $format, $len;
1589
1590 # bypass unshift if we already have the remaining chunk
1591 if ($format + $len <= length $_[0]{rbuf}) {
1592 my $data = substr $_[0]{rbuf}, $format, $len;
1593 substr $_[0]{rbuf}, 0, $format + $len, "";
1594 $cb->($_[0], $data);
1595 } else {
1596 # remove prefix
1597 substr $_[0]{rbuf}, 0, $format, "";
1598
1599 # read remaining chunk
1600 $_[0]->unshift_read (chunk => $len, $cb);
1601 }
1602
1603 1
1604 }
1605};
1606
1607=item json => $cb->($handle, $hash_or_arrayref)
1608
1609Reads a JSON object or array, decodes it and passes it to the
1610callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1611
1612If a C<json> object was passed to the constructor, then that will be used
1613for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1614
1615This read type uses the incremental parser available with JSON version
16162.09 (and JSON::XS version 2.2) and above. You have to provide a
1617dependency on your own: this module will load the JSON module, but
1618AnyEvent does not depend on it itself.
1619
1620Since JSON texts are fully self-delimiting, the C<json> read and write
1621types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1622the C<json> write type description, above, for an actual example.
1623
1624=cut
1625
1626register_read_type json => sub {
1627 my ($self, $cb) = @_;
1628
1629 my $json = $self->{json} ||= json_coder;
1630
1631 my $data;
1632 my $rbuf = \$self->{rbuf};
1633
1634 sub {
1635 my $ref = eval { $json->incr_parse ($self->{rbuf}) };
1636
1637 if ($ref) {
1638 $self->{rbuf} = $json->incr_text;
1639 $json->incr_text = "";
1640 $cb->($self, $ref);
1641
1642 1
1643 } elsif ($@) {
1644 # error case
1645 $json->incr_skip;
1646
1647 $self->{rbuf} = $json->incr_text;
1648 $json->incr_text = "";
1649
1650 $self->_error (Errno::EBADMSG);
1651
1652 ()
1653 } else {
1654 $self->{rbuf} = "";
1655
1656 ()
1657 }
1658 }
1659};
1660
1661=item storable => $cb->($handle, $ref)
1662
1663Deserialises a L<Storable> frozen representation as written by the
1664C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1665data).
1666
1667Raises C<EBADMSG> error if the data could not be decoded.
1668
1669=cut
1670
1671register_read_type storable => sub {
1672 my ($self, $cb) = @_;
1673
1674 require Storable;
1675
1676 sub {
1677 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1678 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1679 or return;
1680
1681 my $format = length pack "w", $len;
1682
1683 # bypass unshift if we already have the remaining chunk
1684 if ($format + $len <= length $_[0]{rbuf}) {
1685 my $data = substr $_[0]{rbuf}, $format, $len;
1686 substr $_[0]{rbuf}, 0, $format + $len, "";
1687 $cb->($_[0], Storable::thaw ($data));
1688 } else {
1689 # remove prefix
1690 substr $_[0]{rbuf}, 0, $format, "";
1691
1692 # read remaining chunk
1693 $_[0]->unshift_read (chunk => $len, sub {
1694 if (my $ref = eval { Storable::thaw ($_[1]) }) {
1695 $cb->($_[0], $ref);
1696 } else {
1697 $self->_error (Errno::EBADMSG);
1698 }
1699 });
1700 }
1701
1702 1
1703 }
1704};
1705
1706=back
1707
1708=item custom read types - Package::anyevent_read_type $handle, $cb, @args
1709
1710Instead of one of the predefined types, you can also specify the name
1711of a package. AnyEvent will try to load the package and then expects to
1712find a function named C<anyevent_read_type> inside. If it isn't found, it
1713progressively tries to load the parent package until it either finds the
1714function (good) or runs out of packages (bad).
1715
1716Whenever this type is used, C<push_read> will invoke the function with the
1717handle object, the original callback and the remaining arguments.
