ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.16 by root, Fri Jun 6 12:57:48 2008 UTC vs.
Revision 1.102 by root, Sat Feb 19 06:46:14 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 = '1.01';	51	our $VERSION = '2.04';
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;
64	our $ACTIVE = 0;	62	our $ACTIVE = 0;
65		63
66	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
67	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
68		66
69	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
70		68
71	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
72	additional parameters.	70	additional parameters and the return value.
73		71
74	=item http_head $url, key => value..., $cb->($data, $headers)	72	=item http_head $url, key => value..., $cb->($data, $headers)
75		73
76	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
77	additional parameters.	75	on additional parameters and the return value.
78		76
79	=item http_post $url, $body, key => value..., $cb->($data, $headers)	77	=item http_post $url, $body, key => value..., $cb->($data, $headers)
80		78
81	Executes an HTTP-POST request with a request body of C<$bod>. See the	79	Executes an HTTP-POST request with a request body of C<$body>. See the
82	http_request function for details on additional parameters.	80	http_request function for details on additional parameters and the return
		81	value.
83		82
84	=item http_request $method => $url, key => value..., $cb->($data, $headers)	83	=item http_request $method => $url, key => value..., $cb->($data, $headers)
85		84
86	Executes a HTTP request of type C<$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
87	must be an absolute http or https URL.	86	must be an absolute http or https URL.
88		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
89	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
90	(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
91	response headers as second argument.	95	(and trailers) as second argument.
92		96
93	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
94	headers, the three "pseudo-headers" C<HTTPVersion>, C<Status> and	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
95	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	$headers->{Redirect}[1]{Redirect} >> will then contain the original
		116	response, and so on.
		117
96	name. If the server sends a header multiple lines, then their contents	118	If the server sends a header multiple times, then their contents will be
97	will be joined together with C<\x00>.	119	joined together with a comma (C<,>), as per the HTTP spec.
98		120
99	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
100	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<599>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
101	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection etsbalishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
102		139
103	A typical callback might look like this:	140	A typical callback might look like this:
104		141
105	sub {	142	sub {
106	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
122	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication
123	retries and so on, and how often to do so.	160	retries and so on, and how often to do so.
124		161
125	=item headers => hashref	162	=item headers => hashref
126		163
127	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
128	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
129	and will provide defaults for C<User-Agent:> and C<Referer:>.	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.
130		173
131	=item timeout => $seconds	174	=item timeout => $seconds
132		175
133	The time-out to use for various stages - each connect attempt will reset	176	The time-out to use for various stages - each connect attempt will reset
134	the timeout, as will read or write activity. Default timeout is 5 minutes.	177	the timeout, as will read or write activity, i.e. this is not an overall
		178	timeout.
		179
		180	Default timeout is 5 minutes.
135		181
136	=item proxy => [$host, $port[, $scheme]] or undef	182	=item proxy => [$host, $port[, $scheme]] or undef
137		183
138	Use the given http proxy for all requests. If not specified, then the	184	Use the given http proxy for all requests, or no proxy if C<undef> is
139	default proxy (as specified by C<$ENV{http_proxy}>) is used.	185	used.
140		186
141	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.
142	HTTPS.	188
		189	If not specified, then the default proxy is used (see
		190	C<AnyEvent::HTTP::set_proxy>).
143		191
144	=item body => $string	192	=item body => $string
145		193
146	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
147	this module might offer more options).	195	this module might offer more options).
148		196
149	=item cookie_jar => $hash_ref	197	=item cookie_jar => $hash_ref
150		198
151	Passing this parameter enables (simplified) cookie-processing, loosely	199	Passing this parameter enables (simplified) cookie-processing, loosely
152	based on the original netscape specification.	200	based on the original netscape specification.
153		201
154	The C<$hash_ref> must be an (initially empty) hash reference which will	202	The C<$hash_ref> must be an (initially empty) hash reference which
155	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
156	persistent storage with something like JSON or Storable, but this is not	204	to persistent storage with something like JSON or Storable - see the
157	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.
158		208
159	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
160	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
161	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
162	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
163	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 => $boolean
		350
		351	Only used when C<persistent> is also true. This parameter decides whether
		352	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		353	(as opposed to only a HTTP/1.1 persistent connection).
		354
		355	The default is true, except when using a proxy, in which case it defaults
		356	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		357
		358	=item handle_params => { key => value ... }
		359
		360	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		361	constructor that is called - not all requests will create a handle, and
		362	sometimes more than one is created, so this parameter is only good for
		363	setting hints.
