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 1.31 by root, Sun May 25 00:08:49 2008 UTC vs.
Revision 1.235 by root, Tue May 8 19:41:22 2012 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 file handles via AnyEvent 3AnyEvent::Handle - non-blocking I/O on streaming handles via AnyEvent
16
17=cut
18
19our $VERSION = '0.04';
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 $handle = 12 my $hdl; $hdl = new AnyEvent::Handle
29 AnyEvent::Handle->new (
30 fh => \*STDIN, 13 fh => \*STDIN,
31 on_eof => sub { 14 on_error => sub {
32 $cv->broadcast; 15 my ($hdl, $fatal, $msg) = @_;
33 }, 16 AE::log error => $msg;
17 $hdl->destroy;
18 $cv->send;
34 ); 19 };
35 20
36 # send some request line 21 # send some request line
37 $handle->push_write ("getinfo\015\012"); 22 $hdl->push_write ("getinfo\015\012");
38 23
39 # read the response line 24 # read the response line
40 $handle->push_read (line => sub { 25 $hdl->push_read (line => sub {
41 my ($handle, $line) = @_; 26 my ($hdl, $line) = @_;
42 warn "read line <$line>\n"; 27 say "got line <$line>";
43 $cv->send; 28 $cv->send;
44 }); 29 });
45 30
46 $cv->recv; 31 $cv->recv;
47 32
48=head1 DESCRIPTION 33=head1 DESCRIPTION
49 34
50This 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
51filehandles. For utility functions for doing non-blocking connects and accepts 36stream-based filehandles (sockets, pipes, and other stream things).
52on sockets see L<AnyEvent::Util>.
53 37
38The L<AnyEvent::Intro> tutorial contains some well-documented
39AnyEvent::Handle examples.
40
54In the following, when the documentation refers to of "bytes" then this 41In the following, where the documentation refers to "bytes", it means
55means characters. As sysread and syswrite are used for all I/O, their 42characters. As sysread and syswrite are used for all I/O, their
56treatment 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.
57 47
58All callbacks will be invoked with the handle object as their first 48All callbacks will be invoked with the handle object as their first
59argument. 49argument.
60 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
61=head1 METHODS 82=head1 METHODS
62 83
63=over 4 84=over 4
64 85
65=item B<new (%args)> 86=item $handle = B<new> AnyEvent::Handle fh => $filehandle, key => value...
66 87
67The constructor supports these arguments (all as key => value pairs). 88The constructor supports these arguments (all as C<< key => value >> pairs).
68 89
69=over 4 90=over 4
70 91
71=item fh => $filehandle [MANDATORY] 92=item fh => $filehandle [C<fh> or C<connect> MANDATORY]
72 93
73The filehandle this L<AnyEvent::Handle> object will operate on. 94The filehandle this L<AnyEvent::Handle> object will operate on.
74
75NOTE: The filehandle will be set to non-blocking (using 95NOTE: The filehandle will be set to non-blocking mode (using
76AnyEvent::Util::fh_nonblocking). 96C<AnyEvent::Util::fh_nonblocking>) by the constructor and needs to stay in
97that mode.
77 98
78=item on_eof => $cb->($self) 99=item connect => [$host, $service] [C<fh> or C<connect> MANDATORY]
79 100
80Set the callback to be called on EOF. 101Try to connect to the specified host and service (port), using
102C<AnyEvent::Socket::tcp_connect>. The C<$host> additionally becomes the
103default C<peername>.
81 104
82While not mandatory, it is highly recommended to set an eof callback, 105You have to specify either this parameter, or C<fh>, above.
83otherwise you might end up with a closed socket while you are still
84waiting for data.
85 106
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
86=item on_error => $cb->($self) 116=item on_prepare => $cb->($handle)
87 117
118This (rarely used) callback is called before a new connection is
119attempted, but after the file handle has been created (you can access that
120file handle via C<< $handle->{fh} >>). It could be used to prepare the
121file handle with parameters required for the actual connect (as opposed to
122settings that can be changed when the connection is already established).
123
124The return value of this callback should be the connect timeout value in
125seconds (or C<0>, or C<undef>, or the empty list, to indicate that the
126default timeout is to be used).
127
128=item on_connect => $cb->($handle, $host, $port, $retry->())
129
130This callback is called when a connection has been successfully established.
131
132The peer's numeric host and port (the socket peername) are passed as
133parameters, together with a retry callback. At the time it is called the
134read and write queues, EOF status, TLS status and similar properties of
135the handle will have been reset.
136
137It is not allowed to use the read or write queues while the handle object
138is connecting.
139
140If, for some reason, the handle is not acceptable, calling C<$retry> will
141continue with the next connection target (in case of multi-homed hosts or
142SRV records there can be multiple connection endpoints). The C<$retry>
143callback can be invoked after the connect callback returns, i.e. one can
144start a handshake and then decide to retry with the next host if the
145handshake fails.
146
147In most cases, you should ignore the C<$retry> parameter.
148
149=item on_connect_error => $cb->($handle, $message)
150
151This callback is called when the connection could not be
152established. C<$!> will contain the relevant error code, and C<$message> a
153message describing it (usually the same as C<"$!">).
154
155If this callback isn't specified, then C<on_error> will be called with a
156fatal error instead.
157
158=back
159
160=item on_error => $cb->($handle, $fatal, $message)
161
88This is the fatal error callback, that is called when, well, a fatal error 162This is the error callback, which is called when, well, some error
89occurs, such as not being able to resolve the hostname, failure to connect 163occured, such as not being able to resolve the hostname, failure to
90or a read error. 164connect, or a read error.
91 165
92The object will not be in a usable state when this callback has been 166Some errors are fatal (which is indicated by C<$fatal> being true). On
93called. 167fatal errors the handle object will be destroyed (by a call to C<< ->
168destroy >>) after invoking the error callback (which means you are free to
169examine the handle object). Examples of fatal errors are an EOF condition
170with active (but unsatisfiable) read watchers (C<EPIPE>) or I/O errors. In
171cases where the other side can close the connection at will, it is
172often easiest to not report C<EPIPE> errors in this callback.
94 173
174AnyEvent::Handle tries to find an appropriate error code for you to check
175against, but in some cases (TLS errors), this does not work well.
176
177If you report the error to the user, it is recommended to always output
178the C<$message> argument in human-readable error messages (you don't need
179to report C<"$!"> if you report C<$message>).
180
181If you want to react programmatically to the error, then looking at C<$!>
182and comparing it against some of the documented C<Errno> values is usually
183better than looking at the C<$message>.
184
185Non-fatal errors can be retried by returning, but it is recommended
186to simply ignore this parameter and instead abondon the handle object
187when this callback is invoked. Examples of non-fatal errors are timeouts
188C<ETIMEDOUT>) or badly-formatted data (C<EBADMSG>).
189
95On callback entrance, the value of C<$!> contains the operating system 190On entry to the callback, the value of C<$!> contains the operating
96error (or C<ENOSPC>, C<EPIPE> or C<EBADMSG>). 191system error code (or C<ENOSPC>, C<EPIPE>, C<ETIMEDOUT>, C<EBADMSG> or
192C<EPROTO>).
97 193
98While not mandatory, it is I<highly> recommended to set this callback, as 194While not mandatory, it is I<highly> recommended to set this callback, as
99you will not be notified of errors otherwise. The default simply calls 195you will not be notified of errors otherwise. The default just calls
100die. 196C<croak>.
101 197
102=item on_read => $cb->($self) 198=item on_read => $cb->($handle)
103 199
104This sets the default read callback, which is called when data arrives 200This sets the default read callback, which is called when data arrives
105and no read request is in the queue. 201and no read request is in the queue (unlike read queue callbacks, this
202callback will only be called when at least one octet of data is in the
203read buffer).
106 204
107To access (and remove data from) the read buffer, use the C<< ->rbuf >> 205To access (and remove data from) the read buffer, use the C<< ->rbuf >>
108method or access the C<$self->{rbuf}> member directly. 206method or access the C<< $handle->{rbuf} >> member directly. Note that you
207must not enlarge or modify the read buffer, you can only remove data at
208the beginning from it.
109 209
210You can also call C<< ->push_read (...) >> or any other function that
211modifies the read queue. Or do both. Or ...
212
110When an EOF condition is detected then AnyEvent::Handle will first try to 213When an EOF condition is detected, AnyEvent::Handle will first try to
111feed all the remaining data to the queued callbacks and C<on_read> before 214feed all the remaining data to the queued callbacks and C<on_read> before
112calling the C<on_eof> callback. If no progress can be made, then a fatal 215calling the C<on_eof> callback. If no progress can be made, then a fatal
113error will be raised (with C<$!> set to C<EPIPE>). 216error will be raised (with C<$!> set to C<EPIPE>).
114 217
218Note that, unlike requests in the read queue, an C<on_read> callback
219doesn't mean you I<require> some data: if there is an EOF and there
220are outstanding read requests then an error will be flagged. With an
221C<on_read> callback, the C<on_eof> callback will be invoked.
222
223=item on_eof => $cb->($handle)
224
225Set the callback to be called when an end-of-file condition is detected,
226i.e. in the case of a socket, when the other side has closed the
227connection cleanly, and there are no outstanding read requests in the
228queue (if there are read requests, then an EOF counts as an unexpected
229connection close and will be flagged as an error).
230
231For sockets, this just means that the other side has stopped sending data,
232you can still try to write data, and, in fact, one can return from the EOF
233callback and continue writing data, as only the read part has been shut
234down.
235
236If an EOF condition has been detected but no C<on_eof> callback has been
237set, then a fatal error will be raised with C<$!> set to <0>.
238
115=item on_drain => $cb->() 239=item on_drain => $cb->($handle)
116 240
117This sets the callback that is called when the write buffer becomes empty 241This sets the callback that is called once when the write buffer becomes
118(or when the callback is set and the buffer is empty already). 242empty (and immediately when the handle object is created).
119 243
120To append to the write buffer, use the C<< ->push_write >> method. 244To append to the write buffer, use the C<< ->push_write >> method.
245
246This callback is useful when you don't want to put all of your write data
247into the queue at once, for example, when you want to write the contents
248of some file to the socket you might not want to read the whole file into
249memory and push it into the queue, but instead only read more data from
250the file when the write queue becomes empty.
251
252=item timeout => $fractional_seconds
253
254=item rtimeout => $fractional_seconds
255
256=item wtimeout => $fractional_seconds
257
258If non-zero, then these enables an "inactivity" timeout: whenever this
259many seconds pass without a successful read or write on the underlying
260file handle (or a call to C<timeout_reset>), the C<on_timeout> callback
261will be invoked (and if that one is missing, a non-fatal C<ETIMEDOUT>
262error will be raised).
263
264There are three variants of the timeouts that work independently of each
265other, for both read and write (triggered when nothing was read I<OR>
266written), just read (triggered when nothing was read), and just write:
267C<timeout>, C<rtimeout> and C<wtimeout>, with corresponding callbacks
268C<on_timeout>, C<on_rtimeout> and C<on_wtimeout>, and reset functions
269C<timeout_reset>, C<rtimeout_reset>, and C<wtimeout_reset>.
270
271Note that timeout processing is active even when you do not have any
272outstanding read or write requests: If you plan to keep the connection
273idle then you should disable the timeout temporarily or ignore the
274timeout in the corresponding C<on_timeout> callback, in which case
275AnyEvent::Handle will simply restart the timeout.
276
277Zero (the default) disables the corresponding timeout.
278
279=item on_timeout => $cb->($handle)
280
281=item on_rtimeout => $cb->($handle)
282
283=item on_wtimeout => $cb->($handle)
284
285Called whenever the inactivity timeout passes. If you return from this
286callback, then the timeout will be reset as if some activity had happened,
287so this condition is not fatal in any way.
121 288
122=item rbuf_max => <bytes> 289=item rbuf_max => <bytes>
123 290
124If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>) 291If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
125when the read buffer ever (strictly) exceeds this size. This is useful to 292when the read buffer ever (strictly) exceeds this size. This is useful to
126avoid denial-of-service attacks. 293avoid some forms of denial-of-service attacks.
127 294
128For example, a server accepting connections from untrusted sources should 295For example, a server accepting connections from untrusted sources should
129be configured to accept only so-and-so much data that it cannot act on 296be configured to accept only so-and-so much data that it cannot act on
130(for example, when expecting a line, an attacker could send an unlimited 297(for example, when expecting a line, an attacker could send an unlimited
131amount of data without a callback ever being called as long as the line 298amount of data without a callback ever being called as long as the line
132isn't finished). 299isn't finished).
133 300
301=item wbuf_max => <bytes>
302
303If defined, then a fatal error will be raised (with C<$!> set to C<ENOSPC>)
304when the write buffer ever (strictly) exceeds this size. This is useful to
305avoid some forms of denial-of-service attacks.
306
307Although the units of this parameter is bytes, this is the I<raw> number
308of bytes not yet accepted by the kernel. This can make a difference when
309you e.g. use TLS, as TLS typically makes your write data larger (but it
310can also make it smaller due to compression).
311
312As an example of when this limit is useful, take a chat server that sends
313chat messages to a client. If the client does not read those in a timely
314manner then the send buffer in the server would grow unbounded.
