ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.7 by root, Wed Jun 4 12:10:27 2008 UTC vs.
Revision 1.139 by root, Fri Aug 5 20:48:14 2022 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 = 2.25;
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>).
		199
		200	Currently, if your proxy requires authorization, you have to specify an
		201	appropriate "Proxy-Authorization" header in every request.
		202
		203	Note that this module will prefer an existing persistent connection,
		204	even if that connection was made using another proxy. If you need to
		205	ensure that a new connection is made in this case, you can either force
		206	C<persistent> to false or e.g. use the proxy address in your C<sessionid>.
123		207
124	=item body => $string	208	=item body => $string
125		209
126	The request body, usually empty. Will be-sent as-is (future versions of	210	The request body, usually empty. Will be sent as-is (future versions of
127	this module might offer more options).	211	this module might offer more options).
128		212
		213	=item cookie_jar => $hash_ref
		214
		215	Passing this parameter enables (simplified) cookie-processing, loosely
		216	based on the original netscape specification.
		217
		218	The C<$hash_ref> must be an (initially empty) hash reference which
		219	will get updated automatically. It is possible to save the cookie jar
		220	to persistent storage with something like JSON or Storable - see the
		221	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		222	expired or session-only cookies, and also for documentation on the format
		223	of the cookie jar.
		224
		225	Note that this cookie implementation is not meant to be complete. If
		226	you want complete cookie management you have to do that on your
		227	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
		228	working. Cookies are a privacy disaster, do not use them unless required
		229	to.
		230
		231	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		232	headers will be set and handled by this module, otherwise they will be
		233	left untouched.
		234
		235	=item tls_ctx => $scheme | $tls_ctx
		236
		237	Specifies the AnyEvent::TLS context to be used for https connections. This
		238	parameter follows the same rules as the C<tls_ctx> parameter to
		239	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		240	C<high> can be specified, which give you a predefined low-security (no
		241	verification, highest compatibility) and high-security (CA and common-name
		242	verification) TLS context.
		243
		244	The default for this option is C<low>, which could be interpreted as "give
		245	me the page, no matter what".
		246
		247	See also the C<sessionid> parameter.
		248
		249	=item sessionid => $string
		250
		251	The module might reuse connections to the same host internally (regardless
		252	of other settings, such as C<tcp_connect> or C<proxy>). Sometimes (e.g.
		253	when using TLS or a specfic proxy), you do not want to reuse connections
		254	from other sessions. This can be achieved by setting this parameter to
		255	some unique ID (such as the address of an object storing your state data
		256	or the TLS context, or the proxy IP) - only connections using the same
		257	unique ID will be reused.
		258
		259	=item on_prepare => $callback->($fh)
		260
		261	In rare cases you need to "tune" the socket before it is used to
		262	connect (for example, to bind it on a given IP address). This parameter
		263	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		264	and behaves exactly the same way (e.g. it has to provide a
		265	timeout). See the description for the C<$prepare_cb> argument of
		266	C<AnyEvent::Socket::tcp_connect> for details.
		267
		268	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		269
		270	In even rarer cases you want total control over how AnyEvent::HTTP
		271	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		272	to do this, but you can provide your own C<tcp_connect> function -
		273	obviously, it has to follow the same calling conventions, except that it
		274	may always return a connection guard object.
		275
		276	The connections made by this hook will be treated as equivalent to
		277	connections made the built-in way, specifically, they will be put into
		278	and taken from the persistent connection cache. If your C<$tcp_connect>
		279	function is incompatible with this kind of re-use, consider switching off
		280	C<persistent> connections and/or providing a C<sessionid> identifier.
		281
		282	There are probably lots of weird uses for this function, starting from
		283	tracing the hosts C<http_request> actually tries to connect, to (inexact
		284	but fast) host => IP address caching or even socks protocol support.
		285
		286	=item on_header => $callback->($headers)
		287
		288	When specified, this callback will be called with the header hash as soon
		289	as headers have been successfully received from the remote server (not on
		290	locally-generated errors).
		291
		292	It has to return either true (in which case AnyEvent::HTTP will continue),
		293	or false, in which case AnyEvent::HTTP will cancel the download (and call
		294	the finish callback with an error code of C<598>).
		295
		296	This callback is useful, among other things, to quickly reject unwanted
		297	content, which, if it is supposed to be rare, can be faster than first
		298	doing a C<HEAD> request.
		299
		300	The downside is that cancelling the request makes it impossible to re-use
		301	the connection. Also, the C<on_header> callback will not receive any
		302	trailer (headers sent after the response body).
		303
		304	Example: cancel the request unless the content-type is "text/html".
		305
		306	on_header => sub {
		307	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		308	},
		309
		310	=item on_body => $callback->($partial_body, $headers)
		311
		312	When specified, all body data will be passed to this callback instead of
		313	to the completion callback. The completion callback will get the empty
		314	string instead of the body data.
		315
		316	It has to return either true (in which case AnyEvent::HTTP will continue),
		317	or false, in which case AnyEvent::HTTP will cancel the download (and call
		318	the completion callback with an error code of C<598>).
		319
		320	The downside to cancelling the request is that it makes it impossible to
		321	re-use the connection.
		322
		323	This callback is useful when the data is too large to be held in memory
		324	(so the callback writes it to a file) or when only some information should
		325	be extracted, or when the body should be processed incrementally.
		326
		327	It is usually preferred over doing your own body handling via
		328	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		329	only used to create a connection, C<want_body_handle> is the better
		330	alternative, as it allows you to install your own event handler, reducing
		331	resource usage.
		332
		333	=item want_body_handle => $enable
		334
		335	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		336	changes considerably: after parsing the headers, and instead of
		337	downloading the body (if any), the completion callback will be
		338	called. Instead of the C<$body> argument containing the body data, the
		339	callback will receive the L<AnyEvent::Handle> object associated with the
		340	connection. In error cases, C<undef> will be passed. When there is no body
		341	(e.g. status C<304>), the empty string will be passed.
		342
		343	The handle object might or might not be in TLS mode, might be connected
		344	to a proxy, be a persistent connection, use chunked transfer encoding
		345	etc., and configured in unspecified ways. The user is responsible for this
		346	handle (it will not be used by this module anymore).