		364
		365	Example: set the maximum read size to 4096, to potentially conserve memory
		366	at the cost of speed.
		367
		368	handle_params => {
		369	max_read_size => 4096,
		370	},
164		371
165	=back	372	=back
166		373
167	Example: make a simple HTTP GET request for http://www.nethype.de/	374	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		375	the response body.
168		376
169	http_request GET => "http://www.nethype.de/", sub {	377	http_request GET => "http://www.nethype.de/", sub {
170	my ($body, $hdr) = @_;	378	my ($body, $hdr) = @_;
171	print "$body\n";	379	print "$body\n";
172	};	380	};
173		381
174	Example: make a HTTP HEAD request on https://www.google.com/, use a	382	Example: do a HTTP HEAD request on https://www.google.com/, use a
175	timeout of 30 seconds.	383	timeout of 30 seconds.
176		384
177	http_request	385	http_request
178	GET => "https://www.google.com",	386	GET => "https://www.google.com",
		387	headers => { "user-agent" => "MySearchClient 1.0" },
179	timeout => 30,	388	timeout => 30,
180	sub {	389	sub {
181	my ($body, $hdr) = @_;	390	my ($body, $hdr) = @_;
182	use Data::Dumper;	391	use Data::Dumper;
183	print Dumper $hdr;	392	print Dumper $hdr;
184	}	393	}
185	;	394	;
186		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 $request;
		405
187	=cut	406	=cut
		407
		408	#############################################################################
		409	# wait queue/slots
188		410
189	sub _slot_schedule;	411	sub _slot_schedule;
190	sub _slot_schedule($) {	412	sub _slot_schedule($) {
191	my $host = shift;	413	my $host = shift;
192		414
…		…
214	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	436	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
215		437
216	_slot_schedule $_[0];	438	_slot_schedule $_[0];
217	}	439	}
218		440
		441	#############################################################################
		442	# cookie handling
		443
		444	# expire cookies
		445	sub cookie_jar_expire($;$) {
		446	my ($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(\%$$) {
		683	my ($state, $cb, $hdr) = @_;
		684
		685	&_destroy_state ($state);
		686
		687	$cb->(undef, $hdr);
		688	()
		689	}
		690
219	sub http_request($$@) {	691	sub http_request($$@) {
220	my $cb = pop;	692	my $cb = pop;
221	my ($method, $url, %arg) = @_;	693	my ($method, $url, %arg) = @_;
222		694
223	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";
224		699
225	$method = uc $method;	700	$method = uc $method;
226		701
227	if (my $hdr = $arg{headers}) {	702	if (my $hdr = $arg{headers}) {
228	while (my ($k, $v) = each %$hdr) {	703	while (my ($k, $v) = each %$hdr) {
229	$hdr{lc $k} = $v;	704	$hdr{lc $k} = $v;
230	}	705	}
231	}	706	}
232		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
233	my $recurse = exists $arg{recurse} ? $arg{recurse} : $MAX_RECURSE;	712	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
234		713
235	return $cb->(undef, { Status => 599, Reason => "recursion limit reached" })	714	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
236	if $recurse < 0;	715	if $recurse < 0;
237		716
238	my $proxy = $arg{proxy} || $PROXY;	717	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
239	my $timeout = $arg{timeout} || $TIMEOUT;	718	my $timeout = $arg{timeout} || $TIMEOUT;
240		719
241	$hdr{"user-agent"} ||= $USERAGENT;
242
243	my ($scheme, $authority, $upath, $query, $fragment) =	720	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
244	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;	721	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
245		722
246	$scheme = lc $scheme;	723	$uscheme = lc $uscheme;
247		724
248	my $uport = $scheme eq "http" ? 80	725	my $uport = $uscheme eq "http" ? 80
249	: $scheme eq "https" ? 443	726	: $uscheme eq "https" ? 443
250	: 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" });
251		728
252	$hdr{referer} ||= "$scheme://$authority$upath"; # leave out fragment and query string, just a heuristic
253
254	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	729	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
255	or return $cb->(undef, { Status => 599, Reason => "unparsable URL" });	730	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
256		731
257	my $uhost = $1;	732	my $uhost = lc $1;
258	$uport = $2 if defined $2;	733	$uport = $2 if defined $2;
259		734
		735	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		736	unless exists $hdr{host};
		737
260	$uhost =~ s/^\[(.*)\]$/$1/;	738	$uhost =~ s/^\[(.*)\]$/$1/;
261	$upath .= "?$query" if length $query;	739	$upath .= $query if length $query;
262		740
263	$upath =~ s%^/?%/%;	741	$upath =~ s%^/?%/%;
264		742
265	# cookie processing	743	# cookie processing