1718
1719The function is supposed to return a callback (usually a closure) that
1720works as a plain read callback (see C<< ->push_read ($cb) >>), so you can
1721mentally treat the function as a "configurable read type to read callback"
1722converter.
1723
1724It should invoke the original callback when it is done reading (remember
1725to pass C<$handle> as first argument as all other callbacks do that,
1726although there is no strict requirement on this).
1727
1728For examples, see the source of this module (F<perldoc -m
1729AnyEvent::Handle>, search for C<register_read_type>)).
600 1730
601=item $handle->stop_read 1731=item $handle->stop_read
602 1732
603=item $handle->start_read 1733=item $handle->start_read
604 1734
605In rare cases you actually do not want to read anything from the 1735In rare cases you actually do not want to read anything from the
606socket. In this case you can call C<stop_read>. Neither C<on_read> no 1736socket. In this case you can call C<stop_read>. Neither C<on_read> nor
607any queued callbacks will be executed then. To start readign again, call 1737any queued callbacks will be executed then. To start reading again, call
608C<start_read>. 1738C<start_read>.
1739
1740Note that AnyEvent::Handle will automatically C<start_read> for you when
1741you change the C<on_read> callback or push/unshift a read callback, and it
1742will automatically C<stop_read> for you when neither C<on_read> is set nor
1743there are any read requests in the queue.
1744
1745These methods will have no effect when in TLS mode (as TLS doesn't support
1746half-duplex connections).
609 1747
610=cut 1748=cut
611 1749
612sub stop_read { 1750sub stop_read {
613 my ($self) = @_; 1751 my ($self) = @_;
614 1752
615 delete $self->{rw}; 1753 delete $self->{_rw} unless $self->{tls};
616} 1754}
617 1755
618sub start_read { 1756sub start_read {
619 my ($self) = @_; 1757 my ($self) = @_;
620 1758
621 unless ($self->{rw} || $self->{eof}) { 1759 unless ($self->{_rw} || $self->{_eof} || !$self->{fh}) {
622 Scalar::Util::weaken $self; 1760 Scalar::Util::weaken $self;
623 1761
624 $self->{rw} = AnyEvent->io (fh => $self->{fh}, poll => "r", cb => sub { 1762 $self->{_rw} = AE::io $self->{fh}, 0, sub {
625 my $rbuf = $self->{filter_r} ? \my $buf : \$self->{rbuf}; 1763 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
626 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size} || 8192, length $$rbuf; 1764 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size}, length $$rbuf;
627 1765
628 if ($len > 0) { 1766 if ($len > 0) {
629 $self->{filter_r} 1767 $self->{_activity} = $self->{_ractivity} = AE::now;
630 ? $self->{filter_r}->($self, $rbuf) 1768
1769 if ($self->{tls}) {
1770 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1771
1772 &_dotls ($self);
1773 } else {
631 : $self->_drain_rbuf; 1774 $self->_drain_rbuf;
1775 }
1776
1777 if ($len == $self->{read_size}) {
1778 $self->{read_size} *= 2;
1779 $self->{read_size} = $self->{max_read_size} || MAX_READ_SIZE
1780 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
1781 }
632 1782
633 } elsif (defined $len) { 1783 } elsif (defined $len) {
634 delete $self->{rw}; 1784 delete $self->{_rw};
635 $self->{eof} = 1; 1785 $self->{_eof} = 1;
636 $self->_drain_rbuf; 1786 $self->_drain_rbuf;
637 1787
638 } elsif ($! != EAGAIN && $! != EINTR) { 1788 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
639 return $self->error; 1789 return $self->_error ($!, 1);
640 } 1790 }
641 }); 1791 };
642 } 1792 }
643} 1793}
644 1794
1795our $ERROR_SYSCALL;
1796our $ERROR_WANT_READ;
1797
1798sub _tls_error {
1799 my ($self, $err) = @_;
1800
1801 return $self->_error ($!, 1)
1802 if $err == Net::SSLeay::ERROR_SYSCALL ();
1803
1804 my $err =Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
1805
1806 # reduce error string to look less scary
1807 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
1808
1809 if ($self->{_on_starttls}) {
1810 (delete $self->{_on_starttls})->($self, undef, $err);
1811 &_freetls;
1812 } else {
1813 &_freetls;
1814 $self->_error (Errno::EPROTO, 1, $err);
1815 }
1816}
1817
1818# poll the write BIO and send the data if applicable
1819# also decode read data if possible
1820# this is basiclaly our TLS state machine
1821# more efficient implementations are possible with openssl,
1822# but not with the buggy and incomplete Net::SSLeay.