315
316=item autocork => <boolean>
317
318When disabled (the default), C<push_write> will try to immediately
319write the data to the handle if possible. This avoids having to register
320a write watcher and wait for the next event loop iteration, but can
321be inefficient if you write multiple small chunks (on the wire, this
322disadvantage is usually avoided by your kernel's nagle algorithm, see
323C<no_delay>, but this option can save costly syscalls).
324
325When enabled, writes will always be queued till the next event loop
326iteration. This is efficient when you do many small writes per iteration,
327but less efficient when you do a single write only per iteration (or when
328the write buffer often is full). It also increases write latency.
329
330=item no_delay => <boolean>
331
332When doing small writes on sockets, your operating system kernel might
333wait a bit for more data before actually sending it out. This is called
334the Nagle algorithm, and usually it is beneficial.
335
336In some situations you want as low a delay as possible, which can be
337accomplishd by setting this option to a true value.
338
339The default is your operating system's default behaviour (most likely
340enabled). This option explicitly enables or disables it, if possible.
341
342=item keepalive => <boolean>
343
344Enables (default disable) the SO_KEEPALIVE option on the stream socket:
345normally, TCP connections have no time-out once established, so TCP
346connections, once established, can stay alive forever even when the other
347side has long gone. TCP keepalives are a cheap way to take down long-lived
348TCP connections when the other side becomes unreachable. While the default
349is OS-dependent, TCP keepalives usually kick in after around two hours,
350and, if the other side doesn't reply, take down the TCP connection some 10
351to 15 minutes later.
352
353It is harmless to specify this option for file handles that do not support
354keepalives, and enabling it on connections that are potentially long-lived
355is usually a good idea.
356
357=item oobinline => <boolean>
358
359BSD majorly fucked up the implementation of TCP urgent data. The result
360is that almost no OS implements TCP according to the specs, and every OS
361implements it slightly differently.
362
363If you want to handle TCP urgent data, then setting this flag (the default
364is enabled) gives you the most portable way of getting urgent data, by
365putting it into the stream.
366
367Since BSD emulation of OOB data on top of TCP's urgent data can have
368security implications, AnyEvent::Handle sets this flag automatically
369unless explicitly specified. Note that setting this flag after
370establishing a connection I<may> be a bit too late (data loss could
371already have occured on BSD systems), but at least it will protect you
372from most attacks.
373
134=item read_size => <bytes> 374=item read_size => <bytes>
135 375
136The default read block size (the amount of bytes this module will try to read 376The initial read block size, the number of bytes this module will try
137on each [loop iteration). Default: C<4096>. 377to read during each loop iteration. Each handle object will consume
378at least this amount of memory for the read buffer as well, so when
379handling many connections watch out for memory requirements). See also
380C<max_read_size>. Default: C<2048>.
381
382=item max_read_size => <bytes>
383
384The maximum read buffer size used by the dynamic adjustment
385algorithm: Each time AnyEvent::Handle can read C<read_size> bytes in
386one go it will double C<read_size> up to the maximum given by this
387option. Default: C<131072> or C<read_size>, whichever is higher.
138 388
139=item low_water_mark => <bytes> 389=item low_water_mark => <bytes>
140 390
141Sets the amount of bytes (default: C<0>) that make up an "empty" write 391Sets the number of bytes (default: C<0>) that make up an "empty" write
142buffer: If the write reaches this size or gets even samller it is 392buffer: If the buffer reaches this size or gets even samller it is
143considered empty. 393considered empty.
144 394
395Sometimes it can be beneficial (for performance reasons) to add data to
396the write buffer before it is fully drained, but this is a rare case, as
397the operating system kernel usually buffers data as well, so the default
398is good in almost all cases.
399
400=item linger => <seconds>
401
402If this is non-zero (default: C<3600>), the destructor of the
403AnyEvent::Handle object will check whether there is still outstanding
404write data and will install a watcher that will write this data to the
405socket. No errors will be reported (this mostly matches how the operating
406system treats outstanding data at socket close time).
407
408This will not work for partial TLS data that could not be encoded
409yet. This data will be lost. Calling the C<stoptls> method in time might
410help.
411
412=item peername => $string
413
414A string used to identify the remote site - usually the DNS hostname
415(I<not> IDN!) used to create the connection, rarely the IP address.
416
417Apart from being useful in error messages, this string is also used in TLS
418peername verification (see C<verify_peername> in L<AnyEvent::TLS>). This
419verification will be skipped when C<peername> is not specified or is
420C<undef>.
421
145=item tls => "accept" | "connect" | Net::SSLeay::SSL object 422=item tls => "accept" | "connect" | Net::SSLeay::SSL object
146 423
147When this parameter is given, it enables TLS (SSL) mode, that means it 424When this parameter is given, it enables TLS (SSL) mode, that means
148will start making tls handshake and will transparently encrypt/decrypt 425AnyEvent will start a TLS handshake as soon as the connection has been
149data. 426established and will transparently encrypt/decrypt data afterwards.
427
428All TLS protocol errors will be signalled as C<EPROTO>, with an
429appropriate error message.
150 430
151TLS mode requires Net::SSLeay to be installed (it will be loaded 431TLS mode requires Net::SSLeay to be installed (it will be loaded
152automatically when you try to create a TLS handle). 432automatically when you try to create a TLS handle): this module doesn't
433have a dependency on that module, so if your module requires it, you have
434to add the dependency yourself. If Net::SSLeay cannot be loaded or is too
435old, you get an C<EPROTO> error.
153 436
154For the TLS server side, use C<accept>, and for the TLS client side of a 437Unlike TCP, TLS has a server and client side: for the TLS server side, use
155connection, use C<connect> mode. 438C<accept>, and for the TLS client side of a connection, use C<connect>
439mode.
156 440
157You can also provide your own TLS connection object, but you have 441You can also provide your own TLS connection object, but you have
158to make sure that you call either C<Net::SSLeay::set_connect_state> 442to make sure that you call either C<Net::SSLeay::set_connect_state>
159or C<Net::SSLeay::set_accept_state> on it before you pass it to 443or C<Net::SSLeay::set_accept_state> on it before you pass it to
160AnyEvent::Handle. 444AnyEvent::Handle. Also, this module will take ownership of this connection
445object.
161 446
447At some future point, AnyEvent::Handle might switch to another TLS
448implementation, then the option to use your own session object will go
449away.
450
451B<IMPORTANT:> since Net::SSLeay "objects" are really only integers,
452passing in the wrong integer will lead to certain crash. This most often
453happens when one uses a stylish C<< tls => 1 >> and is surprised about the
454segmentation fault.
455
162See the C<starttls> method if you need to start TLs negotiation later. 456Use the C<< ->starttls >> method if you need to start TLS negotiation later.
163 457
164=item tls_ctx => $ssl_ctx 458=item tls_ctx => $anyevent_tls
165 459
166Use the given Net::SSLeay::CTX object to create the new TLS connection 460Use the given C<AnyEvent::TLS> object to create the new TLS connection
167(unless a connection object was specified directly). If this parameter is 461(unless a connection object was specified directly). If this
168missing, then AnyEvent::Handle will use C<AnyEvent::Handle::TLS_CTX>. 462parameter is missing (or C<undef>), then AnyEvent::Handle will use
463C<AnyEvent::Handle::TLS_CTX>.
464
465Instead of an object, you can also specify a hash reference with C<< key
466=> value >> pairs. Those will be passed to L<AnyEvent::TLS> to create a
467new TLS context object.
468
469=item on_starttls => $cb->($handle, $success[, $error_message])
470
471This callback will be invoked when the TLS/SSL handshake has finished. If
472C<$success> is true, then the TLS handshake succeeded, otherwise it failed
473(C<on_stoptls> will not be called in this case).
474
475The session in C<< $handle->{tls} >> can still be examined in this
476callback, even when the handshake was not successful.
477
478TLS handshake failures will not cause C<on_error> to be invoked when this
479callback is in effect, instead, the error message will be passed to C<on_starttls>.
480
481Without this callback, handshake failures lead to C<on_error> being
482called as usual.
483
484Note that you cannot just call C<starttls> again in this callback. If you
485need to do that, start an zero-second timer instead whose callback can
486then call C<< ->starttls >> again.
487
488=item on_stoptls => $cb->($handle)
489
490When a SSLv3/TLS shutdown/close notify/EOF is detected and this callback is
491set, then it will be invoked after freeing the TLS session. If it is not,
492then a TLS shutdown condition will be treated like a normal EOF condition
493on the handle.
494
495The session in C<< $handle->{tls} >> can still be examined in this
496callback.
497
498This callback will only be called on TLS shutdowns, not when the
499underlying handle signals EOF.
500
501=item json => JSON or JSON::XS object
502
503This is the json coder object used by the C<json> read and write types.
504
505If you don't supply it, then AnyEvent::Handle will create and use a
506suitable one (on demand), which will write and expect UTF-8 encoded JSON
507texts.
508
509Note that you are responsible to depend on the JSON module if you want to
510use this functionality, as AnyEvent does not have a dependency itself.
169 511
170=back 512=back
171 513
172=cut 514=cut
173 515
174sub new { 516sub new {
175 my $class = shift; 517 my $class = shift;
176
177 my $self = bless { @_ }, $class; 518 my $self = bless { @_ }, $class;
178 519
179 $self->{fh} or Carp::croak "mandatory argument fh is missing"; 520 if ($self->{fh}) {
521 $self->_start;
522 return unless $self->{fh}; # could be gone by now
523
524 } elsif ($self->{connect}) {
525 require AnyEvent::Socket;
526
527 $self->{peername} = $self->{connect}[0]
528 unless exists $self->{peername};
529
530 $self->{_skip_drain_rbuf} = 1;
531
532 {
533 Scalar::Util::weaken (my $self = $self);
534
535 $self->{_connect} =
536 AnyEvent::Socket::tcp_connect (
537 $self->{connect}[0],
538 $self->{connect}[1],
539 sub {
540 my ($fh, $host, $port, $retry) = @_;
541
542 delete $self->{_connect}; # no longer needed
543
544 if ($fh) {
545 $self->{fh} = $fh;
546
547 delete $self->{_skip_drain_rbuf};
548 $self->_start;
549
550 $self->{on_connect}
551 and $self->{on_connect}($self, $host, $port, sub {
552 delete @$self{qw(fh _tw _rtw _wtw _ww _rw _eof _queue rbuf _wbuf tls _tls_rbuf _tls_wbuf)};
553 $self->{_skip_drain_rbuf} = 1;
554 &$retry;
555 });
556
557 } else {
558 if ($self->{on_connect_error}) {
559 $self->{on_connect_error}($self, "$!");
560 $self->destroy if $self;
561 } else {
562 $self->_error ($!, 1);
563 }
564 }
565 },
566 sub {
567 local $self->{fh} = $_[0];
568
569 $self->{on_prepare}
570 ? $self->{on_prepare}->($self)
571 : ()
572 }
573 );
574 }
575
576 } else {
577 Carp::croak "AnyEvent::Handle: either an existing fh or the connect parameter must be specified";
578 }
579
580 $self
581}
582
583sub _start {
584 my ($self) = @_;
585
586 # too many clueless people try to use udp and similar sockets
587 # with AnyEvent::Handle, do them a favour.
588 my $type = getsockopt $self->{fh}, Socket::SOL_SOCKET (), Socket::SO_TYPE ();
589 Carp::croak "AnyEvent::Handle: only stream sockets supported, anything else will NOT work!"
590 if Socket::SOCK_STREAM () != (unpack "I", $type) && defined $type;
180 591
181 AnyEvent::Util::fh_nonblocking $self->{fh}, 1; 592 AnyEvent::Util::fh_nonblocking $self->{fh}, 1;
182 593
183 if ($self->{tls}) { 594 $self->{_activity} =
184 require Net::SSLeay; 595 $self->{_ractivity} =
596 $self->{_wactivity} = AE::now;
597
598 $self->{read_size} ||= 2048;
599 $self->{max_read_size} = $self->{read_size}
600 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
601
602 $self->timeout (delete $self->{timeout} ) if $self->{timeout};
603 $self->rtimeout (delete $self->{rtimeout} ) if $self->{rtimeout};
604 $self->wtimeout (delete $self->{wtimeout} ) if $self->{wtimeout};
605
606 $self->no_delay (delete $self->{no_delay} ) if exists $self->{no_delay} && $self->{no_delay};
607 $self->keepalive (delete $self->{keepalive}) if exists $self->{keepalive} && $self->{keepalive};
608
609 $self->oobinline (exists $self->{oobinline} ? delete $self->{oobinline} : 1);
610
185 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx}); 611 $self->starttls (delete $self->{tls}, delete $self->{tls_ctx})
186 } 612 if $self->{tls};
187 613
188 $self->on_eof (delete $self->{on_eof} ) if $self->{on_eof};
189 $self->on_error (delete $self->{on_error}) if $self->{on_error};
190 $self->on_drain (delete $self->{on_drain}) if $self->{on_drain}; 614 $self->on_drain (delete $self->{on_drain} ) if $self->{on_drain};
191 $self->on_read (delete $self->{on_read} ) if $self->{on_read};
192 615
193 $self->start_read; 616 $self->start_read
617 if $self->{on_read} || @{ $self->{_queue} };
194 618
195 $self 619 $self->_drain_wbuf;
196} 620}
197 621
198sub _shutdown {
199 my ($self) = @_;
200
201 delete $self->{rw};
202 delete $self->{ww};
203 delete $self->{fh};
204}
205
206sub error { 622sub _error {
207 my ($self) = @_; 623 my ($self, $errno, $fatal, $message) = @_;
208 624
209 { 625 $! = $errno;
210 local $!; 626 $message ||= "$!";
211 $self->_shutdown;
212 }
213 627
214 if ($self->{on_error}) { 628 if ($self->{on_error}) {
215 $self->{on_error}($self); 629 $self->{on_error}($self, $fatal, $message);
216 } else { 630 $self->destroy if $fatal;
631 } elsif ($self->{fh} || $self->{connect}) {
632 $self->destroy;
217 Carp::croak "AnyEvent::Handle uncaught fatal error: $!"; 633 Carp::croak "AnyEvent::Handle uncaught error: $message";
218 } 634 }
219} 635}
220 636
221=item $fh = $handle->fh 637=item $fh = $handle->fh
222 638
223This method returns the file handle of the L<AnyEvent::Handle> object. 639This method returns the file handle used to create the L<AnyEvent::Handle> object.