266	if (my $jar = $arg{cookie_jar}) {	744	if (my $jar = $arg{cookie_jar}) {
267	%$jar = () if $jar->{version} < 1;	745	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
268		746
		747	$hdr{cookie} = join "; ", @$cookies
269	my @cookie;	748	if @$cookies;
270		749	}
271	while (my ($chost, $v) = each %$jar) {	750
272	next unless $chost eq substr $uhost, -length $chost;	751	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
273	next unless $chost =~ /^\./;	752
274		753	if ($proxy) {
275	while (my ($cpath, $v) = each %$v) {	754	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
276	next unless $cpath eq substr $upath, 0, length $cpath;	755
277		756	$rscheme = "http" unless defined $rscheme;
278	while (my ($k, $v) = each %$v) {	757
279	next if $scheme ne "https" && exists $v->{secure};	758	# don't support https requests over https-proxy transport,
280	push @cookie, "$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 $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : $idempotent;
		779	my $keepalive10 = exists $arg{keepalive10} ? $arg{keepalive10} : !$proxy;
		780	my $keptalive; # 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	$hdr{connection} = ($keepalive ? $keepalive10 ? "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;
281	}	876	}
282	}	877	}
		878
		879	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
		880	if ($state{handle}) {
		881	# handle keepalive
		882	if (
		883	$keepalive
		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	http_request (
		913	$method => $hdr{location},
		914	%arg,
		915	recurse => $recurse - 1,
		916	Redirect => [$_[0], \%hdr],
		917	$cb
		918	);
		919	} else {
		920	$cb->($_[0], \%hdr);
		921	}
		922	};
		923
		924	$ae_error = 597; # body phase
		925
		926	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		927
		928	my $len = $chunked ? undef : $hdr{"content-length"};
		929
		930	# body handling, many different code paths
		931	# - no body expected
		932	# - want_body_handle
		933	# - te chunked
		934	# - 2x length known (with or without on_body)
		935	# - 2x length not known (with or without on_body)
		936	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		937	$finish->(undef, 598 => "Request cancelled by on_header");
		938	} elsif (
		939	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		940	or $method eq "HEAD"
		941	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		942	) {
		943	# no body
		944	$finish->("", undef, undef, 1);
		945
		946	} elsif (!$redirect && $arg{want_body_handle}) {
		947	$_[0]->on_eof (undef);
		948	$_[0]->on_error (undef);
		949	$_[0]->on_read (undef);
		950
		951	$finish->(delete $state{handle});
		952
		953	} elsif ($chunked) {
		954	my $cl = 0;
		955	my $body = "";
		956	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		957
		958	$state{read_chunk} = sub {
		959	$_[1] =~ /^([0-9a-fA-F]+)/
		960	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		961
		962	my $len = hex $1;
		963
		964	if ($len) {
		965	$cl += $len;
		966
		967	$_[0]->push_read (chunk => $len, sub {
		968	$on_body->($_[1], \%hdr)
		969	or return $finish->(undef, 598 => "Request cancelled by on_body");
		970
		971	$_[0]->push_read (line => sub {
		972	length $_[1]
		973	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		974	$_[0]->push_read (line => $state{read_chunk});
		975	});
		976	});
		977	} else {
		978	$hdr{"content-length"} ||= $cl;
		979
		980	$_[0]->push_read (line => $qr_nlnl, sub {
		981	if (length $_[1]) {
		982	for ("$_[1]") {
		983	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		984
		985	my $hdr = _parse_hdr
		986	or return $finish->(undef, $ae_error => "Garbled response trailers");
		987
		988	%hdr = (%hdr, %$hdr);
		989	}
		990	}
		991
		992	$finish->($body, undef, undef, 1);
		993	});
		994	}
		995	};
		996
		997	$_[0]->push_read (line => $state{read_chunk});
		998
		999	} elsif ($arg{on_body}) {
		1000	if (defined $len) {
		1001	$_[0]->on_read (sub {
		1002	$len -= length $_[0]{rbuf};
		1003
		1004	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1005	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1006
		1007	$len > 0
		1008	or $finish->("", undef, undef, 1);
		1009	});
		1010	} else {
		1011	$_[0]->on_eof (sub {
		1012	$finish->("");
		1013	});
		1014	$_[0]->on_read (sub {
		1015	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1016	or $finish->(undef, 598 => "Request cancelled by on_body");
		1017	});
		1018	}
		1019	} else {
		1020	$_[0]->on_eof (undef);
		1021
		1022	if (defined $len) {
		1023	$_[0]->on_read (sub {
		1024	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1025	if $len <= length $_[0]{rbuf};
		1026	});
		1027	} else {
		1028	$_[0]->on_error (sub {
		1029	($! == Errno::EPIPE || !$!)
		1030	? $finish->(delete $_[0]{rbuf})
		1031	: $finish->(undef, $ae_error => $_[2]);
		1032	});
		1033	$_[0]->on_read (sub { });
		1034	}
		1035	}
		1036	};
		1037
		1038	# if keepalive is enabled, then the server closing the connection
		1039	# before a response can happen legally - we retry on idempotent methods.
		1040	if ($keptalive && $idempotent) {
		1041	my $old_eof = $hdl->{on_eof};
		1042	$hdl->{on_eof} = sub {
		1043	_destroy_state %state;
		1044
		1045	http_request (
		1046	$method => $url,
		1047	%arg,
		1048	keepalive => 0,
		1049	$cb
		1050	);
		1051	};
		1052	$hdl->on_read (sub {
		1053	return unless %state;
		1054
		1055	# as soon as we receive something, a connection close
		1056	# once more becomes a hard error
		1057	$hdl->{on_eof} = $old_eof;
		1058	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1059	});
		1060	} else {
		1061	$hdl->push_read (line => $qr_nlnl, $state{read_response});
283	}	1062	}
284
285	$hdr{cookie} = join "; ", @cookie
286	if @cookie;
287	}	1063	};
288		1064
289	my ($rhost, $rport, $rpath); # request host, port, path	1065	my $prepare_handle = sub {
		1066	my ($hdl) = $state{handle};
290		1067
291	if ($proxy) {	1068	$hdl->timeout ($timeout);
292	($rhost, $rport, $scheme) = @$proxy;	1069	$hdl->on_error (sub {
293	$rpath = $url;	1070	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
294	} else {	1071	});
295	($rhost, $rport, $rpath) = ($uhost, $uport, $upath);	1072	$hdl->on_eof (sub {
296	$hdr{host} = $uhost;	1073	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1074	});
297	}	1075	};
298		1076
299	$hdr{"content-length"} = length $arg{body};	1077	# connected to proxy (or origin server)
		1078	my $connect_cb = sub {
		1079	my $fh = shift
		1080	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
300		1081
301	my %state = (connect_guard => 1);	1082	return unless delete $state{connect_guard};
		1083
		1084	# get handle
		1085	$state{handle} = new AnyEvent::Handle
		1086	%{ $arg{handle_params} },
		1087	fh => $fh,
		1088	peername => $uhost,
		1089	tls_ctx => $arg{tls_ctx},
		1090	;
		1091
		1092	$prepare_handle->();
		1093
		1094	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1095
		1096	# now handle proxy-CONNECT method
		1097	if ($proxy && $uscheme eq "https") {
		1098	# oh dear, we have to wrap it into a connect request
		1099
		1100	# maybe re-use $uauthority with patched port?
		1101	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1102	$state{handle}->push_read (line => $qr_nlnl, sub {
		1103	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1104	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1105
		1106	if ($2 == 200) {
		1107	$rpath = $upath;
		1108	$handle_actual_request->();
		1109	} else {
		1110	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1111	}
		1112	});
		1113	} else {
		1114	$handle_actual_request->();
		1115	}
		1116	};
302		1117
303	_get_slot $uhost, sub {	1118	_get_slot $uhost, sub {
304	$state{slot_guard} = shift;	1119	$state{slot_guard} = shift;
305		1120
306	return unless $state{connect_guard};	1121	return unless $state{connect_guard};
307		1122
308	$state{connect_guard} = AnyEvent::Socket::tcp_connect $rhost, $rport, sub {	1123	# try to use an existing keepalive connection, but only if we, ourselves, plan
309	$state{fh} = shift	1124	# on a keepalive request (in theory, this should be a separate config option).
310	or return $cb->(undef, { Status => 599, Reason => "$!" });	1125	if ($keepalive && $KA_CACHE{$ka_key}) {
		1126	$keptalive = 1;
		1127	$state{handle} = ka_fetch $ka_key;
		1128	$prepare_handle->();
		1129	$handle_actual_request->();
311		1130
312	delete $state{connect_guard}; # reduce memory usage, save a tree
313
314	# get handle
315	$state{handle} = new AnyEvent::Handle
316	fh => $state{fh},
317	($scheme eq "https" ? (tls => "connect") : ());
318
319	# limit the number of persistent connections
320	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
321	++$KA_COUNT{$_[1]};
322	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
323	$hdr{connection} = "keep-alive";
324	delete $hdr{connection}; # keep-alive not yet supported
325	} else {	1131	} else {
326	delete $hdr{connection};	1132	my $tcp_connect = $arg{tcp_connect}
327	}	1133	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
328		1134
329	# (re-)configure handle	1135	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
330	$state{handle}->timeout ($timeout);
331	$state{handle}->on_error (sub {
332	my $errno = "$!";
333	%state = ();
334	$cb->(undef, { Status => 599, Reason => $errno });
335	});
336	$state{handle}->on_eof (sub {
337	%state = ();
338	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file" });
339	});
340
341	# send request
342	$state{handle}->push_write (
343	"$method $rpath HTTP/1.0\015\012"
344	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
345	. "\015\012"
346	. (delete $arg{body})
347	);
348
349	%hdr = (); # reduce memory usage, save a kitten
350
351	# status line
352	$state{handle}->push_read (line => qr/\015?\012/, sub {
353	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix
354	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])" }));
355
356	my %hdr = ( # response headers
357	HTTPVersion => "\x00$1",
358	Status => "\x00$2",
359	Reason => "\x00$3",
360	);
361
362	# headers, could be optimized a bit
363	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
364	for ("$_[1]\012") {
365	# we support spaces in field names, as lotus domino
366	# creates them.
367	$hdr{lc $1} .= "\x00$2"
368	while /\G
369	([^:\000-\037]+):
370	[\011\040]*
371	((?: [^\015\012]+ | \015?\012[\011\040] )*)
372	\015?\012
373	/gxc;
374
375	/\G$/
376	or return (%state = (), $cb->(undef, { Status => 599, Reason => "garbled response headers" }));
377	}
378
379	substr $_, 0, 1, ""
380	for values %hdr;
381
382	my $finish = sub {
383	%state = ();
384
385	# set-cookie processing
386	if ($arg{cookie_jar} && exists $hdr{"set-cookie"}) {
387	for (split /\x00/, $hdr{"set-cookie"}) {
388	my ($cookie, @arg) = split /;\s*/;
389	my ($name, $value) = split /=/, $cookie, 2;
390	my %kv = (value => $value, map { split /=/, $_, 2 } @arg);
391
392	my $cdom = (delete $kv{domain}) || $uhost;
393	my $cpath = (delete $kv{path}) || "/";
394
395	$cdom =~ s/^.?/./; # make sure it starts with a "."
396
397	next if $cdom =~ /\.$/;
398
399	# this is not rfc-like and not netscape-like. go figure.
400	my $ndots = $cdom =~ y/.//;
401	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
402
403	# store it
404	$arg{cookie_jar}{version} = 1;
405	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
406	}
407	}
408
409	if ($_[1]{Status} =~ /^30[12]$/ && $recurse) {
410	# microsoft and other assholes don't give a shit for following standards,
411	# try to support a common form of broken Location header.
412	$_[1]{location} =~ s%^/%$scheme://$uhost:$uport/%;
413
414	http_request ($method, $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
415	} else {
416	$cb->($_[0], $_[1]);
417	}
418	};
419
420	if ($hdr{Status} =~ /^(?:1..|204|304)$/ or $method eq "HEAD") {
421	$finish->(undef, \%hdr);
422	} else {
423	if (exists $hdr{"content-length"}) {
424	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
425	# could cache persistent connection now
426	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
427	# but we don't, due to misdesigns, this is annoyingly complex
428	};
429
430	$finish->($_[1], \%hdr);
431	});
432	} else {
433	# too bad, need to read until we get an error or EOF,
434	# no way to detect winged data.
435	$_[0]->on_error (sub {
436	$finish->($_[0]{rbuf}, \%hdr);
437	});
438	$_[0]->on_eof (undef);
439	$_[0]->on_read (sub { });
440	}
441	}
442	});
443	});
444	}, sub {
445	$timeout
446	};	1136	}
447	};	1137	};
448		1138
449	defined wantarray && AnyEvent::Util::guard { %state = () }	1139	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
450	}	1140	}
451		1141
452	sub http_get($@) {	1142	sub http_get($@) {
453	unshift @_, "GET";	1143	unshift @_, "GET";
454	&http_request	1144	&http_request
…		…
458	unshift @_, "HEAD";	1148	unshift @_, "HEAD";
459	&http_request	1149	&http_request
460	}	1150	}
461		1151
462	sub http_post($$@) {	1152	sub http_post($$@) {
		1153	my $url = shift;
463	unshift @_, "POST", "body";	1154	unshift @_, "POST", $url, "body";
464	&http_request	1155	&http_request
465	}	1156	}
466		1157
467	=back	1158	=back
468		1159
		1160	=head2 DNS CACHING
		1161
		1162	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1163	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1164	hostnames. The latter is a simple stub resolver and does no caching
		1165	on its own. If you want DNS caching, you currently have to provide
		1166	your own default resolver (by storing a suitable resolver object in
		1167	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1168
469	=head2 GLOBAL FUNCTIONS AND VARIABLES	1169	=head2 GLOBAL FUNCTIONS AND VARIABLES
470		1170
471	=over 4	1171	=over 4
472		1172
473	=item AnyEvent::HTTP::set_proxy "proxy-url"	1173	=item AnyEvent::HTTP::set_proxy "proxy-url"
474		1174
475	Sets the default proxy server to use. The proxy-url must begin with a	1175	Sets the default proxy server to use. The proxy-url must begin with a
476	string of the form C<http://host:port> (optionally C<https:...>).	1176	string of the form C<http://host:port>, croaks otherwise.
		1177
		1178	To clear an already-set proxy, use C<undef>.
		1179
		1180	When AnyEvent::HTTP is laoded for the first time it will query the
		1181	default proxy from the operating system, currently by looking at
		1182	C<$ENV{http_proxy>}.
		1183
		1184	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1185
		1186	Remove all cookies from the cookie jar that have been expired. If
		1187	C<$session_end> is given and true, then additionally remove all session
		1188	cookies.
		1189
		1190	You should call this function (with a true C<$session_end>) before you
		1191	save cookies to disk, and you should call this function after loading them
		1192	again. If you have a long-running program you can additonally call this
		1193	function from time to time.
		1194
		1195	A cookie jar is initially an empty hash-reference that is managed by this
		1196	module. It's format is subject to change, but currently it is like this:
		1197
		1198	The key C<version> has to contain C<1>, otherwise the hash gets
		1199	emptied. All other keys are hostnames or IP addresses pointing to
		1200	hash-references. The key for these inner hash references is the
		1201	server path for which this cookie is meant, and the values are again
		1202	hash-references. The keys of those hash-references is the cookie name, and
		1203	the value, you guessed it, is another hash-reference, this time with the
		1204	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1205	which have been replaced by a C<_expires> key that contains the cookie
		1206	expiry timestamp.
		1207
		1208	Here is an example of a cookie jar with a single cookie, so you have a
		1209	chance of understanding the above paragraph:
		1210
		1211	{
		1212	version => 1,
		1213	"10.0.0.1" => {
		1214	"/" => {
		1215	"mythweb_id" => {
		1216	_expires => 1293917923,
		1217	value => "ooRung9dThee3ooyXooM1Ohm",
		1218	},
		1219	},
		1220	},
		1221	}
		1222
		1223	=item $date = AnyEvent::HTTP::format_date $timestamp
		1224
		1225	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1226	Date (RFC 2616).
		1227
		1228	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1229
		1230	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1231	bunch of minor variations of those, and returns the corresponding POSIX
		1232	timestamp, or C<undef> if the date cannot be parsed.
477		1233
478	=item $AnyEvent::HTTP::MAX_RECURSE	1234	=item $AnyEvent::HTTP::MAX_RECURSE
479		1235
480	The default value for the C<recurse> request parameter (default: C<10>).	1236	The default value for the C<recurse> request parameter (default: C<10>).
481		1237
		1238	=item $AnyEvent::HTTP::TIMEOUT
		1239
		1240	The default timeout for conenction operations (default: C<300>).
		1241
482	=item $AnyEvent::HTTP::USERAGENT	1242	=item $AnyEvent::HTTP::USERAGENT
483		1243
484	The default value for the C<User-Agent> header (the default is	1244	The default value for the C<User-Agent> header (the default is
485	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1245	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
486		1246
487	=item $AnyEvent::HTTP::MAX_PERSISTENT	1247	=item $AnyEvent::HTTP::MAX_PER_HOST
488		1248
489	The maximum number of persistent connections to keep open (default: 8).	1249	The maximum number of concurrent connections to the same host (identified
		1250	by the hostname). If the limit is exceeded, then the additional requests
		1251	are queued until previous connections are closed. Both persistent and
		1252	non-persistent connections are counted in this limit.
490		1253
491	Not implemented currently.	1254	The default value for this is C<4>, and it is highly advisable to not
		1255	increase it much.
		1256
		1257	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1258	connections, older browsers used 2, newers (such as firefox 3) typically
		1259	use 6, and Opera uses 8 because like, they have the fastest browser and
		1260	give a shit for everybody else on the planet.
492		1261
493	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1262	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
494		1263
495	The maximum time to cache a persistent connection, in seconds (default: 2).	1264	The time after which idle persistent conenctions get closed by
496		1265	AnyEvent::HTTP (default: C<3>).
497	Not implemented currently.
498		1266
499	=item $AnyEvent::HTTP::ACTIVE	1267	=item $AnyEvent::HTTP::ACTIVE
500		1268
501	The number of active connections. This is not the number of currently	1269	The number of active connections. This is not the number of currently
502	running requests, but the number of currently open and non-idle TCP	1270	running requests, but the number of currently open and non-idle TCP
503	connections. This number of can be useful for load-leveling.	1271	connections. This number can be useful for load-leveling.
504		1272
505	=back	1273	=back
506		1274
507	=cut	1275	=cut
508		1276
		1277	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1278	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1279
		1280	sub format_date($) {
		1281	my ($time) = @_;
		1282
		1283	# RFC 822/1123 format
		1284	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1285
		1286	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1287	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1288	$H, $M, $S;
		1289	}
		1290
		1291	sub parse_date($) {
		1292	my ($date) = @_;
		1293
		1294	my ($d, $m, $y, $H, $M, $S);
		1295
		1296	if ($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$/) {
		1297	# RFC 822/1123, required by RFC 2616 (with " ")
		1298	# cookie dates (with "-")
		1299
		1300	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1301
		1302	} elsif ($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$/) {
		1303	# RFC 850
		1304	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1305
		1306	} elsif ($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])$/) {
		1307	# ISO C's asctime
		1308	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1309	}
		1310	# other formats fail in the loop below
		1311
		1312	for (0..11) {
		1313	if ($m eq $month[$_]) {
		1314	require Time::Local;
		1315	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);
		1316	}
		1317	}
		1318
		1319	undef
		1320	}
		1321
509	sub set_proxy($) {	1322	sub set_proxy($) {
510	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1323	if (length $_[0]) {
		1324	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1325	or Carp::croak "$_[0]: invalid proxy URL";
		1326	$PROXY = [$2, $3 || 3128, $1]
		1327	} else {
		1328	undef $PROXY;
		1329	}
511	}	1330	}
512		1331
513	# initialise proxy from environment	1332	# initialise proxy from environment
		1333	eval {
514	set_proxy $ENV{http_proxy};	1334	set_proxy $ENV{http_proxy};
		1335	};
		1336
		1337	=head2 SHOWCASE
		1338
		1339	This section contaisn some more elaborate "real-world" examples or code
		1340	snippets.
		1341
		1342	=head2 HTTP/1.1 FILE DOWNLOAD
		1343
		1344	Downloading files with HTTP can be quite tricky, especially when something
		1345	goes wrong and you want to resume.
		1346
		1347	Here is a function that initiates and resumes a download. It uses the
		1348	last modified time to check for file content changes, and works with many
		1349	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1350	on older servers.
		1351
		1352	It calls the completion callback with either C<undef>, which means a
		1353	nonretryable error occured, C<0> when the download was partial and should
		1354	be retried, and C<1> if it was successful.
		1355
		1356	use AnyEvent::HTTP;
		1357
		1358	sub download($$$) {
		1359	my ($url, $file, $cb) = @_;
		1360
		1361	open my $fh, "+<", $file
		1362	or die "$file: $!";
		1363
		1364	my %hdr;
		1365	my $ofs = 0;
		1366
		1367	warn stat $fh;
		1368	warn -s _;
		1369	if (stat $fh and -s _) {
		1370	$ofs = -s _;
		1371	warn "-s is ", $ofs;#d#
		1372	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1373	$hdr{"range"} = "bytes=$ofs-";
		1374	}
		1375
		1376	http_get $url,
		1377	headers => \%hdr,
		1378	on_header => sub {
		1379	my ($hdr) = @_;
		1380
		1381	if ($hdr->{Status} == 200 && $ofs) {
		1382	# resume failed
		1383	truncate $fh, $ofs = 0;
		1384	}
		1385
		1386	sysseek $fh, $ofs, 0;
		1387
		1388	1
		1389	},
		1390	on_body => sub {
		1391	my ($data, $hdr) = @_;
		1392
		1393	if ($hdr->{Status} =~ /^2/) {
		1394	length $data == syswrite $fh, $data
		1395	or return; # abort on write errors
		1396	}
		1397
		1398	1
		1399	},
		1400	sub {
		1401	my (undef, $hdr) = @_;
		1402
		1403	my $status = $hdr->{Status};
		1404
		1405	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1406	utime $fh, $time, $time;
		1407	}
		1408
		1409	if ($status == 200 || $status == 206 || $status == 416) {
		1410	# download ok || resume ok || file already fully downloaded
		1411	$cb->(1, $hdr);
		1412
		1413	} elsif ($status == 412) {
		1414	# file has changed while resuming, delete and retry
		1415	unlink $file;
		1416	$cb->(0, $hdr);
		1417
		1418	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1419	# retry later
		1420	$cb->(0, $hdr);
		1421
		1422	} else {
		1423	$cb->(undef, $hdr);
		1424	}
		1425	}
		1426	;
		1427	}
		1428
		1429	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1430	if ($_[0]) {
		1431	print "OK!\n";
		1432	} elsif (defined $_[0]) {
		1433	print "please retry later\n";
		1434	} else {
		1435	print "ERROR\n";
		1436	}
		1437	};
		1438
		1439	=head3 SOCKS PROXIES
		1440
		1441	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1442	compile your perl to support socks, or use an external program such as
		1443	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1444	transparently.
		1445
		1446	Alternatively, for AnyEvent::HTTP only, you can use your own
		1447	C<tcp_connect> function that does the proxy handshake - here is an example
		1448	that works with socks4a proxies:
		1449
		1450	use Errno;
		1451	use AnyEvent::Util;
		1452	use AnyEvent::Socket;
		1453	use AnyEvent::Handle;
		1454
		1455	# host, port and username of/for your socks4a proxy
		1456	my $socks_host = "10.0.0.23";
		1457	my $socks_port = 9050;
		1458	my $socks_user = "";
		1459
		1460	sub socks4a_connect {
		1461	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1462
		1463	my $hdl = new AnyEvent::Handle
		1464	connect => [$socks_host, $socks_port],
		1465	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1466	on_error => sub { $connect_cb->() },
		1467	;
		1468
		1469	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1470
		1471	$hdl->push_read (chunk => 8, sub {
		1472	my ($hdl, $chunk) = @_;
		1473	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1474
		1475	if ($status == 0x5a) {
		1476	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1477	} else {
		1478	$! = Errno::ENXIO; $connect_cb->();
		1479	}
		1480	});
		1481
		1482	$hdl
		1483	}
		1484
		1485	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1486	possibly after switching off other proxy types:
		1487
		1488	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1489
		1490	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1491	my ($data, $headers) = @_;
		1492	...
		1493	};
515		1494
516	=head1 SEE ALSO	1495	=head1 SEE ALSO
517		1496
518	L<AnyEvent>.	1497	L<AnyEvent>.
519		1498
520	=head1 AUTHOR	1499	=head1 AUTHOR
521		1500
522	Marc Lehmann <schmorp@schmorp.de>	1501	Marc Lehmann <schmorp@schmorp.de>
523	http://home.schmorp.de/	1502	http://home.schmorp.de/
		1503
		1504	With many thanks to Дмитрий Шалашов, who provided countless
		1505	testcases and bugreports.
524		1506
525	=cut	1507	=cut
526		1508
527	1	1509	1
528		1510

Diff Legend

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