645sub _dotls { 1823sub _dotls {
646 my ($self) = @_; 1824 my ($self) = @_;
647 1825
1826 my $tmp;
1827
648 if (length $self->{tls_wbuf}) { 1828 if (length $self->{_tls_wbuf}) {
649 my $len = Net::SSLeay::write ($self->{tls}, $self->{tls_wbuf}); 1829 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
650 substr $self->{tls_wbuf}, 0, $len, "" if $len > 0; 1830 substr $self->{_tls_wbuf}, 0, $tmp, "";
651 } 1831 }
652 1832
1833 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1834 return $self->_tls_error ($tmp)
1835 if $tmp != $ERROR_WANT_READ
1836 && ($tmp != $ERROR_SYSCALL || $!);
1837 }
1838
653 if (defined (my $buf = Net::SSLeay::BIO_read ($self->{tls_wbio}))) { 1839 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1840 unless (length $tmp) {
1841 $self->{_on_starttls}
1842 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1843 &_freetls;
1844
1845 if ($self->{on_stoptls}) {
1846 $self->{on_stoptls}($self);
1847 return;
1848 } else {
1849 # let's treat SSL-eof as we treat normal EOF
1850 delete $self->{_rw};
1851 $self->{_eof} = 1;
1852 }
1853 }
1854
1855 $self->{_tls_rbuf} .= $tmp;
1856 $self->_drain_rbuf;
1857 $self->{tls} or return; # tls session might have gone away in callback
1858 }
1859
1860 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
1861 return $self->_tls_error ($tmp)
1862 if $tmp != $ERROR_WANT_READ
1863 && ($tmp != $ERROR_SYSCALL || $!);
1864
1865 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
654 $self->{wbuf} .= $buf; 1866 $self->{wbuf} .= $tmp;
655 $self->_drain_wbuf; 1867 $self->_drain_wbuf;
1868 $self->{tls} or return; # tls session might have gone away in callback
656 } 1869 }
657 1870
658 if (defined (my $buf = Net::SSLeay::read ($self->{tls}))) { 1871 $self->{_on_starttls}
659 $self->{rbuf} .= $buf; 1872 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
660 $self->_drain_rbuf; 1873 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
661 } elsif ( 1874}
662 (my $err = Net::SSLeay::get_error ($self->{tls}, -1)) 1875
1876=item $handle->starttls ($tls[, $tls_ctx])
1877
1878Instead of starting TLS negotiation immediately when the AnyEvent::Handle
1879object is created, you can also do that at a later time by calling
1880C<starttls>.
1881
1882Starting TLS is currently an asynchronous operation - when you push some
1883write data and then call C<< ->starttls >> then TLS negotiation will start
1884immediately, after which the queued write data is then sent.
1885
1886The first argument is the same as the C<tls> constructor argument (either
1887C<"connect">, C<"accept"> or an existing Net::SSLeay object).
1888
1889The second argument is the optional C<AnyEvent::TLS> object that is used
1890when AnyEvent::Handle has to create its own TLS connection object, or
1891a hash reference with C<< key => value >> pairs that will be used to
1892construct a new context.
1893
1894The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1895context in C<< $handle->{tls_ctx} >> after this call and can be used or
1896changed to your liking. Note that the handshake might have already started
1897when this function returns.
1898
1899Due to bugs in OpenSSL, it might or might not be possible to do multiple
1900handshakes on the same stream. It is best to not attempt to use the
1901stream after stopping TLS.
1902
1903This method may invoke callbacks (and therefore the handle might be
1904destroyed after it returns).