224 640
225=cut 641=cut
226 642
227sub fh { $_[0]->{fh} } 643sub fh { $_[0]{fh} }
228 644
229=item $handle->on_error ($cb) 645=item $handle->on_error ($cb)
230 646
231Replace the current C<on_error> callback (see the C<on_error> constructor argument). 647Replace the current C<on_error> callback (see the C<on_error> constructor argument).
232 648
244 660
245sub on_eof { 661sub on_eof {
246 $_[0]{on_eof} = $_[1]; 662 $_[0]{on_eof} = $_[1];
247} 663}
248 664
665=item $handle->on_timeout ($cb)
666
667=item $handle->on_rtimeout ($cb)
668
669=item $handle->on_wtimeout ($cb)
670
671Replace the current C<on_timeout>, C<on_rtimeout> or C<on_wtimeout>
672callback, or disables the callback (but not the timeout) if C<$cb> =
673C<undef>. See the C<timeout> constructor argument and method.
674
675=cut
676
677# see below
678
679=item $handle->autocork ($boolean)
680
681Enables or disables the current autocork behaviour (see C<autocork>
682constructor argument). Changes will only take effect on the next write.
683
684=cut
685
686sub autocork {
687 $_[0]{autocork} = $_[1];
688}
689
690=item $handle->no_delay ($boolean)
691
692Enables or disables the C<no_delay> setting (see constructor argument of
693the same name for details).
694
695=cut
696
697sub no_delay {
698 $_[0]{no_delay} = $_[1];
699
700 setsockopt $_[0]{fh}, Socket::IPPROTO_TCP (), Socket::TCP_NODELAY (), int $_[1]
701 if $_[0]{fh};
702}
703
704=item $handle->keepalive ($boolean)
705
706Enables or disables the C<keepalive> setting (see constructor argument of
707the same name for details).
708
709=cut
710
711sub keepalive {
712 $_[0]{keepalive} = $_[1];
713
714 eval {
715 local $SIG{__DIE__};
716 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
717 if $_[0]{fh};
718 };
719}
720
721=item $handle->oobinline ($boolean)
722
723Enables or disables the C<oobinline> setting (see constructor argument of
724the same name for details).
725
726=cut
727
728sub oobinline {
729 $_[0]{oobinline} = $_[1];
730
731 eval {
732 local $SIG{__DIE__};
733 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_OOBINLINE (), int $_[1]
734 if $_[0]{fh};
735 };
736}
737
738=item $handle->keepalive ($boolean)
739
740Enables or disables the C<keepalive> setting (see constructor argument of
741the same name for details).
742
743=cut
744
745sub keepalive {
746 $_[0]{keepalive} = $_[1];
747
748 eval {
749 local $SIG{__DIE__};
750 setsockopt $_[0]{fh}, Socket::SOL_SOCKET (), Socket::SO_KEEPALIVE (), int $_[1]
751 if $_[0]{fh};
752 };
753}
754
755=item $handle->on_starttls ($cb)
756
757Replace the current C<on_starttls> callback (see the C<on_starttls> constructor argument).
758
759=cut
760
761sub on_starttls {
762 $_[0]{on_starttls} = $_[1];
763}
764
765=item $handle->on_stoptls ($cb)
766
767Replace the current C<on_stoptls> callback (see the C<on_stoptls> constructor argument).
768
769=cut
770
771sub on_stoptls {
772 $_[0]{on_stoptls} = $_[1];
773}
774
775=item $handle->rbuf_max ($max_octets)
776
777Configures the C<rbuf_max> setting (C<undef> disables it).
778
779=item $handle->wbuf_max ($max_octets)
780
781Configures the C<wbuf_max> setting (C<undef> disables it).
782
783=cut
784
785sub rbuf_max {
786 $_[0]{rbuf_max} = $_[1];
787}
788
789sub wbuf_max {
790 $_[0]{wbuf_max} = $_[1];
791}
792
793#############################################################################
794
795=item $handle->timeout ($seconds)
796
797=item $handle->rtimeout ($seconds)
798
799=item $handle->wtimeout ($seconds)
800
801Configures (or disables) the inactivity timeout.
802
803The timeout will be checked instantly, so this method might destroy the
804handle before it returns.
805
806=item $handle->timeout_reset
807
808=item $handle->rtimeout_reset
809
810=item $handle->wtimeout_reset
811
812Reset the activity timeout, as if data was received or sent.
813
814These methods are cheap to call.
815
816=cut
817
818for my $dir ("", "r", "w") {
819 my $timeout = "${dir}timeout";
820 my $tw = "_${dir}tw";
821 my $on_timeout = "on_${dir}timeout";
822 my $activity = "_${dir}activity";
823 my $cb;
824
825 *$on_timeout = sub {
826 $_[0]{$on_timeout} = $_[1];
827 };
828
829 *$timeout = sub {
830 my ($self, $new_value) = @_;
831
832 $new_value >= 0
833 or Carp::croak "AnyEvent::Handle->$timeout called with negative timeout ($new_value), caught";
834
835 $self->{$timeout} = $new_value;
836 delete $self->{$tw}; &$cb;
837 };
838
839 *{"${dir}timeout_reset"} = sub {
840 $_[0]{$activity} = AE::now;
841 };
842
843 # main workhorse:
844 # reset the timeout watcher, as neccessary
845 # also check for time-outs
846 $cb = sub {
847 my ($self) = @_;
848
849 if ($self->{$timeout} && $self->{fh}) {
850 my $NOW = AE::now;
851
852 # when would the timeout trigger?
853 my $after = $self->{$activity} + $self->{$timeout} - $NOW;
854
855 # now or in the past already?
856 if ($after <= 0) {
857 $self->{$activity} = $NOW;
858
859 if ($self->{$on_timeout}) {
860 $self->{$on_timeout}($self);
861 } else {
862 $self->_error (Errno::ETIMEDOUT);
863 }
864
865 # callback could have changed timeout value, optimise
866 return unless $self->{$timeout};
867
868 # calculate new after
869 $after = $self->{$timeout};
870 }
871
872 Scalar::Util::weaken $self;
873 return unless $self; # ->error could have destroyed $self
874
875 $self->{$tw} ||= AE::timer $after, 0, sub {
876 delete $self->{$tw};
877 $cb->($self);
878 };
879 } else {
880 delete $self->{$tw};
881 }
882 }
883}
884
249############################################################################# 885#############################################################################
250 886
251=back 887=back
252 888
253=head2 WRITE QUEUE 889=head2 WRITE QUEUE
257 893
258The write queue is very simple: you can add data to its end, and 894The write queue is very simple: you can add data to its end, and
259AnyEvent::Handle will automatically try to get rid of it for you. 895AnyEvent::Handle will automatically try to get rid of it for you.
260 896
261When data could be written and the write buffer is shorter then the low 897When data could be written and the write buffer is shorter then the low
262water mark, the C<on_drain> callback will be invoked. 898water mark, the C<on_drain> callback will be invoked once.
263 899
264=over 4 900=over 4
265 901
266=item $handle->on_drain ($cb) 902=item $handle->on_drain ($cb)
267 903
268Sets the C<on_drain> callback or clears it (see the description of 904Sets the C<on_drain> callback or clears it (see the description of
269C<on_drain> in the constructor). 905C<on_drain> in the constructor).
270 906
907This method may invoke callbacks (and therefore the handle might be
908destroyed after it returns).
909
271=cut 910=cut
272 911
273sub on_drain { 912sub on_drain {
274 my ($self, $cb) = @_; 913 my ($self, $cb) = @_;
275 914
276 $self->{on_drain} = $cb; 915 $self->{on_drain} = $cb;
277 916
278 $cb->($self) 917 $cb->($self)
279 if $cb && $self->{low_water_mark} >= length $self->{wbuf}; 918 if $cb && $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf});
280} 919}
281 920
282=item $handle->push_write ($data) 921=item $handle->push_write ($data)
283 922
284Queues the given scalar to be written. You can push as much data as you 923Queues the given scalar to be written. You can push as much data as
285want (only limited by the available memory), as C<AnyEvent::Handle> 924you want (only limited by the available memory and C<wbuf_max>), as
286buffers it independently of the kernel. 925C<AnyEvent::Handle> buffers it independently of the kernel.
926
927This method may invoke callbacks (and therefore the handle might be
928destroyed after it returns).
287 929
288=cut 930=cut
289 931
290sub _drain_wbuf { 932sub _drain_wbuf {
291 my ($self) = @_; 933 my ($self) = @_;
292 934
293 if (!$self->{ww} && length $self->{wbuf}) { 935 if (!$self->{_ww} && length $self->{wbuf}) {
936
294 Scalar::Util::weaken $self; 937 Scalar::Util::weaken $self;
938
295 my $cb = sub { 939 my $cb = sub {
296 my $len = syswrite $self->{fh}, $self->{wbuf}; 940 my $len = syswrite $self->{fh}, $self->{wbuf};
297 941
298 if ($len >= 0) { 942 if (defined $len) {
299 substr $self->{wbuf}, 0, $len, ""; 943 substr $self->{wbuf}, 0, $len, "";
300 944
945 $self->{_activity} = $self->{_wactivity} = AE::now;
946
301 $self->{on_drain}($self) 947 $self->{on_drain}($self)
302 if $self->{low_water_mark} >= length $self->{wbuf} 948 if $self->{low_water_mark} >= (length $self->{wbuf}) + (length $self->{_tls_wbuf})
303 && $self->{on_drain}; 949 && $self->{on_drain};
304 950
305 delete $self->{ww} unless length $self->{wbuf}; 951 delete $self->{_ww} unless length $self->{wbuf};
306 } elsif ($! != EAGAIN && $! != EINTR) { 952 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
307 $self->error; 953 $self->_error ($!, 1);
308 } 954 }
309 }; 955 };
310 956
311 $self->{ww} = AnyEvent->io (fh => $self->{fh}, poll => "w", cb => $cb); 957 # try to write data immediately
958 $cb->() unless $self->{autocork};
312 959
313 $cb->($self); 960 # if still data left in wbuf, we need to poll
961 $self->{_ww} = AE::io $self->{fh}, 1, $cb
962 if length $self->{wbuf};
963
964 if (
965 defined $self->{wbuf_max}
966 && $self->{wbuf_max} < length $self->{wbuf}
967 ) {
968 $self->_error (Errno::ENOSPC, 1), return;
969 }
314 }; 970 };
315} 971}
316 972
317our %WH; 973our %WH;
318 974
975# deprecated
319sub register_write_type($$) { 976sub register_write_type($$) {
320 $WH{$_[0]} = $_[1]; 977 $WH{$_[0]} = $_[1];
321} 978}
322 979
323sub push_write { 980sub push_write {
324 my $self = shift; 981 my $self = shift;
325 982
326 if (@_ > 1) { 983 if (@_ > 1) {
327 my $type = shift; 984 my $type = shift;
328 985
986 @_ = ($WH{$type} ||= _load_func "$type\::anyevent_write_type"
329 @_ = ($WH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::push_write") 987 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_write")
330 ->($self, @_); 988 ->($self, @_);
331 } 989 }
332 990
991 # we downgrade here to avoid hard-to-track-down bugs,
992 # and diagnose the problem earlier and better.
993
333 if ($self->{filter_w}) { 994 if ($self->{tls}) {
334 $self->{filter_w}->($self, \$_[0]); 995 utf8::downgrade $self->{_tls_wbuf} .= $_[0];
996 &_dotls ($self) if $self->{fh};
335 } else { 997 } else {
336 $self->{wbuf} .= $_[0]; 998 utf8::downgrade $self->{wbuf} .= $_[0];
337 $self->_drain_wbuf; 999 $self->_drain_wbuf if $self->{fh};
338 } 1000 }
339} 1001}
340 1002
341=item $handle->push_write (type => @args) 1003=item $handle->push_write (type => @args)
342 1004
343=item $handle->unshift_write (type => @args)
344
345Instead of formatting your data yourself, you can also let this module do 1005Instead of formatting your data yourself, you can also let this module
346the job by specifying a type and type-specific arguments. 1006do the job by specifying a type and type-specific arguments. You
1007can also specify the (fully qualified) name of a package, in which
1008case AnyEvent tries to load the package and then expects to find the
1009C<anyevent_write_type> function inside (see "custom write types", below).
347 1010
348Predefined types are (if you have ideas for additional types, feel free to 1011Predefined types are (if you have ideas for additional types, feel free to
349drop by and tell us): 1012drop by and tell us):
350 1013
351=over 4 1014=over 4
353=item netstring => $string 1016=item netstring => $string
354 1017
355Formats the given value as netstring 1018Formats the given value as netstring
356(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them). 1019(http://cr.yp.to/proto/netstrings.txt, this is not a recommendation to use them).
357 1020
358=back
359
360=cut 1021=cut
361 1022
362register_write_type netstring => sub { 1023register_write_type netstring => sub {
363 my ($self, $string) = @_; 1024 my ($self, $string) = @_;
364 1025
365 sprintf "%d:%s,", (length $string), $string 1026 (length $string) . ":$string,"
366}; 1027};
367 1028
368=item AnyEvent::Handle::register_write_type type => $coderef->($self, @args) 1029=item packstring => $format, $data
369 1030
370This function (not method) lets you add your own types to C<push_write>. 1031An octet string prefixed with an encoded length. The encoding C<$format>
1032uses the same format as a Perl C<pack> format, but must specify a single
1033integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1034optional C<!>, C<< < >> or C<< > >> modifier).
1035
1036=cut
1037
1038register_write_type packstring => sub {
1039 my ($self, $format, $string) = @_;
1040
1041 pack "$format/a*", $string
1042};
1043
1044=item json => $array_or_hashref
1045
1046Encodes the given hash or array reference into a JSON object. Unless you
1047provide your own JSON object, this means it will be encoded to JSON text
1048in UTF-8.
1049
1050JSON objects (and arrays) are self-delimiting, so you can write JSON at
1051one end of a handle and read them at the other end without using any
1052additional framing.
1053
1054The generated JSON text is guaranteed not to contain any newlines: While
1055this module doesn't need delimiters after or between JSON texts to be
1056able to read them, many other languages depend on that.
1057
1058A simple RPC protocol that interoperates easily with others is to send
1059JSON arrays (or objects, although arrays are usually the better choice as
1060they mimic how function argument passing works) and a newline after each
1061JSON text:
1062
1063 $handle->push_write (json => ["method", "arg1", "arg2"]); # whatever
1064 $handle->push_write ("\012");
1065
1066An AnyEvent::Handle receiver would simply use the C<json> read type and
1067rely on the fact that the newline will be skipped as leading whitespace:
1068
1069 $handle->push_read (json => sub { my $array = $_[1]; ... });
1070
1071Other languages could read single lines terminated by a newline and pass
1072this line into their JSON decoder of choice.
1073
1074=cut
1075
1076sub json_coder() {
1077 eval { require JSON::XS; JSON::XS->new->utf8 }
1078 || do { require JSON; JSON->new->utf8 }
1079}
1080
1081register_write_type json => sub {
1082 my ($self, $ref) = @_;
1083
1084 my $json = $self->{json} ||= json_coder;
1085
1086 $json->encode ($ref)
1087};
1088
1089=item storable => $reference
1090
1091Freezes the given reference using L<Storable> and writes it to the
1092handle. Uses the C<nfreeze> format.
1093
1094=cut
1095
1096register_write_type storable => sub {
1097 my ($self, $ref) = @_;
1098
1099 require Storable unless $Storable::VERSION;
1100
1101 pack "w/a*", Storable::nfreeze ($ref)
1102};
1103
1104=back
1105
1106=item $handle->push_shutdown
1107
1108Sometimes you know you want to close the socket after writing your data
1109before it was actually written. One way to do that is to replace your
1110C<on_drain> handler by a callback that shuts down the socket (and set
1111C<low_water_mark> to C<0>). This method is a shorthand for just that, and
1112replaces the C<on_drain> callback with:
1113
1114 sub { shutdown $_[0]{fh}, 1 }
1115
1116This simply shuts down the write side and signals an EOF condition to the
1117the peer.
1118
1119You can rely on the normal read queue and C<on_eof> handling
1120afterwards. This is the cleanest way to close a connection.
1121
1122This method may invoke callbacks (and therefore the handle might be
1123destroyed after it returns).
1124
1125=cut
1126
1127sub push_shutdown {
1128 my ($self) = @_;
1129
1130 delete $self->{low_water_mark};
1131 $self->on_drain (sub { shutdown $_[0]{fh}, 1 });
1132}
1133
1134=item custom write types - Package::anyevent_write_type $handle, @args
1135
1136Instead of one of the predefined types, you can also specify the name of
1137a package. AnyEvent will try to load the package and then expects to find
1138a function named C<anyevent_write_type> inside. If it isn't found, it
1139progressively tries to load the parent package until it either finds the
1140function (good) or runs out of packages (bad).
1141
371Whenever the given C<type> is used, C<push_write> will invoke the code 1142Whenever the given C<type> is used, C<push_write> will the function with
372reference with the handle object and the remaining arguments. 1143the handle object and the remaining arguments.
373 1144
374The code reference is supposed to return a single octet string that will 1145The function is supposed to return a single octet string that will be
375be appended to the write buffer. 1146appended to the write buffer, so you can mentally treat this function as a
1147"arguments to on-the-wire-format" converter.
376 1148
377Note that this is a function, and all types registered this way will be 1149Example: implement a custom write type C<join> that joins the remaining
378global, so try to use unique names. 1150arguments using the first one.
1151
1152 $handle->push_write (My::Type => " ", 1,2,3);
1153
1154 # uses the following package, which can be defined in the "My::Type" or in
1155 # the "My" modules to be auto-loaded, or just about anywhere when the
1156 # My::Type::anyevent_write_type is defined before invoking it.
1157
1158 package My::Type;
1159
1160 sub anyevent_write_type {
1161 my ($handle, $delim, @args) = @_;
1162
1163 join $delim, @args
1164 }
379 1165
380=cut 1166=cut
381 1167
382############################################################################# 1168#############################################################################
383 1169
392ways, the "simple" way, using only C<on_read> and the "complex" way, using 1178ways, the "simple" way, using only C<on_read> and the "complex" way, using
393a queue. 1179a queue.
394 1180
395In the simple case, you just install an C<on_read> callback and whenever 1181In the simple case, you just install an C<on_read> callback and whenever
396new data arrives, it will be called. You can then remove some data (if 1182new data arrives, it will be called. You can then remove some data (if
397enough is there) from the read buffer (C<< $handle->rbuf >>) if you want 1183enough is there) from the read buffer (C<< $handle->rbuf >>). Or you can
398or not. 1184leave the data there if you want to accumulate more (e.g. when only a
1185partial message has been received so far), or change the read queue with
1186e.g. C<push_read>.
399 1187
400In the more complex case, you want to queue multiple callbacks. In this 1188In the more complex case, you want to queue multiple callbacks. In this
401case, AnyEvent::Handle will call the first queued callback each time new 1189case, AnyEvent::Handle will call the first queued callback each time new
402data arrives and removes it when it has done its job (see C<push_read>, 1190data arrives (also the first time it is queued) and remove it when it has
403below). 1191done its job (see C<push_read>, below).
404 1192
405This way you can, for example, push three line-reads, followed by reading 1193This way you can, for example, push three line-reads, followed by reading
406a chunk of data, and AnyEvent::Handle will execute them in order. 1194a chunk of data, and AnyEvent::Handle will execute them in order.
407 1195
408Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by 1196Example 1: EPP protocol parser. EPP sends 4 byte length info, followed by
409the specified number of bytes which give an XML datagram. 1197the specified number of bytes which give an XML datagram.
410 1198
411 # in the default state, expect some header bytes 1199 # in the default state, expect some header bytes
412 $handle->on_read (sub { 1200 $handle->on_read (sub {
413 # some data is here, now queue the length-header-read (4 octets) 1201 # some data is here, now queue the length-header-read (4 octets)
414 shift->unshift_read_chunk (4, sub { 1202 shift->unshift_read (chunk => 4, sub {
415 # header arrived, decode 1203 # header arrived, decode
416 my $len = unpack "N", $_[1]; 1204 my $len = unpack "N", $_[1];
417 1205
418 # now read the payload 1206 # now read the payload
419 shift->unshift_read_chunk ($len, sub { 1207 shift->unshift_read (chunk => $len, sub {
420 my $xml = $_[1]; 1208 my $xml = $_[1];
421 # handle xml 1209 # handle xml
422 }); 1210 });
423 }); 1211 });
424 }); 1212 });
425 1213
426Example 2: Implement a client for a protocol that replies either with 1214Example 2: Implement a client for a protocol that replies either with "OK"
427"OK" and another line or "ERROR" for one request, and 64 bytes for the 1215and another line or "ERROR" for the first request that is sent, and 64
428second request. Due tot he availability of a full queue, we can just 1216bytes for the second request. Due to the availability of a queue, we can
429pipeline sending both requests and manipulate the queue as necessary in 1217just pipeline sending both requests and manipulate the queue as necessary
430the callbacks: 1218in the callbacks.
431 1219
432 # request one 1220When the first callback is called and sees an "OK" response, it will
1221C<unshift> another line-read. This line-read will be queued I<before> the
122264-byte chunk callback.
1223
1224 # request one, returns either "OK + extra line" or "ERROR"
433 $handle->push_write ("request 1\015\012"); 1225 $handle->push_write ("request 1\015\012");
434 1226
435 # we expect "ERROR" or "OK" as response, so push a line read 1227 # we expect "ERROR" or "OK" as response, so push a line read
436 $handle->push_read_line (sub { 1228 $handle->push_read (line => sub {
437 # if we got an "OK", we have to _prepend_ another line, 1229 # if we got an "OK", we have to _prepend_ another line,
438 # so it will be read before the second request reads its 64 bytes 1230 # so it will be read before the second request reads its 64 bytes
439 # which are already in the queue when this callback is called 1231 # which are already in the queue when this callback is called
440 # we don't do this in case we got an error 1232 # we don't do this in case we got an error
441 if ($_[1] eq "OK") { 1233 if ($_[1] eq "OK") {
442 $_[0]->unshift_read_line (sub { 1234 $_[0]->unshift_read (line => sub {
443 my $response = $_[1]; 1235 my $response = $_[1];
444 ... 1236 ...
445 }); 1237 });
446 } 1238 }
447 }); 1239 });
448 1240
449 # request two 1241 # request two, simply returns 64 octets
450 $handle->push_write ("request 2\015\012"); 1242 $handle->push_write ("request 2\015\012");
451 1243
452 # simply read 64 bytes, always 1244 # simply read 64 bytes, always
453 $handle->push_read_chunk (64, sub { 1245 $handle->push_read (chunk => 64, sub {
454 my $response = $_[1]; 1246 my $response = $_[1];
455 ... 1247 ...
456 }); 1248 });
457 1249
458=over 4 1250=over 4
459 1251
460=cut 1252=cut
461 1253
462sub _drain_rbuf { 1254sub _drain_rbuf {
463 my ($self) = @_; 1255 my ($self) = @_;
1256
1257 # avoid recursion
1258 return if $self->{_skip_drain_rbuf};
1259 local $self->{_skip_drain_rbuf} = 1;
1260
1261 while () {
1262 # we need to use a separate tls read buffer, as we must not receive data while
1263 # we are draining the buffer, and this can only happen with TLS.