		347
		348	This is useful with some push-type services, where, after the initial
		349	headers, an interactive protocol is used (typical example would be the
		350	push-style twitter API which starts a JSON/XML stream).
		351
		352	If you think you need this, first have a look at C<on_body>, to see if
		353	that doesn't solve your problem in a better way.
		354
		355	=item persistent => $boolean
		356
		357	Try to create/reuse a persistent connection. When this flag is set
		358	(default: true for idempotent requests, false for all others), then
		359	C<http_request> tries to re-use an existing (previously-created)
		360	persistent connection to same host (i.e. identical URL scheme, hostname,
		361	port and sessionid) and, failing that, tries to create a new one.
		362
		363	Requests failing in certain ways will be automatically retried once, which
		364	is dangerous for non-idempotent requests, which is why it defaults to off
		365	for them. The reason for this is because the bozos who designed HTTP/1.1
		366	made it impossible to distinguish between a fatal error and a normal
		367	connection timeout, so you never know whether there was a problem with
		368	your request or not.
		369
		370	When reusing an existent connection, many parameters (such as TLS context)
		371	will be ignored. See the C<sessionid> parameter for a workaround.
		372
		373	=item keepalive => $boolean
		374
		375	Only used when C<persistent> is also true. This parameter decides whether
		376	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		377	(as opposed to only a HTTP/1.1 persistent connection).
		378
		379	The default is true, except when using a proxy, in which case it defaults
		380	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		381
		382	=item handle_params => { key => value ... }
		383
		384	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		385	constructor that is called - not all requests will create a handle, and
		386	sometimes more than one is created, so this parameter is only good for
		387	setting hints.
		388
		389	Example: set the maximum read size to 4096, to potentially conserve memory
		390	at the cost of speed.
		391
		392	handle_params => {
		393	max_read_size => 4096,
		394	},
		395
129	=back	396	=back
130		397
131	=back	398	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		399	the response body.
		400
		401	http_request GET => "http://www.nethype.de/", sub {
		402	my ($body, $hdr) = @_;
		403	print "$body\n";
		404	};
		405
		406	Example: do a HTTP HEAD request on https://www.google.com/, use a
		407	timeout of 30 seconds.
		408
		409	http_request
		410	HEAD => "https://www.google.com",
		411	headers => { "user-agent" => "MySearchClient 1.0" },
		412	timeout => 30,
		413	sub {
		414	my ($body, $hdr) = @_;
		415	use Data::Dumper;
		416	print Dumper $hdr;
		417	}
		418	;
		419
		420	Example: do another simple HTTP GET request, but immediately try to
		421	cancel it.
		422
		423	my $request = http_request GET => "http://www.nethype.de/", sub {
		424	my ($body, $hdr) = @_;
		425	print "$body\n";
		426	};
		427
		428	undef $request;
132		429
133	=cut	430	=cut
134		431
		432	#############################################################################
		433	# wait queue/slots
		434
		435	sub _slot_schedule;
		436	sub _slot_schedule($) {
		437	my $host = shift;
		438
		439	while ($CO_SLOT{$host}[0] < $MAX_PER_HOST) {
		440	if (my $cb = shift @{ $CO_SLOT{$host}[1] }) {
		441	# somebody wants that slot
		442	++$CO_SLOT{$host}[0];
		443	++$ACTIVE;
		444
		445	$cb->(AnyEvent::Util::guard {
		446	--$ACTIVE;
		447	--$CO_SLOT{$host}[0];
		448	_slot_schedule $host;
		449	});
		450	} else {
		451	# nobody wants the slot, maybe we can forget about it
		452	delete $CO_SLOT{$host} unless $CO_SLOT{$host}[0];
		453	last;
		454	}
		455	}
		456	}
		457
		458	# wait for a free slot on host, call callback
		459	sub _get_slot($$) {
		460	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
		461
		462	_slot_schedule $_[0];
		463	}
		464
		465	#############################################################################
		466	# cookie handling
		467
		468	# expire cookies
		469	sub cookie_jar_expire($;$) {
		470	my ($jar, $session_end) = @_;
		471
		472	%$jar = () if $jar->{version} != 2;
		473
		474	my $anow = AE::now;
		475
		476	while (my ($chost, $paths) = each %$jar) {
		477	next unless ref $paths;
		478
		479	while (my ($cpath, $cookies) = each %$paths) {
		480	while (my ($cookie, $kv) = each %$cookies) {
		481	if (exists $kv->{_expires}) {
		482	delete $cookies->{$cookie}
		483	if $anow > $kv->{_expires};
		484	} elsif ($session_end) {
		485	delete $cookies->{$cookie};
		486	}
		487	}
		488
		489	delete $paths->{$cpath}
		490	unless %$cookies;
		491	}
		492
		493	delete $jar->{$chost}
		494	unless %$paths;
		495	}
		496	}
		497
		498	# extract cookies from jar
		499	sub cookie_jar_extract($$$$) {
		500	my ($jar, $scheme, $host, $path) = @_;
		501
		502	%$jar = () if $jar->{version} != 2;
		503
		504	$host = AnyEvent::Util::idn_to_ascii $host
		505	if $host =~ /[^\x00-\x7f]/;
		506
		507	my @cookies;
		508
		509	while (my ($chost, $paths) = each %$jar) {
		510	next unless ref $paths;
		511
		512	# exact match or suffix including . match
		513	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
		514	or next;
		515
		516	while (my ($cpath, $cookies) = each %$paths) {
		517	next unless $cpath eq substr $path, 0, length $cpath;
		518
		519	while (my ($cookie, $kv) = each %$cookies) {
		520	next if $scheme ne "https" && exists $kv->{secure};
		521
		522	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		523	delete $cookies->{$cookie};
		524	next;
		525	}
		526
		527	my $value = $kv->{value};
		528
		529	if ($value =~ /[=;,[:space:]]/) {
		530	$value =~ s/([\\"])/\\$1/g;
		531	$value = "\"$value\"";
		532	}
		533
		534	push @cookies, "$cookie=$value";
		535	}
		536	}
		537	}
		538
		539	\@cookies
		540	}
		541
		542	# parse set_cookie header into jar
		543	sub cookie_jar_set_cookie($$$$) {
		544	my ($jar, $set_cookie, $host, $date) = @_;
		545
		546	%$jar = () if $jar->{version} != 2;
		547
		548	my $anow = int AE::now;