1905
1906=cut
1907
1908our %TLS_CACHE; #TODO not yet documented, should we?
1909
1910sub starttls {
1911 my ($self, $tls, $ctx) = @_;
1912
1913 Carp::croak "It is an error to call starttls on an AnyEvent::Handle object while TLS is already active, caught"
1914 if $self->{tls};
1915
1916 $self->{tls} = $tls;
1917 $self->{tls_ctx} = $ctx if @_ > 2;
1918
1919 return unless $self->{fh};
1920
1921 require Net::SSLeay;
1922
1923 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
663 != Net::SSLeay::ERROR_WANT_READ () 1924 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
664 ) { 1925
665 if ($err == Net::SSLeay::ERROR_SYSCALL ()) { 1926 $tls = delete $self->{tls};
666 $self->error; 1927 $ctx = $self->{tls_ctx};
667 } elsif ($err == Net::SSLeay::ERROR_SSL ()) { 1928
668 $! = &Errno::EIO; 1929 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context or session
669 $self->error; 1930
1931 if ("HASH" eq ref $ctx) {
1932 require AnyEvent::TLS;
1933
1934 if ($ctx->{cache}) {
1935 my $key = $ctx+0;
1936 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
1937 } else {
1938 $ctx = new AnyEvent::TLS %$ctx;
670 } 1939 }
671 1940 }
672 # all others are fine for our purposes
673 } 1941
674} 1942 $self->{tls_ctx} = $ctx || TLS_CTX ();
1943 $self->{tls} = $tls = $self->{tls_ctx}->_get_session ($tls, $self, $self->{peername});
675 1944
676# TODO: maybe document... 1945 # basically, this is deep magic (because SSL_read should have the same issues)
677sub starttls { 1946 # but the openssl maintainers basically said: "trust us, it just works".
678 my ($self, $ssl, $ctx) = @_; 1947 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
679 1948 # and mismaintained ssleay-module doesn't even offer them).
680 if ($ssl eq "accept") { 1949 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
681 $ssl = Net::SSLeay::new ($ctx || TLS_CTX ());
682 Net::SSLeay::set_accept_state ($ssl);
683 } elsif ($ssl eq "connect") {
684 $ssl = Net::SSLeay::new ($ctx || TLS_CTX ());
685 Net::SSLeay::set_connect_state ($ssl);
686 } 1950 #
1951 # in short: this is a mess.
1952 #
1953 # note that we do not try to keep the length constant between writes as we are required to do.
1954 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
1955 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
1956 # have identity issues in that area.
1957# Net::SSLeay::CTX_set_mode ($ssl,
1958# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
1959# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
1960 Net::SSLeay::CTX_set_mode ($tls, 1|2);
687 1961
688 $self->{tls} = $ssl;
689
690 $self->{tls_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ()); 1962 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
691 $self->{tls_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ()); 1963 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
692 1964
1965 Net::SSLeay::BIO_write ($self->{_rbio}, delete $self->{rbuf});
1966
693 Net::SSLeay::set_bio ($ssl, $self->{tls_rbio}, $self->{tls_wbio}); 1967 Net::SSLeay::set_bio ($tls, $self->{_rbio}, $self->{_wbio});
694 1968
695 $self->{filter_w} = sub { 1969 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
696 $_[0]{tls_wbuf} .= ${$_[1]}; 1970 if $self->{on_starttls};
1971
1972 &_dotls; # need to trigger the initial handshake
1973 $self->start_read; # make sure we actually do read
1974}
1975
1976=item $handle->stoptls
1977
1978Shuts down the SSL connection - this makes a proper EOF handshake by
1979sending a close notify to the other side, but since OpenSSL doesn't
1980support non-blocking shut downs, it is not guaranteed that you can re-use
1981the stream afterwards.
1982
1983This method may invoke callbacks (and therefore the handle might be
1984destroyed after it returns).