1264 $self->{rbuf} .= delete $self->{_tls_rbuf}
1265 if exists $self->{_tls_rbuf};
1266
1267 my $len = length $self->{rbuf};
1268
1269 if (my $cb = shift @{ $self->{_queue} }) {
1270 unless ($cb->($self)) {
1271 # no progress can be made
1272 # (not enough data and no data forthcoming)
1273 $self->_error (Errno::EPIPE, 1), return
1274 if $self->{_eof};
1275
1276 unshift @{ $self->{_queue} }, $cb;
1277 last;
1278 }
1279 } elsif ($self->{on_read}) {
1280 last unless $len;
1281
1282 $self->{on_read}($self);
1283
1284 if (
1285 $len == length $self->{rbuf} # if no data has been consumed
1286 && !@{ $self->{_queue} } # and the queue is still empty
1287 && $self->{on_read} # but we still have on_read
1288 ) {
1289 # no further data will arrive
1290 # so no progress can be made
1291 $self->_error (Errno::EPIPE, 1), return
1292 if $self->{_eof};
1293
1294 last; # more data might arrive
1295 }
1296 } else {
1297 # read side becomes idle
1298 delete $self->{_rw} unless $self->{tls};
1299 last;
1300 }
1301 }
1302
1303 if ($self->{_eof}) {
1304 $self->{on_eof}
1305 ? $self->{on_eof}($self)
1306 : $self->_error (0, 1, "Unexpected end-of-file");
1307
1308 return;
1309 }
464 1310
465 if ( 1311 if (
466 defined $self->{rbuf_max} 1312 defined $self->{rbuf_max}
467 && $self->{rbuf_max} < length $self->{rbuf} 1313 && $self->{rbuf_max} < length $self->{rbuf}
468 ) { 1314 ) {
469 $! = &Errno::ENOSPC; return $self->error; 1315 $self->_error (Errno::ENOSPC, 1), return;
470 } 1316 }
471 1317
472 return if $self->{in_drain}; 1318 # may need to restart read watcher
473 local $self->{in_drain} = 1; 1319 unless ($self->{_rw}) {
474 1320 $self->start_read
475 while (my $len = length $self->{rbuf}) { 1321 if $self->{on_read} || @{ $self->{_queue} };
476 no strict 'refs';
477 if (my $cb = shift @{ $self->{queue} }) {
478 unless ($cb->($self)) {
479 if ($self->{eof}) {
480 # no progress can be made (not enough data and no data forthcoming)
481 $! = &Errno::EPIPE; return $self->error;
482 }
483
484 unshift @{ $self->{queue} }, $cb;
485 return;
486 }
487 } elsif ($self->{on_read}) {
488 $self->{on_read}($self);
489
490 if (
491 $self->{eof} # if no further data will arrive
492 && $len == length $self->{rbuf} # and no data has been consumed
493 && !@{ $self->{queue} } # and the queue is still empty
494 && $self->{on_read} # and we still want to read data
495 ) {
496 # then no progress can be made
497 $! = &Errno::EPIPE; return $self->error;
498 }
499 } else {
500 # read side becomes idle
501 delete $self->{rw};
502 return;
503 }
504 }
505
506 if ($self->{eof}) {
507 $self->_shutdown;
508 $self->{on_eof}($self)
509 if $self->{on_eof};
510 } 1322 }
511} 1323}
512 1324
513=item $handle->on_read ($cb) 1325=item $handle->on_read ($cb)
514 1326
515This replaces the currently set C<on_read> callback, or clears it (when 1327This replaces the currently set C<on_read> callback, or clears it (when
516the new callback is C<undef>). See the description of C<on_read> in the 1328the new callback is C<undef>). See the description of C<on_read> in the
517constructor. 1329constructor.
518 1330
1331This method may invoke callbacks (and therefore the handle might be
1332destroyed after it returns).
1333
519=cut 1334=cut
520 1335
521sub on_read { 1336sub on_read {
522 my ($self, $cb) = @_; 1337 my ($self, $cb) = @_;
523 1338
524 $self->{on_read} = $cb; 1339 $self->{on_read} = $cb;
1340 $self->_drain_rbuf if $cb;
525} 1341}
526 1342
527=item $handle->rbuf 1343=item $handle->rbuf
528 1344
529Returns the read buffer (as a modifiable lvalue). 1345Returns the read buffer (as a modifiable lvalue). You can also access the
1346read buffer directly as the C<< ->{rbuf} >> member, if you want (this is
1347much faster, and no less clean).
530 1348
531You can access the read buffer directly as the C<< ->{rbuf} >> member, if 1349The only operation allowed on the read buffer (apart from looking at it)
532you want. 1350is removing data from its beginning. Otherwise modifying or appending to
1351it is not allowed and will lead to hard-to-track-down bugs.
533 1352
534NOTE: The read buffer should only be used or modified if the C<on_read>, 1353NOTE: The read buffer should only be used or modified in the C<on_read>
535C<push_read> or C<unshift_read> methods are used. The other read methods 1354callback or when C<push_read> or C<unshift_read> are used with a single
536automatically manage the read buffer. 1355callback (i.e. untyped). Typed C<push_read> and C<unshift_read> methods
1356will manage the read buffer on their own.
537 1357
538=cut 1358=cut
539 1359
540sub rbuf : lvalue { 1360sub rbuf : lvalue {
541 $_[0]{rbuf} 1361 $_[0]{rbuf}
558 1378
559If enough data was available, then the callback must remove all data it is 1379If enough data was available, then the callback must remove all data it is
560interested in (which can be none at all) and return a true value. After returning 1380interested in (which can be none at all) and return a true value. After returning
561true, it will be removed from the queue. 1381true, it will be removed from the queue.
562 1382
1383These methods may invoke callbacks (and therefore the handle might be
1384destroyed after it returns).
1385
563=cut 1386=cut
564 1387
565our %RH; 1388our %RH;
566 1389
567sub register_read_type($$) { 1390sub register_read_type($$) {
573 my $cb = pop; 1396 my $cb = pop;
574 1397
575 if (@_) { 1398 if (@_) {
576 my $type = shift; 1399 my $type = shift;
577 1400
1401 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
578 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::push_read") 1402 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::push_read")
579 ->($self, $cb, @_); 1403 ->($self, $cb, @_);
580 } 1404 }
581 1405
582 push @{ $self->{queue} }, $cb; 1406 push @{ $self->{_queue} }, $cb;
583 $self->_drain_rbuf; 1407 $self->_drain_rbuf;
584} 1408}
585 1409
586sub unshift_read { 1410sub unshift_read {
587 my $self = shift; 1411 my $self = shift;
588 my $cb = pop; 1412 my $cb = pop;
589 1413
590 if (@_) { 1414 if (@_) {
591 my $type = shift; 1415 my $type = shift;
592 1416
1417 $cb = ($RH{$type} ||= _load_func "$type\::anyevent_read_type"
593 $cb = ($RH{$type} or Carp::croak "unsupported type passed to AnyEvent::Handle::unshift_read") 1418 or Carp::croak "unsupported/unloadable type '$type' passed to AnyEvent::Handle::unshift_read")
594 ->($self, $cb, @_); 1419 ->($self, $cb, @_);
595 } 1420 }
596 1421
597
598 unshift @{ $self->{queue} }, $cb; 1422 unshift @{ $self->{_queue} }, $cb;
599 $self->_drain_rbuf; 1423 $self->_drain_rbuf;
600} 1424}
601 1425
602=item $handle->push_read (type => @args, $cb) 1426=item $handle->push_read (type => @args, $cb)
603 1427
604=item $handle->unshift_read (type => @args, $cb) 1428=item $handle->unshift_read (type => @args, $cb)
605 1429
606Instead of providing a callback that parses the data itself you can chose 1430Instead of providing a callback that parses the data itself you can chose
607between a number of predefined parsing formats, for chunks of data, lines 1431between a number of predefined parsing formats, for chunks of data, lines
608etc. 1432etc. You can also specify the (fully qualified) name of a package, in
1433which case AnyEvent tries to load the package and then expects to find the
1434C<anyevent_read_type> function inside (see "custom read types", below).
609 1435
610Predefined types are (if you have ideas for additional types, feel free to 1436Predefined types are (if you have ideas for additional types, feel free to
611drop by and tell us): 1437drop by and tell us):
612 1438
613=over 4 1439=over 4
614 1440
615=item chunk => $octets, $cb->($self, $data) 1441=item chunk => $octets, $cb->($handle, $data)
616 1442
617Invoke the callback only once C<$octets> bytes have been read. Pass the 1443Invoke the callback only once C<$octets> bytes have been read. Pass the
618data read to the callback. The callback will never be called with less 1444data read to the callback. The callback will never be called with less
619data. 1445data.
620 1446
621Example: read 2 bytes. 1447Example: read 2 bytes.
622 1448
623 $handle->push_read (chunk => 2, sub { 1449 $handle->push_read (chunk => 2, sub {
624 warn "yay ", unpack "H*", $_[1]; 1450 say "yay " . unpack "H*", $_[1];
625 }); 1451 });
626 1452
627=cut 1453=cut
628 1454
629register_read_type chunk => sub { 1455register_read_type chunk => sub {
634 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, ""); 1460 $cb->($_[0], substr $_[0]{rbuf}, 0, $len, "");
635 1 1461 1
636 } 1462 }
637}; 1463};
638 1464
639# compatibility with older API
640sub push_read_chunk {
641 $_[0]->push_read (chunk => $_[1], $_[2]);
642}
643
644sub unshift_read_chunk {
645 $_[0]->unshift_read (chunk => $_[1], $_[2]);
646}
647
648=item line => [$eol, ]$cb->($self, $line, $eol) 1465=item line => [$eol, ]$cb->($handle, $line, $eol)
649 1466
650The callback will be called only once a full line (including the end of 1467The callback will be called only once a full line (including the end of
651line marker, C<$eol>) has been read. This line (excluding the end of line 1468line marker, C<$eol>) has been read. This line (excluding the end of line
652marker) will be passed to the callback as second argument (C<$line>), and 1469marker) will be passed to the callback as second argument (C<$line>), and
653the end of line marker as the third argument (C<$eol>). 1470the end of line marker as the third argument (C<$eol>).
667=cut 1484=cut
668 1485
669register_read_type line => sub { 1486register_read_type line => sub {
670 my ($self, $cb, $eol) = @_; 1487 my ($self, $cb, $eol) = @_;
671 1488
672 $eol = qr|(\015?\012)| if @_ < 3; 1489 if (@_ < 3) {
1490 # this is more than twice as fast as the generic code below
1491 sub {
1492 $_[0]{rbuf} =~ s/^([^\015\012]*)(\015?\012)// or return;
1493
1494 $cb->($_[0], "$1", "$2");
1495 1
1496 }
1497 } else {
673 $eol = quotemeta $eol unless ref $eol; 1498 $eol = quotemeta $eol unless ref $eol;
674 $eol = qr|^(.*?)($eol)|s; 1499 $eol = qr|^(.*?)($eol)|s;
1500
1501 sub {
1502 $_[0]{rbuf} =~ s/$eol// or return;
1503
1504 $cb->($_[0], "$1", "$2");
1505 1
1506 }
1507 }
1508};
1509
1510=item regex => $accept[, $reject[, $skip], $cb->($handle, $data)
1511
1512Makes a regex match against the regex object C<$accept> and returns
1513everything up to and including the match.
1514
1515Example: read a single line terminated by '\n'.
1516
1517 $handle->push_read (regex => qr<\n>, sub { ... });
1518
1519If C<$reject> is given and not undef, then it determines when the data is
1520to be rejected: it is matched against the data when the C<$accept> regex
1521does not match and generates an C<EBADMSG> error when it matches. This is
1522useful to quickly reject wrong data (to avoid waiting for a timeout or a
1523receive buffer overflow).
1524
1525Example: expect a single decimal number followed by whitespace, reject
1526anything else (not the use of an anchor).
1527
1528 $handle->push_read (regex => qr<^[0-9]+\s>, qr<[^0-9]>, sub { ... });
1529
1530If C<$skip> is given and not C<undef>, then it will be matched against
1531the receive buffer when neither C<$accept> nor C<$reject> match,
1532and everything preceding and including the match will be accepted
1533unconditionally. This is useful to skip large amounts of data that you
1534know cannot be matched, so that the C<$accept> or C<$reject> regex do not
1535have to start matching from the beginning. This is purely an optimisation
1536and is usually worth it only when you expect more than a few kilobytes.
1537
1538Example: expect a http header, which ends at C<\015\012\015\012>. Since we
1539expect the header to be very large (it isn't in practice, but...), we use
1540a skip regex to skip initial portions. The skip regex is tricky in that
1541it only accepts something not ending in either \015 or \012, as these are
1542required for the accept regex.
1543
1544 $handle->push_read (regex =>
1545 qr<\015\012\015\012>,
1546 undef, # no reject
1547 qr<^.*[^\015\012]>,
1548 sub { ... });
1549
1550=cut
1551
1552register_read_type regex => sub {
1553 my ($self, $cb, $accept, $reject, $skip) = @_;
1554
1555 my $data;
1556 my $rbuf = \$self->{rbuf};
675 1557
676 sub { 1558 sub {
677 $_[0]{rbuf} =~ s/$eol// or return; 1559 # accept
678 1560 if ($$rbuf =~ $accept) {
1561 $data .= substr $$rbuf, 0, $+[0], "";
679 $cb->($_[0], $1, $2); 1562 $cb->($_[0], $data);
1563 return 1;
1564 }
680 1 1565
1566 # reject
1567 if ($reject && $$rbuf =~ $reject) {
1568 $_[0]->_error (Errno::EBADMSG);
1569 }
1570
1571 # skip
1572 if ($skip && $$rbuf =~ $skip) {
1573 $data .= substr $$rbuf, 0, $+[0], "";
1574 }
1575
1576 ()
681 } 1577 }
682}; 1578};
683 1579
684# compatibility with older API
685sub push_read_line {
686 my $self = shift;
687 $self->push_read (line => @_);
688}
689
690sub unshift_read_line {
691 my $self = shift;
692 $self->unshift_read (line => @_);
693}
694
695=item netstring => $cb->($string) 1580=item netstring => $cb->($handle, $string)
696 1581
697A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement). 1582A netstring (http://cr.yp.to/proto/netstrings.txt, this is not an endorsement).
698 1583
699Throws an error with C<$!> set to EBADMSG on format violations. 1584Throws an error with C<$!> set to EBADMSG on format violations.
700 1585
704 my ($self, $cb) = @_; 1589 my ($self, $cb) = @_;
705 1590
706 sub { 1591 sub {
707 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) { 1592 unless ($_[0]{rbuf} =~ s/^(0|[1-9][0-9]*)://) {
708 if ($_[0]{rbuf} =~ /[^0-9]/) { 1593 if ($_[0]{rbuf} =~ /[^0-9]/) {
709 $! = &Errno::EBADMSG; 1594 $_[0]->_error (Errno::EBADMSG);
710 $self->error;
711 } 1595 }
712 return; 1596 return;
713 } 1597 }
714 1598
715 my $len = $1; 1599 my $len = $1;
716 1600
717 $self->unshift_read (chunk => $len, sub { 1601 $_[0]->unshift_read (chunk => $len, sub {
718 my $string = $_[1]; 1602 my $string = $_[1];
719 $_[0]->unshift_read (chunk => 1, sub { 1603 $_[0]->unshift_read (chunk => 1, sub {
720 if ($_[1] eq ",") { 1604 if ($_[1] eq ",") {
721 $cb->($_[0], $string); 1605 $cb->($_[0], $string);
722 } else { 1606 } else {
723 $! = &Errno::EBADMSG; 1607 $_[0]->_error (Errno::EBADMSG);
724 $self->error;
725 } 1608 }
726 }); 1609 });
727 }); 1610 });
728 1611
729 1 1612 1
730 } 1613 }
731}; 1614};
732 1615
1616=item packstring => $format, $cb->($handle, $string)
1617
1618An octet string prefixed with an encoded length. The encoding C<$format>
1619uses the same format as a Perl C<pack> format, but must specify a single
1620integer only (only one of C<cCsSlLqQiInNvVjJw> is allowed, plus an
1621optional C<!>, C<< < >> or C<< > >> modifier).
1622
1623For example, DNS over TCP uses a prefix of C<n> (2 octet network order),
1624EPP uses a prefix of C<N> (4 octtes).
1625
1626Example: read a block of data prefixed by its length in BER-encoded
1627format (very efficient).
1628
1629 $handle->push_read (packstring => "w", sub {
1630 my ($handle, $data) = @_;
1631 });
1632
1633=cut
1634
1635register_read_type packstring => sub {
1636 my ($self, $cb, $format) = @_;
1637
1638 sub {
1639 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1640 defined (my $len = eval { unpack $format, $_[0]{rbuf} })
1641 or return;
1642
1643 $format = length pack $format, $len;
1644
1645 # bypass unshift if we already have the remaining chunk
1646 if ($format + $len <= length $_[0]{rbuf}) {
1647 my $data = substr $_[0]{rbuf}, $format, $len;
1648 substr $_[0]{rbuf}, 0, $format + $len, "";
1649 $cb->($_[0], $data);
1650 } else {
1651 # remove prefix
1652 substr $_[0]{rbuf}, 0, $format, "";
1653
1654 # read remaining chunk
1655 $_[0]->unshift_read (chunk => $len, $cb);
1656 }
1657
1658 1
1659 }
1660};
1661
1662=item json => $cb->($handle, $hash_or_arrayref)
1663
1664Reads a JSON object or array, decodes it and passes it to the
1665callback. When a parse error occurs, an C<EBADMSG> error will be raised.
1666
1667If a C<json> object was passed to the constructor, then that will be used
1668for the final decode, otherwise it will create a JSON coder expecting UTF-8.
1669
1670This read type uses the incremental parser available with JSON version
16712.09 (and JSON::XS version 2.2) and above. You have to provide a
1672dependency on your own: this module will load the JSON module, but
1673AnyEvent does not depend on it itself.
1674
1675Since JSON texts are fully self-delimiting, the C<json> read and write
1676types are an ideal simple RPC protocol: just exchange JSON datagrams. See
1677the C<json> write type description, above, for an actual example.
1678
1679=cut
1680
1681register_read_type json => sub {
1682 my ($self, $cb) = @_;
1683
1684 my $json = $self->{json} ||= json_coder;
1685
1686 my $data;
1687 my $rbuf = \$self->{rbuf};
1688
1689 sub {
1690 my $ref = eval { $json->incr_parse ($_[0]{rbuf}) };
1691
1692 if ($ref) {
1693 $_[0]{rbuf} = $json->incr_text;
1694 $json->incr_text = "";
1695 $cb->($_[0], $ref);
1696
1697 1
1698 } elsif ($@) {
1699 # error case
1700 $json->incr_skip;
1701
1702 $_[0]{rbuf} = $json->incr_text;
1703 $json->incr_text = "";
1704
1705 $_[0]->_error (Errno::EBADMSG);
1706
1707 ()
1708 } else {
1709 $_[0]{rbuf} = "";
1710
1711 ()
1712 }
1713 }
1714};
1715
1716=item storable => $cb->($handle, $ref)
1717
1718Deserialises a L<Storable> frozen representation as written by the
1719C<storable> write type (BER-encoded length prefix followed by nfreeze'd
1720data).
1721
1722Raises C<EBADMSG> error if the data could not be decoded.
1723
1724=cut
1725
1726register_read_type storable => sub {
1727 my ($self, $cb) = @_;
1728
1729 require Storable unless $Storable::VERSION;
1730
1731 sub {
1732 # when we can use 5.10 we can use ".", but for 5.8 we use the re-pack method
1733 defined (my $len = eval { unpack "w", $_[0]{rbuf} })
1734 or return;
1735
1736 my $format = length pack "w", $len;
1737
1738 # bypass unshift if we already have the remaining chunk
1739 if ($format + $len <= length $_[0]{rbuf}) {
1740 my $data = substr $_[0]{rbuf}, $format, $len;
1741 substr $_[0]{rbuf}, 0, $format + $len, "";
1742
1743 eval { $cb->($_[0], Storable::thaw ($data)); 1 }
1744 or return $_[0]->_error (Errno::EBADMSG);
1745 } else {
1746 # remove prefix
1747 substr $_[0]{rbuf}, 0, $format, "";
1748
1749 # read remaining chunk
1750 $_[0]->unshift_read (chunk => $len, sub {
1751 eval { $cb->($_[0], Storable::thaw ($_[1])); 1 }
1752 or $_[0]->_error (Errno::EBADMSG);
1753 });
1754 }
1755
1756 1
1757 }
1758};
1759
733=back 1760=back
734 1761
735=item AnyEvent::Handle::register_read_type type => $coderef->($self, $cb, @args) 1762=item custom read types - Package::anyevent_read_type $handle, $cb, @args
736 1763
737This function (not method) lets you add your own types to C<push_read>. 1764Instead of one of the predefined types, you can also specify the name
1765of a package. AnyEvent will try to load the package and then expects to
1766find a function named C<anyevent_read_type> inside. If it isn't found, it
1767progressively tries to load the parent package until it either finds the
1768function (good) or runs out of packages (bad).
738 1769
739Whenever the given C<type> is used, C<push_read> will invoke the code 1770Whenever this type is used, C<push_read> will invoke the function with the
740reference with the handle object, the callback and the remaining 1771handle object, the original callback and the remaining arguments.
741arguments.
742 1772
743The code reference is supposed to return a callback (usually a closure) 1773The function is supposed to return a callback (usually a closure) that
744that works as a plain read callback (see C<< ->push_read ($cb) >>). 1774works as a plain read callback (see C<< ->push_read ($cb) >>), so you can
1775mentally treat the function as a "configurable read type to read callback"
1776converter.
745 1777
746It should invoke the passed callback when it is done reading (remember to 1778It should invoke the original callback when it is done reading (remember
747pass C<$self> as first argument as all other callbacks do that). 1779to pass C<$handle> as first argument as all other callbacks do that,
1780although there is no strict requirement on this).
748 1781
749Note that this is a function, and all types registered this way will be
750global, so try to use unique names.
751
752For examples, see the source of this module (F<perldoc -m AnyEvent::Handle>, 1782For examples, see the source of this module (F<perldoc -m
753search for C<register_read_type>)). 1783AnyEvent::Handle>, search for C<register_read_type>)).
754 1784
755=item $handle->stop_read 1785=item $handle->stop_read
756 1786
757=item $handle->start_read 1787=item $handle->start_read
758 1788
759In rare cases you actually do not want to read anything from the 1789In rare cases you actually do not want to read anything from the
760socket. In this case you can call C<stop_read>. Neither C<on_read> no 1790socket. In this case you can call C<stop_read>. Neither C<on_read> nor
761any queued callbacks will be executed then. To start reading again, call 1791any queued callbacks will be executed then. To start reading again, call
762C<start_read>. 1792C<start_read>.
763 1793
1794Note that AnyEvent::Handle will automatically C<start_read> for you when
1795you change the C<on_read> callback or push/unshift a read callback, and it
1796will automatically C<stop_read> for you when neither C<on_read> is set nor
1797there are any read requests in the queue.
1798
1799In older versions of this module (<= 5.3), these methods had no effect,
1800as TLS does not support half-duplex connections. In current versions they
1801work as expected, as this behaviour is required to avoid certain resource
1802attacks, where the program would be forced to read (and buffer) arbitrary
1803amounts of data before being able to send some data. The drawback is that
1804some readings of the the SSL/TLS specifications basically require this
1805attack to be working, as SSL/TLS implementations might stall sending data
1806during a rehandshake.
1807
1808As a guideline, during the initial handshake, you should not stop reading,
1809and as a client, it might cause problems, depending on your application.
1810
764=cut 1811=cut
765 1812
766sub stop_read { 1813sub stop_read {
767 my ($self) = @_; 1814 my ($self) = @_;
768 1815
769 delete $self->{rw}; 1816 delete $self->{_rw};
770} 1817}
771 1818
772sub start_read { 1819sub start_read {
773 my ($self) = @_; 1820 my ($self) = @_;
774 1821
775 unless ($self->{rw} || $self->{eof}) { 1822 unless ($self->{_rw} || $self->{_eof} || !$self->{fh}) {
776 Scalar::Util::weaken $self; 1823 Scalar::Util::weaken $self;
777 1824
778 $self->{rw} = AnyEvent->io (fh => $self->{fh}, poll => "r", cb => sub { 1825 $self->{_rw} = AE::io $self->{fh}, 0, sub {
779 my $rbuf = $self->{filter_r} ? \my $buf : \$self->{rbuf}; 1826 my $rbuf = \($self->{tls} ? my $buf : $self->{rbuf});
780 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size} || 8192, length $$rbuf; 1827 my $len = sysread $self->{fh}, $$rbuf, $self->{read_size}, length $$rbuf;
781 1828
782 if ($len > 0) { 1829 if ($len > 0) {
783 $self->{filter_r} 1830 $self->{_activity} = $self->{_ractivity} = AE::now;
784 ? $self->{filter_r}->($self, $rbuf) 1831
1832 if ($self->{tls}) {
1833 Net::SSLeay::BIO_write ($self->{_rbio}, $$rbuf);
1834
1835 &_dotls ($self);
1836 } else {
785 : $self->_drain_rbuf; 1837 $self->_drain_rbuf;
1838 }
1839
1840 if ($len == $self->{read_size}) {
1841 $self->{read_size} *= 2;
1842 $self->{read_size} = $self->{max_read_size} || MAX_READ_SIZE
1843 if $self->{read_size} > ($self->{max_read_size} || MAX_READ_SIZE);
1844 }
786 1845
787 } elsif (defined $len) { 1846 } elsif (defined $len) {
788 delete $self->{rw}; 1847 delete $self->{_rw};
789 $self->{eof} = 1; 1848 $self->{_eof} = 1;
790 $self->_drain_rbuf; 1849 $self->_drain_rbuf;
791 1850
792 } elsif ($! != EAGAIN && $! != EINTR) { 1851 } elsif ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK) {
793 return $self->error; 1852 return $self->_error ($!, 1);
794 } 1853 }
795 }); 1854 };
796 } 1855 }
797} 1856}
798 1857
1858our $ERROR_SYSCALL;
1859our $ERROR_WANT_READ;
1860
1861sub _tls_error {
1862 my ($self, $err) = @_;
1863
1864 return $self->_error ($!, 1)
1865 if $err == Net::SSLeay::ERROR_SYSCALL ();
1866
1867 my $err = Net::SSLeay::ERR_error_string (Net::SSLeay::ERR_get_error ());
1868
1869 # reduce error string to look less scary
1870 $err =~ s/^error:[0-9a-fA-F]{8}:[^:]+:([^:]+):/\L$1: /;
1871
1872 if ($self->{_on_starttls}) {
1873 (delete $self->{_on_starttls})->($self, undef, $err);
1874 &_freetls;
1875 } else {
1876 &_freetls;
1877 $self->_error (Errno::EPROTO, 1, $err);
1878 }
1879}
1880
1881# poll the write BIO and send the data if applicable
1882# also decode read data if possible
1883# this is basiclaly our TLS state machine
1884# more efficient implementations are possible with openssl,
1885# but not with the buggy and incomplete Net::SSLeay.
799sub _dotls { 1886sub _dotls {
800 my ($self) = @_; 1887 my ($self) = @_;
801 1888
1889 my $tmp;
1890
802 if (length $self->{tls_wbuf}) { 1891 if (length $self->{_tls_wbuf}) {
803 while ((my $len = Net::SSLeay::write ($self->{tls}, $self->{tls_wbuf})) > 0) { 1892 while (($tmp = Net::SSLeay::write ($self->{tls}, $self->{_tls_wbuf})) > 0) {
804 substr $self->{tls_wbuf}, 0, $len, ""; 1893 substr $self->{_tls_wbuf}, 0, $tmp, "";
805 } 1894 }
806 }
807 1895
1896 $tmp = Net::SSLeay::get_error ($self->{tls}, $tmp);
1897 return $self->_tls_error ($tmp)
1898 if $tmp != $ERROR_WANT_READ
1899 && ($tmp != $ERROR_SYSCALL || $!);
1900 }
1901
808 if (defined (my $buf = Net::SSLeay::BIO_read ($self->{tls_wbio}))) { 1902 while (defined ($tmp = Net::SSLeay::read ($self->{tls}))) {
1903 unless (length $tmp) {
1904 $self->{_on_starttls}
1905 and (delete $self->{_on_starttls})->($self, undef, "EOF during handshake"); # ???
1906 &_freetls;
1907
1908 if ($self->{on_stoptls}) {
1909 $self->{on_stoptls}($self);
1910 return;
1911 } else {
1912 # let's treat SSL-eof as we treat normal EOF
1913 delete $self->{_rw};
1914 $self->{_eof} = 1;
1915 }
1916 }
1917
1918 $self->{_tls_rbuf} .= $tmp;
1919 $self->_drain_rbuf;
1920 $self->{tls} or return; # tls session might have gone away in callback
1921 }
1922
1923 $tmp = Net::SSLeay::get_error ($self->{tls}, -1);
1924 return $self->_tls_error ($tmp)
1925 if $tmp != $ERROR_WANT_READ
1926 && ($tmp != $ERROR_SYSCALL || $!);
1927
1928 while (length ($tmp = Net::SSLeay::BIO_read ($self->{_wbio}))) {
809 $self->{wbuf} .= $buf; 1929 $self->{wbuf} .= $tmp;
810 $self->_drain_wbuf; 1930 $self->_drain_wbuf;
1931 $self->{tls} or return; # tls session might have gone away in callback
811 } 1932 }
812 1933
813 while (defined (my $buf = Net::SSLeay::read ($self->{tls}))) { 1934 $self->{_on_starttls}
814 $self->{rbuf} .= $buf; 1935 and Net::SSLeay::state ($self->{tls}) == Net::SSLeay::ST_OK ()
815 $self->_drain_rbuf; 1936 and (delete $self->{_on_starttls})->($self, 1, "TLS/SSL connection established");
816 }
817
818 my $err = Net::SSLeay::get_error ($self->{tls}, -1);
819
820 if ($err!= Net::SSLeay::ERROR_WANT_READ ()) {
821 if ($err == Net::SSLeay::ERROR_SYSCALL ()) {
822 $self->error;
823 } elsif ($err == Net::SSLeay::ERROR_SSL ()) {
824 $! = &Errno::EIO;
825 $self->error;
826 }
827
828 # all others are fine for our purposes
829 }
830} 1937}
831 1938
832=item $handle->starttls ($tls[, $tls_ctx]) 1939=item $handle->starttls ($tls[, $tls_ctx])
833 1940
834Instead of starting TLS negotiation immediately when the AnyEvent::Handle 1941Instead of starting TLS negotiation immediately when the AnyEvent::Handle
835object is created, you can also do that at a later time by calling 1942object is created, you can also do that at a later time by calling
836C<starttls>. 1943C<starttls>. See the C<tls> constructor argument for general info.
1944
1945Starting TLS is currently an asynchronous operation - when you push some
1946write data and then call C<< ->starttls >> then TLS negotiation will start
1947immediately, after which the queued write data is then sent. This might
1948change in future versions, so best make sure you have no outstanding write
1949data when calling this method.
837 1950
838The first argument is the same as the C<tls> constructor argument (either 1951The first argument is the same as the C<tls> constructor argument (either
839C<"connect">, C<"accept"> or an existing Net::SSLeay object). 1952C<"connect">, C<"accept"> or an existing Net::SSLeay object).
840 1953
841The second argument is the optional C<Net::SSLeay::CTX> object that is 1954The second argument is the optional C<AnyEvent::TLS> object that is used
842used when AnyEvent::Handle has to create its own TLS connection object. 1955when AnyEvent::Handle has to create its own TLS connection object, or
1956a hash reference with C<< key => value >> pairs that will be used to
1957construct a new context.
843 1958
844=cut 1959The TLS connection object will end up in C<< $handle->{tls} >>, the TLS
1960context in C<< $handle->{tls_ctx} >> after this call and can be used or
1961changed to your liking. Note that the handshake might have already started
1962when this function returns.
845 1963
846# TODO: maybe document... 1964Due to bugs in OpenSSL, it might or might not be possible to do multiple
1965handshakes on the same stream. It is best to not attempt to use the
1966stream after stopping TLS.
1967
1968This method may invoke callbacks (and therefore the handle might be
1969destroyed after it returns).
1970
1971=cut
1972
1973our %TLS_CACHE; #TODO not yet documented, should we?
1974
847sub starttls { 1975sub starttls {
848 my ($self, $ssl, $ctx) = @_; 1976 my ($self, $tls, $ctx) = @_;
849 1977
850 $self->stoptls; 1978 Carp::croak "It is an error to call starttls on an AnyEvent::Handle object while TLS is already active, caught"
1979 if $self->{tls};
851 1980
852 if ($ssl eq "accept") { 1981 unless (defined $AnyEvent::TLS::VERSION) {
853 $ssl = Net::SSLeay::new ($ctx || TLS_CTX ()); 1982 eval {
854 Net::SSLeay::set_accept_state ($ssl); 1983 require Net::SSLeay;
855 } elsif ($ssl eq "connect") { 1984 require AnyEvent::TLS;
856 $ssl = Net::SSLeay::new ($ctx || TLS_CTX ()); 1985 1
857 Net::SSLeay::set_connect_state ($ssl); 1986 } or return $self->_error (Errno::EPROTO, 1, "TLS support not available on this system");
1987 }
1988
1989 $self->{tls} = $tls;
1990 $self->{tls_ctx} = $ctx if @_ > 2;
1991
1992 return unless $self->{fh};
1993
1994 $ERROR_SYSCALL = Net::SSLeay::ERROR_SYSCALL ();
1995 $ERROR_WANT_READ = Net::SSLeay::ERROR_WANT_READ ();
1996
1997 $tls = delete $self->{tls};
1998 $ctx = $self->{tls_ctx};
1999
2000 local $Carp::CarpLevel = 1; # skip ourselves when creating a new context or session
2001
2002 if ("HASH" eq ref $ctx) {
2003 if ($ctx->{cache}) {
2004 my $key = $ctx+0;
2005 $ctx = $TLS_CACHE{$key} ||= new AnyEvent::TLS %$ctx;
2006 } else {
2007 $ctx = new AnyEvent::TLS %$ctx;
2008 }
2009 }
858 } 2010
859 2011 $self->{tls_ctx} = $ctx || TLS_CTX ();
860 $self->{tls} = $ssl; 2012 $self->{tls} = $tls = $self->{tls_ctx}->_get_session ($tls, $self, $self->{peername});
861 2013
862 # basically, this is deep magic (because SSL_read should have the same issues) 2014 # basically, this is deep magic (because SSL_read should have the same issues)
863 # but the openssl maintainers basically said: "trust us, it just works". 2015 # but the openssl maintainers basically said: "trust us, it just works".
864 # (unfortunately, we have to hardcode constants because the abysmally misdesigned 2016 # (unfortunately, we have to hardcode constants because the abysmally misdesigned
865 # and mismaintained ssleay-module doesn't even offer them). 2017 # and mismaintained ssleay-module doesn't even offer them).
866 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html 2018 # http://www.mail-archive.com/openssl-dev@openssl.org/msg22420.html
2019 #
2020 # in short: this is a mess.
2021 #
2022 # note that we do not try to keep the length constant between writes as we are required to do.
2023 # we assume that most (but not all) of this insanity only applies to non-blocking cases,
2024 # and we drive openssl fully in blocking mode here. Or maybe we don't - openssl seems to
2025 # have identity issues in that area.
867 Net::SSLeay::CTX_set_mode ($self->{tls}, 2026# Net::SSLeay::CTX_set_mode ($ssl,
868 (eval { Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1) 2027# (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ENABLE_PARTIAL_WRITE () } || 1)
869 | (eval { Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2)); 2028# | (eval { local $SIG{__DIE__}; Net::SSLeay::MODE_ACCEPT_MOVING_WRITE_BUFFER () } || 2));
2029 Net::SSLeay::CTX_set_mode ($tls, 1|2);
870 2030
871 $self->{tls_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ()); 2031 $self->{_rbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
872 $self->{tls_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ()); 2032 $self->{_wbio} = Net::SSLeay::BIO_new (Net::SSLeay::BIO_s_mem ());
873 2033
2034 Net::SSLeay::BIO_write ($self->{_rbio}, $self->{rbuf});
2035 $self->{rbuf} = "";
2036
874 Net::SSLeay::set_bio ($ssl, $self->{tls_rbio}, $self->{tls_wbio}); 2037 Net::SSLeay::set_bio ($tls, $self->{_rbio}, $self->{_wbio});
875 2038
876 $self->{filter_w} = sub { 2039 $self->{_on_starttls} = sub { $_[0]{on_starttls}(@_) }
877 $_[0]{tls_wbuf} .= ${$_[1]}; 2040 if $self->{on_starttls};
878 &_dotls; 2041
879 }; 2042 &_dotls; # need to trigger the initial handshake
880 $self->{filter_r} = sub { 2043 $self->start_read; # make sure we actually do read
881 Net::SSLeay::BIO_write ($_[0]{tls_rbio}, ${$_[1]});
882 &_dotls;
883 };
884} 2044}
885 2045
886=item $handle->stoptls 2046=item $handle->stoptls
887 2047
888Destroys the SSL connection, if any. Partial read or write data will be 2048Shuts down the SSL connection - this makes a proper EOF handshake by
889lost. 2049sending a close notify to the other side, but since OpenSSL doesn't
2050support non-blocking shut downs, it is not guaranteed that you can re-use
2051the stream afterwards.
2052
2053This method may invoke callbacks (and therefore the handle might be
2054destroyed after it returns).
890 2055
891=cut 2056=cut
892 2057
893sub stoptls { 2058sub stoptls {
894 my ($self) = @_; 2059 my ($self) = @_;
895 2060
896 Net::SSLeay::free (delete $self->{tls}) if $self->{tls}; 2061 if ($self->{tls} && $self->{fh}) {
897 delete $self->{tls_rbio}; 2062 Net::SSLeay::shutdown ($self->{tls});
898 delete $self->{tls_wbio}; 2063
899 delete $self->{tls_wbuf}; 2064 &_dotls;
900 delete $self->{filter_r}; 2065
901 delete $self->{filter_w}; 2066# # we don't give a shit. no, we do, but we can't. no...#d#
2067# # we, we... have to use openssl :/#d#
2068# &_freetls;#d#
2069 }
902} 2070}
2071
2072sub _freetls {
2073 my ($self) = @_;
2074
2075 return unless $self->{tls};
2076
2077 $self->{tls_ctx}->_put_session (delete $self->{tls})
2078 if $self->{tls} > 0;
2079
2080 delete @$self{qw(_rbio _wbio _tls_wbuf _on_starttls)};
2081}
2082
2083=item $handle->resettls
2084
2085This rarely-used method simply resets and TLS state on the handle, usually
2086causing data loss.
2087
2088One case where it may be useful is when you want to skip over the data in
2089the stream but you are not interested in interpreting it, so data loss is
2090no concern.
2091
2092=cut
2093
2094*resettls = \&_freetls;
903 2095
904sub DESTROY { 2096sub DESTROY {
905 my $self = shift; 2097 my ($self) = @_;
906 2098
907 $self->stoptls; 2099 &_freetls;
2100
2101 my $linger = exists $self->{linger} ? $self->{linger} : 3600;
2102
2103 if ($linger && length $self->{wbuf} && $self->{fh}) {
2104 my $fh = delete $self->{fh};
2105 my $wbuf = delete $self->{wbuf};
2106
2107 my @linger;
2108
2109 push @linger, AE::io $fh, 1, sub {
2110 my $len = syswrite $fh, $wbuf, length $wbuf;
2111
2112 if ($len > 0) {
2113 substr $wbuf, 0, $len, "";
2114 } elsif (defined $len || ($! != EAGAIN && $! != EINTR && $! != WSAEWOULDBLOCK)) {
2115 @linger = (); # end
2116 }
2117 };
2118 push @linger, AE::timer $linger, 0, sub {
2119 @linger = ();
2120 };
2121 }
908} 2122}
2123
2124=item $handle->destroy
2125
2126Shuts down the handle object as much as possible - this call ensures that
2127no further callbacks will be invoked and as many resources as possible
2128will be freed. Any method you will call on the handle object after
2129destroying it in this way will be silently ignored (and it will return the
2130empty list).
2131
2132Normally, you can just "forget" any references to an AnyEvent::Handle
2133object and it will simply shut down. This works in fatal error and EOF
2134callbacks, as well as code outside. It does I<NOT> work in a read or write
2135callback, so when you want to destroy the AnyEvent::Handle object from
2136within such an callback. You I<MUST> call C<< ->destroy >> explicitly in
2137that case.