		549	my $snow; # server-now
		550
		551	for ($set_cookie) {
		552	# parse NAME=VALUE
		553	my @kv;
		554
		555	# expires is not http-compliant in the original cookie-spec,
		556	# we support the official date format and some extensions
		557	while (
		558	m{
		559	\G\s*
		560	(?:
		561	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		562	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
		563	)
		564	}gcxsi
		565	) {
		566	my $name = $2;
		567	my $value = $4;
		568
		569	if (defined $1) {
		570	# expires
		571	$name = "expires";
		572	$value = $1;
		573	} elsif (defined $3) {
		574	# quoted
		575	$value = $3;
		576	$value =~ s/\\(.)/$1/gs;
		577	}
		578
		579	push @kv, @kv ? lc $name : $name, $value;
		580
		581	last unless /\G\s*;/gc;
		582	}
		583
		584	last unless @kv;
		585
		586	my $name = shift @kv;
		587	my %kv = (value => shift @kv, @kv);
		588
		589	if (exists $kv{"max-age"}) {
		590	$kv{_expires} = $anow + delete $kv{"max-age"};
		591	} elsif (exists $kv{expires}) {
		592	$snow ||= parse_date ($date) || $anow;
		593	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		594	} else {
		595	delete $kv{_expires};
		596	}
		597
		598	my $cdom;
		599	my $cpath = (delete $kv{path}) || "/";
		600
		601	if (exists $kv{domain}) {
		602	$cdom = $kv{domain};
		603
		604	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		605
		606	next if $cdom =~ /\.$/;
		607
		608	# this is not rfc-like and not netscape-like. go figure.
		609	my $ndots = $cdom =~ y/.//;
		610	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		611
		612	$cdom = substr $cdom, 1; # remove initial .
		613	} else {
		614	$cdom = $host;
		615	}
		616
		617	# store it
		618	$jar->{version} = 2;
		619	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		620
		621	redo if /\G\s*,/gc;
		622	}
		623	}
		624
		625	#############################################################################
		626	# keepalive/persistent connection cache
		627
		628	# fetch a connection from the keepalive cache
		629	sub ka_fetch($) {
		630	my $ka_key = shift;
		631
		632	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		633	delete $KA_CACHE{$ka_key}
		634	unless @{ $KA_CACHE{$ka_key} };
		635
		636	$hdl
		637	}
		638
		639	sub ka_store($$) {
		640	my ($ka_key, $hdl) = @_;
		641
		642	my $kaa = $KA_CACHE{$ka_key} ||= [];
		643
		644	my $destroy = sub {
		645	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		646
		647	$hdl->destroy;
		648
		649	@ka
		650	? $KA_CACHE{$ka_key} = \@ka
		651	: delete $KA_CACHE{$ka_key};
		652	};
		653
		654	# on error etc., destroy
		655	$hdl->on_error ($destroy);
		656	$hdl->on_eof ($destroy);
		657	$hdl->on_read ($destroy);
		658	$hdl->timeout ($PERSISTENT_TIMEOUT);
		659
		660	push @$kaa, $hdl;
		661	shift @$kaa while @$kaa > $MAX_PER_HOST;
		662	}
		663
		664	#############################################################################
		665	# utilities
		666
		667	# continue to parse $_ for headers and place them into the arg
		668	sub _parse_hdr() {
		669	my %hdr;
		670
		671	# things seen, not parsed:
		672	# p3pP="NON CUR OTPi OUR NOR UNI"
		673
		674	$hdr{lc $1} .= ",$2"
		675	while /\G
		676	([^:\000-\037]*):
		677	[\011\040]*
		678	((?: [^\012]+ | \012[\011\040] )*)
		679	\012
		680	/gxc;
		681
		682	/\G$/
		683	or return;
		684
		685	# remove the "," prefix we added to all headers above
		686	substr $_, 0, 1, ""
		687	for values %hdr;
		688
		689	\%hdr
		690	}
		691
		692	#############################################################################
		693	# http_get
		694
		695	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		696
		697	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		698	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		699
		700	# maybe it should just become a normal object :/
		701
		702	sub _destroy_state(\%) {
		703	my ($state) = @_;
		704
		705	$state->{handle}->destroy if $state->{handle};
		706	%$state = ();
		707	}
		708
		709	sub _error(\%$$) {
		710	my ($state, $cb, $hdr) = @_;
		711
		712	&_destroy_state ($state);
		713
		714	$cb->(undef, $hdr);
		715	()
		716	}
		717
		718	our %IDEMPOTENT = (
		719	DELETE => 1,
		720	GET => 1,
		721	QUERY => 1,
		722	HEAD => 1,
		723	OPTIONS => 1,
		724	PUT => 1,
		725	TRACE => 1,
		726
		727	ACL => 1,
		728	"BASELINE-CONTROL" => 1,
		729	BIND => 1,
		730	CHECKIN => 1,
		731	CHECKOUT => 1,
		732	COPY => 1,
		733	LABEL => 1,
		734	LINK => 1,
		735	MERGE => 1,
		736	MKACTIVITY => 1,
		737	MKCALENDAR => 1,
		738	MKCOL => 1,
		739	MKREDIRECTREF => 1,
		740	MKWORKSPACE => 1,
		741	MOVE => 1,
		742	ORDERPATCH => 1,
		743	PRI => 1,
		744	PROPFIND => 1,
		745	PROPPATCH => 1,
		746	REBIND => 1,
		747	REPORT => 1,
		748	SEARCH => 1,
		749	UNBIND => 1,
		750	UNCHECKOUT => 1,
		751	UNLINK => 1,
		752	UNLOCK => 1,
		753	UPDATE => 1,
		754	UPDATEREDIRECTREF => 1,
		755	"VERSION-CONTROL" => 1,
		756	);
		757
135	sub http_request($$$;@) {	758	sub http_request($$@) {
136	my $cb = pop;	759	my $cb = pop;
137	my ($method, $url, %arg) = @_;	760	my ($method, $url, %arg) = @_;
138		761
139	my %hdr;	762	my %hdr;
140		763
		764	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		765	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
		766
141	$method = uc $method;	767	$method = uc $method;
142		768
143	if (my $hdr = delete $arg{headers}) {	769	if (my $hdr = $arg{headers}) {
144	while (my ($k, $v) = each %$hdr) {	770	while (my ($k, $v) = each %$hdr) {
145	$hdr{lc $k} = $v;	771	$hdr{lc $k} = $v;
146	}	772	}
147	}	773	}
148		774
149	my $proxy = $arg{proxy} || $PROXY;	775	# pseudo headers for all subsequent responses
		776	my @pseudo = (URL => $url);
		777	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		778
		779	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
		780
		781	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
		782	if $recurse < 0;
		783
		784	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
150	my $timeout = $arg{timeout} || $TIMEOUT;	785	my $timeout = $arg{timeout} || $TIMEOUT;
151	my $recurse = exists $arg{recurse} ? $arg{recurse} : $MAX_RECURSE;
152		786
153	$hdr{"user-agent"} ||= $USERAGENT;	787	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
		788	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
154		789
155	my ($host, $port, $path, $scheme);	790	$uscheme = lc $uscheme;
		791
		792	my $uport = $uscheme eq "http" ? 80
		793	: $uscheme eq "https" ? 443
		794	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
		795
		796	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
		797	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
		798
		799	my $uhost = lc $1;
		800	$uport = $2 if defined $2;
		801
		802	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		803	unless exists $hdr{host};
		804
		805	$uhost =~ s/^\[(.*)\]$/$1/;
		806	$upath .= $query if length $query;
		807
		808	$upath =~ s%^/?%/%;
		809
		810	# cookie processing
		811	if (my $jar = $arg{cookie_jar}) {
		812	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
		813
		814	$hdr{cookie} = join "; ", @$cookies
		815	if @$cookies;
		816	}
		817
		818	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
156		819
157	if ($proxy) {	820	if ($proxy) {
158	($host, $port, $scheme) = @$proxy;	821	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
159	$path = $url;	822
		823	$rscheme = "http" unless defined $rscheme;
		824
		825	# don't support https requests over https-proxy transport,
		826	# can't be done with tls as spec'ed, unless you double-encrypt.
		827	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		828
		829	$rhost = lc $rhost;
		830	$rscheme = lc $rscheme;
160	} else {	831	} else {
161	($scheme, my $authority, $path, my $query, my $fragment) =	832	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
162	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;
163
164	$port = $scheme eq "http" ? 80
165	: $scheme eq "https" ? 443
166	: croak "$url: only http and https URLs supported";
167
168	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
169	or croak "$authority: unparsable URL";
170
171	$host = $1;
172	$port = $2 if defined $2;
173
174	$host =~ s/^\[(.*)\]$/$1/;
175	$path .= "?$query" if length $query;
176
177	$path = "/" unless $path;
178
179	$hdr{host} = $host = lc $host;
180	}	833	}
181		834
182	$scheme = lc $scheme;	835	# leave out fragment and query string, just a heuristic
		836	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		837	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
183		838
184	my %state;
185
186	$state{body} = delete $arg{body};
187
188	$hdr{"content-length"} = length $state{body};	839	$hdr{"content-length"} = length $arg{body}
		840	if length $arg{body} || $method ne "GET";
189		841
190	$state{connect_guard} = AnyEvent::Socket::tcp_connect $host, $port, sub {	842	my $idempotent = $IDEMPOTENT{$method};
		843
		844	# default value for keepalive is true iff the request is for an idempotent method
		845	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		846	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		847	my $was_persistent; # true if this is actually a recycled connection
		848
		849	# the key to use in the keepalive cache
		850	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		851
		852	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
		853	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		854
		855	my %state = (connect_guard => 1);
		856
		857	my $ae_error = 595; # connecting
		858
		859	# handle actual, non-tunneled, request
		860	my $handle_actual_request = sub {
		861	$ae_error = 596; # request phase
		862
		863	my $hdl = $state{handle};
		864
		865	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		866
		867	# send request
		868	$hdl->push_write (
		869	"$method $rpath HTTP/1.1\015\012"
		870	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		871	. "\015\012"
		872	. $arg{body}
		873	);
		874
		875	# return if error occurred during push_write()
		876	return unless %state;
		877
		878	# reduce memory usage, save a kitten, also re-use it for the response headers.
		879	%hdr = ();
		880
		881	# status line and headers
		882	$state{read_response} = sub {
		883	return unless %state;
		884
		885	for ("$_[1]") {
		886	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		887
		888	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		889	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		890
		891	# 100 Continue handling
		892	# should not happen as we don't send expect: 100-continue,
		893	# but we handle it just in case.
		894	# since we send the request body regardless, if we get an error
		895	# we are out of-sync, which we currently do NOT handle correctly.
		896	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		897	if $2 eq 100;
		898
		899	push @pseudo,
		900	HTTPVersion => $1,
		901	Status => $2,
		902	Reason => $3,
		903	;
		904
		905	my $hdr = _parse_hdr
		906	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		907
		908	%hdr = (%$hdr, @pseudo);
		909	}
		910
		911	# redirect handling
		912	# relative uri handling forced by microsoft and other shitheads.
		913	# we give our best and fall back to URI if available.
		914	if (exists $hdr{location}) {
		915	my $loc = $hdr{location};
		916
		917	if ($loc =~ m%^//%) { # //
		918	$loc = "$uscheme:$loc";
		919
		920	} elsif ($loc eq "") {
		921	$loc = $url;
		922
		923	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		924	$loc =~ s/^\.\/+//;
		925
		926	if ($loc !~ m%^[.?#]%) {
		927	my $prefix = "$uscheme://$uauthority";
		928
		929	unless ($loc =~ s/^\///) {
		930	$prefix .= $upath;
		931	$prefix =~ s/\/[^\/]*$//;
		932	}
		933
		934	$loc = "$prefix/$loc";
		935
		936	} elsif (eval { require URI }) { # uri
		937	$loc = URI->new_abs ($loc, $url)->as_string;
		938
		939	} else {
		940	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		941	#$hdr{Status} = 599;
		942	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		943	#$recurse = 0;
		944	}
		945	}
		946
		947	$hdr{location} = $loc;
		948	}
		949
		950	my $redirect;
		951
		952	if ($recurse) {
		953	my $status = $hdr{Status};
		954
		955	# industry standard is to redirect POST as GET for
		956	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		957	# also, the UA should ask the user for 301 and 307 and POST,
		958	# industry standard seems to be to simply follow.
		959	# we go with the industry standard. 308 is defined
		960	# by rfc7538
		961	if ($status == 301 or $status == 302 or $status == 303) {
		962	$redirect = 1;
		963	# HTTP/1.1 is unclear on how to mutate the method
		964	unless ($method eq "HEAD") {
		965	$method = "GET";
		966	delete $arg{body};
		967	}
		968	} elsif ($status == 307 or $status == 308) {
		969	$redirect = 1;
		970	}
		971	}
		972
		973	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		974	if ($state{handle}) {
		975	# handle keepalive
		976	if (
		977	$persistent
		978	&& $_[3]
		979	&& ($hdr{HTTPVersion} < 1.1
		980	? $hdr{connection} =~ /\bkeep-?alive\b/i
		981	: $hdr{connection} !~ /\bclose\b/i)
		982	) {
		983	ka_store $ka_key, delete $state{handle};
		984	} else {
		985	# no keepalive, destroy the handle
		986	$state{handle}->destroy;
		987	}
		988	}
		989
		990	%state = ();
		991
		992	if (defined $_[1]) {
		993	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		994	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		995	}
		996
		997	# set-cookie processing
		998	if ($arg{cookie_jar}) {
		999	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		1000	}
		1001
		1002	if ($redirect && exists $hdr{location}) {
		1003	# we ignore any errors, as it is very common to receive
		1004	# Content-Length != 0 but no actual body
		1005	# we also access %hdr, as $_[1] might be an erro
		1006	$state{recurse} =
		1007	http_request (
		1008	$method => $hdr{location},
		1009	%arg,
		1010	recurse => $recurse - 1,
		1011	Redirect => [$_[0], \%hdr],
		1012	sub {
		1013	%state = ();
		1014	&$cb
		1015	},
		1016	);
		1017	} else {
		1018	$cb->($_[0], \%hdr);
		1019	}
		1020	};
		1021
		1022	$ae_error = 597; # body phase
		1023
		1024	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1025
		1026	my $len = $chunked ? undef : $hdr{"content-length"};
		1027
		1028	# body handling, many different code paths
		1029	# - no body expected
		1030	# - want_body_handle
		1031	# - te chunked
		1032	# - 2x length known (with or without on_body)
		1033	# - 2x length not known (with or without on_body)
		1034	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1035	$finish->(undef, 598 => "Request cancelled by on_header");
		1036	} elsif (
		1037	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1038	or $method eq "HEAD"
		1039	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1040	) {
		1041	# no body
		1042	$finish->("", undef, undef, 1);
		1043
		1044	} elsif (!$redirect && $arg{want_body_handle}) {
		1045	$_[0]->on_eof (undef);
		1046	$_[0]->on_error (undef);
		1047	$_[0]->on_read (undef);
		1048
		1049	$finish->(delete $state{handle});
		1050
		1051	} elsif ($chunked) {
		1052	my $cl = 0;
		1053	my $body = "";
		1054	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1055
		1056	$state{read_chunk} = sub {
		1057	$_[1] =~ /^([0-9a-fA-F]+)/
		1058	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1059
		1060	my $len = hex $1;
		1061
		1062	if ($len) {
		1063	$cl += $len;
		1064
		1065	$_[0]->push_read (chunk => $len, sub {
		1066	$on_body->($_[1], \%hdr)
		1067	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1068
		1069	$_[0]->push_read (line => sub {
		1070	length $_[1]
		1071	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1072	$_[0]->push_read (line => $state{read_chunk});
		1073	});
		1074	});
		1075	} else {
		1076	$hdr{"content-length"} ||= $cl;
		1077
		1078	$_[0]->push_read (line => $qr_nlnl, sub {
		1079	if (length $_[1]) {
		1080	for ("$_[1]") {
		1081	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1082
		1083	my $hdr = _parse_hdr
		1084	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1085
		1086	%hdr = (%hdr, %$hdr);
		1087	}
		1088	}
		1089
		1090	$finish->($body, undef, undef, 1);
		1091	});
		1092	}
		1093	};
		1094
		1095	$_[0]->push_read (line => $state{read_chunk});
		1096
		1097	} elsif (!$redirect && $arg{on_body}) {
		1098	if (defined $len) {
		1099	$_[0]->on_read (sub {
		1100	$len -= length $_[0]{rbuf};
		1101
		1102	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1103	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1104
		1105	$len > 0
		1106	or $finish->("", undef, undef, 1);
		1107	});
		1108	} else {
		1109	$_[0]->on_eof (sub {
		1110	$finish->("");
		1111	});
		1112	$_[0]->on_read (sub {
		1113	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1114	or $finish->(undef, 598 => "Request cancelled by on_body");
		1115	});
		1116	}
		1117	} else {
		1118	$_[0]->on_eof (undef);
		1119
		1120	if (defined $len) {
		1121	$_[0]->on_read (sub {
		1122	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1123	if $len <= length $_[0]{rbuf};
		1124	});
		1125	} else {
		1126	$_[0]->on_error (sub {
		1127	($! == Errno::EPIPE || !$!)
		1128	? $finish->(delete $_[0]{rbuf})
		1129	: $finish->(undef, $ae_error => $_[2]);
		1130	});
		1131	$_[0]->on_read (sub { });
		1132	}
		1133	}
		1134	};
		1135
		1136	# if keepalive is enabled, then the server closing the connection
		1137	# before a response can happen legally - we retry on idempotent methods.
		1138	if ($was_persistent && $idempotent) {
		1139	my $old_eof = $hdl->{on_eof};
		1140	$hdl->{on_eof} = sub {
		1141	_destroy_state %state;
		1142
		1143	%state = ();
		1144	$state{recurse} =
		1145	http_request (
		1146	$method => $url,
		1147	%arg,
		1148	recurse => $recurse - 1,
		1149	persistent => 0,
		1150	sub {
		1151	%state = ();
		1152	&$cb
		1153	}
		1154	);
		1155	};
		1156	$hdl->on_read (sub {
		1157	return unless %state;
		1158
		1159	# as soon as we receive something, a connection close
		1160	# once more becomes a hard error
		1161	$hdl->{on_eof} = $old_eof;
		1162	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1163	});
		1164	} else {
		1165	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1166	}
		1167	};
		1168
		1169	my $prepare_handle = sub {
		1170	my ($hdl) = $state{handle};
		1171
		1172	$hdl->on_error (sub {
		1173	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1174	});
		1175	$hdl->on_eof (sub {
		1176	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1177	});
		1178	$hdl->timeout_reset;
		1179	$hdl->timeout ($timeout);
		1180	};
		1181
		1182	# connected to proxy (or origin server)
		1183	my $connect_cb = sub {
191	$state{fh} = shift	1184	my $fh = shift
192	or return $cb->(undef, { Status => 599, Reason => "$!" });	1185	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
193		1186
194	delete $state{connect_guard}; # reduce memory usage, save a tree	1187	return unless delete $state{connect_guard};
195		1188
196	# get handle	1189	# get handle
197	$state{handle} = new AnyEvent::Handle	1190	$state{handle} = new AnyEvent::Handle
198	fh => $state{fh},	1191	%{ $arg{handle_params} },
199	($scheme eq "https" ? (tls => "connect") : ());	1192	fh => $fh,
200		1193	peername => $uhost,
201	# limit the number of persistent connections	1194	tls_ctx => $arg{tls_ctx},
202	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
203	++$KA_COUNT{$_[1]};
204	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
205	$hdr{connection} = "keep-alive";
206	delete $hdr{connection}; # keep-alive not yet supported
207	} else {
208	delete $hdr{connection};
209	}
210
211	# (re-)configure handle
212	$state{handle}->timeout ($timeout);
213	$state{handle}->on_error (sub {
214	%state = ();
215	$cb->(undef, { Status => 599, Reason => "$!" });
216	});
217	$state{handle}->on_eof (sub {
218	%state = ();
219	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file" });
220	});
221
222	# send request
223	$state{handle}->push_write (
224	"$method $path HTTP/1.0\015\012"
225	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
226	. "\015\012"
227	. (delete $state{body})
228	);	1195	;
229		1196
230	%hdr = (); # reduce memory usage, save a kitten	1197	$prepare_handle->();
231		1198
232	# status line	1199	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1200
		1201	# now handle proxy-CONNECT method
		1202	if ($proxy && $uscheme eq "https") {
		1203	# oh dear, we have to wrap it into a connect request
		1204
		1205	my $auth = exists $hdr{"proxy-authorization"}
		1206	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1207	: "";
		1208
		1209	# maybe re-use $uauthority with patched port?
		1210	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
233	$state{handle}->push_read (line => qr/\015?\012/, sub {	1211	$state{handle}->push_read (line => $qr_nlnl, sub {
234	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix	1212	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
235	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])" }));	1213	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
236		1214
237	my %hdr = ( # response headers	1215	if ($2 == 200) {
238	HTTPVersion => ",$1",	1216	$rpath = $upath;
239	Status => ",$2",	1217	$handle_actual_request->();
240	Reason => ",$3",
241	);
242
243	# headers, could be optimized a bit
244	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
245	for ("$_[1]\012") {
246	# we support spaces in field names, as lotus domino
247	# creates them.
248	$hdr{lc $1} .= ",$2"
249	while /\G
250	([^:\000-\037]+):
251	[\011\040]*
252	((?: [^\015\012]+ | \015?\012[\011\040] )*)
253	\015?\012
254	/gxc;
255
256	/\G$/
257	or return $cb->(undef, { Status => 599, Reason => "garbled response headers" });
258	}
259
260	substr $_, 0, 1, ""
261	for values %hdr;
262
263	if ($method eq "HEAD") {
264	%state = ();
265	$cb->(undef, \%hdr);
266	} else {	1218	} else {
267	if (exists $hdr{"content-length"}) {	1219	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
268	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
269	# could cache persistent connection now
270	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
271	# but we don't, due to misdesigns, this is annoyingly complex
272	};
273
274	%state = ();
275	$cb->($_[1], \%hdr);
276	});
277	} else {
278	# too bad, need to read until we get an error or EOF,
279	# no way to detect winged data.
280	$_[0]->on_error (sub {
281	%state = ();
282	$cb->($_[0]{rbuf}, \%hdr);
283	});
284	$_[0]->on_eof (undef);
285	$_[0]->on_read (sub { });
286	}
287	}	1220	}
288	});	1221	});
		1222	} else {
		1223	delete $hdr{"proxy-authorization"} unless $proxy;
		1224
		1225	$handle_actual_request->();
289	});	1226	}
290	}, sub {
291	$timeout
292	};	1227	};
293		1228
		1229	_get_slot $uhost, sub {
		1230	$state{slot_guard} = shift;
		1231
		1232	return unless $state{connect_guard};
		1233
		1234	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1235	# on a keepalive request (in theory, this should be a separate config option).
		1236	if ($persistent && $KA_CACHE{$ka_key}) {
		1237	$was_persistent = 1;
		1238
		1239	$state{handle} = ka_fetch $ka_key;
		1240	# $state{handle}->destroyed
		1241	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1242	$prepare_handle->();
		1243	# $state{handle}->destroyed
		1244	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1245	$rpath = $upath;
		1246	$handle_actual_request->();
		1247
		1248	} else {
		1249	my $tcp_connect = $arg{tcp_connect}
		1250	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1251
		1252	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
		1253	}
		1254	};
		1255
294	defined wantarray && AnyEvent::Util::guard { %state = () }	1256	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
295	}	1257	}
296		1258
297	sub http_get($$;@) {	1259	sub http_get($@) {
298	unshift @_, "GET";	1260	unshift @_, "GET";
299	&http_request	1261	&http_request
300	}	1262	}
301		1263
302	sub http_head($$;@) {	1264	sub http_head($@) {
303	unshift @_, "HEAD";	1265	unshift @_, "HEAD";
304	&http_request	1266	&http_request
305	}	1267	}
306		1268
307	sub http_post($$$;@) {	1269	sub http_post($$@) {
		1270	my $url = shift;
308	unshift @_, "POST", "body";	1271	unshift @_, "POST", $url, "body";
309	&http_request	1272	&http_request
310	}	1273	}
311		1274
		1275	=back
		1276
		1277	=head2 DNS CACHING
		1278
		1279	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1280	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1281	hostnames. The latter is a simple stub resolver and does no caching
		1282	on its own. If you want DNS caching, you currently have to provide
		1283	your own default resolver (by storing a suitable resolver object in
		1284	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1285
312	=head2 GLOBAL FUNCTIONS AND VARIABLES	1286	=head2 GLOBAL FUNCTIONS AND VARIABLES
313		1287
314	=over 4	1288	=over 4
315		1289
316	=item AnyEvent::HTTP::set_proxy "proxy-url"	1290	=item AnyEvent::HTTP::set_proxy "proxy-url"
317		1291
318	Sets the default proxy server to use. The proxy-url must begin with a	1292	Sets the default proxy server to use. The proxy-url must begin with a
319	string of the form C<http://host:port> (optionally C<https:...>).	1293	string of the form C<http://host:port>, croaks otherwise.
		1294
		1295	To clear an already-set proxy, use C<undef>.
		1296
		1297	When AnyEvent::HTTP is loaded for the first time it will query the
		1298	default proxy from the operating system, currently by looking at
		1299	C<$ENV{http_proxy>}.
		1300
		1301	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1302
		1303	Remove all cookies from the cookie jar that have been expired. If
		1304	C<$session_end> is given and true, then additionally remove all session
		1305	cookies.
		1306
		1307	You should call this function (with a true C<$session_end>) before you
		1308	save cookies to disk, and you should call this function after loading them
		1309	again. If you have a long-running program you can additionally call this
		1310	function from time to time.
		1311
		1312	A cookie jar is initially an empty hash-reference that is managed by this
		1313	module. Its format is subject to change, but currently it is as follows:
		1314
		1315	The key C<version> has to contain C<2>, otherwise the hash gets
		1316	cleared. All other keys are hostnames or IP addresses pointing to
		1317	hash-references. The key for these inner hash references is the
		1318	server path for which this cookie is meant, and the values are again
		1319	hash-references. Each key of those hash-references is a cookie name, and
		1320	the value, you guessed it, is another hash-reference, this time with the
		1321	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1322	which have been replaced by a C<_expires> key that contains the cookie
		1323	expiry timestamp. Session cookies are indicated by not having an
		1324	C<_expires> key.
		1325
		1326	Here is an example of a cookie jar with a single cookie, so you have a
		1327	chance of understanding the above paragraph:
		1328
		1329	{
		1330	version => 2,
		1331	"10.0.0.1" => {
		1332	"/" => {
		1333	"mythweb_id" => {
		1334	_expires => 1293917923,
		1335	value => "ooRung9dThee3ooyXooM1Ohm",
		1336	},
		1337	},
		1338	},
		1339	}
		1340
		1341	=item $date = AnyEvent::HTTP::format_date $timestamp
		1342
		1343	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1344	Date (RFC 2616).
		1345
		1346	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1347
		1348	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1349	bunch of minor variations of those, and returns the corresponding POSIX
		1350	timestamp, or C<undef> if the date cannot be parsed.
320		1351
321	=item $AnyEvent::HTTP::MAX_RECURSE	1352	=item $AnyEvent::HTTP::MAX_RECURSE
322		1353
323	The default value for the C<recurse> request parameter (default: C<10>).	1354	The default value for the C<recurse> request parameter (default: C<10>).
324		1355
		1356	=item $AnyEvent::HTTP::TIMEOUT
		1357
		1358	The default timeout for connection operations (default: C<300>).
		1359
325	=item $AnyEvent::HTTP::USERAGENT	1360	=item $AnyEvent::HTTP::USERAGENT
326		1361
327	The default value for the C<User-Agent> header (the default is	1362	The default value for the C<User-Agent> header (the default is
328	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1363	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
329		1364
330	=item $AnyEvent::HTTP::MAX_PERSISTENT	1365	=item $AnyEvent::HTTP::MAX_PER_HOST
331		1366
332	The maximum number of persistent connections to keep open (default: 8).	1367	The maximum number of concurrent connections to the same host (identified
		1368	by the hostname). If the limit is exceeded, then additional requests
		1369	are queued until previous connections are closed. Both persistent and
		1370	non-persistent connections are counted in this limit.
333		1371
334	Not implemented currently.	1372	The default value for this is C<4>, and it is highly advisable to not
		1373	increase it much.
		1374
		1375	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1376	connections, older browsers used 2, newer ones (such as firefox 3)
		1377	typically use 6, and Opera uses 8 because like, they have the fastest
		1378	browser and give a shit for everybody else on the planet.
335		1379
336	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1380	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
337		1381
338	The maximum time to cache a persistent connection, in seconds (default: 2).	1382	The time after which idle persistent connections get closed by
		1383	AnyEvent::HTTP (default: C<3>).
339		1384
340	Not implemented currently.	1385	=item $AnyEvent::HTTP::ACTIVE
		1386
		1387	The number of active connections. This is not the number of currently
		1388	running requests, but the number of currently open and non-idle TCP
		1389	connections. This number can be useful for load-leveling.
341		1390
342	=back	1391	=back
343		1392
344	=cut	1393	=cut
345		1394
		1395	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1396	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1397
		1398	sub format_date($) {
		1399	my ($time) = @_;
		1400
		1401	# RFC 822/1123 format
		1402	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1403
		1404	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1405	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1406	$H, $M, $S;
		1407	}
		1408
		1409	sub parse_date($) {
		1410	my ($date) = @_;
		1411
		1412	my ($d, $m, $y, $H, $M, $S);
		1413
		1414	if ($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$/) {
		1415	# RFC 822/1123, required by RFC 2616 (with " ")
		1416	# cookie dates (with "-")
		1417
		1418	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1419
		1420	} elsif ($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$/) {
		1421	# RFC 850
		1422	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1423
		1424	} elsif ($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])$/) {
		1425	# ISO C's asctime
		1426	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1427	}
		1428	# other formats fail in the loop below
		1429
		1430	for (0..11) {
		1431	if ($m eq $month[$_]) {
		1432	require Time::Local;
		1433	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
		1434	}
		1435	}
		1436
		1437	undef
		1438	}
		1439
346	sub set_proxy($) {	1440	sub set_proxy($) {
347	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1441	if (length $_[0]) {
		1442	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1443	or Carp::croak "$_[0]: invalid proxy URL";
		1444	$PROXY = [$2, $3 || 3128, $1]
		1445	} else {
		1446	undef $PROXY;
		1447	}
348	}	1448	}
349		1449
350	# initialise proxy from environment	1450	# initialise proxy from environment
		1451	eval {
351	set_proxy $ENV{http_proxy};	1452	set_proxy $ENV{http_proxy};
		1453	};
		1454
		1455	=head2 SHOWCASE
		1456
		1457	This section contains some more elaborate "real-world" examples or code
		1458	snippets.
		1459
		1460	=head2 HTTP/1.1 FILE DOWNLOAD
		1461
		1462	Downloading files with HTTP can be quite tricky, especially when something
		1463	goes wrong and you want to resume.
		1464
		1465	Here is a function that initiates and resumes a download. It uses the
		1466	last modified time to check for file content changes, and works with many
		1467	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1468	on older servers.
		1469
		1470	It calls the completion callback with either C<undef>, which means a
		1471	nonretryable error occurred, C<0> when the download was partial and should
		1472	be retried, and C<1> if it was successful.
		1473
		1474	use AnyEvent::HTTP;
		1475
		1476	sub download($$$) {
		1477	my ($url, $file, $cb) = @_;
		1478
		1479	open my $fh, "+<", $file
		1480	or die "$file: $!";
		1481
		1482	my %hdr;
		1483	my $ofs = 0;
		1484
		1485	if (stat $fh and -s _) {
		1486	$ofs = -s _;
		1487	warn "-s is ", $ofs;
		1488	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1489	$hdr{"range"} = "bytes=$ofs-";
		1490	}
		1491
		1492	http_get $url,
		1493	headers => \%hdr,
		1494	on_header => sub {
		1495	my ($hdr) = @_;
		1496
		1497	if ($hdr->{Status} == 200 && $ofs) {
		1498	# resume failed
		1499	truncate $fh, $ofs = 0;
		1500	}
		1501
		1502	sysseek $fh, $ofs, 0;
		1503
		1504	1
		1505	},
		1506	on_body => sub {
		1507	my ($data, $hdr) = @_;
		1508
		1509	if ($hdr->{Status} =~ /^2/) {
		1510	length $data == syswrite $fh, $data
		1511	or return; # abort on write errors
		1512	}
		1513
		1514	1
		1515	},
		1516	sub {
		1517	my (undef, $hdr) = @_;
		1518
		1519	my $status = $hdr->{Status};
		1520
		1521	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1522	utime $time, $time, $fh;
		1523	}
		1524
		1525	if ($status == 200 || $status == 206 || $status == 416) {
		1526	# download ok || resume ok || file already fully downloaded
		1527	$cb->(1, $hdr);
		1528
		1529	} elsif ($status == 412) {
		1530	# file has changed while resuming, delete and retry
		1531	unlink $file;
		1532	$cb->(0, $hdr);
		1533
		1534	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1535	# retry later
		1536	$cb->(0, $hdr);
		1537
		1538	} else {
		1539	$cb->(undef, $hdr);
		1540	}
		1541	}
		1542	;
		1543	}
		1544
		1545	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1546	if ($_[0]) {
		1547	print "OK!\n";
		1548	} elsif (defined $_[0]) {
		1549	print "please retry later\n";
		1550	} else {
		1551	print "ERROR\n";
		1552	}
		1553	};
		1554
		1555	=head3 SOCKS PROXIES
		1556
		1557	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1558	compile your perl to support socks, or use an external program such as
		1559	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1560	transparently.
		1561
		1562	Alternatively, for AnyEvent::HTTP only, you can use your own
		1563	C<tcp_connect> function that does the proxy handshake - here is an example
		1564	that works with socks4a proxies:
		1565
		1566	use Errno;
		1567	use AnyEvent::Util;
		1568	use AnyEvent::Socket;
		1569	use AnyEvent::Handle;
		1570
		1571	# host, port and username of/for your socks4a proxy
		1572	my $socks_host = "10.0.0.23";
		1573	my $socks_port = 9050;
		1574	my $socks_user = "";
		1575
		1576	sub socks4a_connect {
		1577	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1578
		1579	my $hdl = new AnyEvent::Handle
		1580	connect => [$socks_host, $socks_port],
		1581	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1582	on_error => sub { $connect_cb->() },
		1583	;
		1584
		1585	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1586
		1587	$hdl->push_read (chunk => 8, sub {
		1588	my ($hdl, $chunk) = @_;
		1589	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1590
		1591	if ($status == 0x5a) {
		1592	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1593	} else {
		1594	$! = Errno::ENXIO; $connect_cb->();
		1595	}
		1596	});
		1597
		1598	$hdl
		1599	}
		1600
		1601	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1602	possibly after switching off other proxy types:
		1603
		1604	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1605
		1606	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1607	my ($data, $headers) = @_;
		1608	...
		1609	};
352		1610
353	=head1 SEE ALSO	1611	=head1 SEE ALSO
354		1612
355	L<AnyEvent>.	1613	L<AnyEvent>.
356		1614
357	=head1 AUTHOR	1615	=head1 AUTHOR
358		1616
359	Marc Lehmann <schmorp@schmorp.de>	1617	Marc Lehmann <schmorp@schmorp.de>
360	http://home.schmorp.de/	1618	http://home.schmorp.de/
		1619
		1620	With many thanks to Дмитрий Шалашов, who provided countless
		1621	testcases and bugreports.
361		1622
362	=cut	1623	=cut
363		1624
364	1	1625	1
365		1626

Diff Legend

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