1985
1986=cut
1987
1988sub stoptls {
1989 my ($self) = @_;
1990
1991 if ($self->{tls} && $self->{fh}) {
1992 Net::SSLeay::shutdown ($self->{tls});
1993
697 &_dotls; 1994 &_dotls;
1995
1996# # we don't give a shit. no, we do, but we can't. no...#d#
1997# # we, we... have to use openssl :/#d#
1998# &_freetls;#d#
1999 }
2000}
2001
2002sub _freetls {
2003 my ($self) = @_;
2004
2005 return unless $self->{tls};
2006
2007 $self->{tls_ctx}->_put_session (delete $self->{tls})
2008 if $self->{tls} > 0;
2009
2010 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
2011}
2012
2013sub DESTROY {
2014 my ($self) = @_;
2015
2016 &_freetls;
2017
2018 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
2019
2020 if ($linger && length $self->{wbuf} && $self->{fh}) {
2021 my $fh = delete $self->{fh};
2022 my $wbuf = delete $self->{wbuf};
2023
2024 my @linger;
2025
2026 push @linger, AE::io $fh, 1, sub {
2027 my $len = syswrite $fh, $wbuf, length $wbuf;
2028
2029 if ($len > 0) {
2030 substr $wbuf, 0, $len, "";
2031 } elsif (defined $len || ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK)) {
2032 @linger = (); # end
2033 }
2034 };
2035 push @linger, AE::timer $linger, 0, sub {
2036 @linger = ();
2037 };
2038 }
2039}
2040
2041=item $handle->destroy
2042
2043Shuts down the handle object as much as possible - this call ensures that
2044no further callbacks will be invoked and as many resources as possible
2045will be freed. Any method you will call on the handle object after
2046destroying it in this way will be silently ignored (and it will return the
2047empty list).
2048
2049Normally, you can just "forget" any references to an AnyEvent::Handle
2050object and it will simply shut down. This works in fatal error and EOF
2051callbacks, as well as code outside. It does I<NOT> work in a read or write
2052callback, so when you want to destroy the AnyEvent::Handle object from
2053within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
2054that case.
2055
2056Destroying the handle object in this way has the advantage that callbacks
2057will be removed as well, so if those are the only reference holders (as
2058is common), then one doesn't need to do anything special to break any
2059reference cycles.
2060
2061The handle might still linger in the background and write out remaining
2062data, as specified by the C<linger> option, however.
2063
2064=cut
2065
2066sub destroy {
2067 my ($self) = @_;
2068
2069 $self->DESTROY;
2070 %$self = ();
2071 bless $self, "AnyEvent::Handle::destroyed";
2072}
2073
2074sub AnyEvent::Handle::destroyed::AUTOLOAD {
2075 #nop
2076}
2077
2078=item $handle->destroyed
2079
2080Returns false as long as the handle hasn't been destroyed by a call to C<<
2081->destroy >>, true otherwise.
2082
2083Can be useful to decide whether the handle is still valid after some
2084callback possibly destroyed the handle. For example, C<< ->push_write >>,
2085C<< ->starttls >> and other methods can call user callbacks, which in turn
2086can destroy the handle, so work can be avoided by checking sometimes:
2087
2088 $hdl->starttls ("accept");
2089 return if $hdl->destroyed;
2090 $hdl->push_write (...
2091
2092Note that the call to C<push_write> will silently be ignored if the handle
2093has been destroyed, so often you can just ignore the possibility of the
2094handle being destroyed.
2095
2096=cut
2097
2098sub destroyed { 0 }
2099sub AnyEvent::Handle::destroyed::destroyed { 1 }
2100
2101=item AnyEvent::Handle::TLS_CTX
2102
2103This function creates and returns the AnyEvent::TLS object used by default
2104for TLS mode.
2105
2106The context is created by calling L<AnyEvent::TLS> without any arguments.
2107
2108=cut
2109
2110our $TLS_CTX;
2111
2112sub TLS_CTX() {
2113 $TLS_CTX ||= do {
2114 require AnyEvent::TLS;
2115
2116 new AnyEvent::TLS
2117 }
2118}
2119
2120=back
2121
2122
2123=head1 NONFREQUENTLY ASKED QUESTIONS
2124
2125=over 4
2126
2127=item I C<undef> the AnyEvent::Handle reference inside my callback and
2128still get further invocations!
2129
2130That's because AnyEvent::Handle keeps a reference to itself when handling
2131read or write callbacks.
2132
2133It is only safe to "forget" the reference inside EOF or error callbacks,
2134from within all other callbacks, you need to explicitly call the C<<
2135->destroy >> method.
2136
2137=item I get different callback invocations in TLS mode/Why can't I pause
2138reading?
2139
2140Unlike, say, TCP, TLS connections do not consist of two independent
2141communication channels, one for each direction. Or put differently, the
2142read and write directions are not independent of each other: you cannot
2143write data unless you are also prepared to read, and vice versa.
2144
2145This means that, in TLS mode, you might get C<on_error> or C<on_eof>
2146callback invocations when you are not expecting any read data - the reason
2147is that AnyEvent::Handle always reads in TLS mode.
2148
2149During the connection, you have to make sure that you always have a
2150non-empty read-queue, or an C<on_read> watcher. At the end of the
2151connection (or when you no longer want to use it) you can call the
2152C<destroy> method.
2153
2154=item How do I read data until the other side closes the connection?
2155
2156If you just want to read your data into a perl scalar, the easiest way
2157to achieve this is by setting an C<on_read> callback that does nothing,
2158clearing the C<on_eof> callback and in the C<on_error> callback, the data
2159will be in C<$_[0]{rbuf}>:
2160
2161 $handle->on_read (sub { });
2162 $handle->on_eof (undef);
2163 $handle->on_error (sub {
2164 my $data = delete $_[0]{rbuf};
2165 });
2166
2167The reason to use C<on_error> is that TCP connections, due to latencies
2168and packets loss, might get closed quite violently with an error, when in
2169fact all data has been received.
2170
2171It is usually better to use acknowledgements when transferring data,
2172to make sure the other side hasn't just died and you got the data
2173intact. This is also one reason why so many internet protocols have an
2174explicit QUIT command.
2175
2176=item I don't want to destroy the handle too early - how do I wait until
2177all data has been written?
2178
2179After writing your last bits of data, set the C<on_drain> callback
2180and destroy the handle in there - with the default setting of
2181C<low_water_mark> this will be called precisely when all data has been
2182written to the socket:
2183
2184 $handle->push_write (...);
2185 $handle->on_drain (sub {
2186 warn "all data submitted to the kernel\n";
2187 undef $handle;
2188 });
2189
2190If you just want to queue some data and then signal EOF to the other side,
2191consider using C<< ->push_shutdown >> instead.
2192
2193=item I want to contact a TLS/SSL server, I don't care about security.
2194
2195If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
2196connect to it and then create the AnyEvent::Handle with the C<tls>
2197parameter:
2198
2199 tcp_connect $host, $port, sub {
2200 my ($fh) = @_;
2201
2202 my $handle = new AnyEvent::Handle
2203 fh => $fh,
2204 tls => "connect",
2205 on_error => sub { ... };
2206
2207 $handle->push_write (...);
698 }; 2208 };
699 $self->{filter_r} = sub {
700 Net::SSLeay::BIO_write ($_[0]{tls_rbio}, ${$_[1]});
701 &_dotls;
702 };
703}
704 2209
705sub DESTROY { 2210=item I want to contact a TLS/SSL server, I do care about security.
706 my $self = shift;
707 2211
708 Net::SSLeay::free (delete $self->{tls}) if $self->{tls}; 2212Then you should additionally enable certificate verification, including
709} 2213peername verification, if the protocol you use supports it (see
2214L<AnyEvent::TLS>, C<verify_peername>).
710 2215
711=item AnyEvent::Handle::TLS_CTX 2216E.g. for HTTPS:
712 2217
713This function creates and returns the Net::SSLeay::CTX object used by 2218 tcp_connect $host, $port, sub {
714default for TLS mode. 2219 my ($fh) = @_;
715 2220
716The context is created like this: 2221 my $handle = new AnyEvent::Handle
2222 fh => $fh,
2223 peername => $host,
2224 tls => "connect",
2225 tls_ctx => { verify => 1, verify_peername => "https" },
2226 ...
