ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.11 by root, Thu Jun 5 15:34:00 2008 UTC vs.
Revision 1.103 by root, Thu Feb 24 12:13:11 2011 UTC

…		…
3	AnyEvent::HTTP - simple but non-blocking HTTP/HTTPS client	3	AnyEvent::HTTP - simple but non-blocking HTTP/HTTPS client
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use AnyEvent::HTTP;	7	use AnyEvent::HTTP;
		8
		9	http_get "http://www.nethype.de/", sub { print $_[1] };
		10
		11	# ... do something else here
8		12
9	=head1 DESCRIPTION	13	=head1 DESCRIPTION
10		14
11	This module is an L<AnyEvent> user, you need to make sure that you use and	15	This module is an L<AnyEvent> user, you need to make sure that you use and
12	run a supported event loop.	16	run a supported event loop.
13		17
14	This module implements a simple, stateless and non-blocking HTTP	18	This module implements a simple, stateless and non-blocking HTTP
15	client. It supports GET, POST and other request methods, cookies and more,	19	client. It supports GET, POST and other request methods, cookies and more,
16	all on a very low level. It can follow redirects supports proxies and	20	all on a very low level. It can follow redirects, supports proxies, and
17	automatically limits the number of connections to the values specified in	21	automatically limits the number of connections to the values specified in
18	the RFC.	22	the RFC.
19		23
20	It should generally be a "good client" that is enough for most HTTP	24	It should generally be a "good client" that is enough for most HTTP
21	tasks. Simple tasks should be simple, but complex tasks should still be	25	tasks. Simple tasks should be simple, but complex tasks should still be
…		…
32		36
33	=cut	37	=cut
34		38
35	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
36		40
37	use strict;	41	use common::sense;
38	no warnings;
39		42
40	use Carp;	43	use Errno ();
41		44
42	use AnyEvent ();	45	use AnyEvent 5.0 ();
43	use AnyEvent::Util ();	46	use AnyEvent::Util ();
44	use AnyEvent::Socket ();
45	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
46		48
47	use base Exporter::;	49	use base Exporter::;
48		50
49	our $VERSION = '1.0';	51	our $VERSION = '2.1';
50		52
51	our @EXPORT = qw(http_get http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
52		54
53	our $USERAGENT = "Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";	55	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
54	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
55	our $MAX_PERSISTENT = 8;
56	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
57	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
58		59	our $MAX_PER_HOST = 4; # changing this is evil
59	# changing these is evil
60	our $MAX_PERSISTENT_PER_HOST = 2;
61	our $MAX_PER_HOST = 4;
62		60
63	our $PROXY;	61	our $PROXY;
		62	our $ACTIVE = 0;
64		63
65	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
66	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
67		66
68	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
69		68
70	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
71	additional parameters.	70	additional parameters and the return value.
72		71
73	=item http_head $url, key => value..., $cb->($data, $headers)	72	=item http_head $url, key => value..., $cb->($data, $headers)
74		73
75	Executes an HTTP-HEAD request. See the http_request function for details on	74	Executes an HTTP-HEAD request. See the http_request function for details
76	additional parameters.	75	on additional parameters and the return value.
77		76
78	=item http_post $url, $body, key => value..., $cb->($data, $headers)	77	=item http_post $url, $body, key => value..., $cb->($data, $headers)
79		78
80	Executes an HTTP-POST request with a request body of C<$bod>. See the	79	Executes an HTTP-POST request with a request body of C<$body>. See the
81	http_request function for details on additional parameters.	80	http_request function for details on additional parameters and the return
		81	value.
82		82
83	=item http_request $method => $url, key => value..., $cb->($data, $headers)	83	=item http_request $method => $url, key => value..., $cb->($data, $headers)
84		84
85	Executes a HTTP request of type C<$method> (e.g. C<GET>, C<POST>). The URL	85	Executes a HTTP request of type C<$method> (e.g. C<GET>, C<POST>). The URL
86	must be an absolute http or https URL.	86	must be an absolute http or https URL.
87		87
		88	When called in void context, nothing is returned. In other contexts,
		89	C<http_request> returns a "cancellation guard" - you have to keep the
		90	object at least alive until the callback get called. If the object gets
		91	destroyed before the callback is called, the request will be cancelled.
		92
88	The callback will be called with the response data as first argument	93	The callback will be called with the response body data as first argument
89	(or C<undef> if it wasn't available due to errors), and a hash-ref with	94	(or C<undef> if an error occured), and a hash-ref with response headers
90	response headers as second argument.	95	(and trailers) as second argument.
91		96
92	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
93	headers, the three "pseudo-headers" C<HTTPVersion>, C<Status> and	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
94	C<Reason> contain the three parts of the HTTP Status-Line of the same	99	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
		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>.
		104
		105	The pseudo-header C<URL> contains the actual URL (which can differ from
		106	the requested URL when following redirects - for example, you might get
		107	an error that your URL scheme is not supported even though your URL is a
		108	valid http URL because it redirected to an ftp URL, in which case you can
		109	look at the URL pseudo header).
		110
		111	The pseudo-header C<Redirect> only exists when the request was a result
		112	of an internal redirect. In that case it is an array reference with
		113	the C<($data, $headers)> from the redirect response. Note that this
		114	response could in turn be the result of a redirect itself, and C<<
		115	$headers->{Redirect}[1]{Redirect} >> will then contain the original
		116	response, and so on.
		117
95	name. If the server sends a header multiple lines, then their contents	118	If the server sends a header multiple times, then their contents will be
96	will be joined together with C<\x00>.	119	joined together with a comma (C<,>), as per the HTTP spec.
97		120
98	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
99	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<599>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
100	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection etsbalishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
101		139
102	A typical callback might look like this:	140	A typical callback might look like this:
103		141
104	sub {	142	sub {
105	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
121	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication
122	retries and so on, and how often to do so.	160	retries and so on, and how often to do so.
123		161
124	=item headers => hashref	162	=item headers => hashref
125		163
126	The request headers to use.	164	The request headers to use. Currently, C<http_request> may provide its own
		165	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
		166	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
		167	(this can be suppressed by using C<undef> for these headers in which case
		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.
127		173
128	=item timeout => $seconds	174	=item timeout => $seconds
129		175
130	The time-out to use for various stages - each connect attempt will reset	176	The time-out to use for various stages - each connect attempt will reset
131	the timeout, as will read or write activity. Default timeout is 5 minutes.	177	the timeout, as will read or write activity, i.e. this is not an overall
		178	timeout.
		179
		180	Default timeout is 5 minutes.
132		181
133	=item proxy => [$host, $port[, $scheme]] or undef	182	=item proxy => [$host, $port[, $scheme]] or undef
134		183
135	Use the given http proxy for all requests. If not specified, then the	184	Use the given http proxy for all requests, or no proxy if C<undef> is
136	default proxy (as specified by C<$ENV{http_proxy}>) is used.	185	used.
137		186
138	C<$scheme> must be either missing or C<http> for HTTP, or C<https> for	187	C<$scheme> must be either missing or must be C<http> for HTTP.
139	HTTPS.	188
		189	If not specified, then the default proxy is used (see
		190	C<AnyEvent::HTTP::set_proxy>).
140		191
141	=item body => $string	192	=item body => $string
142		193
143	The request body, usually empty. Will be-sent as-is (future versions of	194	The request body, usually empty. Will be sent as-is (future versions of
144	this module might offer more options).	195	this module might offer more options).
145		196
146	=item cookie_jar => $hash_ref	197	=item cookie_jar => $hash_ref
147		198
148	Passing this parameter enables (simplified) cookie-processing, loosely	199	Passing this parameter enables (simplified) cookie-processing, loosely
149	based on the original netscape specification.	200	based on the original netscape specification.
150		201
151	The C<$hash_ref> must be an (initially empty) hash reference which will	202	The C<$hash_ref> must be an (initially empty) hash reference which
152	get updated automatically. It is possible to save the cookie_jar to	203	will get updated automatically. It is possible to save the cookie jar
153	persistent storage with something like JSON or Storable, but this is not	204	to persistent storage with something like JSON or Storable - see the
154	recommended, as expire times are currently being ignored.	205	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		206	expired or session-only cookies, and also for documentation on the format
		207	of the cookie jar.
155		208
156	Note that this cookie implementation is not of very high quality, nor	209	Note that this cookie implementation is not meant to be complete. If
157	meant to be complete. If you want complete cookie management you have to	210	you want complete cookie management you have to do that on your
158	do that on your own. C<cookie_jar> is meant as a quick fix to get some	211	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
159	cookie-using sites working. Cookies are a privacy disaster, do not use	212	working. Cookies are a privacy disaster, do not use them unless required
160	them unless required to.	213	to.
		214
		215	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		216	headers will be set and handled by this module, otherwise they will be
		217	left untouched.
		218
		219	=item tls_ctx => $scheme | $tls_ctx
		220
		221	Specifies the AnyEvent::TLS context to be used for https connections. This
		222	parameter follows the same rules as the C<tls_ctx> parameter to
		223	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		224	C<high> can be specified, which give you a predefined low-security (no
		225	verification, highest compatibility) and high-security (CA and common-name
		226	verification) TLS context.
		227
		228	The default for this option is C<low>, which could be interpreted as "give
		229	me the page, no matter what".
		230
		231	See also the C<sessionid> parameter.
		232
		233	=item session => $string
		234
		235	The module might reuse connections to the same host internally. Sometimes
		236	(e.g. when using TLS), you do not want to reuse connections from other
		237	sessions. This can be achieved by setting this parameter to some unique
		238	ID (such as the address of an object storing your state data, or the TLS
		239	context) - only connections using the same unique ID will be reused.
		240
		241	=item on_prepare => $callback->($fh)
		242
		243	In rare cases you need to "tune" the socket before it is used to
		244	connect (for exmaple, to bind it on a given IP address). This parameter
		245	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		246	and behaves exactly the same way (e.g. it has to provide a
		247	timeout). See the description for the C<$prepare_cb> argument of
		248	C<AnyEvent::Socket::tcp_connect> for details.
		249
		250	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		251
		252	In even rarer cases you want total control over how AnyEvent::HTTP
		253	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		254	to do this, but you can provide your own C<tcp_connect> function -
		255	obviously, it has to follow the same calling conventions, except that it
		256	may always return a connection guard object.
		257
		258	There are probably lots of weird uses for this function, starting from
		259	tracing the hosts C<http_request> actually tries to connect, to (inexact
		260	but fast) host => IP address caching or even socks protocol support.
		261
		262	=item on_header => $callback->($headers)
		263
		264	When specified, this callback will be called with the header hash as soon
		265	as headers have been successfully received from the remote server (not on
		266	locally-generated errors).
		267
		268	It has to return either true (in which case AnyEvent::HTTP will continue),
		269	or false, in which case AnyEvent::HTTP will cancel the download (and call
		270	the finish callback with an error code of C<598>).
		271
		272	This callback is useful, among other things, to quickly reject unwanted
		273	content, which, if it is supposed to be rare, can be faster than first
		274	doing a C<HEAD> request.
		275
		276	The downside is that cancelling the request makes it impossible to re-use
		277	the connection. Also, the C<on_header> callback will not receive any
		278	trailer (headers sent after the response body).
		279
		280	Example: cancel the request unless the content-type is "text/html".
		281
		282	on_header => sub {
		283	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		284	},
		285
		286	=item on_body => $callback->($partial_body, $headers)
		287
		288	When specified, all body data will be passed to this callback instead of
		289	to the completion callback. The completion callback will get the empty
		290	string instead of the body data.
		291
		292	It has to return either true (in which case AnyEvent::HTTP will continue),
		293	or false, in which case AnyEvent::HTTP will cancel the download (and call
		294	the completion callback with an error code of C<598>).
		295
		296	The downside to cancelling the request is that it makes it impossible to
		297	re-use the connection.
		298
		299	This callback is useful when the data is too large to be held in memory
		300	(so the callback writes it to a file) or when only some information should
		301	be extracted, or when the body should be processed incrementally.
		302
		303	It is usually preferred over doing your own body handling via
		304	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		305	only used to create a connection, C<want_body_handle> is the better
		306	alternative, as it allows you to install your own event handler, reducing
		307	resource usage.
		308
		309	=item want_body_handle => $enable
		310
		311	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		312	changes considerably: after parsing the headers, and instead of
		313	downloading the body (if any), the completion callback will be
		314	called. Instead of the C<$body> argument containing the body data, the
		315	callback will receive the L<AnyEvent::Handle> object associated with the
		316	connection. In error cases, C<undef> will be passed. When there is no body
		317	(e.g. status C<304>), the empty string will be passed.
		318
		319	The handle object might or might not be in TLS mode, might be connected
		320	to a proxy, be a persistent connection, use chunked transfer encoding
		321	etc., and configured in unspecified ways. The user is responsible for this
		322	handle (it will not be used by this module anymore).
		323
		324	This is useful with some push-type services, where, after the initial
		325	headers, an interactive protocol is used (typical example would be the
		326	push-style twitter API which starts a JSON/XML stream).
		327
		328	If you think you need this, first have a look at C<on_body>, to see if
		329	that doesn't solve your problem in a better way.
		330
		331	=item persistent => $boolean
		332
		333	Try to create/reuse a persistent connection. When this flag is set
		334	(default: true for idempotent requests, false for all others), then
		335	C<http_request> tries to re-use an existing (previously-created)
		336	persistent connection to the host and, failing that, tries to create a new
		337	one.
		338
		339	Requests failing in certain ways will be automatically retried once, which
		340	is dangerous for non-idempotent requests, which is why it defaults to off
		341	for them. The reason for this is because the bozos who designed HTTP/1.1
		342	made it impossible to distinguish between a fatal error and a normal
		343	connection timeout, so you never know whether there was a problem with
		344	your request or not.
		345
		346	When reusing an existent connection, many parameters (such as TLS context)
		347	will be ignored. See the C<session> parameter for a workaround.
		348
		349	=item keepalive => $boolean
		350
		351	Only used when C<persistent> is also true. This parameter decides whether
		352	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		353	(as opposed to only a HTTP/1.1 persistent connection).
		354
		355	The default is true, except when using a proxy, in which case it defaults
		356	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		357
		358	=item handle_params => { key => value ... }
		359
		360	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		361	constructor that is called - not all requests will create a handle, and
		362	sometimes more than one is created, so this parameter is only good for
		363	setting hints.
		364
		365	Example: set the maximum read size to 4096, to potentially conserve memory
		366	at the cost of speed.
		367
		368	handle_params => {
		369	max_read_size => 4096,
		370	},
161		371
162	=back	372	=back
163		373
164	Example: make a simple HTTP GET request for http://www.nethype.de/	374	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		375	the response body.
165		376
166	http_request GET => "http://www.nethype.de/", sub {	377	http_request GET => "http://www.nethype.de/", sub {
167	my ($body, $hdr) = @_;	378	my ($body, $hdr) = @_;
168	print "$body\n";	379	print "$body\n";
169	};	380	};
170		381
171	Example: make a HTTP HEAD request on https://www.google.com/, use a	382	Example: do a HTTP HEAD request on https://www.google.com/, use a
172	timeout of 30 seconds.	383	timeout of 30 seconds.
173		384
174	http_request	385	http_request
175	GET => "https://www.google.com",	386	GET => "https://www.google.com",
		387	headers => { "user-agent" => "MySearchClient 1.0" },
176	timeout => 30,	388	timeout => 30,
177	sub {	389	sub {
178	my ($body, $hdr) = @_;	390	my ($body, $hdr) = @_;
179	use Data::Dumper;	391	use Data::Dumper;
180	print Dumper $hdr;	392	print Dumper $hdr;
181	}	393	}
182	;	394	;
183		395
		396	Example: do another simple HTTP GET request, but immediately try to
		397	cancel it.
		398
		399	my $request = http_request GET => "http://www.nethype.de/", sub {
		400	my ($body, $hdr) = @_;
		401	print "$body\n";
		402	};
		403
		404	undef $request;
		405
184	=cut	406	=cut
185		407
		408	#############################################################################
		409	# wait queue/slots
		410
		411	sub _slot_schedule;
186	sub _slot_schedule($) {	412	sub _slot_schedule($) {
187	my $host = shift;	413	my $host = shift;
188		414
189	while ($CO_SLOT{$host}[0] < $MAX_PER_HOST) {	415	while ($CO_SLOT{$host}[0] < $MAX_PER_HOST) {
190	if (my $cb = shift @{ $CO_SLOT{$host}[1] }) {	416	if (my $cb = shift @{ $CO_SLOT{$host}[1] }) {
191	# somebody wnats that slot	417	# somebody wants that slot
192	++$CO_SLOT{$host}[0];	418	++$CO_SLOT{$host}[0];
		419	++$ACTIVE;
193		420
194	$cb->(AnyEvent::Util::guard {	421	$cb->(AnyEvent::Util::guard {
		422	--$ACTIVE;
195	--$CO_SLOT{$host}[0];	423	--$CO_SLOT{$host}[0];
196	_slot_schedule $host;	424	_slot_schedule $host;
197	});	425	});
198	} else {	426	} else {
199	# nobody wants the slot, maybe we can forget about it	427	# nobody wants the slot, maybe we can forget about it
200	delete $CO_SLOT{$host} unless $CO_SLOT{$host}[0];	428	delete $CO_SLOT{$host} unless $CO_SLOT{$host}[0];
201	warn "$host deleted" unless $CO_SLOT{$host}[0];#d#
202	last;	429	last;
203	}	430	}
204	}	431	}
205	}	432	}
206		433
…		…
209	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	436	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
210		437
211	_slot_schedule $_[0];	438	_slot_schedule $_[0];
212	}	439	}
213		440
		441	#############################################################################
		442	# cookie handling
		443
		444	# expire cookies
		445	sub cookie_jar_expire($;$) {
		446	my ($jar, $session_end) = @_;
		447
		448	%$jar = () if $jar->{version} != 1;
		449
		450	my $anow = AE::now;
		451
		452	while (my ($chost, $paths) = each %$jar) {
		453	next unless ref $paths;
		454
		455	while (my ($cpath, $cookies) = each %$paths) {
		456	while (my ($cookie, $kv) = each %$cookies) {
		457	if (exists $kv->{_expires}) {
		458	delete $cookies->{$cookie}
		459	if $anow > $kv->{_expires};
		460	} elsif ($session_end) {
		461	delete $cookies->{$cookie};
		462	}
		463	}
		464
		465	delete $paths->{$cpath}
		466	unless %$cookies;
		467	}
		468
		469	delete $jar->{$chost}
		470	unless %$paths;
		471	}
		472	}
		473
		474	# extract cookies from jar
		475	sub cookie_jar_extract($$$$) {
		476	my ($jar, $scheme, $host, $path) = @_;
		477
		478	%$jar = () if $jar->{version} != 1;
		479
		480	my @cookies;
		481
		482	while (my ($chost, $paths) = each %$jar) {
		483	next unless ref $paths;
		484
		485	if ($chost =~ /^\./) {
		486	next unless $chost eq substr $host, -length $chost;
		487	} elsif ($chost =~ /\./) {
		488	next unless $chost eq $host;
		489	} else {
		490	next;
		491	}
		492
		493	while (my ($cpath, $cookies) = each %$paths) {
		494	next unless $cpath eq substr $path, 0, length $cpath;
		495
		496	while (my ($cookie, $kv) = each %$cookies) {
		497	next if $scheme ne "https" && exists $kv->{secure};
		498
		499	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		500	delete $cookies->{$cookie};
		501	next;
		502	}
		503
		504	my $value = $kv->{value};
		505
		506	if ($value =~ /[=;,[:space:]]/) {
		507	$value =~ s/([\\"])/\\$1/g;
		508	$value = "\"$value\"";
		509	}
		510
		511	push @cookies, "$cookie=$value";
		512	}
		513	}
		514	}
		515
		516	\@cookies
		517	}
		518
		519	# parse set_cookie header into jar
		520	sub cookie_jar_set_cookie($$$$) {
		521	my ($jar, $set_cookie, $host, $date) = @_;
		522
		523	my $anow = int AE::now;
		524	my $snow; # server-now
		525
		526	for ($set_cookie) {
		527	# parse NAME=VALUE
		528	my @kv;
		529
		530	# expires is not http-compliant in the original cookie-spec,
		531	# we support the official date format and some extensions
		532	while (
		533	m{
		534	\G\s*
		535	(?:
		536	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		537	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
		538	)
		539	}gcxsi
		540	) {
		541	my $name = $2;
		542	my $value = $4;
		543
		544	if (defined $1) {
		545	# expires
		546	$name = "expires";
		547	$value = $1;
		548	} elsif (defined $3) {
		549	# quoted
		550	$value = $3;
		551	$value =~ s/\\(.)/$1/gs;
		552	}
		553
		554	push @kv, @kv ? lc $name : $name, $value;
		555
		556	last unless /\G\s*;/gc;
		557	}
		558
		559	last unless @kv;
		560
		561	my $name = shift @kv;
		562	my %kv = (value => shift @kv, @kv);
		563
		564	if (exists $kv{"max-age"}) {
		565	$kv{_expires} = $anow + delete $kv{"max-age"};
		566	} elsif (exists $kv{expires}) {
		567	$snow ||= parse_date ($date) || $anow;
		568	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		569	} else {
		570	delete $kv{_expires};
		571	}
		572
		573	my $cdom;
		574	my $cpath = (delete $kv{path}) || "/";
		575
		576	if (exists $kv{domain}) {
		577	$cdom = delete $kv{domain};
		578
		579	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		580
		581	next if $cdom =~ /\.$/;
		582
		583	# this is not rfc-like and not netscape-like. go figure.
		584	my $ndots = $cdom =~ y/.//;
		585	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		586	} else {
		587	$cdom = $host;
		588	}
		589
		590	# store it
		591	$jar->{version} = 1;
		592	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		593
		594	redo if /\G\s*,/gc;
		595	}
		596	}
		597
		598	#############################################################################
		599	# keepalive/persistent connection cache
		600
		601	# fetch a connection from the keepalive cache
		602	sub ka_fetch($) {
		603	my $ka_key = shift;
		604
		605	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		606	delete $KA_CACHE{$ka_key}
		607	unless @{ $KA_CACHE{$ka_key} };
		608
		609	$hdl
		610	}
		611
		612	sub ka_store($$) {
		613	my ($ka_key, $hdl) = @_;
		614
		615	my $kaa = $KA_CACHE{$ka_key} ||= [];
		616
		617	my $destroy = sub {
		618	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		619
		620	$hdl->destroy;
		621
		622	@ka
		623	? $KA_CACHE{$ka_key} = \@ka
		624	: delete $KA_CACHE{$ka_key};
		625	};
		626
		627	# on error etc., destroy
		628	$hdl->on_error ($destroy);
		629	$hdl->on_eof ($destroy);
		630	$hdl->on_read ($destroy);
		631	$hdl->timeout ($PERSISTENT_TIMEOUT);
		632
		633	push @$kaa, $hdl;
		634	shift @$kaa while @$kaa > $MAX_PER_HOST;
		635	}
		636
		637	#############################################################################
		638	# utilities
		639
		640	# continue to parse $_ for headers and place them into the arg
		641	sub _parse_hdr() {
		642	my %hdr;
		643
		644	# things seen, not parsed:
		645	# p3pP="NON CUR OTPi OUR NOR UNI"
		646
		647	$hdr{lc $1} .= ",$2"
		648	while /\G
		649	([^:\000-\037]*):
		650	[\011\040]*
		651	((?: [^\012]+ | \012[\011\040] )*)
		652	\012
		653	/gxc;
		654
		655	/\G$/
		656	or return;
		657
		658	# remove the "," prefix we added to all headers above
		659	substr $_, 0, 1, ""
		660	for values %hdr;
		661
		662	\%hdr
		663	}
		664
		665	#############################################################################
		666	# http_get
		667
		668	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		669
		670	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		671	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		672
		673	# maybe it should just become a normal object :/
		674
		675	sub _destroy_state(\%) {
		676	my ($state) = @_;
		677
		678	$state->{handle}->destroy if $state->{handle};
		679	%$state = ();
		680	}
		681
		682	sub _error(\%$$) {
		683	my ($state, $cb, $hdr) = @_;
		684
		685	&_destroy_state ($state);
		686
		687	$cb->(undef, $hdr);
		688	()
		689	}
		690
214	sub http_request($$$;@) {	691	sub http_request($$@) {
215	my $cb = pop;	692	my $cb = pop;
216	my ($method, $url, %arg) = @_;	693	my ($method, $url, %arg) = @_;
217		694
218	my %hdr;	695	my %hdr;
		696
		697	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		698	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
219		699
220	$method = uc $method;	700	$method = uc $method;
221		701
222	if (my $hdr = $arg{headers}) {	702	if (my $hdr = $arg{headers}) {
223	while (my ($k, $v) = each %$hdr) {	703	while (my ($k, $v) = each %$hdr) {
224	$hdr{lc $k} = $v;	704	$hdr{lc $k} = $v;
225	}	705	}
226	}	706	}
227		707
		708	# pseudo headers for all subsequent responses
		709	my @pseudo = (URL => $url);
		710	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		711
228	my $recurse = exists $arg{recurse} ? $arg{recurse} : $MAX_RECURSE;	712	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
229		713
230	return $cb->(undef, { Status => 599, Reason => "recursion limit reached" })	714	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
231	if $recurse < 0;	715	if $recurse < 0;
232		716
233	my $proxy = $arg{proxy} || $PROXY;	717	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
234	my $timeout = $arg{timeout} || $TIMEOUT;	718	my $timeout = $arg{timeout} || $TIMEOUT;
235		719
236	$hdr{"user-agent"} ||= $USERAGENT;
237
238	my ($scheme, $authority, $upath, $query, $fragment) =	720	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
239	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;	721	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
240		722
241	$scheme = lc $scheme;	723	$uscheme = lc $uscheme;
242		724
243	my $uport = $scheme eq "http" ? 80	725	my $uport = $uscheme eq "http" ? 80
244	: $scheme eq "https" ? 443	726	: $uscheme eq "https" ? 443
245	: return $cb->(undef, { Status => 599, Reason => "only http and https URL schemes supported" });	727	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
246		728
247	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	729	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
248	or return $cb->(undef, { Status => 599, Reason => "unparsable URL" });	730	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
249		731
250	my $uhost = $1;	732	my $uhost = lc $1;
251	$uport = $2 if defined $2;	733	$uport = $2 if defined $2;
252		734
		735	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		736	unless exists $hdr{host};
		737
253	$uhost =~ s/^\[(.*)\]$/$1/;	738	$uhost =~ s/^\[(.*)\]$/$1/;
254	$upath .= "?$query" if length $query;	739	$upath .= $query if length $query;
255		740
256	$upath =~ s%^/?%/%;	741	$upath =~ s%^/?%/%;
257		742
258	# cookie processing	743	# cookie processing
259	if (my $jar = $arg{cookie_jar}) {	744	if (my $jar = $arg{cookie_jar}) {
260	%$jar = () if $jar->{version} < 1;	745	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
261		746
		747	$hdr{cookie} = join "; ", @$cookies
262	my @cookie;	748	if @$cookies;
263		749	}
264	while (my ($chost, $v) = each %$jar) {	750
265	next unless $chost eq substr $uhost, -length $chost;	751	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
266	next unless $chost =~ /^\./;	752
267		753	if ($proxy) {
268	while (my ($cpath, $v) = each %$v) {	754	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
269	next unless $cpath eq substr $upath, 0, length $cpath;	755
270		756	$rscheme = "http" unless defined $rscheme;
271	while (my ($k, $v) = each %$v) {	757
272	next if $scheme ne "https" && exists $v->{secure};	758	# don't support https requests over https-proxy transport,
273	push @cookie, "$k=$v->{value}";	759	# can't be done with tls as spec'ed, unless you double-encrypt.
		760	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		761
		762	$rhost = lc $rhost;
		763	$rscheme = lc $rscheme;
		764	} else {
		765	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
		766	}
		767
		768	# leave out fragment and query string, just a heuristic
		769	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		770	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
		771
		772	$hdr{"content-length"} = length $arg{body}
		773	if length $arg{body} || $method ne "GET";
		774
		775	my $idempotent = $method =~ /^(?:GET|HEAD|PUT|DELETE|OPTIONS|TRACE)$/;
		776
		777	# default value for keepalive is true iff the request is for an idempotent method
		778	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		779	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		780	my $was_persistent; # true if this is actually a recycled connection
		781
		782	# the key to use in the keepalive cache
		783	my $ka_key = "$uhost\x00$arg{sessionid}";
		784
		785	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive " : "" : "close ") . "Te"; #1.1
		786	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		787
		788	my %state = (connect_guard => 1);
		789
		790	my $ae_error = 595; # connecting
		791
		792	# handle actual, non-tunneled, request
		793	my $handle_actual_request = sub {
		794	$ae_error = 596; # request phase
		795
		796	my $hdl = $state{handle};
		797
		798	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		799
		800	# send request
		801	$hdl->push_write (
		802	"$method $rpath HTTP/1.1\015\012"
		803	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		804	. "\015\012"
		805	. (delete $arg{body})
		806	);
		807
		808	# return if error occured during push_write()
		809	return unless %state;
		810
		811	# reduce memory usage, save a kitten, also re-use it for the response headers.
		812	%hdr = ();
		813
		814	# status line and headers
		815	$state{read_response} = sub {
		816	return unless %state;
		817
		818	for ("$_[1]") {
		819	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		820
		821	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		822	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		823
		824	# 100 Continue handling
		825	# should not happen as we don't send expect: 100-continue,
		826	# but we handle it just in case.
		827	# since we send the request body regardless, if we get an error
		828	# we are out of-sync, which we currently do NOT handle correctly.
		829	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		830	if $2 eq 100;
		831
		832	push @pseudo,
		833	HTTPVersion => $1,
		834	Status => $2,
		835	Reason => $3,
		836	;
		837
		838	my $hdr = _parse_hdr
		839	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		840
		841	%hdr = (%$hdr, @pseudo);
		842	}
		843
		844	# redirect handling
		845	# microsoft and other shitheads don't give a shit for following standards,
		846	# try to support some common forms of broken Location headers.
		847	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
		848	$hdr{location} =~ s/^\.\/+//;
		849
		850	my $url = "$rscheme://$uhost:$uport";
		851
		852	unless ($hdr{location} =~ s/^\///) {
		853	$url .= $upath;
		854	$url =~ s/\/[^\/]*$//;
		855	}
		856
		857	$hdr{location} = "$url/$hdr{location}";
		858	}
		859
		860	my $redirect;
		861
		862	if ($recurse) {
		863	my $status = $hdr{Status};
		864
		865	# industry standard is to redirect POST as GET for
		866	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		867	# also, the UA should ask the user for 301 and 307 and POST,
		868	# industry standard seems to be to simply follow.
		869	# we go with the industry standard.
		870	if ($status == 301 or $status == 302 or $status == 303) {
		871	# HTTP/1.1 is unclear on how to mutate the method
		872	$method = "GET" unless $method eq "HEAD";
		873	$redirect = 1;
		874	} elsif ($status == 307) {
		875	$redirect = 1;
274	}	876	}
275	}	877	}
		878
		879	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		880	if ($state{handle}) {
		881	# handle keepalive
		882	if (
		883	$persistent
		884	&& $_[3]
		885	&& ($hdr{HTTPVersion} < 1.1
		886	? $hdr{connection} =~ /\bkeep-?alive\b/i
		887	: $hdr{connection} !~ /\bclose\b/i)
		888	) {
		889	ka_store $ka_key, delete $state{handle};
		890	} else {
		891	# no keepalive, destroy the handle
		892	$state{handle}->destroy;
		893	}
		894	}
		895
		896	%state = ();
		897
		898	if (defined $_[1]) {
		899	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		900	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		901	}
		902
		903	# set-cookie processing
		904	if ($arg{cookie_jar}) {
		905	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		906	}
		907
		908	if ($redirect && exists $hdr{location}) {
		909	# we ignore any errors, as it is very common to receive
		910	# Content-Length != 0 but no actual body
		911	# we also access %hdr, as $_[1] might be an erro
		912	$state{recurse} =
		913	http_request (
		914	$method => $hdr{location},
		915	%arg,
		916	recurse => $recurse - 1,
		917	Redirect => [$_[0], \%hdr],
		918	sub {
		919	%state = ();
		920	&$cb
		921	},
		922	);
		923	} else {
		924	$cb->($_[0], \%hdr);
		925	}
		926	};
		927
		928	$ae_error = 597; # body phase
		929
		930	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		931
		932	my $len = $chunked ? undef : $hdr{"content-length"};
		933
		934	# body handling, many different code paths
		935	# - no body expected
		936	# - want_body_handle
		937	# - te chunked
		938	# - 2x length known (with or without on_body)
		939	# - 2x length not known (with or without on_body)
		940	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		941	$finish->(undef, 598 => "Request cancelled by on_header");
		942	} elsif (
		943	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		944	or $method eq "HEAD"
		945	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		946	) {
		947	# no body
		948	$finish->("", undef, undef, 1);
		949
		950	} elsif (!$redirect && $arg{want_body_handle}) {
		951	$_[0]->on_eof (undef);
		952	$_[0]->on_error (undef);
		953	$_[0]->on_read (undef);
		954
		955	$finish->(delete $state{handle});
		956
		957	} elsif ($chunked) {
		958	my $cl = 0;
		959	my $body = "";
		960	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		961
		962	$state{read_chunk} = sub {
		963	$_[1] =~ /^([0-9a-fA-F]+)/
		964	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		965
		966	my $len = hex $1;
		967
		968	if ($len) {
		969	$cl += $len;
		970
		971	$_[0]->push_read (chunk => $len, sub {
		972	$on_body->($_[1], \%hdr)
		973	or return $finish->(undef, 598 => "Request cancelled by on_body");
		974
		975	$_[0]->push_read (line => sub {
		976	length $_[1]
		977	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		978	$_[0]->push_read (line => $state{read_chunk});
		979	});
		980	});
		981	} else {
		982	$hdr{"content-length"} ||= $cl;
		983
		984	$_[0]->push_read (line => $qr_nlnl, sub {
		985	if (length $_[1]) {
		986	for ("$_[1]") {
		987	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		988
		989	my $hdr = _parse_hdr
		990	or return $finish->(undef, $ae_error => "Garbled response trailers");
		991
		992	%hdr = (%hdr, %$hdr);
		993	}
		994	}
		995
		996	$finish->($body, undef, undef, 1);
		997	});
		998	}
		999	};
		1000
		1001	$_[0]->push_read (line => $state{read_chunk});
		1002
		1003	} elsif ($arg{on_body}) {
		1004	if (defined $len) {
		1005	$_[0]->on_read (sub {
		1006	$len -= length $_[0]{rbuf};
		1007
		1008	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1009	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1010
		1011	$len > 0
		1012	or $finish->("", undef, undef, 1);
		1013	});
		1014	} else {
		1015	$_[0]->on_eof (sub {
		1016	$finish->("");
		1017	});
		1018	$_[0]->on_read (sub {
		1019	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1020	or $finish->(undef, 598 => "Request cancelled by on_body");
		1021	});
		1022	}
		1023	} else {
		1024	$_[0]->on_eof (undef);
		1025
		1026	if (defined $len) {
		1027	$_[0]->on_read (sub {
		1028	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1029	if $len <= length $_[0]{rbuf};
		1030	});
		1031	} else {
		1032	$_[0]->on_error (sub {
		1033	($! == Errno::EPIPE || !$!)
		1034	? $finish->(delete $_[0]{rbuf})
		1035	: $finish->(undef, $ae_error => $_[2]);
		1036	});
		1037	$_[0]->on_read (sub { });
		1038	}
		1039	}
		1040	};
		1041
		1042	# if keepalive is enabled, then the server closing the connection
		1043	# before a response can happen legally - we retry on idempotent methods.
		1044	if ($was_persistent && $idempotent) {
		1045	my $old_eof = $hdl->{on_eof};
		1046	$hdl->{on_eof} = sub {
		1047	_destroy_state %state;
		1048
		1049	%state = ();
		1050	$state{recurse} =
		1051	http_request (
		1052	$method => $url,
		1053	%arg,
		1054	keepalive => 0,
		1055	sub {
		1056	%state = ();
		1057	&$cb
		1058	}
		1059	);
		1060	};
		1061	$hdl->on_read (sub {
		1062	return unless %state;
		1063
		1064	# as soon as we receive something, a connection close
		1065	# once more becomes a hard error
		1066	$hdl->{on_eof} = $old_eof;
		1067	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1068	});
		1069	} else {
		1070	$hdl->push_read (line => $qr_nlnl, $state{read_response});
276	}	1071	}
277
278	$hdr{cookie} = join "; ", @cookie
279	if @cookie;
280	}	1072	};
281		1073
282	my ($rhost, $rport, $rpath); # request host, port, path	1074	my $prepare_handle = sub {
		1075	my ($hdl) = $state{handle};
283		1076
284	if ($proxy) {	1077	$hdl->on_error (sub {
285	($rhost, $rport, $scheme) = @$proxy;	1078	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
286	$rpath = $url;	1079	});
287	} else {	1080	$hdl->on_eof (sub {
288	($rhost, $rport, $rpath) = ($uhost, $uport, $upath);	1081	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
289	$hdr{host} = $uhost;	1082	});
		1083	$hdl->timeout_reset;
		1084	$hdl->timeout ($timeout);
290	}	1085	};
291		1086
292	$hdr{"content-length"} = length $arg{body};	1087	# connected to proxy (or origin server)
		1088	my $connect_cb = sub {
		1089	my $fh = shift
		1090	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
293		1091
294	my %state = (connect_guard => 1);	1092	return unless delete $state{connect_guard};
		1093
		1094	# get handle
		1095	$state{handle} = new AnyEvent::Handle
		1096	%{ $arg{handle_params} },
		1097	fh => $fh,
		1098	peername => $uhost,
		1099	tls_ctx => $arg{tls_ctx},
		1100	;
		1101
		1102	$prepare_handle->();
		1103
		1104	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1105
		1106	# now handle proxy-CONNECT method
		1107	if ($proxy && $uscheme eq "https") {
		1108	# oh dear, we have to wrap it into a connect request
		1109
		1110	# maybe re-use $uauthority with patched port?
		1111	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1112	$state{handle}->push_read (line => $qr_nlnl, sub {
		1113	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1114	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1115
		1116	if ($2 == 200) {
		1117	$rpath = $upath;
		1118	$handle_actual_request->();
		1119	} else {
		1120	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1121	}
		1122	});
		1123	} else {
		1124	$handle_actual_request->();
		1125	}
		1126	};
295		1127
296	_get_slot $uhost, sub {	1128	_get_slot $uhost, sub {
297	$state{slot_guard} = shift;	1129	$state{slot_guard} = shift;
298		1130
299	return unless $state{connect_guard};	1131	return unless $state{connect_guard};
300		1132
301	$state{connect_guard} = AnyEvent::Socket::tcp_connect $rhost, $rport, sub {	1133	# try to use an existing keepalive connection, but only if we, ourselves, plan
302	$state{fh} = shift	1134	# on a keepalive request (in theory, this should be a separate config option).
303	or return $cb->(undef, { Status => 599, Reason => "$!" });	1135	if ($persistent && $KA_CACHE{$ka_key}) {
		1136	$was_persistent = 1;
304		1137
305	delete $state{connect_guard}; # reduce memory usage, save a tree	1138	$state{handle} = ka_fetch $ka_key;
		1139	$state{handle}->destroyed
		1140	and die "got a destructed habndle. pah\n";#d#
		1141	$prepare_handle->();
		1142	$state{handle}->destroyed
		1143	and die "got a destructed habndle. pa2\n";#d#
		1144	$handle_actual_request->();
		1145	$state{handle}->destroyed
		1146	and die "got a destructed habndle. pa3\n";#d#
306		1147
307	# get handle
308	$state{handle} = new AnyEvent::Handle
309	fh => $state{fh},
310	($scheme eq "https" ? (tls => "connect") : ());
311
312	# limit the number of persistent connections
313	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
314	++$KA_COUNT{$_[1]};
315	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
316	$hdr{connection} = "keep-alive";
317	delete $hdr{connection}; # keep-alive not yet supported
318	} else {	1148	} else {
319	delete $hdr{connection};	1149	my $tcp_connect = $arg{tcp_connect}
320	}	1150	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
321		1151
322	# (re-)configure handle	1152	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
323	$state{handle}->timeout ($timeout);
324	$state{handle}->on_error (sub {
325	%state = ();
326	$cb->(undef, { Status => 599, Reason => "$!" });
327	});
328	$state{handle}->on_eof (sub {
329	%state = ();
330	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file" });
331	});
332
333	# send request
334	$state{handle}->push_write (
335	"$method $rpath HTTP/1.0\015\012"
336	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
337	. "\015\012"
338	. (delete $arg{body})
339	);
340
341	%hdr = (); # reduce memory usage, save a kitten
342
343	# status line
344	$state{handle}->push_read (line => qr/\015?\012/, sub {
345	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix
346	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])" }));
347
348	my %hdr = ( # response headers
349	HTTPVersion => "\x00$1",
350	Status => "\x00$2",
351	Reason => "\x00$3",
352	);
353
354	# headers, could be optimized a bit
355	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
356	for ("$_[1]\012") {
357	# we support spaces in field names, as lotus domino
358	# creates them.
359	$hdr{lc $1} .= "\x00$2"
360	while /\G
361	([^:\000-\037]+):
362	[\011\040]*
363	((?: [^\015\012]+ | \015?\012[\011\040] )*)
364	\015?\012
365	/gxc;
366
367	/\G$/
368	or return (%state = (), $cb->(undef, { Status => 599, Reason => "garbled response headers" }));
369	}
370
371	substr $_, 0, 1, ""
372	for values %hdr;
373
374	my $finish = sub {
375	%state = ();
376
377	# set-cookie processing
378	if ($arg{cookie_jar} && exists $hdr{"set-cookie"}) {
379	for (split /\x00/, $hdr{"set-cookie"}) {
380	my ($cookie, @arg) = split /;\s*/;
381	my ($name, $value) = split /=/, $cookie, 2;
382	my %kv = (value => $value, map { split /=/, $_, 2 } @arg);
383
384	my $cdom = (delete $kv{domain}) || $uhost;
385	my $cpath = (delete $kv{path}) || "/";
386
387	$cdom =~ s/^.?/./; # make sure it starts with a "."
388
389	next if $cdom =~ /\.$/;
390
391	# this is not rfc-like and not netscape-like. go figure.
392	my $ndots = $cdom =~ y/.//;
393	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
394
395	# store it
396	$arg{cookie_jar}{version} = 1;
397	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
398	}
399	}
400
401	if ($_[1]{Status} =~ /^x30[12]$/ && $recurse) {
402	# microsoft and other assholes don't give a shit for following standards,
403	# try to support a common form of broken Location header.
404	$_[1]{location} =~ s%^/%$scheme://$uhost:$uport/%;
405
406	http_request ($method, $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
407	} else {
408	$cb->($_[0], $_[1]);
409	}
410	};
411
412	if ($hdr{Status} =~ /^(?:1..|204|304)$/ or $method eq "HEAD") {
413	$finish->(undef, \%hdr);
414	} else {
415	if (exists $hdr{"content-length"}) {
416	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
417	# could cache persistent connection now
418	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
419	# but we don't, due to misdesigns, this is annoyingly complex
420	};
421
422	$finish->($_[1], \%hdr);
423	});
424	} else {
425	# too bad, need to read until we get an error or EOF,
426	# no way to detect winged data.
427	$_[0]->on_error (sub {
428	$finish->($_[0]{rbuf}, \%hdr);
429	});
430	$_[0]->on_eof (undef);
431	$_[0]->on_read (sub { });
432	}
433	}
434	});
435	});
436	}, sub {
437	$timeout
438	};	1153	}
439	};	1154	};
440		1155
441	defined wantarray && AnyEvent::Util::guard { %state = () }	1156	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
442	}	1157	}
443		1158
444	sub http_get($$;@) {	1159	sub http_get($@) {
445	unshift @_, "GET";	1160	unshift @_, "GET";
446	&http_request	1161	&http_request
447	}	1162	}
448		1163
449	sub http_head($$;@) {	1164	sub http_head($@) {
450	unshift @_, "HEAD";	1165	unshift @_, "HEAD";
451	&http_request	1166	&http_request
452	}	1167	}
453		1168
454	sub http_post($$$;@) {	1169	sub http_post($$@) {
		1170	my $url = shift;
455	unshift @_, "POST", "body";	1171	unshift @_, "POST", $url, "body";
456	&http_request	1172	&http_request
457	}	1173	}
458		1174
459	=back	1175	=back
460		1176
		1177	=head2 DNS CACHING
		1178
		1179	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1180	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1181	hostnames. The latter is a simple stub resolver and does no caching
		1182	on its own. If you want DNS caching, you currently have to provide
		1183	your own default resolver (by storing a suitable resolver object in
		1184	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1185
461	=head2 GLOBAL FUNCTIONS AND VARIABLES	1186	=head2 GLOBAL FUNCTIONS AND VARIABLES
462		1187
463	=over 4	1188	=over 4
464		1189
465	=item AnyEvent::HTTP::set_proxy "proxy-url"	1190	=item AnyEvent::HTTP::set_proxy "proxy-url"
466		1191
467	Sets the default proxy server to use. The proxy-url must begin with a	1192	Sets the default proxy server to use. The proxy-url must begin with a
468	string of the form C<http://host:port> (optionally C<https:...>).	1193	string of the form C<http://host:port>, croaks otherwise.
		1194
		1195	To clear an already-set proxy, use C<undef>.
		1196
		1197	When AnyEvent::HTTP is laoded for the first time it will query the
		1198	default proxy from the operating system, currently by looking at
		1199	C<$ENV{http_proxy>}.
		1200
		1201	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1202
		1203	Remove all cookies from the cookie jar that have been expired. If
		1204	C<$session_end> is given and true, then additionally remove all session
		1205	cookies.
		1206
		1207	You should call this function (with a true C<$session_end>) before you
		1208	save cookies to disk, and you should call this function after loading them
		1209	again. If you have a long-running program you can additonally call this
		1210	function from time to time.
		1211
		1212	A cookie jar is initially an empty hash-reference that is managed by this
		1213	module. It's format is subject to change, but currently it is like this:
		1214
		1215	The key C<version> has to contain C<1>, otherwise the hash gets
		1216	emptied. All other keys are hostnames or IP addresses pointing to
		1217	hash-references. The key for these inner hash references is the
		1218	server path for which this cookie is meant, and the values are again
		1219	hash-references. The keys of those hash-references is the cookie name, and
		1220	the value, you guessed it, is another hash-reference, this time with the
		1221	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1222	which have been replaced by a C<_expires> key that contains the cookie
		1223	expiry timestamp.
		1224
		1225	Here is an example of a cookie jar with a single cookie, so you have a
		1226	chance of understanding the above paragraph:
		1227
		1228	{
		1229	version => 1,
		1230	"10.0.0.1" => {
		1231	"/" => {
		1232	"mythweb_id" => {
		1233	_expires => 1293917923,
		1234	value => "ooRung9dThee3ooyXooM1Ohm",
		1235	},
		1236	},
		1237	},
		1238	}
		1239
		1240	=item $date = AnyEvent::HTTP::format_date $timestamp
		1241
		1242	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1243	Date (RFC 2616).
		1244
		1245	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1246
		1247	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1248	bunch of minor variations of those, and returns the corresponding POSIX
		1249	timestamp, or C<undef> if the date cannot be parsed.
469		1250
470	=item $AnyEvent::HTTP::MAX_RECURSE	1251	=item $AnyEvent::HTTP::MAX_RECURSE
471		1252
472	The default value for the C<recurse> request parameter (default: C<10>).	1253	The default value for the C<recurse> request parameter (default: C<10>).
473		1254
		1255	=item $AnyEvent::HTTP::TIMEOUT
		1256
		1257	The default timeout for conenction operations (default: C<300>).
		1258
474	=item $AnyEvent::HTTP::USERAGENT	1259	=item $AnyEvent::HTTP::USERAGENT
475		1260
476	The default value for the C<User-Agent> header (the default is	1261	The default value for the C<User-Agent> header (the default is
477	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1262	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
478		1263
479	=item $AnyEvent::HTTP::MAX_PERSISTENT	1264	=item $AnyEvent::HTTP::MAX_PER_HOST
480		1265
481	The maximum number of persistent connections to keep open (default: 8).	1266	The maximum number of concurrent connections to the same host (identified
		1267	by the hostname). If the limit is exceeded, then the additional requests
		1268	are queued until previous connections are closed. Both persistent and
		1269	non-persistent connections are counted in this limit.
482		1270
483	Not implemented currently.	1271	The default value for this is C<4>, and it is highly advisable to not
		1272	increase it much.
		1273
		1274	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1275	connections, older browsers used 2, newers (such as firefox 3) typically
		1276	use 6, and Opera uses 8 because like, they have the fastest browser and
		1277	give a shit for everybody else on the planet.
484		1278
485	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1279	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
486		1280
487	The maximum time to cache a persistent connection, in seconds (default: 2).	1281	The time after which idle persistent conenctions get closed by
		1282	AnyEvent::HTTP (default: C<3>).
488		1283
489	Not implemented currently.	1284	=item $AnyEvent::HTTP::ACTIVE
		1285
		1286	The number of active connections. This is not the number of currently
		1287	running requests, but the number of currently open and non-idle TCP
		1288	connections. This number can be useful for load-leveling.
490		1289
491	=back	1290	=back
492		1291
493	=cut	1292	=cut
494		1293
		1294	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1295	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1296
		1297	sub format_date($) {
		1298	my ($time) = @_;
		1299
		1300	# RFC 822/1123 format
		1301	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1302
		1303	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1304	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1305	$H, $M, $S;
		1306	}
		1307
		1308	sub parse_date($) {
		1309	my ($date) = @_;
		1310
		1311	my ($d, $m, $y, $H, $M, $S);
		1312
		1313	if ($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$/) {
		1314	# RFC 822/1123, required by RFC 2616 (with " ")
		1315	# cookie dates (with "-")
		1316
		1317	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1318
		1319	} elsif ($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$/) {
		1320	# RFC 850
		1321	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1322
		1323	} elsif ($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])$/) {
		1324	# ISO C's asctime
		1325	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1326	}
		1327	# other formats fail in the loop below
		1328
		1329	for (0..11) {
		1330	if ($m eq $month[$_]) {
		1331	require Time::Local;
		1332	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);
		1333	}
		1334	}
		1335
		1336	undef
		1337	}
		1338
495	sub set_proxy($) {	1339	sub set_proxy($) {
496	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1340	if (length $_[0]) {
		1341	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1342	or Carp::croak "$_[0]: invalid proxy URL";
		1343	$PROXY = [$2, $3 || 3128, $1]
		1344	} else {
		1345	undef $PROXY;
		1346	}
497	}	1347	}
498		1348
499	# initialise proxy from environment	1349	# initialise proxy from environment
		1350	eval {
500	set_proxy $ENV{http_proxy};	1351	set_proxy $ENV{http_proxy};
		1352	};
		1353
		1354	=head2 SHOWCASE
		1355
		1356	This section contaisn some more elaborate "real-world" examples or code
		1357	snippets.
		1358
		1359	=head2 HTTP/1.1 FILE DOWNLOAD
		1360
		1361	Downloading files with HTTP can be quite tricky, especially when something
		1362	goes wrong and you want to resume.
		1363
		1364	Here is a function that initiates and resumes a download. It uses the
		1365	last modified time to check for file content changes, and works with many
		1366	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1367	on older servers.
		1368
		1369	It calls the completion callback with either C<undef>, which means a
		1370	nonretryable error occured, C<0> when the download was partial and should
		1371	be retried, and C<1> if it was successful.
		1372
		1373	use AnyEvent::HTTP;
		1374
		1375	sub download($$$) {
		1376	my ($url, $file, $cb) = @_;
		1377
		1378	open my $fh, "+<", $file
		1379	or die "$file: $!";
		1380
		1381	my %hdr;
		1382	my $ofs = 0;
		1383
		1384	warn stat $fh;
		1385	warn -s _;
		1386	if (stat $fh and -s _) {
		1387	$ofs = -s _;
		1388	warn "-s is ", $ofs;#d#
		1389	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1390	$hdr{"range"} = "bytes=$ofs-";
		1391	}
		1392
		1393	http_get $url,
		1394	headers => \%hdr,
		1395	on_header => sub {
		1396	my ($hdr) = @_;
		1397
		1398	if ($hdr->{Status} == 200 && $ofs) {
		1399	# resume failed
		1400	truncate $fh, $ofs = 0;
		1401	}
		1402
		1403	sysseek $fh, $ofs, 0;
		1404
		1405	1
		1406	},
		1407	on_body => sub {
		1408	my ($data, $hdr) = @_;
		1409
		1410	if ($hdr->{Status} =~ /^2/) {
		1411	length $data == syswrite $fh, $data
		1412	or return; # abort on write errors
		1413	}
		1414
		1415	1
		1416	},
		1417	sub {
		1418	my (undef, $hdr) = @_;
		1419
		1420	my $status = $hdr->{Status};
		1421
		1422	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1423	utime $fh, $time, $time;
		1424	}
		1425
		1426	if ($status == 200 || $status == 206 || $status == 416) {
		1427	# download ok || resume ok || file already fully downloaded
		1428	$cb->(1, $hdr);
		1429
		1430	} elsif ($status == 412) {
		1431	# file has changed while resuming, delete and retry
		1432	unlink $file;
		1433	$cb->(0, $hdr);
		1434
		1435	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1436	# retry later
		1437	$cb->(0, $hdr);
		1438
		1439	} else {
		1440	$cb->(undef, $hdr);
		1441	}
		1442	}
		1443	;
		1444	}
		1445
		1446	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1447	if ($_[0]) {
		1448	print "OK!\n";
		1449	} elsif (defined $_[0]) {
		1450	print "please retry later\n";
		1451	} else {
		1452	print "ERROR\n";
		1453	}
		1454	};
		1455
		1456	=head3 SOCKS PROXIES
		1457
		1458	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1459	compile your perl to support socks, or use an external program such as
		1460	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1461	transparently.
		1462
		1463	Alternatively, for AnyEvent::HTTP only, you can use your own
		1464	C<tcp_connect> function that does the proxy handshake - here is an example
		1465	that works with socks4a proxies:
		1466
		1467	use Errno;
		1468	use AnyEvent::Util;
		1469	use AnyEvent::Socket;
		1470	use AnyEvent::Handle;
		1471
		1472	# host, port and username of/for your socks4a proxy
		1473	my $socks_host = "10.0.0.23";
		1474	my $socks_port = 9050;
		1475	my $socks_user = "";
		1476
		1477	sub socks4a_connect {
		1478	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1479
		1480	my $hdl = new AnyEvent::Handle
		1481	connect => [$socks_host, $socks_port],
		1482	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1483	on_error => sub { $connect_cb->() },
		1484	;
		1485
		1486	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1487
		1488	$hdl->push_read (chunk => 8, sub {
		1489	my ($hdl, $chunk) = @_;
		1490	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1491
		1492	if ($status == 0x5a) {
		1493	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1494	} else {
		1495	$! = Errno::ENXIO; $connect_cb->();
		1496	}
		1497	});
		1498
		1499	$hdl
		1500	}
		1501
		1502	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1503	possibly after switching off other proxy types:
		1504
		1505	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1506
		1507	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1508	my ($data, $headers) = @_;
		1509	...
		1510	};
501		1511
502	=head1 SEE ALSO	1512	=head1 SEE ALSO
503		1513
504	L<AnyEvent>.	1514	L<AnyEvent>.
505		1515
506	=head1 AUTHOR	1516	=head1 AUTHOR
507		1517
508	Marc Lehmann <schmorp@schmorp.de>	1518	Marc Lehmann <schmorp@schmorp.de>
509	http://home.schmorp.de/	1519	http://home.schmorp.de/
		1520
		1521	With many thanks to Дмитрий Шалашов, who provided countless
		1522	testcases and bugreports.
510		1523
511	=cut	1524	=cut
512		1525
513	1	1526	1
514		1527

Diff Legend

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