2138
2139Destroying the handle object in this way has the advantage that callbacks
2140will be removed as well, so if those are the only reference holders (as
2141is common), then one doesn't need to do anything special to break any
2142reference cycles.
2143
2144The handle might still linger in the background and write out remaining
2145data, as specified by the C<linger> option, however.
2146
2147=cut
2148
2149sub destroy {
2150 my ($self) = @_;
2151
2152 $self->DESTROY;
2153 %$self = ();
2154 bless $self, "AnyEvent::Handle::destroyed";
2155}
2156
2157sub AnyEvent::Handle::destroyed::AUTOLOAD {
2158 #nop
2159}
2160
2161=item $handle->destroyed
2162
2163Returns false as long as the handle hasn't been destroyed by a call to C<<
2164->destroy >>, true otherwise.
2165
2166Can be useful to decide whether the handle is still valid after some
2167callback possibly destroyed the handle. For example, C<< ->push_write >>,
2168C<< ->starttls >> and other methods can call user callbacks, which in turn
2169can destroy the handle, so work can be avoided by checking sometimes:
2170
2171 $hdl->starttls ("accept");
2172 return if $hdl->destroyed;
2173 $hdl->push_write (...
2174
2175Note that the call to C<push_write> will silently be ignored if the handle
2176has been destroyed, so often you can just ignore the possibility of the
2177handle being destroyed.
2178
2179=cut
2180
2181sub destroyed { 0 }
2182sub AnyEvent::Handle::destroyed::destroyed { 1 }
909 2183
910=item AnyEvent::Handle::TLS_CTX 2184=item AnyEvent::Handle::TLS_CTX
911 2185
912This function creates and returns the Net::SSLeay::CTX object used by 2186This function creates and returns the AnyEvent::TLS object used by default
913default for TLS mode. 2187for TLS mode.
914 2188
915The context is created like this: 2189The context is created by calling L<AnyEvent::TLS> without any arguments.
916
917 Net::SSLeay::load_error_strings;
918 Net::SSLeay::SSLeay_add_ssl_algorithms;
919 Net::SSLeay::randomize;
920
921 my $CTX = Net::SSLeay::CTX_new;
922
923 Net::SSLeay::CTX_set_options $CTX, Net::SSLeay::OP_ALL
924 2190
925=cut 2191=cut
926 2192
927our $TLS_CTX; 2193our $TLS_CTX;
928 2194
929sub TLS_CTX() { 2195sub TLS_CTX() {
930 $TLS_CTX || do { 2196 $TLS_CTX ||= do {
931 require Net::SSLeay; 2197 require AnyEvent::TLS;
932 2198
933 Net::SSLeay::load_error_strings (); 2199 new AnyEvent::TLS
934 Net::SSLeay::SSLeay_add_ssl_algorithms ();
935 Net::SSLeay::randomize ();
936
937 $TLS_CTX = Net::SSLeay::CTX_new ();
938
939 Net::SSLeay::CTX_set_options ($TLS_CTX, Net::SSLeay::OP_ALL ());
940
941 $TLS_CTX
942 } 2200 }
943} 2201}
944 2202
945=back 2203=back
946 2204
2205
2206=head1 NONFREQUENTLY ASKED QUESTIONS
2207
2208=over 4
2209
2210=item I C<undef> the AnyEvent::Handle reference inside my callback and
2211still get further invocations!
2212
2213That's because AnyEvent::Handle keeps a reference to itself when handling
2214read or write callbacks.
2215
2216It is only safe to "forget" the reference inside EOF or error callbacks,
2217from within all other callbacks, you need to explicitly call the C<<
2218->destroy >> method.
2219
2220=item Why is my C<on_eof> callback never called?
2221
2222Probably because your C<on_error> callback is being called instead: When
2223you have outstanding requests in your read queue, then an EOF is
2224considered an error as you clearly expected some data.
2225
2226To avoid this, make sure you have an empty read queue whenever your handle
2227is supposed to be "idle" (i.e. connection closes are O.K.). You can set
2228an C<on_read> handler that simply pushes the first read requests in the
2229queue.
2230
2231See also the next question, which explains this in a bit more detail.
2232
2233=item How can I serve requests in a loop?
2234
2235Most protocols consist of some setup phase (authentication for example)
2236followed by a request handling phase, where the server waits for requests
2237and handles them, in a loop.
2238
2239There are two important variants: The first (traditional, better) variant
2240handles requests until the server gets some QUIT command, causing it to
2241close the connection first (highly desirable for a busy TCP server). A
2242client dropping the connection is an error, which means this variant can
2243detect an unexpected detection close.
2244
2245To handle this case, always make sure you have a non-empty read queue, by
2246pushing the "read request start" handler on it:
2247
2248 # we assume a request starts with a single line
2249 my @start_request; @start_request = (line => sub {
2250 my ($hdl, $line) = @_;
2251
2252 ... handle request
2253
2254 # push next request read, possibly from a nested callback
2255 $hdl->push_read (@start_request);
2256 });
2257
2258 # auth done, now go into request handling loop
2259 # now push the first @start_request
2260 $hdl->push_read (@start_request);
2261
2262By always having an outstanding C<push_read>, the handle always expects
2263some data and raises the C<EPIPE> error when the connction is dropped
2264unexpectedly.
2265
2266The second variant is a protocol where the client can drop the connection
2267at any time. For TCP, this means that the server machine may run out of
2268sockets easier, and in general, it means you cannot distinguish a protocl
2269failure/client crash from a normal connection close. Nevertheless, these
2270kinds of protocols are common (and sometimes even the best solution to the
2271problem).
2272
2273Having an outstanding read request at all times is possible if you ignore
2274C<EPIPE> errors, but this doesn't help with when the client drops the
2275connection during a request, which would still be an error.
2276
2277A better solution is to push the initial request read in an C<on_read>
2278callback. This avoids an error, as when the server doesn't expect data
2279(i.e. is idly waiting for the next request, an EOF will not raise an
2280error, but simply result in an C<on_eof> callback. It is also a bit slower
2281and simpler:
2282
2283 # auth done, now go into request handling loop
2284 $hdl->on_read (sub {
2285 my ($hdl) = @_;
2286
2287 # called each time we receive data but the read queue is empty
2288 # simply start read the request
2289
2290 $hdl->push_read (line => sub {
2291 my ($hdl, $line) = @_;
2292
2293 ... handle request
2294
2295 # do nothing special when the request has been handled, just
2296 # let the request queue go empty.
2297 });
2298 });
2299
2300=item I get different callback invocations in TLS mode/Why can't I pause
2301reading?
2302
2303Unlike, say, TCP, TLS connections do not consist of two independent
2304communication channels, one for each direction. Or put differently, the
2305read and write directions are not independent of each other: you cannot
2306write data unless you are also prepared to read, and vice versa.
2307
2308This means that, in TLS mode, you might get C<on_error> or C<on_eof>
2309callback invocations when you are not expecting any read data - the reason
2310is that AnyEvent::Handle always reads in TLS mode.
2311
2312During the connection, you have to make sure that you always have a
2313non-empty read-queue, or an C<on_read> watcher. At the end of the
2314connection (or when you no longer want to use it) you can call the
2315C<destroy> method.
2316
2317=item How do I read data until the other side closes the connection?
2318
2319If you just want to read your data into a perl scalar, the easiest way
2320to achieve this is by setting an C<on_read> callback that does nothing,
2321clearing the C<on_eof> callback and in the C<on_error> callback, the data
2322will be in C<$_[0]{rbuf}>:
2323
2324 $handle->on_read (sub { });
2325 $handle->on_eof (undef);
2326 $handle->on_error (sub {
2327 my $data = delete $_[0]{rbuf};
2328 });
2329
2330Note that this example removes the C<rbuf> member from the handle object,
2331which is not normally allowed by the API. It is expressly permitted in
2332this case only, as the handle object needs to be destroyed afterwards.
2333
2334The reason to use C<on_error> is that TCP connections, due to latencies
2335and packets loss, might get closed quite violently with an error, when in
2336fact all data has been received.
2337
2338It is usually better to use acknowledgements when transferring data,
2339to make sure the other side hasn't just died and you got the data
2340intact. This is also one reason why so many internet protocols have an
2341explicit QUIT command.
2342
2343=item I don't want to destroy the handle too early - how do I wait until
2344all data has been written?
2345
2346After writing your last bits of data, set the C<on_drain> callback
2347and destroy the handle in there - with the default setting of
2348C<low_water_mark> this will be called precisely when all data has been
2349written to the socket:
2350
2351 $handle->push_write (...);
2352 $handle->on_drain (sub {
2353 AE::log debug => "All data submitted to the kernel.";
2354 undef $handle;
2355 });
2356
2357If you just want to queue some data and then signal EOF to the other side,
2358consider using C<< ->push_shutdown >> instead.
2359
2360=item I want to contact a TLS/SSL server, I don't care about security.
2361
2362If your TLS server is a pure TLS server (e.g. HTTPS) that only speaks TLS,
2363connect to it and then create the AnyEvent::Handle with the C<tls>
2364parameter:
2365
2366 tcp_connect $host, $port, sub {
2367 my ($fh) = @_;
2368
2369 my $handle = new AnyEvent::Handle
2370 fh => $fh,
2371 tls => "connect",
2372 on_error => sub { ... };
2373
2374 $handle->push_write (...);
2375 };
2376
2377=item I want to contact a TLS/SSL server, I do care about security.
2378
2379Then you should additionally enable certificate verification, including
2380peername verification, if the protocol you use supports it (see
2381L<AnyEvent::TLS>, C<verify_peername>).
2382
2383E.g. for HTTPS:
2384
2385 tcp_connect $host, $port, sub {
2386 my ($fh) = @_;
2387
2388 my $handle = new AnyEvent::Handle
2389 fh => $fh,
2390 peername => $host,
2391 tls => "connect",
2392 tls_ctx => { verify => 1, verify_peername => "https" },
2393 ...
2394
2395Note that you must specify the hostname you connected to (or whatever
2396"peername" the protocol needs) as the C<peername> argument, otherwise no
2397peername verification will be done.
2398
2399The above will use the system-dependent default set of trusted CA
2400certificates. If you want to check against a specific CA, add the
2401C<ca_file> (or C<ca_cert>) arguments to C<tls_ctx>:
2402
2403 tls_ctx => {
2404 verify => 1,
2405 verify_peername => "https",
2406 ca_file => "my-ca-cert.pem",
2407 },
2408
2409=item I want to create a TLS/SSL server, how do I do that?
2410
2411Well, you first need to get a server certificate and key. You have
2412three options: a) ask a CA (buy one, use cacert.org etc.) b) create a
2413self-signed certificate (cheap. check the search engine of your choice,
2414there are many tutorials on the net) or c) make your own CA (tinyca2 is a
2415nice program for that purpose).
2416
2417Then create a file with your private key (in PEM format, see
2418L<AnyEvent::TLS>), followed by the certificate (also in PEM format). The
2419file should then look like this:
2420
2421 -----BEGIN RSA PRIVATE KEY-----
2422 ...header data
2423 ... lots of base64'y-stuff
2424 -----END RSA PRIVATE KEY-----
2425
2426 -----BEGIN CERTIFICATE-----
2427 ... lots of base64'y-stuff
2428 -----END CERTIFICATE-----
2429
2430The important bits are the "PRIVATE KEY" and "CERTIFICATE" parts. Then
2431specify this file as C<cert_file>:
2432
2433 tcp_server undef, $port, sub {
2434 my ($fh) = @_;
2435
2436 my $handle = new AnyEvent::Handle
2437 fh => $fh,
2438 tls => "accept",
2439 tls_ctx => { cert_file => "my-server-keycert.pem" },
2440 ...
2441
2442When you have intermediate CA certificates that your clients might not
2443know about, just append them to the C<cert_file>.
2444
2445=back
2446
2447=head1 SUBCLASSING AnyEvent::Handle
2448
2449In many cases, you might want to subclass AnyEvent::Handle.
2450
2451To make this easier, a given version of AnyEvent::Handle uses these
2452conventions:
2453
2454=over 4
2455
2456=item * all constructor arguments become object members.
2457
2458At least initially, when you pass a C<tls>-argument to the constructor it
2459will end up in C<< $handle->{tls} >>. Those members might be changed or
2460mutated later on (for example C<tls> will hold the TLS connection object).
2461
2462=item * other object member names are prefixed with an C<_>.
2463
2464All object members not explicitly documented (internal use) are prefixed
2465with an underscore character, so the remaining non-C<_>-namespace is free
2466for use for subclasses.
2467
2468=item * all members not documented here and not prefixed with an underscore
2469are free to use in subclasses.
2470
2471Of course, new versions of AnyEvent::Handle may introduce more "public"
2472member variables, but that's just life. At least it is documented.
2473
2474=back
2475
947=head1 AUTHOR 2476=head1 AUTHOR
948 2477
949Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>. 2478Robin Redeker C<< <elmex at ta-sa.org> >>, Marc Lehmann <schmorp@schmorp.de>.
950 2479
951=cut 2480=cut
952 2481
9531; # End of AnyEvent::Handle 24821
2483

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines