ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.58 by root, Sun Nov 14 20:23:00 2010 UTC vs.
Revision 1.92 by root, Tue Jan 4 08:17:59 2011 UTC

…		…
36		36
37	=cut	37	=cut
38		38
39	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
40		40
41	use strict;	41	use common::sense;
42	no warnings;
43		42
44	use Errno ();	43	use Errno ();
45		44
46	use AnyEvent 5.0 ();	45	use AnyEvent 5.0 ();
47	use AnyEvent::Util ();	46	use AnyEvent::Util ();
48	use AnyEvent::Socket ();
49	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
50		48
51	use base Exporter::;	49	use base Exporter::;
52		50
53	our $VERSION = '1.46';	51	our $VERSION = '2.0';
54		52
55	our @EXPORT = qw(http_get http_post http_head http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
56		54
57	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";	55	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
58	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
59	our $MAX_PERSISTENT = 8;
60	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
61	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
62		59	our $MAX_PER_HOST = 4; # changing this is evil
63	# changing these is evil
64	our $MAX_PERSISTENT_PER_HOST = 0;
65	our $MAX_PER_HOST = 4;
66		60
67	our $PROXY;	61	our $PROXY;
68	our $ACTIVE = 0;	62	our $ACTIVE = 0;
69		63
70	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
71	my %CO_SLOT; # number of open connections, and wait queue, per host	65	my %CO_SLOT; # number of open connections, and wait queue, per host
72		66
73	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
74		68
75	Executes an HTTP-GET request. See the http_request function for details on	69	Executes an HTTP-GET request. See the http_request function for details on
…		…
95	C<http_request> returns a "cancellation guard" - you have to keep the	89	C<http_request> returns a "cancellation guard" - you have to keep the
96	object at least alive until the callback get called. If the object gets	90	object at least alive until the callback get called. If the object gets
97	destroyed before the callback is called, the request will be cancelled.	91	destroyed before the callback is called, the request will be cancelled.
98		92
99	The callback will be called with the response body data as first argument	93	The callback will be called with the response body data as first argument
100	(or C<undef> if an error occured), and a hash-ref with response headers as	94	(or C<undef> if an error occured), and a hash-ref with response headers
101	second argument.	95	(and trailers) as second argument.
102		96
103	All the headers in that hash are lowercased. In addition to the response	97	All the headers in that hash are lowercased. In addition to the response
104	headers, the "pseudo-headers" (uppercase to avoid clashing with possible	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
105	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the	99	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
106	three parts of the HTTP Status-Line of the same name.	100	three parts of the HTTP Status-Line of the same name. If an error occurs
		101	during the body phase of a request, then the original C<Status> and
		102	C<Reason> values from the header are available as C<OrigStatus> and
		103	C<OrigReason>.
107		104
108	The pseudo-header C<URL> contains the actual URL (which can differ from	105	The pseudo-header C<URL> contains the actual URL (which can differ from
109	the requested URL when following redirects - for example, you might get	106	the requested URL when following redirects - for example, you might get
110	an error that your URL scheme is not supported even though your URL is a	107	an error that your URL scheme is not supported even though your URL is a
111	valid http URL because it redirected to an ftp URL, in which case you can	108	valid http URL because it redirected to an ftp URL, in which case you can
…		…
120		117
121	If the server sends a header multiple times, then their contents will be	118	If the server sends a header multiple times, then their contents will be
122	joined together with a comma (C<,>), as per the HTTP spec.	119	joined together with a comma (C<,>), as per the HTTP spec.
123		120
124	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
125	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<59x>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
126	(usually C<599>) and the C<Reason> pseudo-header will contain an error	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
127	message.	124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection etsbalishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
128		139
129	A typical callback might look like this:	140	A typical callback might look like this:
130		141
131	sub {	142	sub {
132	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
148	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication
149	retries and so on, and how often to do so.	160	retries and so on, and how often to do so.
150		161
151	=item headers => hashref	162	=item headers => hashref
152		163
153	The request headers to use. Currently, C<http_request> may provide its	164	The request headers to use. Currently, C<http_request> may provide its own
154	own C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers	165	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
155	and will provide defaults for C<User-Agent:> and C<Referer:> (this can be	166	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
156	suppressed by using C<undef> for these headers in which case they won't be	167	(this can be suppressed by using C<undef> for these headers in which case
157	sent at all).	168	they won't be sent at all).
		169
		170	You really should provide your own C<User-Agent:> header value that is
		171	appropriate for your program - I wouldn't be surprised if the default
		172	AnyEvent string gets blocked by webservers sooner or later.
158		173
159	=item timeout => $seconds	174	=item timeout => $seconds
160		175
161	The time-out to use for various stages - each connect attempt will reset	176	The time-out to use for various stages - each connect attempt will reset
162	the timeout, as will read or write activity, i.e. this is not an overall	177	the timeout, as will read or write activity, i.e. this is not an overall
…		…
167	=item proxy => [$host, $port[, $scheme]] or undef	182	=item proxy => [$host, $port[, $scheme]] or undef
168		183
169	Use the given http proxy for all requests. If not specified, then the	184	Use the given http proxy for all requests. If not specified, then the
170	default proxy (as specified by C<$ENV{http_proxy}>) is used.	185	default proxy (as specified by C<$ENV{http_proxy}>) is used.
171		186
172	C<$scheme> must be either missing, C<http> for HTTP or C<https> for	187	C<$scheme> must be either missing or must be C<http> for HTTP.
173	HTTPS.
174		188
175	=item body => $string	189	=item body => $string
176		190
177	The request body, usually empty. Will be-sent as-is (future versions of	191	The request body, usually empty. Will be sent as-is (future versions of
178	this module might offer more options).	192	this module might offer more options).
179		193
180	=item cookie_jar => $hash_ref	194	=item cookie_jar => $hash_ref
181		195
182	Passing this parameter enables (simplified) cookie-processing, loosely	196	Passing this parameter enables (simplified) cookie-processing, loosely
183	based on the original netscape specification.	197	based on the original netscape specification.
184		198
185	The C<$hash_ref> must be an (initially empty) hash reference which will	199	The C<$hash_ref> must be an (initially empty) hash reference which
186	get updated automatically. It is possible to save the cookie_jar to	200	will get updated automatically. It is possible to save the cookie jar
187	persistent storage with something like JSON or Storable, but this is not	201	to persistent storage with something like JSON or Storable - see the
188	recommended, as expiry times are currently being ignored.	202	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		203	expired or session-only cookies, and also for documentation on the format
		204	of the cookie jar.
189		205
190	Note that this cookie implementation is not of very high quality, nor	206	Note that this cookie implementation is not meant to be complete. If
191	meant to be complete. If you want complete cookie management you have to	207	you want complete cookie management you have to do that on your
192	do that on your own. C<cookie_jar> is meant as a quick fix to get some	208	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
193	cookie-using sites working. Cookies are a privacy disaster, do not use	209	working. Cookies are a privacy disaster, do not use them unless required
194	them unless required to.	210	to.
		211
		212	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		213	headers will be set and handled by this module, otherwise they will be
		214	left untouched.
195		215
196	=item tls_ctx => $scheme | $tls_ctx	216	=item tls_ctx => $scheme | $tls_ctx
197		217
198	Specifies the AnyEvent::TLS context to be used for https connections. This	218	Specifies the AnyEvent::TLS context to be used for https connections. This
199	parameter follows the same rules as the C<tls_ctx> parameter to	219	parameter follows the same rules as the C<tls_ctx> parameter to
…		…
203	verification) TLS context.	223	verification) TLS context.
204		224
205	The default for this option is C<low>, which could be interpreted as "give	225	The default for this option is C<low>, which could be interpreted as "give
206	me the page, no matter what".	226	me the page, no matter what".
207		227
		228	See also the C<sessionid> parameter.
		229
		230	=item session => $string
		231
		232	The module might reuse connections to the same host internally. Sometimes
		233	(e.g. when using TLS), you do not want to reuse connections from other
		234	sessions. This can be achieved by setting this parameter to some unique
		235	ID (such as the address of an object storing your state data, or the TLS
		236	context) - only connections using the same unique ID will be reused.
		237
208	=item on_prepare => $callback->($fh)	238	=item on_prepare => $callback->($fh)
209		239
210	In rare cases you need to "tune" the socket before it is used to	240	In rare cases you need to "tune" the socket before it is used to
211	connect (for exmaple, to bind it on a given IP address). This parameter	241	connect (for exmaple, to bind it on a given IP address). This parameter
212	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>	242	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
213	and behaves exactly the same way (e.g. it has to provide a	243	and behaves exactly the same way (e.g. it has to provide a
214	timeout). See the description for the C<$prepare_cb> argument of	244	timeout). See the description for the C<$prepare_cb> argument of
215	C<AnyEvent::Socket::tcp_connect> for details.	245	C<AnyEvent::Socket::tcp_connect> for details.
216		246
		247	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		248
		249	In even rarer cases you want total control over how AnyEvent::HTTP
		250	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		251	to do this, but you can provide your own C<tcp_connect> function -
		252	obviously, it has to follow the same calling conventions, except that it
		253	may always return a connection guard object.
		254
		255	There are probably lots of weird uses for this function, starting from
		256	tracing the hosts C<http_request> actually tries to connect, to (inexact
		257	but fast) host => IP address caching or even socks protocol support.
		258
217	=item on_header => $callback->($headers)	259	=item on_header => $callback->($headers)
218		260
219	When specified, this callback will be called with the header hash as soon	261	When specified, this callback will be called with the header hash as soon
220	as headers have been successfully received from the remote server (not on	262	as headers have been successfully received from the remote server (not on
221	locally-generated errors).	263	locally-generated errors).
…		…
226		268
227	This callback is useful, among other things, to quickly reject unwanted	269	This callback is useful, among other things, to quickly reject unwanted
228	content, which, if it is supposed to be rare, can be faster than first	270	content, which, if it is supposed to be rare, can be faster than first
229	doing a C<HEAD> request.	271	doing a C<HEAD> request.
230		272
		273	The downside is that cancelling the request makes it impossible to re-use
		274	the connection. Also, the C<on_header> callback will not receive any
		275	trailer (headers sent after the response body).
		276
231	Example: cancel the request unless the content-type is "text/html".	277	Example: cancel the request unless the content-type is "text/html".
232		278
233	on_header => sub {	279	on_header => sub {
234	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/	280	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
235	},	281	},
…		…
241	string instead of the body data.	287	string instead of the body data.
242		288
243	It has to return either true (in which case AnyEvent::HTTP will continue),	289	It has to return either true (in which case AnyEvent::HTTP will continue),
244	or false, in which case AnyEvent::HTTP will cancel the download (and call	290	or false, in which case AnyEvent::HTTP will cancel the download (and call
245	the completion callback with an error code of C<598>).	291	the completion callback with an error code of C<598>).
		292
		293	The downside to cancelling the request is that it makes it impossible to
		294	re-use the connection.
246		295
247	This callback is useful when the data is too large to be held in memory	296	This callback is useful when the data is too large to be held in memory
248	(so the callback writes it to a file) or when only some information should	297	(so the callback writes it to a file) or when only some information should
249	be extracted, or when the body should be processed incrementally.	298	be extracted, or when the body should be processed incrementally.
250		299
…		…
262	called. Instead of the C<$body> argument containing the body data, the	311	called. Instead of the C<$body> argument containing the body data, the
263	callback will receive the L<AnyEvent::Handle> object associated with the	312	callback will receive the L<AnyEvent::Handle> object associated with the
264	connection. In error cases, C<undef> will be passed. When there is no body	313	connection. In error cases, C<undef> will be passed. When there is no body
265	(e.g. status C<304>), the empty string will be passed.	314	(e.g. status C<304>), the empty string will be passed.
266		315
267	The handle object might or might not be in TLS mode, might be connected to	316	The handle object might or might not be in TLS mode, might be connected
268	a proxy, be a persistent connection etc., and configured in unspecified	317	to a proxy, be a persistent connection, use chunked transfer encoding
269	ways. The user is responsible for this handle (it will not be used by this	318	etc., and configured in unspecified ways. The user is responsible for this
270	module anymore).	319	handle (it will not be used by this module anymore).
271		320
272	This is useful with some push-type services, where, after the initial	321	This is useful with some push-type services, where, after the initial
273	headers, an interactive protocol is used (typical example would be the	322	headers, an interactive protocol is used (typical example would be the
274	push-style twitter API which starts a JSON/XML stream).	323	push-style twitter API which starts a JSON/XML stream).
275		324
276	If you think you need this, first have a look at C<on_body>, to see if	325	If you think you need this, first have a look at C<on_body>, to see if
277	that doesn't solve your problem in a better way.	326	that doesn't solve your problem in a better way.
278		327
		328	=item persistent => $boolean
		329
		330	Try to create/reuse a persistent connection. When this flag is set
		331	(default: true for idempotent requests, false for all others), then
		332	C<http_request> tries to re-use an existing (previously-created)
		333	persistent connection to the host and, failing that, tries to create a new
		334	one.
		335
		336	Requests failing in certain ways will be automatically retried once, which
		337	is dangerous for non-idempotent requests, which is why it defaults to off
		338	for them. The reason for this is because the bozos who designed HTTP/1.1
		339	made it impossible to distinguish between a fatal error and a normal
		340	connection timeout, so you never know whether there was a problem with
		341	your request or not.
		342
		343	When reusing an existent connection, many parameters (such as TLS context)
		344	will be ignored. See the C<session> parameter for a workaround.
		345
		346	=item keepalive => $boolean
		347
		348	Only used when C<persistent> is also true. This parameter decides whether
		349	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		350	(as opposed to only a HTTP/1.1 persistent connection).
		351
		352	The default is true, except when using a proxy, in which case it defaults
		353	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		354
		355	=item handle_params => { key => value ... }
		356
		357	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		358	constructor that is called - not all requests will create a handle, and
		359	sometimes more than one is created, so this parameter is only good for
		360	setting hints.
		361
		362	Example: set the maximum read size to 4096, to potentially conserve memory
		363	at the cost of speed.
		364
		365	handle_params => {
		366	max_read_size => 4096,
		367	},
		368
279	=back	369	=back
280		370
281	Example: make a simple HTTP GET request for http://www.nethype.de/	371	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		372	the response body.
282		373
283	http_request GET => "http://www.nethype.de/", sub {	374	http_request GET => "http://www.nethype.de/", sub {
284	my ($body, $hdr) = @_;	375	my ($body, $hdr) = @_;
285	print "$body\n";	376	print "$body\n";
286	};	377	};
287		378
288	Example: make a HTTP HEAD request on https://www.google.com/, use a	379	Example: do a HTTP HEAD request on https://www.google.com/, use a
289	timeout of 30 seconds.	380	timeout of 30 seconds.
290		381
291	http_request	382	http_request
292	GET => "https://www.google.com",	383	GET => "https://www.google.com",
		384	headers => { "user-agent" => "MySearchClient 1.0" },
293	timeout => 30,	385	timeout => 30,
294	sub {	386	sub {
295	my ($body, $hdr) = @_;	387	my ($body, $hdr) = @_;
296	use Data::Dumper;	388	use Data::Dumper;
297	print Dumper $hdr;	389	print Dumper $hdr;
298	}	390	}
299	;	391	;
300		392
301	Example: make another simple HTTP GET request, but immediately try to	393	Example: do another simple HTTP GET request, but immediately try to
302	cancel it.	394	cancel it.
303		395
304	my $request = http_request GET => "http://www.nethype.de/", sub {	396	my $request = http_request GET => "http://www.nethype.de/", sub {
305	my ($body, $hdr) = @_;	397	my ($body, $hdr) = @_;
306	print "$body\n";	398	print "$body\n";
307	};	399	};
308		400
309	undef $request;	401	undef $request;
310		402
311	=cut	403	=cut
		404
		405	#############################################################################
		406	# wait queue/slots
312		407
313	sub _slot_schedule;	408	sub _slot_schedule;
314	sub _slot_schedule($) {	409	sub _slot_schedule($) {
315	my $host = shift;	410	my $host = shift;
316		411
…		…
338	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	433	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
339		434
340	_slot_schedule $_[0];	435	_slot_schedule $_[0];
341	}	436	}
342		437
		438	#############################################################################
		439	# cookie handling
		440
		441	# expire cookies
		442	sub cookie_jar_expire($;$) {
		443	my ($jar, $session_end) = @_;
		444
		445	%$jar = () if $jar->{version} != 1;
		446
		447	my $anow = AE::now;
		448
		449	while (my ($chost, $paths) = each %$jar) {
		450	next unless ref $paths;
		451
		452	while (my ($cpath, $cookies) = each %$paths) {
		453	while (my ($cookie, $kv) = each %$cookies) {
		454	if (exists $kv->{_expires}) {
		455	delete $cookies->{$cookie}
		456	if $anow > $kv->{_expires};
		457	} elsif ($session_end) {
		458	delete $cookies->{$cookie};
		459	}
		460	}
		461
		462	delete $paths->{$cpath}
		463	unless %$cookies;
		464	}
		465
		466	delete $jar->{$chost}
		467	unless %$paths;
		468	}
		469	}
		470
		471	# extract cookies from jar
		472	sub cookie_jar_extract($$$$) {
		473	my ($jar, $scheme, $host, $path) = @_;
		474
		475	%$jar = () if $jar->{version} != 1;
		476
		477	my @cookies;
		478
		479	while (my ($chost, $paths) = each %$jar) {
		480	next unless ref $paths;
		481
		482	if ($chost =~ /^\./) {
		483	next unless $chost eq substr $host, -length $chost;
		484	} elsif ($chost =~ /\./) {
		485	next unless $chost eq $host;
		486	} else {
		487	next;
		488	}
		489
		490	while (my ($cpath, $cookies) = each %$paths) {
		491	next unless $cpath eq substr $path, 0, length $cpath;
		492
		493	while (my ($cookie, $kv) = each %$cookies) {
		494	next if $scheme ne "https" && exists $kv->{secure};
		495
		496	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		497	delete $cookies->{$cookie};
		498	next;
		499	}
		500
		501	my $value = $kv->{value};
		502
		503	if ($value =~ /[=;,[:space:]]/) {
		504	$value =~ s/([\\"])/\\$1/g;
		505	$value = "\"$value\"";
		506	}
		507
		508	push @cookies, "$cookie=$value";
		509	}
		510	}
		511	}
		512
		513	\@cookies
		514	}
		515
		516	# parse set_cookie header into jar
		517	sub cookie_jar_set_cookie($$$$) {
		518	my ($jar, $set_cookie, $host, $date) = @_;
		519
		520	my $anow = int AE::now;
		521	my $snow; # server-now
		522
		523	for ($set_cookie) {
		524	# parse NAME=VALUE
		525	my @kv;
		526
		527	# expires is not http-compliant in the original cookie-spec,
		528	# we support the official date format and some extensions
		529	while (
		530	m{
		531	\G\s*
		532	(?:
		533	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		534	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) ) )?
		535	)
		536	}gcxsi
		537	) {
		538	my $name = $2;
		539	my $value = $4;
		540
		541	if (defined $1) {
		542	# expires
		543	$name = "expires";
		544	$value = $1;
		545	} elsif (defined $3) {
		546	# quoted
		547	$value = $3;
		548	$value =~ s/\\(.)/$1/gs;
		549	}
		550
		551	push @kv, lc $name, $value;
		552
		553	last unless /\G\s*;/gc;
		554	}
		555
		556	last unless @kv;
		557
		558	my $name = shift @kv;
		559	my %kv = (value => shift @kv, @kv);
		560
		561	if (exists $kv{"max-age"}) {
		562	$kv{_expires} = $anow + delete $kv{"max-age"};
		563	} elsif (exists $kv{expires}) {
		564	$snow ||= parse_date ($date) || $anow;
		565	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		566	} else {
		567	delete $kv{_expires};
		568	}
		569
		570	my $cdom;
		571	my $cpath = (delete $kv{path}) || "/";
		572
		573	if (exists $kv{domain}) {
		574	$cdom = delete $kv{domain};
		575
		576	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		577
		578	next if $cdom =~ /\.$/;
		579
		580	# this is not rfc-like and not netscape-like. go figure.
		581	my $ndots = $cdom =~ y/.//;
		582	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		583	} else {
		584	$cdom = $host;
		585	}
		586
		587	# store it
		588	$jar->{version} = 1;
		589	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		590
		591	redo if /\G\s*,/gc;
		592	}
		593	}
		594
		595	#############################################################################
		596	# keepalive/persistent connection cache
		597
		598	# fetch a connection from the keepalive cache
		599	sub ka_fetch($) {
		600	my $ka_key = shift;
		601
		602	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		603	delete $KA_CACHE{$ka_key}
		604	unless @{ $KA_CACHE{$ka_key} };
		605
		606	$hdl
		607	}
		608
		609	sub ka_store($$) {
		610	my ($ka_key, $hdl) = @_;
		611
		612	my $kaa = $KA_CACHE{$ka_key} ||= [];
		613
		614	my $destroy = sub {
		615	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		616
		617	$hdl->destroy;
		618
		619	@ka
		620	? $KA_CACHE{$ka_key} = \@ka
		621	: delete $KA_CACHE{$ka_key};
		622	};
		623
		624	# on error etc., destroy
		625	$hdl->on_error ($destroy);
		626	$hdl->on_eof ($destroy);
		627	$hdl->on_read ($destroy);
		628	$hdl->timeout ($PERSISTENT_TIMEOUT);
		629
		630	push @$kaa, $hdl;
		631	shift @$kaa while @$kaa > $MAX_PER_HOST;
		632	}
		633
		634	#############################################################################
		635	# utilities
		636
		637	# continue to parse $_ for headers and place them into the arg
		638	sub _parse_hdr() {
		639	my %hdr;
		640
		641	# things seen, not parsed:
		642	# p3pP="NON CUR OTPi OUR NOR UNI"
		643
		644	$hdr{lc $1} .= ",$2"
		645	while /\G
		646	([^:\000-\037]*):
		647	[\011\040]*
		648	((?: [^\012]+ | \012[\011\040] )*)
		649	\012
		650	/gxc;
		651
		652	/\G$/
		653	or return;
		654
		655	# remove the "," prefix we added to all headers above
		656	substr $_, 0, 1, ""
		657	for values %hdr;
		658
		659	\%hdr
		660	}
		661
		662	#############################################################################
		663	# http_get
		664
343	our $qr_nlnl = qr{(?<![^\012])\015?\012};	665	our $qr_nlnl = qr{(?<![^\012])\015?\012};
344		666
345	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	667	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
346	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };	668	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		669
		670	# maybe it should just become a normal object :/
		671
		672	sub _destroy_state(\%) {
		673	my ($state) = @_;
		674
		675	$state->{handle}->destroy if $state->{handle};
		676	%$state = ();
		677	}
		678
		679	sub _error(\%$$) {
		680	my ($state, $cb, $hdr) = @_;
		681
		682	&_destroy_state ($state);
		683
		684	$cb->(undef, $hdr);
		685	()
		686	}
347		687
348	sub http_request($$@) {	688	sub http_request($$@) {
349	my $cb = pop;	689	my $cb = pop;
350	my ($method, $url, %arg) = @_;	690	my ($method, $url, %arg) = @_;
351		691
…		…
366	my @pseudo = (URL => $url);	706	my @pseudo = (URL => $url);
367	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};	707	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
368		708
369	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	709	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
370		710
371	return $cb->(undef, { Status => 599, Reason => "Too many redirections", @pseudo })	711	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
372	if $recurse < 0;	712	if $recurse < 0;
373		713
374	my $proxy = $arg{proxy} || $PROXY;	714	my $proxy = $arg{proxy} || $PROXY;
375	my $timeout = $arg{timeout} || $TIMEOUT;	715	my $timeout = $arg{timeout} || $TIMEOUT;
376		716
377	my ($uscheme, $uauthority, $upath, $query, $fragment) =	717	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
378	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	718	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;
379		719
380	$uscheme = lc $uscheme;	720	$uscheme = lc $uscheme;
381		721
382	my $uport = $uscheme eq "http" ? 80	722	my $uport = $uscheme eq "http" ? 80
383	: $uscheme eq "https" ? 443	723	: $uscheme eq "https" ? 443
384	: return $cb->(undef, { Status => 599, Reason => "Only http and https URL schemes supported", @pseudo });	724	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
385		725
386	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	726	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
387	or return $cb->(undef, { Status => 599, Reason => "Unparsable URL", @pseudo });	727	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
388		728
389	my $uhost = $1;	729	my $uhost = lc $1;
390	$uport = $2 if defined $2;	730	$uport = $2 if defined $2;
391		731
392	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	732	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
393	unless exists $hdr{host};	733	unless exists $hdr{host};
394		734
…		…
397		737
398	$upath =~ s%^/?%/%;	738	$upath =~ s%^/?%/%;
399		739
400	# cookie processing	740	# cookie processing
401	if (my $jar = $arg{cookie_jar}) {	741	if (my $jar = $arg{cookie_jar}) {
402	%$jar = () if $jar->{version} != 1;	742	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
403		743
404	my @cookie;
405
406	while (my ($chost, $v) = each %$jar) {
407	if ($chost =~ /^\./) {
408	next unless $chost eq substr $uhost, -length $chost;
409	} elsif ($chost =~ /\./) {
410	next unless $chost eq $uhost;
411	} else {
412	next;
413	}
414
415	while (my ($cpath, $v) = each %$v) {
416	next unless $cpath eq substr $upath, 0, length $cpath;
417
418	while (my ($k, $v) = each %$v) {
419	next if $uscheme ne "https" && exists $v->{secure};
420	my $value = $v->{value};
421	$value =~ s/([\\"])/\\$1/g;
422	push @cookie, "$k=\"$value\"";
423	}
424	}
425	}
426
427	$hdr{cookie} = join "; ", @cookie	744	$hdr{cookie} = join "; ", @$cookies
428	if @cookie;	745	if @$cookies;
429	}	746	}
430		747
431	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path	748	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
432		749
433	if ($proxy) {	750	if ($proxy) {
…		…
436	$rscheme = "http" unless defined $rscheme;	753	$rscheme = "http" unless defined $rscheme;
437		754
438	# don't support https requests over https-proxy transport,	755	# don't support https requests over https-proxy transport,
439	# can't be done with tls as spec'ed, unless you double-encrypt.	756	# can't be done with tls as spec'ed, unless you double-encrypt.
440	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";	757	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		758
		759	$rhost = lc $rhost;
		760	$rscheme = lc $rscheme;
441	} else {	761	} else {
442	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	762	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
443	}	763	}
444		764
445	# leave out fragment and query string, just a heuristic	765	# leave out fragment and query string, just a heuristic
446	$hdr{referer} ||= "$uscheme://$uauthority$upath" unless exists $hdr{referer};	766	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
447	$hdr{"user-agent"} ||= $USERAGENT unless exists $hdr{"user-agent"};	767	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
448		768
449	$hdr{"content-length"} = length $arg{body}	769	$hdr{"content-length"} = length $arg{body}
450	if length $arg{body} || $method ne "GET";	770	if length $arg{body} || $method ne "GET";
451		771
		772	my $idempotent = $method =~ /^(?:GET|HEAD|PUT|DELETE|OPTIONS|TRACE)$/;
		773
		774	# default value for keepalive is true iff the request is for an idempotent method
		775	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : $idempotent;
		776	my $keepalive10 = exists $arg{keepalive10} ? $arg{keepalive10} : !$proxy;
		777	my $keptalive; # true if this is actually a recycled connection
		778
		779	# the key to use in the keepalive cache
		780	my $ka_key = "$uhost\x00$arg{sessionid}";
		781
		782	$hdr{connection} = ($keepalive ? $keepalive10 ? "keep-alive " : "" : "close ") . "Te"; #1.1
		783	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		784
452	my %state = (connect_guard => 1);	785	my %state = (connect_guard => 1);
		786
		787	my $ae_error = 595; # connecting
		788
		789	# handle actual, non-tunneled, request
		790	my $handle_actual_request = sub {
		791	$ae_error = 596; # request phase
		792
		793	my $hdl = $state{handle};
		794
		795	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		796
		797	# send request
		798	$hdl->push_write (
		799	"$method $rpath HTTP/1.1\015\012"
		800	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		801	. "\015\012"
		802	. (delete $arg{body})
		803	);
		804
		805	# return if error occured during push_write()
		806	return unless %state;
		807
		808	# reduce memory usage, save a kitten, also re-use it for the response headers.
		809	%hdr = ();
		810
		811	# status line and headers
		812	$state{read_response} = sub {
		813	return unless %state;
		814
		815	for ("$_[1]") {
		816	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		817
		818	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		819	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		820
		821	# 100 Continue handling
		822	# should not happen as we don't send expect: 100-continue,
		823	# but we handle it just in case.
		824	# since we send the request body regardless, if we get an error
		825	# we are out of-sync, which we currently do NOT handle correctly.
		826	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		827	if $2 eq 100;
		828
		829	push @pseudo,
		830	HTTPVersion => $1,
		831	Status => $2,
		832	Reason => $3,
		833	;
		834
		835	my $hdr = _parse_hdr
		836	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		837
		838	%hdr = (%$hdr, @pseudo);
		839	}
		840
		841	# redirect handling
		842	# microsoft and other shitheads don't give a shit for following standards,
		843	# try to support some common forms of broken Location headers.
		844	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
		845	$hdr{location} =~ s/^\.\/+//;
		846
		847	my $url = "$rscheme://$uhost:$uport";
		848
		849	unless ($hdr{location} =~ s/^\///) {
		850	$url .= $upath;
		851	$url =~ s/\/[^\/]*$//;
		852	}
		853
		854	$hdr{location} = "$url/$hdr{location}";
		855	}
		856
		857	my $redirect;
		858
		859	if ($recurse) {
		860	my $status = $hdr{Status};
		861
		862	# industry standard is to redirect POST as GET for
		863	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		864	# also, the UA should ask the user for 301 and 307 and POST,
		865	# industry standard seems to be to simply follow.
		866	# we go with the industry standard.
		867	if ($status == 301 or $status == 302 or $status == 303) {
		868	# HTTP/1.1 is unclear on how to mutate the method
		869	$method = "GET" unless $method eq "HEAD";
		870	$redirect = 1;
		871	} elsif ($status == 307) {
		872	$redirect = 1;
		873	}
		874	}
		875
		876	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
		877	if ($state{handle}) {
		878	# handle keepalive
		879	if (
		880	$keepalive
		881	&& $_[3]
		882	&& ($hdr{HTTPVersion} < 1.1
		883	? $hdr{connection} =~ /\bkeep-?alive\b/i
		884	: $hdr{connection} !~ /\bclose\b/i)
		885	) {
		886	ka_store $ka_key, delete $state{handle};
		887	} else {
		888	# no keepalive, destroy the handle
		889	$state{handle}->destroy;
		890	}
		891	}
		892
		893	%state = ();
		894
		895	if (defined $_[1]) {
		896	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		897	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		898	}
		899
		900	# set-cookie processing
		901	if ($arg{cookie_jar}) {
		902	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		903	}
		904
		905	if ($redirect && exists $hdr{location}) {
		906	# we ignore any errors, as it is very common to receive
		907	# Content-Length != 0 but no actual body
		908	# we also access %hdr, as $_[1] might be an erro
		909	http_request (
		910	$method => $hdr{location},
		911	%arg,
		912	recurse => $recurse - 1,
		913	Redirect => [$_[0], \%hdr],
		914	$cb
		915	);
		916	} else {
		917	$cb->($_[0], \%hdr);
		918	}
		919	};
		920
		921	$ae_error = 597; # body phase
		922
		923	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		924
		925	my $len = $chunked ? undef : $hdr{"content-length"};
		926
		927	# body handling, many different code paths
		928	# - no body expected
		929	# - want_body_handle
		930	# - te chunked
		931	# - 2x length known (with or without on_body)
		932	# - 2x length not known (with or without on_body)
		933	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		934	$finish->(undef, 598 => "Request cancelled by on_header");
		935	} elsif (
		936	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		937	or $method eq "HEAD"
		938	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		939	) {
		940	# no body
		941	$finish->("", undef, undef, 1);
		942
		943	} elsif (!$redirect && $arg{want_body_handle}) {
		944	$_[0]->on_eof (undef);
		945	$_[0]->on_error (undef);
		946	$_[0]->on_read (undef);
		947
		948	$finish->(delete $state{handle});
		949
		950	} elsif ($chunked) {
		951	my $cl = 0;
		952	my $body = "";
		953	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		954
		955	$state{read_chunk} = sub {
		956	$_[1] =~ /^([0-9a-fA-F]+)/
		957	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		958
		959	my $len = hex $1;
		960
		961	if ($len) {
		962	$cl += $len;
		963
		964	$_[0]->push_read (chunk => $len, sub {
		965	$on_body->($_[1], \%hdr)
		966	or return $finish->(undef, 598 => "Request cancelled by on_body");
		967
		968	$_[0]->push_read (line => sub {
		969	length $_[1]
		970	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		971	$_[0]->push_read (line => $state{read_chunk});
		972	});
		973	});
		974	} else {
		975	$hdr{"content-length"} ||= $cl;
		976
		977	$_[0]->push_read (line => $qr_nlnl, sub {
		978	if (length $_[1]) {
		979	for ("$_[1]") {
		980	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		981
		982	my $hdr = _parse_hdr
		983	or return $finish->(undef, $ae_error => "Garbled response trailers");
		984
		985	%hdr = (%hdr, %$hdr);
		986	}
		987	}
		988
		989	$finish->($body, undef, undef, 1);
		990	});
		991	}
		992	};
		993
		994	$_[0]->push_read (line => $state{read_chunk});
		995
		996	} elsif ($arg{on_body}) {
		997	if (defined $len) {
		998	$_[0]->on_read (sub {
		999	$len -= length $_[0]{rbuf};
		1000
		1001	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1002	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1003
		1004	$len > 0
		1005	or $finish->("", undef, undef, 1);
		1006	});
		1007	} else {
		1008	$_[0]->on_eof (sub {
		1009	$finish->("");
		1010	});
		1011	$_[0]->on_read (sub {
		1012	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1013	or $finish->(undef, 598 => "Request cancelled by on_body");
		1014	});
		1015	}
		1016	} else {
		1017	$_[0]->on_eof (undef);
		1018
		1019	if (defined $len) {
		1020	$_[0]->on_read (sub {
		1021	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1022	if $len <= length $_[0]{rbuf};
		1023	});
		1024	} else {
		1025	$_[0]->on_error (sub {
		1026	($! == Errno::EPIPE || !$!)
		1027	? $finish->(delete $_[0]{rbuf})
		1028	: $finish->(undef, $ae_error => $_[2]);
		1029	});
		1030	$_[0]->on_read (sub { });
		1031	}
		1032	}
		1033	};
		1034
		1035	# if keepalive is enabled, then the server closing the connection
		1036	# before a response can happen legally - we retry on idempotent methods.
		1037	if ($keptalive && $idempotent) {
		1038	my $old_eof = $hdl->{on_eof};
		1039	$hdl->{on_eof} = sub {
		1040	_destroy_state %state;
		1041
		1042	http_request (
		1043	$method => $url,
		1044	%arg,
		1045	keepalive => 0,
		1046	$cb
		1047	);
		1048	};
		1049	$hdl->on_read (sub {
		1050	return unless %state;
		1051
		1052	# as soon as we receive something, a connection close
		1053	# once more becomes a hard error
		1054	$hdl->{on_eof} = $old_eof;
		1055	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1056	});
		1057	} else {
		1058	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1059	}
		1060	};
		1061
		1062	my $prepare_handle = sub {
		1063	my ($hdl) = $state{handle};
		1064
		1065	$hdl->timeout ($timeout);
		1066	$hdl->on_error (sub {
		1067	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1068	});
		1069	$hdl->on_eof (sub {
		1070	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1071	});
		1072	};
		1073
		1074	# connected to proxy (or origin server)
		1075	my $connect_cb = sub {
		1076	my $fh = shift
		1077	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1078
		1079	return unless delete $state{connect_guard};
		1080
		1081	# get handle
		1082	$state{handle} = new AnyEvent::Handle
		1083	%{ $arg{handle_params} },
		1084	fh => $fh,
		1085	peername => $uhost,
		1086	tls_ctx => $arg{tls_ctx},
		1087	;
		1088
		1089	$prepare_handle->();
		1090
		1091	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1092
		1093	# now handle proxy-CONNECT method
		1094	if ($proxy && $uscheme eq "https") {
		1095	# oh dear, we have to wrap it into a connect request
		1096
		1097	# maybe re-use $uauthority with patched port?
		1098	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1099	$state{handle}->push_read (line => $qr_nlnl, sub {
		1100	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1101	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1102
		1103	if ($2 == 200) {
		1104	$rpath = $upath;
		1105	$handle_actual_request->();
		1106	} else {
		1107	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1108	}
		1109	});
		1110	} else {
		1111	$handle_actual_request->();
		1112	}
		1113	};
453		1114
454	_get_slot $uhost, sub {	1115	_get_slot $uhost, sub {
455	$state{slot_guard} = shift;	1116	$state{slot_guard} = shift;
456		1117
457	return unless $state{connect_guard};	1118	return unless $state{connect_guard};
458		1119
459	$state{connect_guard} = AnyEvent::Socket::tcp_connect $rhost, $rport, sub {	1120	# try to use an existing keepalive connection, but only if we, ourselves, plan
460	$state{fh} = shift	1121	# on a keepalive request (in theory, this should be a separate config option).
461	or do {	1122	if ($keepalive && $KA_CACHE{$ka_key}) {
462	my $err = "$!";	1123	$keptalive = 1;
463	%state = ();	1124	$state{handle} = ka_fetch $ka_key;
464	return $cb->(undef, { Status => 599, Reason => $err, @pseudo });	1125	$prepare_handle->();
465	};
466
467	pop; # free memory, save a tree
468
469	return unless delete $state{connect_guard};
470
471	# get handle
472	$state{handle} = new AnyEvent::Handle
473	fh => $state{fh},
474	peername => $rhost,
475	tls_ctx => $arg{tls_ctx},
476	# these need to be reconfigured on keepalive handles
477	timeout => $timeout,
478	on_error => sub {
479	%state = ();
480	$cb->(undef, { Status => 599, Reason => $_[2], @pseudo });
481	},
482	on_eof => sub {
483	%state = ();
484	$cb->(undef, { Status => 599, Reason => "Unexpected end-of-file", @pseudo });
485	},
486	;
487
488	# limit the number of persistent connections
489	# keepalive not yet supported
490	# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
491	# ++$KA_COUNT{$_[1]};
492	# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
493	# --$KA_COUNT{$_[1]}
494	# };
495	# $hdr{connection} = "keep-alive";
496	# } else {
497	delete $hdr{connection};
498	# }
499
500	$state{handle}->starttls ("connect") if $rscheme eq "https";
501
502	# handle actual, non-tunneled, request
503	my $handle_actual_request = sub {
504	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};
505
506	# send request
507	$state{handle}->push_write (
508	"$method $rpath HTTP/1.0\015\012"
509	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
510	. "\015\012"
511	. (delete $arg{body})
512	);
513
514	# return if error occured during push_write()
515	return unless %state;
516
517	%hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
518
519	# status line and headers
520	$state{handle}->push_read (line => $qr_nlnl, sub {
521	for ("$_[1]") {
522	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
523
524	/^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )? \015?\012/igxc
525	or return (%state = (), $cb->(undef, { Status => 599, Reason => "Invalid server response", @pseudo }));
526
527	push @pseudo,
528	HTTPVersion => $1,
529	Status => $2,
530	Reason => $3,
531	;
532
533	# things seen, not parsed:
534	# p3pP="NON CUR OTPi OUR NOR UNI"
535
536	$hdr{lc $1} .= ",$2"
537	while /\G
538	([^:\000-\037]*):
539	[\011\040]*
540	((?: [^\012]+ | \012[\011\040] )*)
541	\012
542	/gxc;
543
544	/\G$/
545	or return (%state = (), $cb->(undef, { Status => 599, Reason => "Garbled response headers", @pseudo }));
546	}
547
548	# remove the "," prefix we added to all headers above
549	substr $_, 0, 1, ""
550	for values %hdr;
551
552	# patch in all pseudo headers
553	%hdr = (%hdr, @pseudo);
554
555	# redirect handling
556	# microsoft and other shitheads don't give a shit for following standards,
557	# try to support some common forms of broken Location headers.
558	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
559	$hdr{location} =~ s/^\.\/+//;
560
561	my $url = "$rscheme://$uhost:$uport";
562
563	unless ($hdr{location} =~ s/^\///) {
564	$url .= $upath;
565	$url =~ s/\/[^\/]*$//;
566	}
567
568	$hdr{location} = "$url/$hdr{location}";
569	}
570
571	my $redirect;
572
573	if ($recurse) {
574	my $status = $hdr{Status};
575
576	# industry standard is to redirect POST as GET for
577	# 301, 302 and 303, in contrast to http/1.0 and 1.1.
578	# also, the UA should ask the user for 301 and 307 and POST,
579	# industry standard seems to be to simply follow.
580	# we go with the industry standard.
581	if ($status == 301 or $status == 302 or $status == 303) {
582	# HTTP/1.1 is unclear on how to mutate the method
583	$method = "GET" unless $method eq "HEAD";
584	$redirect = 1;
585	} elsif ($status == 307) {
586	$redirect = 1;
587	}
588	}
589
590	my $finish = sub {
591	$state{handle}->destroy if $state{handle};
592	%state = ();
593
594	# set-cookie processing
595	if ($arg{cookie_jar}) {
596	for ($_[1]{"set-cookie"}) {
597	# parse NAME=VALUE
598	my @kv;
599
600	while (/\G\s* ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )/gcxs) {
601	my $name = $1;
602	my $value = $3;
603
604	unless ($value) {
605	$value = $2;
606	$value =~ s/\\(.)/$1/gs;
607	}
608
609	push @kv, $name => $value;
610
611	last unless /\G\s*;/gc;
612	}
613
614	last unless @kv;
615
616	my $name = shift @kv;
617	my %kv = (value => shift @kv, @kv);
618
619	my $cdom;
620	my $cpath = (delete $kv{path}) || "/";
621
622	if (exists $kv{domain}) {
623	$cdom = delete $kv{domain};
624
625	$cdom =~ s/^\.?/./; # make sure it starts with a "."
626
627	next if $cdom =~ /\.$/;
628
629	# this is not rfc-like and not netscape-like. go figure.
630	my $ndots = $cdom =~ y/.//;
631	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
632	} else {
633	$cdom = $uhost;
634	}
635
636	# store it
637	$arg{cookie_jar}{version} = 1;
638	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
639
640	redo if /\G\s*,/gc;
641	}
642	}
643
644	if ($redirect && exists $hdr{location}) {
645	# we ignore any errors, as it is very common to receive
646	# Content-Length != 0 but no actual body
647	# we also access %hdr, as $_[1] might be an erro
648	http_request (
649	$method => $hdr{location},
650	%arg,
651	recurse => $recurse - 1,
652	Redirect => \@_,
653	$cb);
654	} else {
655	$cb->($_[0], $_[1]);
656	}
657	};
658
659	my $len = $hdr{"content-length"};
660
661	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
662	$finish->(undef, { Status => 598, Reason => "Request cancelled by on_header", @pseudo });
663	} elsif (
664	$hdr{Status} =~ /^(?:1..|[23]04)$/
665	or $method eq "HEAD"
666	or (defined $len && !$len)
667	) {
668	# no body
669	$finish->("", \%hdr);
670	} else {
671	# body handling, four different code paths
672	# for want_body_handle, on_body (2x), normal (2x)
673	# we might read too much here, but it does not matter yet (no pers. connections)
674	if (!$redirect && $arg{want_body_handle}) {
675	$_[0]->on_eof (undef);
676	$_[0]->on_error (undef);
677	$_[0]->on_read (undef);
678
679	$finish->(delete $state{handle}, \%hdr);
680
681	} elsif ($arg{on_body}) {
682	$_[0]->on_error (sub { $finish->(undef, { Status => 599, Reason => $_[2], @pseudo }) });
683	if ($len) {
684	$_[0]->on_eof (undef);
685	$_[0]->on_read (sub {
686	$len -= length $_[0]{rbuf};
687
688	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
689	or $finish->(undef, { Status => 598, Reason => "Request cancelled by on_body", @pseudo });
690
691	$len > 0
692	or $finish->("", \%hdr);
693	});
694	} else {
695	$_[0]->on_eof (sub {
696	$finish->("", \%hdr);
697	});
698	$_[0]->on_read (sub {
699	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
700	or $finish->(undef, { Status => 598, Reason => "Request cancelled by on_body", @pseudo });
701	});
702	}
703	} else {
704	$_[0]->on_eof (undef);
705
706	if ($len) {
707	$_[0]->on_error (sub { $finish->(undef, { Status => 599, Reason => $_[2], @pseudo }) });
708	$_[0]->on_read (sub {
709	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), \%hdr)
710	if $len <= length $_[0]{rbuf};
711	});
712	} else {
713	$_[0]->on_error (sub {
714	($! == Errno::EPIPE || !$!)
715	? $finish->(delete $_[0]{rbuf}, \%hdr)
716	: $finish->(undef, { Status => 599, Reason => $_[2], @pseudo });
717	});
718	$_[0]->on_read (sub { });
719	}
720	}
721	}
722	});
723	};
724
725	# now handle proxy-CONNECT method
726	if ($proxy && $uscheme eq "https") {
727	# oh dear, we have to wrap it into a connect request
728
729	# maybe re-use $uauthority with patched port?
730	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
731	$state{handle}->push_read (line => $qr_nlnl, sub {
732	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
733	or return (%state = (), $cb->(undef, { Status => 599, Reason => "Invalid proxy connect response ($_[1])", @pseudo }));
734
735	if ($2 == 200) {
736	$rpath = $upath;
737	&$handle_actual_request;	1126	$handle_actual_request->();
738	} else {	1127
739	%state = ();
740	$cb->(undef, { Status => $2, Reason => $3, @pseudo });
741	}
742	});
743	} else {	1128	} else {
744	&$handle_actual_request;	1129	my $tcp_connect = $arg{tcp_connect}
		1130	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1131
		1132	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
745	}	1133	}
746
747	}, $arg{on_prepare} || sub { $timeout };
748	};	1134	};
749		1135
750	defined wantarray && AnyEvent::Util::guard { %state = () }	1136	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
751	}	1137	}
752		1138
753	sub http_get($@) {	1139	sub http_get($@) {
754	unshift @_, "GET";	1140	unshift @_, "GET";
755	&http_request	1141	&http_request
…		…
773	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1159	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
774	the actual connection, which in turn uses AnyEvent::DNS to resolve	1160	the actual connection, which in turn uses AnyEvent::DNS to resolve
775	hostnames. The latter is a simple stub resolver and does no caching	1161	hostnames. The latter is a simple stub resolver and does no caching
776	on its own. If you want DNS caching, you currently have to provide	1162	on its own. If you want DNS caching, you currently have to provide
777	your own default resolver (by storing a suitable resolver object in	1163	your own default resolver (by storing a suitable resolver object in
778	C<$AnyEvent::DNS::RESOLVER>).	1164	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
779		1165
780	=head2 GLOBAL FUNCTIONS AND VARIABLES	1166	=head2 GLOBAL FUNCTIONS AND VARIABLES
781		1167
782	=over 4	1168	=over 4
783		1169
784	=item AnyEvent::HTTP::set_proxy "proxy-url"	1170	=item AnyEvent::HTTP::set_proxy "proxy-url"
785		1171
786	Sets the default proxy server to use. The proxy-url must begin with a	1172	Sets the default proxy server to use. The proxy-url must begin with a
787	string of the form C<http://host:port> (optionally C<https:...>), croaks	1173	string of the form C<http://host:port>, croaks otherwise.
788	otherwise.
789		1174
790	To clear an already-set proxy, use C<undef>.	1175	To clear an already-set proxy, use C<undef>.
791		1176
		1177	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1178
		1179	Remove all cookies from the cookie jar that have been expired. If
		1180	C<$session_end> is given and true, then additionally remove all session
		1181	cookies.
		1182
		1183	You should call this function (with a true C<$session_end>) before you
		1184	save cookies to disk, and you should call this function after loading them
		1185	again. If you have a long-running program you can additonally call this
		1186	function from time to time.
		1187
		1188	A cookie jar is initially an empty hash-reference that is managed by this
		1189	module. It's format is subject to change, but currently it is like this:
		1190
		1191	The key C<version> has to contain C<1>, otherwise the hash gets
		1192	emptied. All other keys are hostnames or IP addresses pointing to
		1193	hash-references. The key for these inner hash references is the
		1194	server path for which this cookie is meant, and the values are again
		1195	hash-references. The keys of those hash-references is the cookie name, and
		1196	the value, you guessed it, is another hash-reference, this time with the
		1197	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1198	which have been replaced by a C<_expires> key that contains the cookie
		1199	expiry timestamp.
		1200
		1201	Here is an example of a cookie jar with a single cookie, so you have a
		1202	chance of understanding the above paragraph:
		1203
		1204	{
		1205	version => 1,
		1206	"10.0.0.1" => {
		1207	"/" => {
		1208	"mythweb_id" => {
		1209	_expires => 1293917923,
		1210	value => "ooRung9dThee3ooyXooM1Ohm",
		1211	},
		1212	},
		1213	},
		1214	}
		1215
		1216	=item $date = AnyEvent::HTTP::format_date $timestamp
		1217
		1218	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1219	Date (RFC 2616).
		1220
		1221	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1222
		1223	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1224	bunch of minor variations of those, and returns the corresponding POSIX
		1225	timestamp, or C<undef> if the date cannot be parsed.
		1226
792	=item $AnyEvent::HTTP::MAX_RECURSE	1227	=item $AnyEvent::HTTP::MAX_RECURSE
793		1228
794	The default value for the C<recurse> request parameter (default: C<10>).	1229	The default value for the C<recurse> request parameter (default: C<10>).
		1230
		1231	=item $AnyEvent::HTTP::TIMEOUT
		1232
		1233	The default timeout for conenction operations (default: C<300>).
795		1234
796	=item $AnyEvent::HTTP::USERAGENT	1235	=item $AnyEvent::HTTP::USERAGENT
797		1236
798	The default value for the C<User-Agent> header (the default is	1237	The default value for the C<User-Agent> header (the default is
799	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1238	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
800		1239
801	=item $AnyEvent::HTTP::MAX_PER_HOST	1240	=item $AnyEvent::HTTP::MAX_PER_HOST
802		1241
803	The maximum number of concurrent connections to the same host (identified	1242	The maximum number of concurrent connections to the same host (identified
804	by the hostname). If the limit is exceeded, then the additional requests	1243	by the hostname). If the limit is exceeded, then the additional requests
805	are queued until previous connections are closed.	1244	are queued until previous connections are closed. Both persistent and
		1245	non-persistent connections are counted in this limit.
806		1246
807	The default value for this is C<4>, and it is highly advisable to not	1247	The default value for this is C<4>, and it is highly advisable to not
808	increase it.	1248	increase it much.
		1249
		1250	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1251	connections, older browsers used 2, newers (such as firefox 3) typically
		1252	use 6, and Opera uses 8 because like, they have the fastest browser and
		1253	give a shit for everybody else on the planet.
		1254
		1255	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1256
		1257	The time after which idle persistent conenctions get closed by
		1258	AnyEvent::HTTP (default: C<3>).
809		1259
810	=item $AnyEvent::HTTP::ACTIVE	1260	=item $AnyEvent::HTTP::ACTIVE
811		1261
812	The number of active connections. This is not the number of currently	1262	The number of active connections. This is not the number of currently
813	running requests, but the number of currently open and non-idle TCP	1263	running requests, but the number of currently open and non-idle TCP
814	connections. This number of can be useful for load-leveling.	1264	connections. This number can be useful for load-leveling.
815		1265
816	=back	1266	=back
817		1267
818	=cut	1268	=cut
		1269
		1270	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1271	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1272
		1273	sub format_date($) {
		1274	my ($time) = @_;
		1275
		1276	# RFC 822/1123 format
		1277	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1278
		1279	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1280	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1281	$H, $M, $S;
		1282	}
		1283
		1284	sub parse_date($) {
		1285	my ($date) = @_;
		1286
		1287	my ($d, $m, $y, $H, $M, $S);
		1288
		1289	if ($date =~ /^[A-Z][a-z][a-z]+, ([0-9][0-9]?)[\- ]([A-Z][a-z][a-z])[\- ]([0-9][0-9][0-9][0-9]) ([0-9][0-9]?):([0-9][0-9]?):([0-9][0-9]?) GMT$/) {
		1290	# RFC 822/1123, required by RFC 2616 (with " ")
		1291	# cookie dates (with "-")
		1292
		1293	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1294
		1295	} elsif ($date =~ /^[A-Z][a-z][a-z]+, ([0-9][0-9]?)-([A-Z][a-z][a-z])-([0-9][0-9]) ([0-9][0-9]?):([0-9][0-9]?):([0-9][0-9]?) GMT$/) {
		1296	# RFC 850
		1297	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1298
		1299	} elsif ($date =~ /^[A-Z][a-z][a-z]+ ([A-Z][a-z][a-z]) ([0-9 ]?[0-9]) ([0-9][0-9]?):([0-9][0-9]?):([0-9][0-9]?) ([0-9][0-9][0-9][0-9])$/) {
		1300	# ISO C's asctime
		1301	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1302	}
		1303	# other formats fail in the loop below
		1304
		1305	for (0..11) {
		1306	if ($m eq $month[$_]) {
		1307	require Time::Local;
		1308	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);
		1309	}
		1310	}
		1311
		1312	undef
		1313	}
819		1314
820	sub set_proxy($) {	1315	sub set_proxy($) {
821	if (length $_[0]) {	1316	if (length $_[0]) {
822	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1317	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
823	or Carp::croak "$_[0]: invalid proxy URL";	1318	or Carp::croak "$_[0]: invalid proxy URL";
824	$PROXY = [$2, $3 || 3128, $1]	1319	$PROXY = [$2, $3 || 3128, $1]
825	} else {	1320	} else {
826	undef $PROXY;	1321	undef $PROXY;
827	}	1322	}
…		…
830	# initialise proxy from environment	1325	# initialise proxy from environment
831	eval {	1326	eval {
832	set_proxy $ENV{http_proxy};	1327	set_proxy $ENV{http_proxy};
833	};	1328	};
834		1329
		1330	=head2 SOCKS PROXIES
		1331
		1332	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1333	compile your perl to support socks, or use an external program such as
		1334	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1335	transparently.
		1336
		1337	Alternatively, for AnyEvent::HTTP only, you can use your own
		1338	C<tcp_connect> function that does the proxy handshake - here is an example
		1339	that works with socks4a proxies:
		1340
		1341	use Errno;
		1342	use AnyEvent::Util;
		1343	use AnyEvent::Socket;
		1344	use AnyEvent::Handle;
		1345
		1346	# host, port and username of/for your socks4a proxy
		1347	my $socks_host = "10.0.0.23";
		1348	my $socks_port = 9050;
		1349	my $socks_user = "";
		1350
		1351	sub socks4a_connect {
		1352	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1353
		1354	my $hdl = new AnyEvent::Handle
		1355	connect => [$socks_host, $socks_port],
		1356	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1357	on_error => sub { $connect_cb->() },
		1358	;
		1359
		1360	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1361
		1362	$hdl->push_read (chunk => 8, sub {
		1363	my ($hdl, $chunk) = @_;
		1364	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1365
		1366	if ($status == 0x5a) {
		1367	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1368	} else {
		1369	$! = Errno::ENXIO; $connect_cb->();
		1370	}
		1371	});
		1372
		1373	$hdl
		1374	}
		1375
		1376	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1377	possibly after switching off other proxy types:
		1378
		1379	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1380
		1381	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1382	my ($data, $headers) = @_;
		1383	...
		1384	};
		1385
835	=head1 SEE ALSO	1386	=head1 SEE ALSO
836		1387
837	L<AnyEvent>.	1388	L<AnyEvent>.
838		1389
839	=head1 AUTHOR	1390	=head1 AUTHOR

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines