ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.9 by root, Wed Jun 4 13:51:53 2008 UTC vs.
Revision 1.121 by root, Tue Apr 7 01:18:20 2015 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
		18	This module implements a simple, stateless and non-blocking HTTP
		19	client. It supports GET, POST and other request methods, cookies and more,
		20	all on a very low level. It can follow redirects, supports proxies, and
		21	automatically limits the number of connections to the values specified in
		22	the RFC.
		23
		24	It should generally be a "good client" that is enough for most HTTP
		25	tasks. Simple tasks should be simple, but complex tasks should still be
		26	possible as the user retains control over request and response headers.
		27
		28	The caller is responsible for authentication management, cookies (if
		29	the simplistic implementation in this module doesn't suffice), referer
		30	and other high-level protocol details for which this module offers only
		31	limited support.
		32
14	=head2 METHODS	33	=head2 METHODS
15		34
16	=over 4	35	=over 4
17		36
18	=cut	37	=cut
19		38
20	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
21		40
22	use strict;	41	use common::sense;
23	no warnings;
24		42
25	use Carp;	43	use Errno ();
26		44
27	use AnyEvent ();	45	use AnyEvent 5.0 ();
28	use AnyEvent::Util ();	46	use AnyEvent::Util ();
29	use AnyEvent::Socket ();
30	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
31		48
32	use base Exporter::;	49	use base Exporter::;
33		50
34	our $VERSION = '1.0';	51	our $VERSION = 2.21;
35		52
36	our @EXPORT = qw(http_get http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
37		54
38	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)";
39	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
40	our $MAX_PERSISTENT = 8;
41	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
42	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
43		59	our $MAX_PER_HOST = 4; # changing this is evil
44	# changing these is evil
45	our $MAX_PERSISTENT_PER_HOST = 2;
46	our $MAX_PER_HOST = 4; # not respected yet :(
47		60
48	our $PROXY;	61	our $PROXY;
		62	our $ACTIVE = 0;
49		63
50	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
		65	my %CO_SLOT; # number of open connections, and wait queue, per host
51		66
52	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
53		68
54	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
55	additional parameters.	70	additional parameters and the return value.
56		71
57	=item http_head $url, key => value..., $cb->($data, $headers)	72	=item http_head $url, key => value..., $cb->($data, $headers)
58		73
59	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
60	additional parameters.	75	on additional parameters and the return value.
61		76
62	=item http_post $url, $body, key => value..., $cb->($data, $headers)	77	=item http_post $url, $body, key => value..., $cb->($data, $headers)
63		78
64	Executes an HTTP-POST request with a request body of C<$bod>. See the	79	Executes an HTTP-POST request with a request body of C<$body>. See the
65	http_request function for details on additional parameters.	80	http_request function for details on additional parameters and the return
		81	value.
66		82
67	=item http_request $method => $url, key => value..., $cb->($data, $headers)	83	=item http_request $method => $url, key => value..., $cb->($data, $headers)
68		84
69	Executes a HTTP request of type C<$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
70	must be an absolute http or https URL.	86	must be an absolute http or https URL.
71		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
72	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
73	(or C<undef> if it wasn't available due to errors), and a hash-ref with	94	(or C<undef> if an error occurred), and a hash-ref with response headers
74	response headers as second argument.	95	(and trailers) as second argument.
75		96
76	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
77	headers, the three "pseudo-headers" C<HTTPVersion>, C<Status> and	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
78	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
79	name.	100	three parts of the HTTP Status-Line of the same name. If an error occurs
		101	during the body phase of a request, then the original C<Status> and
		102	C<Reason> values from the header are available as C<OrigStatus> and
		103	C<OrigReason>.
		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	$headers->{Redirect}[1]{Redirect} >> will then contain the original
		116	response, and so on.
		117
		118	If the server sends a header multiple times, then their contents will be
		119	joined together with a comma (C<,>), as per the HTTP spec.
80		120
81	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
82	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<599>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
83	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection establishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
84		139
85	A typical callback might look like this:	140	A typical callback might look like this:
86		141
87	sub {	142	sub {
88	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
99		154
100	=over 4	155	=over 4
101		156
102	=item recurse => $count (default: $MAX_RECURSE)	157	=item recurse => $count (default: $MAX_RECURSE)
103		158
104	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
105	retries and so on, and how often to do so.	160	other retries and so on, and how often to do so.
		161
		162	Only redirects to http and https URLs are supported. While most common
		163	redirection forms are handled entirely within this module, some require
		164	the use of the optional L<URI> module. If it is required but missing, then
		165	the request will fail with an error.
106		166
107	=item headers => hashref	167	=item headers => hashref
108		168
109	The request headers to use.	169	The request headers to use. Currently, C<http_request> may provide its own
		170	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
		171	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
		172	(this can be suppressed by using C<undef> for these headers in which case
		173	they won't be sent at all).
		174
		175	You really should provide your own C<User-Agent:> header value that is
		176	appropriate for your program - I wouldn't be surprised if the default
		177	AnyEvent string gets blocked by webservers sooner or later.
		178
		179	Also, make sure that your headers names and values do not contain any
		180	embedded newlines.
110		181
111	=item timeout => $seconds	182	=item timeout => $seconds
112		183
113	The time-out to use for various stages - each connect attempt will reset	184	The time-out to use for various stages - each connect attempt will reset
114	the timeout, as will read or write activity. Default timeout is 5 minutes.	185	the timeout, as will read or write activity, i.e. this is not an overall
		186	timeout.
		187
		188	Default timeout is 5 minutes.
115		189
116	=item proxy => [$host, $port[, $scheme]] or undef	190	=item proxy => [$host, $port[, $scheme]] or undef
117		191
118	Use the given http proxy for all requests. If not specified, then the	192	Use the given http proxy for all requests, or no proxy if C<undef> is
119	default proxy (as specified by C<$ENV{http_proxy}>) is used.	193	used.
120		194
121	C<$scheme> must be either missing or C<http> for HTTP, or C<https> for	195	C<$scheme> must be either missing or must be C<http> for HTTP.
122	HTTPS.	196
		197	If not specified, then the default proxy is used (see
		198	C<AnyEvent::HTTP::set_proxy>).
123		199
124	=item body => $string	200	=item body => $string
125		201
126	The request body, usually empty. Will be-sent as-is (future versions of	202	The request body, usually empty. Will be sent as-is (future versions of
127	this module might offer more options).	203	this module might offer more options).
128		204
		205	=item cookie_jar => $hash_ref
		206
		207	Passing this parameter enables (simplified) cookie-processing, loosely
		208	based on the original netscape specification.
		209
		210	The C<$hash_ref> must be an (initially empty) hash reference which
		211	will get updated automatically. It is possible to save the cookie jar
		212	to persistent storage with something like JSON or Storable - see the
		213	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		214	expired or session-only cookies, and also for documentation on the format
		215	of the cookie jar.
		216
		217	Note that this cookie implementation is not meant to be complete. If
		218	you want complete cookie management you have to do that on your
		219	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
		220	working. Cookies are a privacy disaster, do not use them unless required
		221	to.
		222
		223	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		224	headers will be set and handled by this module, otherwise they will be
		225	left untouched.
		226
		227	=item tls_ctx => $scheme | $tls_ctx
		228
		229	Specifies the AnyEvent::TLS context to be used for https connections. This
		230	parameter follows the same rules as the C<tls_ctx> parameter to
		231	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		232	C<high> can be specified, which give you a predefined low-security (no
		233	verification, highest compatibility) and high-security (CA and common-name
		234	verification) TLS context.
		235
		236	The default for this option is C<low>, which could be interpreted as "give
		237	me the page, no matter what".
		238
		239	See also the C<sessionid> parameter.
		240
		241	=item session => $string
		242
		243	The module might reuse connections to the same host internally. Sometimes
		244	(e.g. when using TLS), you do not want to reuse connections from other
		245	sessions. This can be achieved by setting this parameter to some unique
		246	ID (such as the address of an object storing your state data, or the TLS
		247	context) - only connections using the same unique ID will be reused.
		248
		249	=item on_prepare => $callback->($fh)
		250
		251	In rare cases you need to "tune" the socket before it is used to
		252	connect (for example, to bind it on a given IP address). This parameter
		253	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		254	and behaves exactly the same way (e.g. it has to provide a
		255	timeout). See the description for the C<$prepare_cb> argument of
		256	C<AnyEvent::Socket::tcp_connect> for details.
		257
		258	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		259
		260	In even rarer cases you want total control over how AnyEvent::HTTP
		261	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		262	to do this, but you can provide your own C<tcp_connect> function -
		263	obviously, it has to follow the same calling conventions, except that it
		264	may always return a connection guard object.
		265
		266	There are probably lots of weird uses for this function, starting from
		267	tracing the hosts C<http_request> actually tries to connect, to (inexact
		268	but fast) host => IP address caching or even socks protocol support.
		269
		270	=item on_header => $callback->($headers)
		271
		272	When specified, this callback will be called with the header hash as soon
		273	as headers have been successfully received from the remote server (not on
		274	locally-generated errors).
		275
		276	It has to return either true (in which case AnyEvent::HTTP will continue),
		277	or false, in which case AnyEvent::HTTP will cancel the download (and call
		278	the finish callback with an error code of C<598>).
		279
		280	This callback is useful, among other things, to quickly reject unwanted
		281	content, which, if it is supposed to be rare, can be faster than first
		282	doing a C<HEAD> request.
		283
		284	The downside is that cancelling the request makes it impossible to re-use
		285	the connection. Also, the C<on_header> callback will not receive any
		286	trailer (headers sent after the response body).
		287
		288	Example: cancel the request unless the content-type is "text/html".
		289
		290	on_header => sub {
		291	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		292	},
		293
		294	=item on_body => $callback->($partial_body, $headers)
		295
		296	When specified, all body data will be passed to this callback instead of
		297	to the completion callback. The completion callback will get the empty
		298	string instead of the body data.
		299
		300	It has to return either true (in which case AnyEvent::HTTP will continue),
		301	or false, in which case AnyEvent::HTTP will cancel the download (and call
		302	the completion callback with an error code of C<598>).
		303
		304	The downside to cancelling the request is that it makes it impossible to
		305	re-use the connection.
		306
		307	This callback is useful when the data is too large to be held in memory
		308	(so the callback writes it to a file) or when only some information should
		309	be extracted, or when the body should be processed incrementally.
		310
		311	It is usually preferred over doing your own body handling via
		312	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		313	only used to create a connection, C<want_body_handle> is the better
		314	alternative, as it allows you to install your own event handler, reducing
		315	resource usage.
		316
		317	=item want_body_handle => $enable
		318
		319	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		320	changes considerably: after parsing the headers, and instead of
		321	downloading the body (if any), the completion callback will be
		322	called. Instead of the C<$body> argument containing the body data, the
		323	callback will receive the L<AnyEvent::Handle> object associated with the
		324	connection. In error cases, C<undef> will be passed. When there is no body
		325	(e.g. status C<304>), the empty string will be passed.
		326
		327	The handle object might or might not be in TLS mode, might be connected
		328	to a proxy, be a persistent connection, use chunked transfer encoding
		329	etc., and configured in unspecified ways. The user is responsible for this
		330	handle (it will not be used by this module anymore).
		331
		332	This is useful with some push-type services, where, after the initial
		333	headers, an interactive protocol is used (typical example would be the
		334	push-style twitter API which starts a JSON/XML stream).
		335
		336	If you think you need this, first have a look at C<on_body>, to see if
		337	that doesn't solve your problem in a better way.
		338
		339	=item persistent => $boolean
		340
		341	Try to create/reuse a persistent connection. When this flag is set
		342	(default: true for idempotent requests, false for all others), then
		343	C<http_request> tries to re-use an existing (previously-created)
		344	persistent connection to the host and, failing that, tries to create a new
		345	one.
		346
		347	Requests failing in certain ways will be automatically retried once, which
		348	is dangerous for non-idempotent requests, which is why it defaults to off
		349	for them. The reason for this is because the bozos who designed HTTP/1.1
		350	made it impossible to distinguish between a fatal error and a normal
		351	connection timeout, so you never know whether there was a problem with
		352	your request or not.
		353
		354	When reusing an existent connection, many parameters (such as TLS context)
		355	will be ignored. See the C<session> parameter for a workaround.
		356
		357	=item keepalive => $boolean
		358
		359	Only used when C<persistent> is also true. This parameter decides whether
		360	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		361	(as opposed to only a HTTP/1.1 persistent connection).
		362
		363	The default is true, except when using a proxy, in which case it defaults
		364	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		365
		366	=item handle_params => { key => value ... }
		367
		368	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		369	constructor that is called - not all requests will create a handle, and
		370	sometimes more than one is created, so this parameter is only good for
		371	setting hints.
		372
		373	Example: set the maximum read size to 4096, to potentially conserve memory
		374	at the cost of speed.
		375
		376	handle_params => {
		377	max_read_size => 4096,
		378	},
		379
129	=back	380	=back
130		381
131	Example: make a simple HTTP GET request for http://www.nethype.de/	382	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		383	the response body.
132		384
133	http_request GET => "http://www.nethype.de/", sub {	385	http_request GET => "http://www.nethype.de/", sub {
134	my ($body, $hdr) = @_;	386	my ($body, $hdr) = @_;
135	print "$body\n";	387	print "$body\n";
136	};	388	};
137		389
138	Example: make a HTTP HEAD request on https://www.google.com/, use a	390	Example: do a HTTP HEAD request on https://www.google.com/, use a
139	timeout of 30 seconds.	391	timeout of 30 seconds.
140		392
141	http_request	393	http_request
142	GET => "https://www.google.com",	394	HEAD => "https://www.google.com",
		395	headers => { "user-agent" => "MySearchClient 1.0" },
143	timeout => 30,	396	timeout => 30,
144	sub {	397	sub {
145	my ($body, $hdr) = @_;	398	my ($body, $hdr) = @_;
146	use Data::Dumper;	399	use Data::Dumper;
147	print Dumper $hdr;	400	print Dumper $hdr;
148	}	401	}
149	;	402	;
150		403
		404	Example: do another simple HTTP GET request, but immediately try to
		405	cancel it.
		406
		407	my $request = http_request GET => "http://www.nethype.de/", sub {
		408	my ($body, $hdr) = @_;
		409	print "$body\n";
		410	};
		411
		412	undef $request;
		413
151	=cut	414	=cut
152		415
		416	#############################################################################
		417	# wait queue/slots
		418
		419	sub _slot_schedule;
		420	sub _slot_schedule($) {
		421	my $host = shift;
		422
		423	while ($CO_SLOT{$host}[0] < $MAX_PER_HOST) {
		424	if (my $cb = shift @{ $CO_SLOT{$host}[1] }) {
		425	# somebody wants that slot
		426	++$CO_SLOT{$host}[0];
		427	++$ACTIVE;
		428
		429	$cb->(AnyEvent::Util::guard {
		430	--$ACTIVE;
		431	--$CO_SLOT{$host}[0];
		432	_slot_schedule $host;
		433	});
		434	} else {
		435	# nobody wants the slot, maybe we can forget about it
		436	delete $CO_SLOT{$host} unless $CO_SLOT{$host}[0];
		437	last;
		438	}
		439	}
		440	}
		441
		442	# wait for a free slot on host, call callback
		443	sub _get_slot($$) {
		444	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
		445
		446	_slot_schedule $_[0];
		447	}
		448
		449	#############################################################################
		450	# cookie handling
		451
		452	# expire cookies
		453	sub cookie_jar_expire($;$) {
		454	my ($jar, $session_end) = @_;
		455
		456	%$jar = () if $jar->{version} != 1;
		457
		458	my $anow = AE::now;
		459
		460	while (my ($chost, $paths) = each %$jar) {
		461	next unless ref $paths;
		462
		463	while (my ($cpath, $cookies) = each %$paths) {
		464	while (my ($cookie, $kv) = each %$cookies) {
		465	if (exists $kv->{_expires}) {
		466	delete $cookies->{$cookie}
		467	if $anow > $kv->{_expires};
		468	} elsif ($session_end) {
		469	delete $cookies->{$cookie};
		470	}
		471	}
		472
		473	delete $paths->{$cpath}
		474	unless %$cookies;
		475	}
		476
		477	delete $jar->{$chost}
		478	unless %$paths;
		479	}
		480	}
		481
		482	# extract cookies from jar
		483	sub cookie_jar_extract($$$$) {
		484	my ($jar, $scheme, $host, $path) = @_;
		485
		486	%$jar = () if $jar->{version} != 1;
		487
		488	my @cookies;
		489
		490	while (my ($chost, $paths) = each %$jar) {
		491	next unless ref $paths;
		492
		493	if ($chost =~ /^\./) {
		494	next unless $chost eq substr $host, -length $chost;
		495	} elsif ($chost =~ /\./) {
		496	next unless $chost eq $host;
		497	} else {
		498	next;
		499	}
		500
		501	while (my ($cpath, $cookies) = each %$paths) {
		502	next unless $cpath eq substr $path, 0, length $cpath;
		503
		504	while (my ($cookie, $kv) = each %$cookies) {
		505	next if $scheme ne "https" && exists $kv->{secure};
		506
		507	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		508	delete $cookies->{$cookie};
		509	next;
		510	}
		511
		512	my $value = $kv->{value};
		513
		514	if ($value =~ /[=;,[:space:]]/) {
		515	$value =~ s/([\\"])/\\$1/g;
		516	$value = "\"$value\"";
		517	}
		518
		519	push @cookies, "$cookie=$value";
		520	}
		521	}
		522	}
		523
		524	\@cookies
		525	}
		526
		527	# parse set_cookie header into jar
		528	sub cookie_jar_set_cookie($$$$) {
		529	my ($jar, $set_cookie, $host, $date) = @_;
		530
		531	my $anow = int AE::now;
		532	my $snow; # server-now
		533
		534	for ($set_cookie) {
		535	# parse NAME=VALUE
		536	my @kv;
		537
		538	# expires is not http-compliant in the original cookie-spec,
		539	# we support the official date format and some extensions
		540	while (
		541	m{
		542	\G\s*
		543	(?:
		544	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		545	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
		546	)
		547	}gcxsi
		548	) {
		549	my $name = $2;
		550	my $value = $4;
		551
		552	if (defined $1) {
		553	# expires
		554	$name = "expires";
		555	$value = $1;
		556	} elsif (defined $3) {
		557	# quoted
		558	$value = $3;
		559	$value =~ s/\\(.)/$1/gs;
		560	}
		561
		562	push @kv, @kv ? lc $name : $name, $value;
		563
		564	last unless /\G\s*;/gc;
		565	}
		566
		567	last unless @kv;
		568
		569	my $name = shift @kv;
		570	my %kv = (value => shift @kv, @kv);
		571
		572	if (exists $kv{"max-age"}) {
		573	$kv{_expires} = $anow + delete $kv{"max-age"};
		574	} elsif (exists $kv{expires}) {
		575	$snow ||= parse_date ($date) || $anow;
		576	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		577	} else {
		578	delete $kv{_expires};
		579	}
		580
		581	my $cdom;
		582	my $cpath = (delete $kv{path}) || "/";
		583
		584	if (exists $kv{domain}) {
		585	$cdom = delete $kv{domain};
		586
		587	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		588
		589	next if $cdom =~ /\.$/;
		590
		591	# this is not rfc-like and not netscape-like. go figure.
		592	my $ndots = $cdom =~ y/.//;
		593	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		594	} else {
		595	$cdom = $host;
		596	}
		597
		598	# store it
		599	$jar->{version} = 1;
		600	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		601
		602	redo if /\G\s*,/gc;
		603	}
		604	}
		605
		606	#############################################################################
		607	# keepalive/persistent connection cache
		608
		609	# fetch a connection from the keepalive cache
		610	sub ka_fetch($) {
		611	my $ka_key = shift;
		612
		613	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		614	delete $KA_CACHE{$ka_key}
		615	unless @{ $KA_CACHE{$ka_key} };
		616
		617	$hdl
		618	}
		619
		620	sub ka_store($$) {
		621	my ($ka_key, $hdl) = @_;
		622
		623	my $kaa = $KA_CACHE{$ka_key} ||= [];
		624
		625	my $destroy = sub {
		626	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		627
		628	$hdl->destroy;
		629
		630	@ka
		631	? $KA_CACHE{$ka_key} = \@ka
		632	: delete $KA_CACHE{$ka_key};
		633	};
		634
		635	# on error etc., destroy
		636	$hdl->on_error ($destroy);
		637	$hdl->on_eof ($destroy);
		638	$hdl->on_read ($destroy);
		639	$hdl->timeout ($PERSISTENT_TIMEOUT);
		640
		641	push @$kaa, $hdl;
		642	shift @$kaa while @$kaa > $MAX_PER_HOST;
		643	}
		644
		645	#############################################################################
		646	# utilities
		647
		648	# continue to parse $_ for headers and place them into the arg
		649	sub _parse_hdr() {
		650	my %hdr;
		651
		652	# things seen, not parsed:
		653	# p3pP="NON CUR OTPi OUR NOR UNI"
		654
		655	$hdr{lc $1} .= ",$2"
		656	while /\G
		657	([^:\000-\037]*):
		658	[\011\040]*
		659	((?: [^\012]+ | \012[\011\040] )*)
		660	\012
		661	/gxc;
		662
		663	/\G$/
		664	or return;
		665
		666	# remove the "," prefix we added to all headers above
		667	substr $_, 0, 1, ""
		668	for values %hdr;
		669
		670	\%hdr
		671	}
		672
		673	#############################################################################
		674	# http_get
		675
		676	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		677
		678	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		679	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		680
		681	# maybe it should just become a normal object :/
		682
		683	sub _destroy_state(\%) {
		684	my ($state) = @_;
		685
		686	$state->{handle}->destroy if $state->{handle};
		687	%$state = ();
		688	}
		689
		690	sub _error(\%$$) {
		691	my ($state, $cb, $hdr) = @_;
		692
		693	&_destroy_state ($state);
		694
		695	$cb->(undef, $hdr);
		696	()
		697	}
		698
		699	our %IDEMPOTENT = (
		700	DELETE => 1,
		701	GET => 1,
		702	HEAD => 1,
		703	OPTIONS => 1,
		704	PUT => 1,
		705	TRACE => 1,
		706
		707	ACL => 1,
		708	"BASELINE-CONTROL" => 1,
		709	BIND => 1,
		710	CHECKIN => 1,
		711	CHECKOUT => 1,
		712	COPY => 1,
		713	LABEL => 1,
		714	LINK => 1,
		715	MERGE => 1,
		716	MKACTIVITY => 1,
		717	MKCALENDAR => 1,
		718	MKCOL => 1,
		719	MKREDIRECTREF => 1,
		720	MKWORKSPACE => 1,
		721	MOVE => 1,
		722	ORDERPATCH => 1,
		723	PROPFIND => 1,
		724	PROPPATCH => 1,
		725	REBIND => 1,
		726	REPORT => 1,
		727	SEARCH => 1,
		728	UNBIND => 1,
		729	UNCHECKOUT => 1,
		730	UNLINK => 1,
		731	UNLOCK => 1,
		732	UPDATE => 1,
		733	UPDATEREDIRECTREF => 1,
		734	"VERSION-CONTROL" => 1,
		735	);
		736
153	sub http_request($$$;@) {	737	sub http_request($$@) {
154	my $cb = pop;	738	my $cb = pop;
155	my ($method, $url, %arg) = @_;	739	my ($method, $url, %arg) = @_;
156		740
157	my %hdr;	741	my %hdr;
		742
		743	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		744	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
158		745
159	$method = uc $method;	746	$method = uc $method;
160		747
161	if (my $hdr = $arg{headers}) {	748	if (my $hdr = $arg{headers}) {
162	while (my ($k, $v) = each %$hdr) {	749	while (my ($k, $v) = each %$hdr) {
163	$hdr{lc $k} = $v;	750	$hdr{lc $k} = $v;
164	}	751	}
165	}	752	}
166		753
		754	# pseudo headers for all subsequent responses
		755	my @pseudo = (URL => $url);
		756	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		757
167	my $recurse = exists $arg{recurse} ? $arg{recurse} : $MAX_RECURSE;	758	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
168		759
169	return $cb->(undef, { Status => 599, Reason => "recursion limit reached" })	760	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
170	if $recurse < 0;	761	if $recurse < 0;
171		762
172	my $proxy = $arg{proxy} || $PROXY;	763	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
173	my $timeout = $arg{timeout} || $TIMEOUT;	764	my $timeout = $arg{timeout} || $TIMEOUT;
174		765
175	$hdr{"user-agent"} ||= $USERAGENT;	766	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
		767	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
176		768
177	my ($host, $port, $path, $scheme);	769	$uscheme = lc $uscheme;
		770
		771	my $uport = $uscheme eq "http" ? 80
		772	: $uscheme eq "https" ? 443
		773	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
		774
		775	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
		776	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
		777
		778	my $uhost = lc $1;
		779	$uport = $2 if defined $2;
		780
		781	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		782	unless exists $hdr{host};
		783
		784	$uhost =~ s/^\[(.*)\]$/$1/;
		785	$upath .= $query if length $query;
		786
		787	$upath =~ s%^/?%/%;
		788
		789	# cookie processing
		790	if (my $jar = $arg{cookie_jar}) {
		791	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
		792
		793	$hdr{cookie} = join "; ", @$cookies
		794	if @$cookies;
		795	}
		796
		797	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
178		798
179	if ($proxy) {	799	if ($proxy) {
180	($host, $port, $scheme) = @$proxy;	800	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
181	$path = $url;	801
		802	$rscheme = "http" unless defined $rscheme;
		803
		804	# don't support https requests over https-proxy transport,
		805	# can't be done with tls as spec'ed, unless you double-encrypt.
		806	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		807
		808	$rhost = lc $rhost;
		809	$rscheme = lc $rscheme;
182	} else {	810	} else {
183	($scheme, my $authority, $path, my $query, my $fragment) =	811	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
184	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;
185
186	$port = $scheme eq "http" ? 80
187	: $scheme eq "https" ? 443
188	: return $cb->(undef, { Status => 599, Reason => "$url: only http and https URLs supported" });
189
190	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
191	or return $cb->(undef, { Status => 599, Reason => "$url: unparsable URL" });
192
193	$host = $1;
194	$port = $2 if defined $2;
195
196	$host =~ s/^\[(.*)\]$/$1/;
197	$path .= "?$query" if length $query;
198
199	$path = "/" unless $path;
200
201	$hdr{host} = $host = lc $host;
202	}	812	}
203		813
204	$scheme = lc $scheme;	814	# leave out fragment and query string, just a heuristic
		815	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		816	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
205		817
206	my %state;
207
208	$hdr{"content-length"} = length $arg{body};	818	$hdr{"content-length"} = length $arg{body}
		819	if length $arg{body} || $method ne "GET";
209		820
210	$state{connect_guard} = AnyEvent::Socket::tcp_connect $host, $port, sub {	821	my $idempotent = $IDEMPOTENT{$method};
		822
		823	# default value for keepalive is true iff the request is for an idempotent method
		824	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		825	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		826	my $was_persistent; # true if this is actually a recycled connection
		827
		828	# the key to use in the keepalive cache
		829	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		830
		831	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
		832	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		833
		834	my %state = (connect_guard => 1);
		835
		836	my $ae_error = 595; # connecting
		837
		838	# handle actual, non-tunneled, request
		839	my $handle_actual_request = sub {
		840	$ae_error = 596; # request phase
		841
		842	my $hdl = $state{handle};
		843
		844	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		845
		846	# send request
		847	$hdl->push_write (
		848	"$method $rpath HTTP/1.1\015\012"
		849	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		850	. "\015\012"
		851	. $arg{body}
		852	);
		853
		854	# return if error occurred during push_write()
		855	return unless %state;
		856
		857	# reduce memory usage, save a kitten, also re-use it for the response headers.
		858	%hdr = ();
		859
		860	# status line and headers
		861	$state{read_response} = sub {
		862	return unless %state;
		863
		864	for ("$_[1]") {
		865	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		866
		867	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		868	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		869
		870	# 100 Continue handling
		871	# should not happen as we don't send expect: 100-continue,
		872	# but we handle it just in case.
		873	# since we send the request body regardless, if we get an error
		874	# we are out of-sync, which we currently do NOT handle correctly.
		875	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		876	if $2 eq 100;
		877
		878	push @pseudo,
		879	HTTPVersion => $1,
		880	Status => $2,
		881	Reason => $3,
		882	;
		883
		884	my $hdr = _parse_hdr
		885	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		886
		887	%hdr = (%$hdr, @pseudo);
		888	}
		889
		890	# redirect handling
		891	# relative uri handling forced by microsoft and other shitheads.
		892	# we give our best and fall back to URI if available.
		893	if (exists $hdr{location}) {
		894	my $loc = $hdr{location};
		895
		896	if ($loc =~ m%^//%) { # //
		897	$loc = "$rscheme:$loc";
		898
		899	} elsif ($loc eq "") {
		900	$loc = $url;
		901
		902	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		903	$loc =~ s/^\.\/+//;
		904
		905	if ($loc !~ m%^[.?#]%) {
		906	my $prefix = "$rscheme://$uhost:$uport";
		907
		908	unless ($loc =~ s/^\///) {
		909	$prefix .= $upath;
		910	$prefix =~ s/\/[^\/]*$//;
		911	}
		912
		913	$loc = "$prefix/$loc";
		914
		915	} elsif (eval { require URI }) { # uri
		916	$loc = URI->new_abs ($loc, $url)->as_string;
		917
		918	} else {
		919	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		920	#$hdr{Status} = 599;
		921	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		922	#$recurse = 0;
		923	}
		924	}
		925
		926	$hdr{location} = $loc;
		927	}
		928
		929	my $redirect;
		930
		931	if ($recurse) {
		932	my $status = $hdr{Status};
		933
		934	# industry standard is to redirect POST as GET for
		935	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		936	# also, the UA should ask the user for 301 and 307 and POST,
		937	# industry standard seems to be to simply follow.
		938	# we go with the industry standard. 308 is defined
		939	# by rfc7538
		940	if ($status == 301 or $status == 302 or $status == 303) {
		941	# HTTP/1.1 is unclear on how to mutate the method
		942	$method = "GET" unless $method eq "HEAD";
		943	$redirect = 1;
		944	} elsif ($status == 307 or $status == 308) {
		945	$redirect = 1;
		946	}
		947	}
		948
		949	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		950	if ($state{handle}) {
		951	# handle keepalive
		952	if (
		953	$persistent
		954	&& $_[3]
		955	&& ($hdr{HTTPVersion} < 1.1
		956	? $hdr{connection} =~ /\bkeep-?alive\b/i
		957	: $hdr{connection} !~ /\bclose\b/i)
		958	) {
		959	ka_store $ka_key, delete $state{handle};
		960	} else {
		961	# no keepalive, destroy the handle
		962	$state{handle}->destroy;
		963	}
		964	}
		965
		966	%state = ();
		967
		968	if (defined $_[1]) {
		969	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		970	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		971	}
		972
		973	# set-cookie processing
		974	if ($arg{cookie_jar}) {
		975	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		976	}
		977
		978	if ($redirect && exists $hdr{location}) {
		979	# we ignore any errors, as it is very common to receive
		980	# Content-Length != 0 but no actual body
		981	# we also access %hdr, as $_[1] might be an erro
		982	$state{recurse} =
		983	http_request (
		984	$method => $hdr{location},
		985	%arg,
		986	recurse => $recurse - 1,
		987	Redirect => [$_[0], \%hdr],
		988	sub {
		989	%state = ();
		990	&$cb
		991	},
		992	);
		993	} else {
		994	$cb->($_[0], \%hdr);
		995	}
		996	};
		997
		998	$ae_error = 597; # body phase
		999
		1000	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1001
		1002	my $len = $chunked ? undef : $hdr{"content-length"};
		1003
		1004	# body handling, many different code paths
		1005	# - no body expected
		1006	# - want_body_handle
		1007	# - te chunked
		1008	# - 2x length known (with or without on_body)
		1009	# - 2x length not known (with or without on_body)
		1010	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1011	$finish->(undef, 598 => "Request cancelled by on_header");
		1012	} elsif (
		1013	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1014	or $method eq "HEAD"
		1015	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1016	) {
		1017	# no body
		1018	$finish->("", undef, undef, 1);
		1019
		1020	} elsif (!$redirect && $arg{want_body_handle}) {
		1021	$_[0]->on_eof (undef);
		1022	$_[0]->on_error (undef);
		1023	$_[0]->on_read (undef);
		1024
		1025	$finish->(delete $state{handle});
		1026
		1027	} elsif ($chunked) {
		1028	my $cl = 0;
		1029	my $body = "";
		1030	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		1031
		1032	$state{read_chunk} = sub {
		1033	$_[1] =~ /^([0-9a-fA-F]+)/
		1034	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1035
		1036	my $len = hex $1;
		1037
		1038	if ($len) {
		1039	$cl += $len;
		1040
		1041	$_[0]->push_read (chunk => $len, sub {
		1042	$on_body->($_[1], \%hdr)
		1043	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1044
		1045	$_[0]->push_read (line => sub {
		1046	length $_[1]
		1047	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1048	$_[0]->push_read (line => $state{read_chunk});
		1049	});
		1050	});
		1051	} else {
		1052	$hdr{"content-length"} ||= $cl;
		1053
		1054	$_[0]->push_read (line => $qr_nlnl, sub {
		1055	if (length $_[1]) {
		1056	for ("$_[1]") {
		1057	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1058
		1059	my $hdr = _parse_hdr
		1060	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1061
		1062	%hdr = (%hdr, %$hdr);
		1063	}
		1064	}
		1065
		1066	$finish->($body, undef, undef, 1);
		1067	});
		1068	}
		1069	};
		1070
		1071	$_[0]->push_read (line => $state{read_chunk});
		1072
		1073	} elsif ($arg{on_body}) {
		1074	if (defined $len) {
		1075	$_[0]->on_read (sub {
		1076	$len -= length $_[0]{rbuf};
		1077
		1078	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1079	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1080
		1081	$len > 0
		1082	or $finish->("", undef, undef, 1);
		1083	});
		1084	} else {
		1085	$_[0]->on_eof (sub {
		1086	$finish->("");
		1087	});
		1088	$_[0]->on_read (sub {
		1089	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1090	or $finish->(undef, 598 => "Request cancelled by on_body");
		1091	});
		1092	}
		1093	} else {
		1094	$_[0]->on_eof (undef);
		1095
		1096	if (defined $len) {
		1097	$_[0]->on_read (sub {
		1098	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1099	if $len <= length $_[0]{rbuf};
		1100	});
		1101	} else {
		1102	$_[0]->on_error (sub {
		1103	($! == Errno::EPIPE || !$!)
		1104	? $finish->(delete $_[0]{rbuf})
		1105	: $finish->(undef, $ae_error => $_[2]);
		1106	});
		1107	$_[0]->on_read (sub { });
		1108	}
		1109	}
		1110	};
		1111
		1112	# if keepalive is enabled, then the server closing the connection
		1113	# before a response can happen legally - we retry on idempotent methods.
		1114	if ($was_persistent && $idempotent) {
		1115	my $old_eof = $hdl->{on_eof};
		1116	$hdl->{on_eof} = sub {
		1117	_destroy_state %state;
		1118
		1119	%state = ();
		1120	$state{recurse} =
		1121	http_request (
		1122	$method => $url,
		1123	%arg,
		1124	recurse => $recurse - 1,
		1125	keepalive => 0,
		1126	sub {
		1127	%state = ();
		1128	&$cb
		1129	}
		1130	);
		1131	};
		1132	$hdl->on_read (sub {
		1133	return unless %state;
		1134
		1135	# as soon as we receive something, a connection close
		1136	# once more becomes a hard error
		1137	$hdl->{on_eof} = $old_eof;
		1138	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1139	});
		1140	} else {
		1141	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1142	}
		1143	};
		1144
		1145	my $prepare_handle = sub {
		1146	my ($hdl) = $state{handle};
		1147
		1148	$hdl->on_error (sub {
		1149	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1150	});
		1151	$hdl->on_eof (sub {
		1152	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1153	});
		1154	$hdl->timeout_reset;
		1155	$hdl->timeout ($timeout);
		1156	};
		1157
		1158	# connected to proxy (or origin server)
		1159	my $connect_cb = sub {
211	$state{fh} = shift	1160	my $fh = shift
212	or return $cb->(undef, { Status => 599, Reason => "$!" });	1161	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
213		1162
214	delete $state{connect_guard}; # reduce memory usage, save a tree	1163	return unless delete $state{connect_guard};
215		1164
216	# get handle	1165	# get handle
217	$state{handle} = new AnyEvent::Handle	1166	$state{handle} = new AnyEvent::Handle
218	fh => $state{fh},	1167	%{ $arg{handle_params} },
219	($scheme eq "https" ? (tls => "connect") : ());	1168	fh => $fh,
220		1169	peername => $uhost,
221	# limit the number of persistent connections	1170	tls_ctx => $arg{tls_ctx},
222	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
223	++$KA_COUNT{$_[1]};
224	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
225	$hdr{connection} = "keep-alive";
226	delete $hdr{connection}; # keep-alive not yet supported
227	} else {
228	delete $hdr{connection};
229	}
230
231	# (re-)configure handle
232	$state{handle}->timeout ($timeout);
233	$state{handle}->on_error (sub {
234	%state = ();
235	$cb->(undef, { Status => 599, Reason => "$!" });
236	});
237	$state{handle}->on_eof (sub {
238	%state = ();
239	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file" });
240	});
241
242	# send request
243	$state{handle}->push_write (
244	"$method $path HTTP/1.0\015\012"
245	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
246	. "\015\012"
247	. (delete $arg{body})
248	);	1171	;
249		1172
250	%hdr = (); # reduce memory usage, save a kitten	1173	$prepare_handle->();
251		1174
252	# status line	1175	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1176
		1177	# now handle proxy-CONNECT method
		1178	if ($proxy && $uscheme eq "https") {
		1179	# oh dear, we have to wrap it into a connect request
		1180
		1181	# maybe re-use $uauthority with patched port?
		1182	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
253	$state{handle}->push_read (line => qr/\015?\012/, sub {	1183	$state{handle}->push_read (line => $qr_nlnl, sub {
254	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix	1184	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
255	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])" }));	1185	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
256		1186
257	my %hdr = ( # response headers	1187	if ($2 == 200) {
258	HTTPVersion => ",$1",	1188	$rpath = $upath;
259	Status => ",$2",	1189	$handle_actual_request->();
260	Reason => ",$3",
261	);
262
263	# headers, could be optimized a bit
264	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
265	for ("$_[1]\012") {
266	# we support spaces in field names, as lotus domino
267	# creates them.
268	$hdr{lc $1} .= ",$2"
269	while /\G
270	([^:\000-\037]+):
271	[\011\040]*
272	((?: [^\015\012]+ | \015?\012[\011\040] )*)
273	\015?\012
274	/gxc;
275
276	/\G$/
277	or return $cb->(undef, { Status => 599, Reason => "garbled response headers" });
278	}
279
280	substr $_, 0, 1, ""
281	for values %hdr;
282
283	my $finish = sub {
284	if ($_[1]{Status} =~ /^30[12]$/ && $recurse) {
285	http_request ($method, $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
286	} else {
287	$cb->($_[0], $_[1]);
288	}
289	};
290
291	if ($hdr{Status} =~ /^(?:1..|204|304)$/ or $method eq "HEAD") {
292	%state = ();
293	$finish->(undef, \%hdr);
294	} else {	1190	} else {
295	if (exists $hdr{"content-length"}) {	1191	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
296	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
297	# could cache persistent connection now
298	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
299	# but we don't, due to misdesigns, this is annoyingly complex
300	};
301
302	%state = ();
303	$finish->($_[1], \%hdr);
304	});
305	} else {
306	# too bad, need to read until we get an error or EOF,
307	# no way to detect winged data.
308	$_[0]->on_error (sub {
309	%state = ();
310	$finish->($_[0]{rbuf}, \%hdr);
311	});
312	$_[0]->on_eof (undef);
313	$_[0]->on_read (sub { });
314	}
315	}	1192	}
316	});	1193	});
		1194	} else {
		1195	$handle_actual_request->();
317	});	1196	}
318	}, sub {
319	$timeout
320	};	1197	};
321		1198
		1199	_get_slot $uhost, sub {
		1200	$state{slot_guard} = shift;
		1201
		1202	return unless $state{connect_guard};
		1203
		1204	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1205	# on a keepalive request (in theory, this should be a separate config option).
		1206	if ($persistent && $KA_CACHE{$ka_key}) {
		1207	$was_persistent = 1;
		1208
		1209	$state{handle} = ka_fetch $ka_key;
		1210	$state{handle}->destroyed
		1211	and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1212	$prepare_handle->();
		1213	$state{handle}->destroyed
		1214	and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1215	$handle_actual_request->();
		1216
		1217	} else {
		1218	my $tcp_connect = $arg{tcp_connect}
		1219	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1220
		1221	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
		1222	}
		1223	};
		1224
322	defined wantarray && AnyEvent::Util::guard { %state = () }	1225	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
323	}	1226	}
324		1227
325	sub http_get($$;@) {	1228	sub http_get($@) {
326	unshift @_, "GET";	1229	unshift @_, "GET";
327	&http_request	1230	&http_request
328	}	1231	}
329		1232
330	sub http_head($$;@) {	1233	sub http_head($@) {
331	unshift @_, "HEAD";	1234	unshift @_, "HEAD";
332	&http_request	1235	&http_request
333	}	1236	}
334		1237
335	sub http_post($$$;@) {	1238	sub http_post($$@) {
		1239	my $url = shift;
336	unshift @_, "POST", "body";	1240	unshift @_, "POST", $url, "body";
337	&http_request	1241	&http_request
338	}	1242	}
339		1243
340	=back	1244	=back
341		1245
		1246	=head2 DNS CACHING
		1247
		1248	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1249	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1250	hostnames. The latter is a simple stub resolver and does no caching
		1251	on its own. If you want DNS caching, you currently have to provide
		1252	your own default resolver (by storing a suitable resolver object in
		1253	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1254
342	=head2 GLOBAL FUNCTIONS AND VARIABLES	1255	=head2 GLOBAL FUNCTIONS AND VARIABLES
343		1256
344	=over 4	1257	=over 4
345		1258
346	=item AnyEvent::HTTP::set_proxy "proxy-url"	1259	=item AnyEvent::HTTP::set_proxy "proxy-url"
347		1260
348	Sets the default proxy server to use. The proxy-url must begin with a	1261	Sets the default proxy server to use. The proxy-url must begin with a
349	string of the form C<http://host:port> (optionally C<https:...>).	1262	string of the form C<http://host:port>, croaks otherwise.
		1263
		1264	To clear an already-set proxy, use C<undef>.
		1265
		1266	When AnyEvent::HTTP is loaded for the first time it will query the
		1267	default proxy from the operating system, currently by looking at
		1268	C<$ENV{http_proxy>}.
		1269
		1270	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1271
		1272	Remove all cookies from the cookie jar that have been expired. If
		1273	C<$session_end> is given and true, then additionally remove all session
		1274	cookies.
		1275
		1276	You should call this function (with a true C<$session_end>) before you
		1277	save cookies to disk, and you should call this function after loading them
		1278	again. If you have a long-running program you can additionally call this
		1279	function from time to time.
		1280
		1281	A cookie jar is initially an empty hash-reference that is managed by this
		1282	module. Its format is subject to change, but currently it is as follows:
		1283
		1284	The key C<version> has to contain C<1>, otherwise the hash gets
		1285	emptied. All other keys are hostnames or IP addresses pointing to
		1286	hash-references. The key for these inner hash references is the
		1287	server path for which this cookie is meant, and the values are again
		1288	hash-references. Each key of those hash-references is a cookie name, and
		1289	the value, you guessed it, is another hash-reference, this time with the
		1290	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1291	which have been replaced by a C<_expires> key that contains the cookie
		1292	expiry timestamp. Session cookies are indicated by not having an
		1293	C<_expires> key.
		1294
		1295	Here is an example of a cookie jar with a single cookie, so you have a
		1296	chance of understanding the above paragraph:
		1297
		1298	{
		1299	version => 1,
		1300	"10.0.0.1" => {
		1301	"/" => {
		1302	"mythweb_id" => {
		1303	_expires => 1293917923,
		1304	value => "ooRung9dThee3ooyXooM1Ohm",
		1305	},
		1306	},
		1307	},
		1308	}
		1309
		1310	=item $date = AnyEvent::HTTP::format_date $timestamp
		1311
		1312	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1313	Date (RFC 2616).
		1314
		1315	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1316
		1317	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1318	bunch of minor variations of those, and returns the corresponding POSIX
		1319	timestamp, or C<undef> if the date cannot be parsed.
350		1320
351	=item $AnyEvent::HTTP::MAX_RECURSE	1321	=item $AnyEvent::HTTP::MAX_RECURSE
352		1322
353	The default value for the C<recurse> request parameter (default: C<10>).	1323	The default value for the C<recurse> request parameter (default: C<10>).
354		1324
		1325	=item $AnyEvent::HTTP::TIMEOUT
		1326
		1327	The default timeout for connection operations (default: C<300>).
		1328
355	=item $AnyEvent::HTTP::USERAGENT	1329	=item $AnyEvent::HTTP::USERAGENT
356		1330
357	The default value for the C<User-Agent> header (the default is	1331	The default value for the C<User-Agent> header (the default is
358	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1332	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
359		1333
360	=item $AnyEvent::HTTP::MAX_PERSISTENT	1334	=item $AnyEvent::HTTP::MAX_PER_HOST
361		1335
362	The maximum number of persistent connections to keep open (default: 8).	1336	The maximum number of concurrent connections to the same host (identified
		1337	by the hostname). If the limit is exceeded, then additional requests
		1338	are queued until previous connections are closed. Both persistent and
		1339	non-persistent connections are counted in this limit.
363		1340
364	Not implemented currently.	1341	The default value for this is C<4>, and it is highly advisable to not
		1342	increase it much.
		1343
		1344	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1345	connections, older browsers used 2, newer ones (such as firefox 3)
		1346	typically use 6, and Opera uses 8 because like, they have the fastest
		1347	browser and give a shit for everybody else on the planet.
365		1348
366	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1349	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
367		1350
368	The maximum time to cache a persistent connection, in seconds (default: 2).	1351	The time after which idle persistent connections get closed by
		1352	AnyEvent::HTTP (default: C<3>).
369		1353
370	Not implemented currently.	1354	=item $AnyEvent::HTTP::ACTIVE
		1355
		1356	The number of active connections. This is not the number of currently
		1357	running requests, but the number of currently open and non-idle TCP
		1358	connections. This number can be useful for load-leveling.
371		1359
372	=back	1360	=back
373		1361
374	=cut	1362	=cut
375		1363
		1364	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1365	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1366
		1367	sub format_date($) {
		1368	my ($time) = @_;
		1369
		1370	# RFC 822/1123 format
		1371	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1372
		1373	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1374	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1375	$H, $M, $S;
		1376	}
		1377
		1378	sub parse_date($) {
		1379	my ($date) = @_;
		1380
		1381	my ($d, $m, $y, $H, $M, $S);
		1382
		1383	if ($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$/) {
		1384	# RFC 822/1123, required by RFC 2616 (with " ")
		1385	# cookie dates (with "-")
		1386
		1387	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1388
		1389	} elsif ($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$/) {
		1390	# RFC 850
		1391	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1392
		1393	} elsif ($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])$/) {
		1394	# ISO C's asctime
		1395	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1396	}
		1397	# other formats fail in the loop below
		1398
		1399	for (0..11) {
		1400	if ($m eq $month[$_]) {
		1401	require Time::Local;
		1402	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
		1403	}
		1404	}
		1405
		1406	undef
		1407	}
		1408
376	sub set_proxy($) {	1409	sub set_proxy($) {
377	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1410	if (length $_[0]) {
		1411	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1412	or Carp::croak "$_[0]: invalid proxy URL";
		1413	$PROXY = [$2, $3 || 3128, $1]
		1414	} else {
		1415	undef $PROXY;
		1416	}
378	}	1417	}
379		1418
380	# initialise proxy from environment	1419	# initialise proxy from environment
		1420	eval {
381	set_proxy $ENV{http_proxy};	1421	set_proxy $ENV{http_proxy};
		1422	};
		1423
		1424	=head2 SHOWCASE
		1425
		1426	This section contains some more elaborate "real-world" examples or code
		1427	snippets.
		1428
		1429	=head2 HTTP/1.1 FILE DOWNLOAD
		1430
		1431	Downloading files with HTTP can be quite tricky, especially when something
		1432	goes wrong and you want to resume.
		1433
		1434	Here is a function that initiates and resumes a download. It uses the
		1435	last modified time to check for file content changes, and works with many
		1436	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1437	on older servers.
		1438
		1439	It calls the completion callback with either C<undef>, which means a
		1440	nonretryable error occurred, C<0> when the download was partial and should
		1441	be retried, and C<1> if it was successful.
		1442
		1443	use AnyEvent::HTTP;
		1444
		1445	sub download($$$) {
		1446	my ($url, $file, $cb) = @_;
		1447
		1448	open my $fh, "+<", $file
		1449	or die "$file: $!";
		1450
		1451	my %hdr;
		1452	my $ofs = 0;
		1453
		1454	warn stat $fh;
		1455	warn -s _;
		1456	if (stat $fh and -s _) {
		1457	$ofs = -s _;
		1458	warn "-s is ", $ofs;
		1459	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1460	$hdr{"range"} = "bytes=$ofs-";
		1461	}
		1462
		1463	http_get $url,
		1464	headers => \%hdr,
		1465	on_header => sub {
		1466	my ($hdr) = @_;
		1467
		1468	if ($hdr->{Status} == 200 && $ofs) {
		1469	# resume failed
		1470	truncate $fh, $ofs = 0;
		1471	}
		1472
		1473	sysseek $fh, $ofs, 0;
		1474
		1475	1
		1476	},
		1477	on_body => sub {
		1478	my ($data, $hdr) = @_;
		1479
		1480	if ($hdr->{Status} =~ /^2/) {
		1481	length $data == syswrite $fh, $data
		1482	or return; # abort on write errors
		1483	}
		1484
		1485	1
		1486	},
		1487	sub {
		1488	my (undef, $hdr) = @_;
		1489
		1490	my $status = $hdr->{Status};
		1491
		1492	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1493	utime $fh, $time, $time;
		1494	}
		1495
		1496	if ($status == 200 || $status == 206 || $status == 416) {
		1497	# download ok || resume ok || file already fully downloaded
		1498	$cb->(1, $hdr);
		1499
		1500	} elsif ($status == 412) {
		1501	# file has changed while resuming, delete and retry
		1502	unlink $file;
		1503	$cb->(0, $hdr);
		1504
		1505	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1506	# retry later
		1507	$cb->(0, $hdr);
		1508
		1509	} else {
		1510	$cb->(undef, $hdr);
		1511	}
		1512	}
		1513	;
		1514	}
		1515
		1516	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1517	if ($_[0]) {
		1518	print "OK!\n";
		1519	} elsif (defined $_[0]) {
		1520	print "please retry later\n";
		1521	} else {
		1522	print "ERROR\n";
		1523	}
		1524	};
		1525
		1526	=head3 SOCKS PROXIES
		1527
		1528	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1529	compile your perl to support socks, or use an external program such as
		1530	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1531	transparently.
		1532
		1533	Alternatively, for AnyEvent::HTTP only, you can use your own
		1534	C<tcp_connect> function that does the proxy handshake - here is an example
		1535	that works with socks4a proxies:
		1536
		1537	use Errno;
		1538	use AnyEvent::Util;
		1539	use AnyEvent::Socket;
		1540	use AnyEvent::Handle;
		1541
		1542	# host, port and username of/for your socks4a proxy
		1543	my $socks_host = "10.0.0.23";
		1544	my $socks_port = 9050;
		1545	my $socks_user = "";
		1546
		1547	sub socks4a_connect {
		1548	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1549
		1550	my $hdl = new AnyEvent::Handle
		1551	connect => [$socks_host, $socks_port],
		1552	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1553	on_error => sub { $connect_cb->() },
		1554	;
		1555
		1556	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1557
		1558	$hdl->push_read (chunk => 8, sub {
		1559	my ($hdl, $chunk) = @_;
		1560	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1561
		1562	if ($status == 0x5a) {
		1563	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1564	} else {
		1565	$! = Errno::ENXIO; $connect_cb->();
		1566	}
		1567	});
		1568
		1569	$hdl
		1570	}
		1571
		1572	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1573	possibly after switching off other proxy types:
		1574
		1575	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1576
		1577	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1578	my ($data, $headers) = @_;
		1579	...
		1580	};
382		1581
383	=head1 SEE ALSO	1582	=head1 SEE ALSO
384		1583
385	L<AnyEvent>.	1584	L<AnyEvent>.
386		1585
387	=head1 AUTHOR	1586	=head1 AUTHOR
388		1587
389	Marc Lehmann <schmorp@schmorp.de>	1588	Marc Lehmann <schmorp@schmorp.de>
390	http://home.schmorp.de/	1589	http://home.schmorp.de/
		1590
		1591	With many thanks to Дмитрий Шалашов, who provided countless
		1592	testcases and bugreports.
391		1593
392	=cut	1594	=cut
393		1595
394	1	1596	1
395		1597

Diff Legend

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