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 1.106 by root, Tue Jun 14 05:20:13 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 = '2.11';
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.
		173
		174	Also, make sure that your headers names and values do not contain any
		175	embedded newlines.
130		176
131	=item timeout => $seconds	177	=item timeout => $seconds
132		178
133	The time-out to use for various stages - each connect attempt will reset	179	The time-out to use for various stages - each connect attempt will reset
134	the timeout, as will read or write activity. Default timeout is 5 minutes.	180	the timeout, as will read or write activity, i.e. this is not an overall
		181	timeout.
		182
		183	Default timeout is 5 minutes.
135		184
136	=item proxy => [$host, $port[, $scheme]] or undef	185	=item proxy => [$host, $port[, $scheme]] or undef
137		186
138	Use the given http proxy for all requests. If not specified, then the	187	Use the given http proxy for all requests, or no proxy if C<undef> is
139	default proxy (as specified by C<$ENV{http_proxy}>) is used.	188	used.
140		189
141	C<$scheme> must be either missing or C<http> for HTTP, or C<https> for	190	C<$scheme> must be either missing or must be C<http> for HTTP.
142	HTTPS.	191
		192	If not specified, then the default proxy is used (see
		193	C<AnyEvent::HTTP::set_proxy>).
143		194
144	=item body => $string	195	=item body => $string
145		196
146	The request body, usually empty. Will be-sent as-is (future versions of	197	The request body, usually empty. Will be sent as-is (future versions of
147	this module might offer more options).	198	this module might offer more options).
148		199
149	=item cookie_jar => $hash_ref	200	=item cookie_jar => $hash_ref
150		201
151	Passing this parameter enables (simplified) cookie-processing, loosely	202	Passing this parameter enables (simplified) cookie-processing, loosely
152	based on the original netscape specification.	203	based on the original netscape specification.
153		204
154	The C<$hash_ref> must be an (initially empty) hash reference which will	205	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	206	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	207	to persistent storage with something like JSON or Storable - see the
157	recommended, as expire times are currently being ignored.	208	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		209	expired or session-only cookies, and also for documentation on the format
		210	of the cookie jar.
158		211
159	Note that this cookie implementation is not of very high quality, nor	212	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	213	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	214	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	215	working. Cookies are a privacy disaster, do not use them unless required
163	them unless required to.	216	to.
		217
		218	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		219	headers will be set and handled by this module, otherwise they will be
		220	left untouched.
		221
		222	=item tls_ctx => $scheme | $tls_ctx
		223
		224	Specifies the AnyEvent::TLS context to be used for https connections. This
		225	parameter follows the same rules as the C<tls_ctx> parameter to
		226	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		227	C<high> can be specified, which give you a predefined low-security (no
		228	verification, highest compatibility) and high-security (CA and common-name
		229	verification) TLS context.
		230
		231	The default for this option is C<low>, which could be interpreted as "give
		232	me the page, no matter what".
		233
		234	See also the C<sessionid> parameter.
		235
		236	=item session => $string
		237
		238	The module might reuse connections to the same host internally. Sometimes
		239	(e.g. when using TLS), you do not want to reuse connections from other
		240	sessions. This can be achieved by setting this parameter to some unique
		241	ID (such as the address of an object storing your state data, or the TLS
		242	context) - only connections using the same unique ID will be reused.
		243
		244	=item on_prepare => $callback->($fh)
		245
		246	In rare cases you need to "tune" the socket before it is used to
		247	connect (for exmaple, to bind it on a given IP address). This parameter
		248	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		249	and behaves exactly the same way (e.g. it has to provide a
		250	timeout). See the description for the C<$prepare_cb> argument of
		251	C<AnyEvent::Socket::tcp_connect> for details.
		252
		253	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		254
		255	In even rarer cases you want total control over how AnyEvent::HTTP
		256	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		257	to do this, but you can provide your own C<tcp_connect> function -
		258	obviously, it has to follow the same calling conventions, except that it
		259	may always return a connection guard object.
		260
		261	There are probably lots of weird uses for this function, starting from
		262	tracing the hosts C<http_request> actually tries to connect, to (inexact
		263	but fast) host => IP address caching or even socks protocol support.
		264
		265	=item on_header => $callback->($headers)
		266
		267	When specified, this callback will be called with the header hash as soon
		268	as headers have been successfully received from the remote server (not on
		269	locally-generated errors).
		270
		271	It has to return either true (in which case AnyEvent::HTTP will continue),
		272	or false, in which case AnyEvent::HTTP will cancel the download (and call
		273	the finish callback with an error code of C<598>).
		274
		275	This callback is useful, among other things, to quickly reject unwanted
		276	content, which, if it is supposed to be rare, can be faster than first
		277	doing a C<HEAD> request.
		278
		279	The downside is that cancelling the request makes it impossible to re-use
		280	the connection. Also, the C<on_header> callback will not receive any
		281	trailer (headers sent after the response body).
		282
		283	Example: cancel the request unless the content-type is "text/html".
		284
		285	on_header => sub {
		286	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		287	},
		288
		289	=item on_body => $callback->($partial_body, $headers)
		290
		291	When specified, all body data will be passed to this callback instead of
		292	to the completion callback. The completion callback will get the empty
		293	string instead of the body data.
		294
		295	It has to return either true (in which case AnyEvent::HTTP will continue),
		296	or false, in which case AnyEvent::HTTP will cancel the download (and call
		297	the completion callback with an error code of C<598>).
		298
		299	The downside to cancelling the request is that it makes it impossible to
		300	re-use the connection.
		301
		302	This callback is useful when the data is too large to be held in memory
		303	(so the callback writes it to a file) or when only some information should
		304	be extracted, or when the body should be processed incrementally.
		305
		306	It is usually preferred over doing your own body handling via
		307	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		308	only used to create a connection, C<want_body_handle> is the better
		309	alternative, as it allows you to install your own event handler, reducing
		310	resource usage.
		311
		312	=item want_body_handle => $enable
		313
		314	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		315	changes considerably: after parsing the headers, and instead of
		316	downloading the body (if any), the completion callback will be
		317	called. Instead of the C<$body> argument containing the body data, the
		318	callback will receive the L<AnyEvent::Handle> object associated with the
		319	connection. In error cases, C<undef> will be passed. When there is no body
		320	(e.g. status C<304>), the empty string will be passed.
		321
		322	The handle object might or might not be in TLS mode, might be connected
		323	to a proxy, be a persistent connection, use chunked transfer encoding
		324	etc., and configured in unspecified ways. The user is responsible for this
		325	handle (it will not be used by this module anymore).
		326
		327	This is useful with some push-type services, where, after the initial
		328	headers, an interactive protocol is used (typical example would be the
		329	push-style twitter API which starts a JSON/XML stream).
		330
		331	If you think you need this, first have a look at C<on_body>, to see if
		332	that doesn't solve your problem in a better way.
		333
		334	=item persistent => $boolean
		335
		336	Try to create/reuse a persistent connection. When this flag is set
		337	(default: true for idempotent requests, false for all others), then
		338	C<http_request> tries to re-use an existing (previously-created)
		339	persistent connection to the host and, failing that, tries to create a new
		340	one.
		341
		342	Requests failing in certain ways will be automatically retried once, which
		343	is dangerous for non-idempotent requests, which is why it defaults to off
		344	for them. The reason for this is because the bozos who designed HTTP/1.1
		345	made it impossible to distinguish between a fatal error and a normal
		346	connection timeout, so you never know whether there was a problem with
		347	your request or not.
		348
		349	When reusing an existent connection, many parameters (such as TLS context)
		350	will be ignored. See the C<session> parameter for a workaround.
		351
		352	=item keepalive => $boolean
		353
		354	Only used when C<persistent> is also true. This parameter decides whether
		355	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		356	(as opposed to only a HTTP/1.1 persistent connection).
		357
		358	The default is true, except when using a proxy, in which case it defaults
		359	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		360
		361	=item handle_params => { key => value ... }
		362
		363	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		364	constructor that is called - not all requests will create a handle, and
		365	sometimes more than one is created, so this parameter is only good for
		366	setting hints.
		367
		368	Example: set the maximum read size to 4096, to potentially conserve memory
		369	at the cost of speed.
		370
		371	handle_params => {
		372	max_read_size => 4096,
		373	},
164		374
165	=back	375	=back
166		376
167	Example: make a simple HTTP GET request for http://www.nethype.de/	377	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		378	the response body.
168		379
169	http_request GET => "http://www.nethype.de/", sub {	380	http_request GET => "http://www.nethype.de/", sub {
170	my ($body, $hdr) = @_;	381	my ($body, $hdr) = @_;
171	print "$body\n";	382	print "$body\n";
172	};	383	};
173		384
174	Example: make a HTTP HEAD request on https://www.google.com/, use a	385	Example: do a HTTP HEAD request on https://www.google.com/, use a
175	timeout of 30 seconds.	386	timeout of 30 seconds.
176		387
177	http_request	388	http_request
178	GET => "https://www.google.com",	389	GET => "https://www.google.com",
		390	headers => { "user-agent" => "MySearchClient 1.0" },
179	timeout => 30,	391	timeout => 30,
180	sub {	392	sub {
181	my ($body, $hdr) = @_;	393	my ($body, $hdr) = @_;
182	use Data::Dumper;	394	use Data::Dumper;
183	print Dumper $hdr;	395	print Dumper $hdr;
184	}	396	}
185	;	397	;
186		398
		399	Example: do another simple HTTP GET request, but immediately try to
		400	cancel it.
		401
		402	my $request = http_request GET => "http://www.nethype.de/", sub {
		403	my ($body, $hdr) = @_;
		404	print "$body\n";
		405	};
		406
		407	undef $request;
		408
187	=cut	409	=cut
		410
		411	#############################################################################
		412	# wait queue/slots
188		413
189	sub _slot_schedule;	414	sub _slot_schedule;
190	sub _slot_schedule($) {	415	sub _slot_schedule($) {
191	my $host = shift;	416	my $host = shift;
192		417
…		…
214	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	439	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
215		440
216	_slot_schedule $_[0];	441	_slot_schedule $_[0];
217	}	442	}
218		443
		444	#############################################################################
		445	# cookie handling
		446
		447	# expire cookies
		448	sub cookie_jar_expire($;$) {
		449	my ($jar, $session_end) = @_;
		450
		451	%$jar = () if $jar->{version} != 1;
		452
		453	my $anow = AE::now;
		454
		455	while (my ($chost, $paths) = each %$jar) {
		456	next unless ref $paths;
		457
		458	while (my ($cpath, $cookies) = each %$paths) {
		459	while (my ($cookie, $kv) = each %$cookies) {
		460	if (exists $kv->{_expires}) {
		461	delete $cookies->{$cookie}
		462	if $anow > $kv->{_expires};
		463	} elsif ($session_end) {
		464	delete $cookies->{$cookie};
		465	}
		466	}
		467
		468	delete $paths->{$cpath}
		469	unless %$cookies;
		470	}
		471
		472	delete $jar->{$chost}
		473	unless %$paths;
		474	}
		475	}
		476
		477	# extract cookies from jar
		478	sub cookie_jar_extract($$$$) {
		479	my ($jar, $scheme, $host, $path) = @_;
		480
		481	%$jar = () if $jar->{version} != 1;
		482
		483	my @cookies;
		484
		485	while (my ($chost, $paths) = each %$jar) {
		486	next unless ref $paths;
		487
		488	if ($chost =~ /^\./) {
		489	next unless $chost eq substr $host, -length $chost;
		490	} elsif ($chost =~ /\./) {
		491	next unless $chost eq $host;
		492	} else {
		493	next;
		494	}
		495
		496	while (my ($cpath, $cookies) = each %$paths) {
		497	next unless $cpath eq substr $path, 0, length $cpath;
		498
		499	while (my ($cookie, $kv) = each %$cookies) {
		500	next if $scheme ne "https" && exists $kv->{secure};
		501
		502	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		503	delete $cookies->{$cookie};
		504	next;
		505	}
		506
		507	my $value = $kv->{value};
		508
		509	if ($value =~ /[=;,[:space:]]/) {
		510	$value =~ s/([\\"])/\\$1/g;
		511	$value = "\"$value\"";
		512	}
		513
		514	push @cookies, "$cookie=$value";
		515	}
		516	}
		517	}
		518
		519	\@cookies
		520	}
		521
		522	# parse set_cookie header into jar
		523	sub cookie_jar_set_cookie($$$$) {
		524	my ($jar, $set_cookie, $host, $date) = @_;
		525
		526	my $anow = int AE::now;
		527	my $snow; # server-now
		528
		529	for ($set_cookie) {
		530	# parse NAME=VALUE
		531	my @kv;
		532
		533	# expires is not http-compliant in the original cookie-spec,
		534	# we support the official date format and some extensions
		535	while (
		536	m{
		537	\G\s*
		538	(?:
		539	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		540	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
		541	)
		542	}gcxsi
		543	) {
		544	my $name = $2;
		545	my $value = $4;
		546
		547	if (defined $1) {
		548	# expires
		549	$name = "expires";
		550	$value = $1;
		551	} elsif (defined $3) {
		552	# quoted
		553	$value = $3;
		554	$value =~ s/\\(.)/$1/gs;
		555	}
		556
		557	push @kv, @kv ? lc $name : $name, $value;
		558
		559	last unless /\G\s*;/gc;
		560	}
		561
		562	last unless @kv;
		563
		564	my $name = shift @kv;
		565	my %kv = (value => shift @kv, @kv);
		566
		567	if (exists $kv{"max-age"}) {
		568	$kv{_expires} = $anow + delete $kv{"max-age"};
		569	} elsif (exists $kv{expires}) {
		570	$snow ||= parse_date ($date) || $anow;
		571	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		572	} else {
		573	delete $kv{_expires};
		574	}
		575
		576	my $cdom;
		577	my $cpath = (delete $kv{path}) || "/";
		578
		579	if (exists $kv{domain}) {
		580	$cdom = delete $kv{domain};
		581
		582	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		583
		584	next if $cdom =~ /\.$/;
		585
		586	# this is not rfc-like and not netscape-like. go figure.
		587	my $ndots = $cdom =~ y/.//;
		588	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		589	} else {
		590	$cdom = $host;
		591	}
		592
		593	# store it
		594	$jar->{version} = 1;
		595	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		596
		597	redo if /\G\s*,/gc;
		598	}
		599	}
		600
		601	#############################################################################
		602	# keepalive/persistent connection cache
		603
		604	# fetch a connection from the keepalive cache
		605	sub ka_fetch($) {
		606	my $ka_key = shift;
		607
		608	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		609	delete $KA_CACHE{$ka_key}
		610	unless @{ $KA_CACHE{$ka_key} };
		611
		612	$hdl
		613	}
		614
		615	sub ka_store($$) {
		616	my ($ka_key, $hdl) = @_;
		617
		618	my $kaa = $KA_CACHE{$ka_key} ||= [];
		619
		620	my $destroy = sub {
		621	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		622
		623	$hdl->destroy;
		624
		625	@ka
		626	? $KA_CACHE{$ka_key} = \@ka
		627	: delete $KA_CACHE{$ka_key};
		628	};
		629
		630	# on error etc., destroy
		631	$hdl->on_error ($destroy);
		632	$hdl->on_eof ($destroy);
		633	$hdl->on_read ($destroy);
		634	$hdl->timeout ($PERSISTENT_TIMEOUT);
		635
		636	push @$kaa, $hdl;
		637	shift @$kaa while @$kaa > $MAX_PER_HOST;
		638	}
		639
		640	#############################################################################
		641	# utilities
		642
		643	# continue to parse $_ for headers and place them into the arg
		644	sub _parse_hdr() {
		645	my %hdr;
		646
		647	# things seen, not parsed:
		648	# p3pP="NON CUR OTPi OUR NOR UNI"
		649
		650	$hdr{lc $1} .= ",$2"
		651	while /\G
		652	([^:\000-\037]*):
		653	[\011\040]*
		654	((?: [^\012]+ | \012[\011\040] )*)
		655	\012
		656	/gxc;
		657
		658	/\G$/
		659	or return;
		660
		661	# remove the "," prefix we added to all headers above
		662	substr $_, 0, 1, ""
		663	for values %hdr;
		664
		665	\%hdr
		666	}
		667
		668	#############################################################################
		669	# http_get
		670
		671	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		672
		673	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		674	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		675
		676	# maybe it should just become a normal object :/
		677
		678	sub _destroy_state(\%) {
		679	my ($state) = @_;
		680
		681	$state->{handle}->destroy if $state->{handle};
		682	%$state = ();
		683	}
		684
		685	sub _error(\%$$) {
		686	my ($state, $cb, $hdr) = @_;
		687
		688	&_destroy_state ($state);
		689
		690	$cb->(undef, $hdr);
		691	()
		692	}
		693
219	sub http_request($$@) {	694	sub http_request($$@) {
220	my $cb = pop;	695	my $cb = pop;
221	my ($method, $url, %arg) = @_;	696	my ($method, $url, %arg) = @_;
222		697
223	my %hdr;	698	my %hdr;
		699
		700	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		701	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
224		702
225	$method = uc $method;	703	$method = uc $method;
226		704
227	if (my $hdr = $arg{headers}) {	705	if (my $hdr = $arg{headers}) {
228	while (my ($k, $v) = each %$hdr) {	706	while (my ($k, $v) = each %$hdr) {
229	$hdr{lc $k} = $v;	707	$hdr{lc $k} = $v;
230	}	708	}
231	}	709	}
232		710
		711	# pseudo headers for all subsequent responses
		712	my @pseudo = (URL => $url);
		713	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		714
233	my $recurse = exists $arg{recurse} ? $arg{recurse} : $MAX_RECURSE;	715	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
234		716
235	return $cb->(undef, { Status => 599, Reason => "recursion limit reached" })	717	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
236	if $recurse < 0;	718	if $recurse < 0;
237		719
238	my $proxy = $arg{proxy} || $PROXY;	720	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
239	my $timeout = $arg{timeout} || $TIMEOUT;	721	my $timeout = $arg{timeout} || $TIMEOUT;
240		722
241	$hdr{"user-agent"} ||= $USERAGENT;
242
243	my ($scheme, $authority, $upath, $query, $fragment) =	723	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
244	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;	724	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
245		725
246	$scheme = lc $scheme;	726	$uscheme = lc $uscheme;
247		727
248	my $uport = $scheme eq "http" ? 80	728	my $uport = $uscheme eq "http" ? 80
249	: $scheme eq "https" ? 443	729	: $uscheme eq "https" ? 443
250	: return $cb->(undef, { Status => 599, Reason => "only http and https URL schemes supported" });	730	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
251		731
252	$hdr{referer} ||= "$scheme://$authority$upath"; # leave out fragment and query string, just a heuristic
253
254	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	732	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
255	or return $cb->(undef, { Status => 599, Reason => "unparsable URL" });	733	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
256		734
257	my $uhost = $1;	735	my $uhost = lc $1;
258	$uport = $2 if defined $2;	736	$uport = $2 if defined $2;
259		737
		738	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		739	unless exists $hdr{host};
		740
260	$uhost =~ s/^\[(.*)\]$/$1/;	741	$uhost =~ s/^\[(.*)\]$/$1/;
261	$upath .= "?$query" if length $query;	742	$upath .= $query if length $query;
262		743
263	$upath =~ s%^/?%/%;	744	$upath =~ s%^/?%/%;
264		745
265	# cookie processing	746	# cookie processing
266	if (my $jar = $arg{cookie_jar}) {	747	if (my $jar = $arg{cookie_jar}) {
267	%$jar = () if $jar->{version} < 1;	748	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
268		749
		750	$hdr{cookie} = join "; ", @$cookies
269	my @cookie;	751	if @$cookies;
270		752	}
271	while (my ($chost, $v) = each %$jar) {	753
272	next unless $chost eq substr $uhost, -length $chost;	754	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
273	next unless $chost =~ /^\./;	755
274		756	if ($proxy) {
275	while (my ($cpath, $v) = each %$v) {	757	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
276	next unless $cpath eq substr $upath, 0, length $cpath;	758
277		759	$rscheme = "http" unless defined $rscheme;
278	while (my ($k, $v) = each %$v) {	760
279	next if $scheme ne "https" && exists $v->{secure};	761	# don't support https requests over https-proxy transport,
280	push @cookie, "$k=$v->{value}";	762	# can't be done with tls as spec'ed, unless you double-encrypt.
		763	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		764
		765	$rhost = lc $rhost;
		766	$rscheme = lc $rscheme;
		767	} else {
		768	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
		769	}
		770
		771	# leave out fragment and query string, just a heuristic
		772	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		773	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
		774
		775	$hdr{"content-length"} = length $arg{body}
		776	if length $arg{body} || $method ne "GET";
		777
		778	my $idempotent = $method =~ /^(?:GET|HEAD|PUT|DELETE|OPTIONS|TRACE)$/;
		779
		780	# default value for keepalive is true iff the request is for an idempotent method
		781	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		782	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		783	my $was_persistent; # true if this is actually a recycled connection
		784
		785	# the key to use in the keepalive cache
		786	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		787
		788	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive " : "" : "close ") . "Te"; #1.1
		789	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		790
		791	my %state = (connect_guard => 1);
		792
		793	my $ae_error = 595; # connecting
		794
		795	# handle actual, non-tunneled, request
		796	my $handle_actual_request = sub {
		797	$ae_error = 596; # request phase
		798
		799	my $hdl = $state{handle};
		800
		801	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		802
		803	# send request
		804	$hdl->push_write (
		805	"$method $rpath HTTP/1.1\015\012"
		806	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		807	. "\015\012"
		808	. (delete $arg{body})
		809	);
		810
		811	# return if error occured during push_write()
		812	return unless %state;
		813
		814	# reduce memory usage, save a kitten, also re-use it for the response headers.
		815	%hdr = ();
		816
		817	# status line and headers
		818	$state{read_response} = sub {
		819	return unless %state;
		820
		821	for ("$_[1]") {
		822	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		823
		824	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		825	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		826
		827	# 100 Continue handling
		828	# should not happen as we don't send expect: 100-continue,
		829	# but we handle it just in case.
		830	# since we send the request body regardless, if we get an error
		831	# we are out of-sync, which we currently do NOT handle correctly.
		832	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		833	if $2 eq 100;
		834
		835	push @pseudo,
		836	HTTPVersion => $1,
		837	Status => $2,
		838	Reason => $3,
		839	;
		840
		841	my $hdr = _parse_hdr
		842	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		843
		844	%hdr = (%$hdr, @pseudo);
		845	}
		846
		847	# redirect handling
		848	# microsoft and other shitheads don't give a shit for following standards,
		849	# try to support some common forms of broken Location headers.
		850	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
		851	$hdr{location} =~ s/^\.\/+//;
		852
		853	my $url = "$rscheme://$uhost:$uport";
		854
		855	unless ($hdr{location} =~ s/^\///) {
		856	$url .= $upath;
		857	$url =~ s/\/[^\/]*$//;
		858	}
		859
		860	$hdr{location} = "$url/$hdr{location}";
		861	}
		862
		863	my $redirect;
		864
		865	if ($recurse) {
		866	my $status = $hdr{Status};
		867
		868	# industry standard is to redirect POST as GET for
		869	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		870	# also, the UA should ask the user for 301 and 307 and POST,
		871	# industry standard seems to be to simply follow.
		872	# we go with the industry standard.
		873	if ($status == 301 or $status == 302 or $status == 303) {
		874	# HTTP/1.1 is unclear on how to mutate the method
		875	$method = "GET" unless $method eq "HEAD";
		876	$redirect = 1;
		877	} elsif ($status == 307) {
		878	$redirect = 1;
281	}	879	}
282	}	880	}
		881
		882	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		883	if ($state{handle}) {
		884	# handle keepalive
		885	if (
		886	$persistent
		887	&& $_[3]
		888	&& ($hdr{HTTPVersion} < 1.1
		889	? $hdr{connection} =~ /\bkeep-?alive\b/i
		890	: $hdr{connection} !~ /\bclose\b/i)
		891	) {
		892	ka_store $ka_key, delete $state{handle};
		893	} else {
		894	# no keepalive, destroy the handle
		895	$state{handle}->destroy;
		896	}
		897	}
		898
		899	%state = ();
		900
		901	if (defined $_[1]) {
		902	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		903	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		904	}
		905
		906	# set-cookie processing
		907	if ($arg{cookie_jar}) {
		908	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		909	}
		910
		911	if ($redirect && exists $hdr{location}) {
		912	# we ignore any errors, as it is very common to receive
		913	# Content-Length != 0 but no actual body
		914	# we also access %hdr, as $_[1] might be an erro
		915	$state{recurse} =
		916	http_request (
		917	$method => $hdr{location},
		918	%arg,
		919	recurse => $recurse - 1,
		920	Redirect => [$_[0], \%hdr],
		921	sub {
		922	%state = ();
		923	&$cb
		924	},
		925	);
		926	} else {
		927	$cb->($_[0], \%hdr);
		928	}
		929	};
		930
		931	$ae_error = 597; # body phase
		932
		933	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		934
		935	my $len = $chunked ? undef : $hdr{"content-length"};
		936
		937	# body handling, many different code paths
		938	# - no body expected
		939	# - want_body_handle
		940	# - te chunked
		941	# - 2x length known (with or without on_body)
		942	# - 2x length not known (with or without on_body)
		943	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		944	$finish->(undef, 598 => "Request cancelled by on_header");
		945	} elsif (
		946	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		947	or $method eq "HEAD"
		948	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		949	) {
		950	# no body
		951	$finish->("", undef, undef, 1);
		952
		953	} elsif (!$redirect && $arg{want_body_handle}) {
		954	$_[0]->on_eof (undef);
		955	$_[0]->on_error (undef);
		956	$_[0]->on_read (undef);
		957
		958	$finish->(delete $state{handle});
		959
		960	} elsif ($chunked) {
		961	my $cl = 0;
		962	my $body = "";
		963	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		964
		965	$state{read_chunk} = sub {
		966	$_[1] =~ /^([0-9a-fA-F]+)/
		967	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		968
		969	my $len = hex $1;
		970
		971	if ($len) {
		972	$cl += $len;
		973
		974	$_[0]->push_read (chunk => $len, sub {
		975	$on_body->($_[1], \%hdr)
		976	or return $finish->(undef, 598 => "Request cancelled by on_body");
		977
		978	$_[0]->push_read (line => sub {
		979	length $_[1]
		980	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		981	$_[0]->push_read (line => $state{read_chunk});
		982	});
		983	});
		984	} else {
		985	$hdr{"content-length"} ||= $cl;
		986
		987	$_[0]->push_read (line => $qr_nlnl, sub {
		988	if (length $_[1]) {
		989	for ("$_[1]") {
		990	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		991
		992	my $hdr = _parse_hdr
		993	or return $finish->(undef, $ae_error => "Garbled response trailers");
		994
		995	%hdr = (%hdr, %$hdr);
		996	}
		997	}
		998
		999	$finish->($body, undef, undef, 1);
		1000	});
		1001	}
		1002	};
		1003
		1004	$_[0]->push_read (line => $state{read_chunk});
		1005
		1006	} elsif ($arg{on_body}) {
		1007	if (defined $len) {
		1008	$_[0]->on_read (sub {
		1009	$len -= length $_[0]{rbuf};
		1010
		1011	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1012	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1013
		1014	$len > 0
		1015	or $finish->("", undef, undef, 1);
		1016	});
		1017	} else {
		1018	$_[0]->on_eof (sub {
		1019	$finish->("");
		1020	});
		1021	$_[0]->on_read (sub {
		1022	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1023	or $finish->(undef, 598 => "Request cancelled by on_body");
		1024	});
		1025	}
		1026	} else {
		1027	$_[0]->on_eof (undef);
		1028
		1029	if (defined $len) {
		1030	$_[0]->on_read (sub {
		1031	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1032	if $len <= length $_[0]{rbuf};
		1033	});
		1034	} else {
		1035	$_[0]->on_error (sub {
		1036	($! == Errno::EPIPE || !$!)
		1037	? $finish->(delete $_[0]{rbuf})
		1038	: $finish->(undef, $ae_error => $_[2]);
		1039	});
		1040	$_[0]->on_read (sub { });
		1041	}
		1042	}
		1043	};
		1044
		1045	# if keepalive is enabled, then the server closing the connection
		1046	# before a response can happen legally - we retry on idempotent methods.
		1047	if ($was_persistent && $idempotent) {
		1048	my $old_eof = $hdl->{on_eof};
		1049	$hdl->{on_eof} = sub {
		1050	_destroy_state %state;
		1051
		1052	%state = ();
		1053	$state{recurse} =
		1054	http_request (
		1055	$method => $url,
		1056	%arg,
		1057	keepalive => 0,
		1058	sub {
		1059	%state = ();
		1060	&$cb
		1061	}
		1062	);
		1063	};
		1064	$hdl->on_read (sub {
		1065	return unless %state;
		1066
		1067	# as soon as we receive something, a connection close
		1068	# once more becomes a hard error
		1069	$hdl->{on_eof} = $old_eof;
		1070	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1071	});
		1072	} else {
		1073	$hdl->push_read (line => $qr_nlnl, $state{read_response});
283	}	1074	}
284
285	$hdr{cookie} = join "; ", @cookie
286	if @cookie;
287	}	1075	};
288		1076
289	my ($rhost, $rport, $rpath); # request host, port, path	1077	my $prepare_handle = sub {
		1078	my ($hdl) = $state{handle};
290		1079
291	if ($proxy) {	1080	$hdl->on_error (sub {
292	($rhost, $rport, $scheme) = @$proxy;	1081	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
293	$rpath = $url;	1082	});
294	} else {	1083	$hdl->on_eof (sub {
295	($rhost, $rport, $rpath) = ($uhost, $uport, $upath);	1084	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
296	$hdr{host} = $uhost;	1085	});
		1086	$hdl->timeout_reset;
		1087	$hdl->timeout ($timeout);
297	}	1088	};
298		1089
299	$hdr{"content-length"} = length $arg{body};	1090	# connected to proxy (or origin server)
		1091	my $connect_cb = sub {
		1092	my $fh = shift
		1093	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
300		1094
301	my %state = (connect_guard => 1);	1095	return unless delete $state{connect_guard};
		1096
		1097	# get handle
		1098	$state{handle} = new AnyEvent::Handle
		1099	%{ $arg{handle_params} },
		1100	fh => $fh,
		1101	peername => $uhost,
		1102	tls_ctx => $arg{tls_ctx},
		1103	;
		1104
		1105	$prepare_handle->();
		1106
		1107	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1108
		1109	# now handle proxy-CONNECT method
		1110	if ($proxy && $uscheme eq "https") {
		1111	# oh dear, we have to wrap it into a connect request
		1112
		1113	# maybe re-use $uauthority with patched port?
		1114	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1115	$state{handle}->push_read (line => $qr_nlnl, sub {
		1116	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1117	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1118
		1119	if ($2 == 200) {
		1120	$rpath = $upath;
		1121	$handle_actual_request->();
		1122	} else {
		1123	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1124	}
		1125	});
		1126	} else {
		1127	$handle_actual_request->();
		1128	}
		1129	};
302		1130
303	_get_slot $uhost, sub {	1131	_get_slot $uhost, sub {
304	$state{slot_guard} = shift;	1132	$state{slot_guard} = shift;
305		1133
306	return unless $state{connect_guard};	1134	return unless $state{connect_guard};
307		1135
308	$state{connect_guard} = AnyEvent::Socket::tcp_connect $rhost, $rport, sub {	1136	# try to use an existing keepalive connection, but only if we, ourselves, plan
309	$state{fh} = shift	1137	# on a keepalive request (in theory, this should be a separate config option).
310	or return $cb->(undef, { Status => 599, Reason => "$!" });	1138	if ($persistent && $KA_CACHE{$ka_key}) {
		1139	$was_persistent = 1;
311		1140
312	delete $state{connect_guard}; # reduce memory usage, save a tree	1141	$state{handle} = ka_fetch $ka_key;
		1142	$state{handle}->destroyed
		1143	and die "got a destructed handle. pah\n";#d#
		1144	$prepare_handle->();
		1145	$state{handle}->destroyed
		1146	and die "got a destructed handle. pa2\n";#d#
		1147	$handle_actual_request->();
		1148	$state{handle}->destroyed
		1149	and die "got a destructed handle. pa3\n";#d#
313		1150
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 {	1151	} else {
326	delete $hdr{connection};	1152	my $tcp_connect = $arg{tcp_connect}
327	}	1153	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
328		1154
329	# (re-)configure handle	1155	$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	$timeout
446	};	1156	}
447	};	1157	};
448		1158
449	defined wantarray && AnyEvent::Util::guard { %state = () }	1159	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
450	}	1160	}
451		1161
452	sub http_get($@) {	1162	sub http_get($@) {
453	unshift @_, "GET";	1163	unshift @_, "GET";
454	&http_request	1164	&http_request
…		…
458	unshift @_, "HEAD";	1168	unshift @_, "HEAD";
459	&http_request	1169	&http_request
460	}	1170	}
461		1171
462	sub http_post($$@) {	1172	sub http_post($$@) {
		1173	my $url = shift;
463	unshift @_, "POST", "body";	1174	unshift @_, "POST", $url, "body";
464	&http_request	1175	&http_request
465	}	1176	}
466		1177
467	=back	1178	=back
468		1179
		1180	=head2 DNS CACHING
		1181
		1182	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1183	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1184	hostnames. The latter is a simple stub resolver and does no caching
		1185	on its own. If you want DNS caching, you currently have to provide
		1186	your own default resolver (by storing a suitable resolver object in
		1187	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1188
469	=head2 GLOBAL FUNCTIONS AND VARIABLES	1189	=head2 GLOBAL FUNCTIONS AND VARIABLES
470		1190
471	=over 4	1191	=over 4
472		1192
473	=item AnyEvent::HTTP::set_proxy "proxy-url"	1193	=item AnyEvent::HTTP::set_proxy "proxy-url"
474		1194
475	Sets the default proxy server to use. The proxy-url must begin with a	1195	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:...>).	1196	string of the form C<http://host:port>, croaks otherwise.
		1197
		1198	To clear an already-set proxy, use C<undef>.
		1199
		1200	When AnyEvent::HTTP is laoded for the first time it will query the
		1201	default proxy from the operating system, currently by looking at
		1202	C<$ENV{http_proxy>}.
		1203
		1204	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1205
		1206	Remove all cookies from the cookie jar that have been expired. If
		1207	C<$session_end> is given and true, then additionally remove all session
		1208	cookies.
		1209
		1210	You should call this function (with a true C<$session_end>) before you
		1211	save cookies to disk, and you should call this function after loading them
		1212	again. If you have a long-running program you can additonally call this
		1213	function from time to time.
		1214
		1215	A cookie jar is initially an empty hash-reference that is managed by this
		1216	module. It's format is subject to change, but currently it is like this:
		1217
		1218	The key C<version> has to contain C<1>, otherwise the hash gets
		1219	emptied. All other keys are hostnames or IP addresses pointing to
		1220	hash-references. The key for these inner hash references is the
		1221	server path for which this cookie is meant, and the values are again
		1222	hash-references. The keys of those hash-references is the cookie name, and
		1223	the value, you guessed it, is another hash-reference, this time with the
		1224	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1225	which have been replaced by a C<_expires> key that contains the cookie
		1226	expiry timestamp.
		1227
		1228	Here is an example of a cookie jar with a single cookie, so you have a
		1229	chance of understanding the above paragraph:
		1230
		1231	{
		1232	version => 1,
		1233	"10.0.0.1" => {
		1234	"/" => {
		1235	"mythweb_id" => {
		1236	_expires => 1293917923,
		1237	value => "ooRung9dThee3ooyXooM1Ohm",
		1238	},
		1239	},
		1240	},
		1241	}
		1242
		1243	=item $date = AnyEvent::HTTP::format_date $timestamp
		1244
		1245	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1246	Date (RFC 2616).
		1247
		1248	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1249
		1250	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1251	bunch of minor variations of those, and returns the corresponding POSIX
		1252	timestamp, or C<undef> if the date cannot be parsed.
477		1253
478	=item $AnyEvent::HTTP::MAX_RECURSE	1254	=item $AnyEvent::HTTP::MAX_RECURSE
479		1255
480	The default value for the C<recurse> request parameter (default: C<10>).	1256	The default value for the C<recurse> request parameter (default: C<10>).
481		1257
		1258	=item $AnyEvent::HTTP::TIMEOUT
		1259
		1260	The default timeout for conenction operations (default: C<300>).
		1261
482	=item $AnyEvent::HTTP::USERAGENT	1262	=item $AnyEvent::HTTP::USERAGENT
483		1263
484	The default value for the C<User-Agent> header (the default is	1264	The default value for the C<User-Agent> header (the default is
485	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1265	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
486		1266
487	=item $AnyEvent::HTTP::MAX_PERSISTENT	1267	=item $AnyEvent::HTTP::MAX_PER_HOST
488		1268
489	The maximum number of persistent connections to keep open (default: 8).	1269	The maximum number of concurrent connections to the same host (identified
		1270	by the hostname). If the limit is exceeded, then the additional requests
		1271	are queued until previous connections are closed. Both persistent and
		1272	non-persistent connections are counted in this limit.
490		1273
491	Not implemented currently.	1274	The default value for this is C<4>, and it is highly advisable to not
		1275	increase it much.
		1276
		1277	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1278	connections, older browsers used 2, newers (such as firefox 3) typically
		1279	use 6, and Opera uses 8 because like, they have the fastest browser and
		1280	give a shit for everybody else on the planet.
492		1281
493	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1282	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
494		1283
495	The maximum time to cache a persistent connection, in seconds (default: 2).	1284	The time after which idle persistent conenctions get closed by
496		1285	AnyEvent::HTTP (default: C<3>).
497	Not implemented currently.
498		1286
499	=item $AnyEvent::HTTP::ACTIVE	1287	=item $AnyEvent::HTTP::ACTIVE
500		1288
501	The number of active connections. This is not the number of currently	1289	The number of active connections. This is not the number of currently
502	running requests, but the number of currently open and non-idle TCP	1290	running requests, but the number of currently open and non-idle TCP
503	connections. This number of can be useful for load-leveling.	1291	connections. This number can be useful for load-leveling.
504		1292
505	=back	1293	=back
506		1294
507	=cut	1295	=cut
508		1296
		1297	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1298	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1299
		1300	sub format_date($) {
		1301	my ($time) = @_;
		1302
		1303	# RFC 822/1123 format
		1304	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1305
		1306	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1307	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1308	$H, $M, $S;
		1309	}
		1310
		1311	sub parse_date($) {
		1312	my ($date) = @_;
		1313
		1314	my ($d, $m, $y, $H, $M, $S);
		1315
		1316	if ($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$/) {
		1317	# RFC 822/1123, required by RFC 2616 (with " ")
		1318	# cookie dates (with "-")
		1319
		1320	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1321
		1322	} elsif ($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$/) {
		1323	# RFC 850
		1324	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1325
		1326	} elsif ($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])$/) {
		1327	# ISO C's asctime
		1328	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1329	}
		1330	# other formats fail in the loop below
		1331
		1332	for (0..11) {
		1333	if ($m eq $month[$_]) {
		1334	require Time::Local;
		1335	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);
		1336	}
		1337	}
		1338
		1339	undef
		1340	}
		1341
509	sub set_proxy($) {	1342	sub set_proxy($) {
510	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1343	if (length $_[0]) {
		1344	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1345	or Carp::croak "$_[0]: invalid proxy URL";
		1346	$PROXY = [$2, $3 || 3128, $1]
		1347	} else {
		1348	undef $PROXY;
		1349	}
511	}	1350	}
512		1351
513	# initialise proxy from environment	1352	# initialise proxy from environment
		1353	eval {
514	set_proxy $ENV{http_proxy};	1354	set_proxy $ENV{http_proxy};
		1355	};
		1356
		1357	=head2 SHOWCASE
		1358
		1359	This section contaisn some more elaborate "real-world" examples or code
		1360	snippets.
		1361
		1362	=head2 HTTP/1.1 FILE DOWNLOAD
		1363
		1364	Downloading files with HTTP can be quite tricky, especially when something
		1365	goes wrong and you want to resume.
		1366
		1367	Here is a function that initiates and resumes a download. It uses the
		1368	last modified time to check for file content changes, and works with many
		1369	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1370	on older servers.
		1371
		1372	It calls the completion callback with either C<undef>, which means a
		1373	nonretryable error occured, C<0> when the download was partial and should
		1374	be retried, and C<1> if it was successful.
		1375
		1376	use AnyEvent::HTTP;
		1377
		1378	sub download($$$) {
		1379	my ($url, $file, $cb) = @_;
		1380
		1381	open my $fh, "+<", $file
		1382	or die "$file: $!";
		1383
		1384	my %hdr;
		1385	my $ofs = 0;
		1386
		1387	warn stat $fh;
		1388	warn -s _;
		1389	if (stat $fh and -s _) {
		1390	$ofs = -s _;
		1391	warn "-s is ", $ofs;#d#
		1392	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1393	$hdr{"range"} = "bytes=$ofs-";
		1394	}
		1395
		1396	http_get $url,
		1397	headers => \%hdr,
		1398	on_header => sub {
		1399	my ($hdr) = @_;
		1400
		1401	if ($hdr->{Status} == 200 && $ofs) {
		1402	# resume failed
		1403	truncate $fh, $ofs = 0;
		1404	}
		1405
		1406	sysseek $fh, $ofs, 0;
		1407
		1408	1
		1409	},
		1410	on_body => sub {
		1411	my ($data, $hdr) = @_;
		1412
		1413	if ($hdr->{Status} =~ /^2/) {
		1414	length $data == syswrite $fh, $data
		1415	or return; # abort on write errors
		1416	}
		1417
		1418	1
		1419	},
		1420	sub {
		1421	my (undef, $hdr) = @_;
		1422
		1423	my $status = $hdr->{Status};
		1424
		1425	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1426	utime $fh, $time, $time;
		1427	}
		1428
		1429	if ($status == 200 || $status == 206 || $status == 416) {
		1430	# download ok || resume ok || file already fully downloaded
		1431	$cb->(1, $hdr);
		1432
		1433	} elsif ($status == 412) {
		1434	# file has changed while resuming, delete and retry
		1435	unlink $file;
		1436	$cb->(0, $hdr);
		1437
		1438	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1439	# retry later
		1440	$cb->(0, $hdr);
		1441
		1442	} else {
		1443	$cb->(undef, $hdr);
		1444	}
		1445	}
		1446	;
		1447	}
		1448
		1449	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1450	if ($_[0]) {
		1451	print "OK!\n";
		1452	} elsif (defined $_[0]) {
		1453	print "please retry later\n";
		1454	} else {
		1455	print "ERROR\n";
		1456	}
		1457	};
		1458
		1459	=head3 SOCKS PROXIES
		1460
		1461	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1462	compile your perl to support socks, or use an external program such as
		1463	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1464	transparently.
		1465
		1466	Alternatively, for AnyEvent::HTTP only, you can use your own
		1467	C<tcp_connect> function that does the proxy handshake - here is an example
		1468	that works with socks4a proxies:
		1469
		1470	use Errno;
		1471	use AnyEvent::Util;
		1472	use AnyEvent::Socket;
		1473	use AnyEvent::Handle;
		1474
		1475	# host, port and username of/for your socks4a proxy
		1476	my $socks_host = "10.0.0.23";
		1477	my $socks_port = 9050;
		1478	my $socks_user = "";
		1479
		1480	sub socks4a_connect {
		1481	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1482
		1483	my $hdl = new AnyEvent::Handle
		1484	connect => [$socks_host, $socks_port],
		1485	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1486	on_error => sub { $connect_cb->() },
		1487	;
		1488
		1489	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1490
		1491	$hdl->push_read (chunk => 8, sub {
		1492	my ($hdl, $chunk) = @_;
		1493	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1494
		1495	if ($status == 0x5a) {
		1496	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1497	} else {
		1498	$! = Errno::ENXIO; $connect_cb->();
		1499	}
		1500	});
		1501
		1502	$hdl
		1503	}
		1504
		1505	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1506	possibly after switching off other proxy types:
		1507
		1508	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1509
		1510	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1511	my ($data, $headers) = @_;
		1512	...
		1513	};
515		1514
516	=head1 SEE ALSO	1515	=head1 SEE ALSO
517		1516
518	L<AnyEvent>.	1517	L<AnyEvent>.
519		1518
520	=head1 AUTHOR	1519	=head1 AUTHOR
521		1520
522	Marc Lehmann <schmorp@schmorp.de>	1521	Marc Lehmann <schmorp@schmorp.de>
523	http://home.schmorp.de/	1522	http://home.schmorp.de/
		1523
		1524	With many thanks to Дмитрий Шалашов, who provided countless
		1525	testcases and bugreports.
524		1526
525	=cut	1527	=cut
526		1528
527	1	1529	1
528		1530

Diff Legend

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