717 2227
718 Net::SSLeay::load_error_strings; 2228Note that you must specify the hostname you connected to (or whatever
719 Net::SSLeay::SSLeay_add_ssl_algorithms; 2229"peername" the protocol needs) as the C<peername> argument, otherwise no
720 Net::SSLeay::randomize; 2230peername verification will be done.
721 2231
722 my $CTX = Net::SSLeay::CTX_new; 2232The above will use the system-dependent default set of trusted CA
2233certificates. If you want to check against a specific CA, add the
2234C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
723 2235
724 Net::SSLeay::CTX_set_options $CTX, Net::SSLeay::OP_ALL 2236 tls_ctx => {
2237 verify => 1,
2238 verify_peername => "https",
2239 ca_file => "my-ca-cert.pem",
2240 },
725 2241
726=cut 2242=item I want to create a TLS/SSL server, how do I do that?
727 2243
728our $TLS_CTX; 2244Well, you first need to get a server certificate and key. You have
2245three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
2246self-signed certificate (cheap. check the search engine of your choice,
2247there are many tutorials on the net) or c) make your own CA (tinyca2 is a
2248nice program for that purpose).
729 2249
730sub TLS_CTX() { 2250Then create a file with your private key (in PEM format, see
731 $TLS_CTX || do { 2251L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
732 require Net::SSLeay; 2252file should then look like this:
733 2253
734 Net::SSLeay::load_error_strings (); 2254 -----BEGIN RSA PRIVATE KEY-----
735 Net::SSLeay::SSLeay_add_ssl_algorithms (); 2255 ...header data
736 Net::SSLeay::randomize (); 2256 ... lots of base64'y-stuff
2257 -----END RSA PRIVATE KEY-----
737 2258
738 $TLS_CTX = Net::SSLeay::CTX_new (); 2259 -----BEGIN CERTIFICATE-----
2260 ... lots of base64'y-stuff
2261 -----END CERTIFICATE-----
739 2262
740 Net::SSLeay::CTX_set_options ($TLS_CTX, Net::SSLeay::OP_ALL ()); 2263The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
2264specify this file as C<cert_file>:
741 2265
742 $TLS_CTX 2266 tcp_server undef, $port, sub {
743 } 2267 my ($fh) = @_;
744} 2268
2269 my $handle = new AnyEvent::Handle
2270 fh => $fh,
2271 tls => "accept",
2272 tls_ctx => { cert_file => "my-server-keycert.pem" },
2273 ...
2274
2275When you have intermediate CA certificates that your clients might not
2276know about, just append them to the C<cert_file>.
745 2277
746=back 2278=back
747 2279
2280
2281=head1 SUBCLASSING AnyEvent::Handle
2282
2283In many cases, you might want to subclass AnyEvent::Handle.
2284
2285To make this easier, a given version of AnyEvent::Handle uses these
2286conventions:
2287
2288=over 4
2289
2290=item * all constructor arguments become object members.
2291
2292At least initially, when you pass a C<tls>-argument to the constructor it
2293will end up in C<< $handle->{tls} >>. Those members might be changed or
2294mutated later on (for example C<tls> will hold the TLS connection object).
2295
2296=item * other object member names are prefixed with an C<_>.
2297
2298All object members not explicitly documented (internal use) are prefixed
2299with an underscore character, so the remaining non-C<_>-namespace is free
2300for use for subclasses.
2301
2302=item * all members not documented here and not prefixed with an underscore
2303are free to use in subclasses.
2304
2305Of course, new versions of AnyEvent::Handle may introduce more "public"
2306member variables, but that's just life. At least it is documented.
2307
2308=back
2309
748=head1 AUTHOR 2310=head1 AUTHOR
749 2311
750Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>. 2312Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
751 2313
752=cut 2314=cut

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines