ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.5 by root, Wed Jun 4 12:03:47 2008 UTC vs.
Revision 1.136 by root, Wed Oct 16 01:20:02 2019 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 = 2.24;
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, $headers)
63		78
64	Executes an HTTP-POST request with a requets body of C<$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 has 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<< $headers->{Status} >> will be
83	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection establishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
		139
		140	A typical callback might look like this:
		141
		142	sub {
		143	my ($body, $hdr) = @_;
		144
		145	if ($hdr->{Status} =~ /^2/) {
		146	... everything should be ok
		147	} else {
		148	print "error, $hdr->{Status} $hdr->{Reason}\n";
		149	}
		150	}
84		151
85	Additional parameters are key-value pairs, and are fully optional. They	152	Additional parameters are key-value pairs, and are fully optional. They
86	include:	153	include:
87		154
88	=over 4	155	=over 4
89		156
90	=item recurse => $count (default: $MAX_RECURSE)	157	=item recurse => $count (default: $MAX_RECURSE)
91		158
92	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
93	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.
94		166
95	=item headers => hashref	167	=item headers => hashref
96		168
97	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.
98		181
99	=item timeout => $seconds	182	=item timeout => $seconds
100		183
101	The time-out to use for various stages - each connect attempt will reset	184	The time-out to use for various stages - each connect attempt will reset
102	the timeout, as will read or write activity. Default timeout is 5 minutes.	185	the timeout, as will read or write activity, i.e. this is not an overall
		186	timeout.
		187
		188	Default timeout is 5 minutes.
103		189
104	=item proxy => [$host, $port[, $scheme]] or undef	190	=item proxy => [$host, $port[, $scheme]] or undef
105		191
106	Use the given http proxy for all requests. If not specified, then the	192	Use the given http proxy for all requests, or no proxy if C<undef> is
107	default proxy (as specified by C<$ENV{http_proxy}>) is used.	193	used.
108		194
109	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.
110	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>.
111		207
112	=item body => $string	208	=item body => $string
113		209
114	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
115	this module might offer more options).	211	this module might offer more options).
116		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 => $boolean
		374
		375	Only used when C<persistent> is also true. This parameter decides whether
		376	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		377	(as opposed to only a HTTP/1.1 persistent connection).
		378
		379	The default is true, except when using a proxy, in which case it defaults
		380	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		381
		382	=item handle_params => { key => value ... }
		383
		384	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		385	constructor that is called - not all requests will create a handle, and
		386	sometimes more than one is created, so this parameter is only good for
		387	setting hints.
		388
		389	Example: set the maximum read size to 4096, to potentially conserve memory
		390	at the cost of speed.
		391
		392	handle_params => {
		393	max_read_size => 4096,
		394	},
		395
117	=back	396	=back
118		397
119	=back	398	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		399	the response body.
		400
		401	http_request GET => "http://www.nethype.de/", sub {
		402	my ($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 $request;
120		429
121	=cut	430	=cut
122		431
		432	#############################################################################
		433	# wait queue/slots
		434
		435	sub _slot_schedule;
		436	sub _slot_schedule($) {
		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(\%$$) {
		710	my ($state, $cb, $hdr) = @_;
		711
		712	&_destroy_state ($state);
		713
		714	$cb->(undef, $hdr);
		715	()
		716	}
		717
		718	our %IDEMPOTENT = (
		719	DELETE => 1,
		720	GET => 1,
		721	HEAD => 1,
		722	OPTIONS => 1,
		723	PUT => 1,
		724	TRACE => 1,
		725
		726	ACL => 1,
		727	"BASELINE-CONTROL" => 1,
		728	BIND => 1,
		729	CHECKIN => 1,
		730	CHECKOUT => 1,
		731	COPY => 1,
		732	LABEL => 1,
		733	LINK => 1,
		734	MERGE => 1,
		735	MKACTIVITY => 1,
		736	MKCALENDAR => 1,
		737	MKCOL => 1,
		738	MKREDIRECTREF => 1,
		739	MKWORKSPACE => 1,
		740	MOVE => 1,
		741	ORDERPATCH => 1,
		742	PROPFIND => 1,
		743	PROPPATCH => 1,
		744	REBIND => 1,
		745	REPORT => 1,
		746	SEARCH => 1,
		747	UNBIND => 1,
		748	UNCHECKOUT => 1,
		749	UNLINK => 1,
		750	UNLOCK => 1,
		751	UPDATE => 1,
		752	UPDATEREDIRECTREF => 1,
		753	"VERSION-CONTROL" => 1,
		754	);
		755
123	sub http_request($$$;@) {	756	sub http_request($$@) {
124	my $cb = pop;	757	my $cb = pop;
125	my ($method, $url, %arg) = @_;	758	my ($method, $url, %arg) = @_;
126		759
127	my %hdr;	760	my %hdr;
128		761
		762	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		763	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
		764
129	$method = uc $method;	765	$method = uc $method;
130		766
131	if (my $hdr = delete $arg{headers}) {	767	if (my $hdr = $arg{headers}) {
132	while (my ($k, $v) = each %$hdr) {	768	while (my ($k, $v) = each %$hdr) {
133	$hdr{lc $k} = $v;	769	$hdr{lc $k} = $v;
134	}	770	}
135	}	771	}
136		772
137	my $proxy = $arg{proxy} || $PROXY;	773	# pseudo headers for all subsequent responses
		774	my @pseudo = (URL => $url);
		775	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		776
		777	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
		778
		779	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
		780	if $recurse < 0;
		781
		782	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
138	my $timeout = $arg{timeout} || $TIMEOUT;	783	my $timeout = $arg{timeout} || $TIMEOUT;
139	my $recurse = exists $arg{recurse} ? $arg{recurse} : $MAX_RECURSE;
140		784
141	$hdr{"user-agent"} ||= $USERAGENT;	785	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
		786	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
142		787
143	my ($host, $port, $path, $scheme);	788	$uscheme = lc $uscheme;
		789
		790	my $uport = $uscheme eq "http" ? 80
		791	: $uscheme eq "https" ? 443
		792	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
		793
		794	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
		795	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
		796
		797	my $uhost = lc $1;
		798	$uport = $2 if defined $2;
		799
		800	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		801	unless exists $hdr{host};
		802
		803	$uhost =~ s/^\[(.*)\]$/$1/;
		804	$upath .= $query if length $query;
		805
		806	$upath =~ s%^/?%/%;
		807
		808	# cookie processing
		809	if (my $jar = $arg{cookie_jar}) {
		810	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
		811
		812	$hdr{cookie} = join "; ", @$cookies
		813	if @$cookies;
		814	}
		815
		816	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
144		817
145	if ($proxy) {	818	if ($proxy) {
146	($host, $port, $scheme) = @$proxy;	819	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
147	$path = $url;	820
		821	$rscheme = "http" unless defined $rscheme;
		822
		823	# don't support https requests over https-proxy transport,
		824	# can't be done with tls as spec'ed, unless you double-encrypt.
		825	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		826
		827	$rhost = lc $rhost;
		828	$rscheme = lc $rscheme;
148	} else {	829	} else {
149	($scheme, my $authority, $path, my $query, my $fragment) =	830	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
150	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;
151
152	$port = $scheme eq "http" ? 80
153	: $scheme eq "https" ? 443
154	: croak "$url: only http and https URLs supported";
155
156	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
157	or croak "$authority: unparsable URL";
158
159	$host = $1;
160	$port = $2 if defined $2;
161
162	$host =~ s/^\[(.*)\]$/$1/;
163	$path .= "?$query" if length $query;
164
165	$path = "/" unless $path;
166
167	$hdr{host} = $host = lc $host;
168	}	831	}
169		832
170	$scheme = lc $scheme;	833	# leave out fragment and query string, just a heuristic
		834	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		835	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
171		836
172	my %state;
173
174	$state{body} = delete $arg{body};
175
176	$hdr{"content-length"} = length $state{body};	837	$hdr{"content-length"} = length $arg{body}
		838	if length $arg{body} || $method ne "GET";
177		839
178	$state{connect_guard} = AnyEvent::Socket::tcp_connect $host, $port, sub {	840	my $idempotent = $IDEMPOTENT{$method};
		841
		842	# default value for keepalive is true iff the request is for an idempotent method
		843	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		844	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		845	my $was_persistent; # true if this is actually a recycled connection
		846
		847	# the key to use in the keepalive cache
		848	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		849
		850	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
		851	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		852
		853	my %state = (connect_guard => 1);
		854
		855	my $ae_error = 595; # connecting
		856
		857	# handle actual, non-tunneled, request
		858	my $handle_actual_request = sub {
		859	$ae_error = 596; # request phase
		860
		861	my $hdl = $state{handle};
		862
		863	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		864
		865	# send request
		866	$hdl->push_write (
		867	"$method $rpath HTTP/1.1\015\012"
		868	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		869	. "\015\012"
		870	. $arg{body}
		871	);
		872
		873	# return if error occurred during push_write()
		874	return unless %state;
		875
		876	# reduce memory usage, save a kitten, also re-use it for the response headers.
		877	%hdr = ();
		878
		879	# status line and headers
		880	$state{read_response} = sub {
		881	return unless %state;
		882
		883	for ("$_[1]") {
		884	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		885
		886	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		887	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		888
		889	# 100 Continue handling
		890	# should not happen as we don't send expect: 100-continue,
		891	# but we handle it just in case.
		892	# since we send the request body regardless, if we get an error
		893	# we are out of-sync, which we currently do NOT handle correctly.
		894	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		895	if $2 eq 100;
		896
		897	push @pseudo,
		898	HTTPVersion => $1,
		899	Status => $2,
		900	Reason => $3,
		901	;
		902
		903	my $hdr = _parse_hdr
		904	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		905
		906	%hdr = (%$hdr, @pseudo);
		907	}
		908
		909	# redirect handling
		910	# relative uri handling forced by microsoft and other shitheads.
		911	# we give our best and fall back to URI if available.
		912	if (exists $hdr{location}) {
		913	my $loc = $hdr{location};
		914
		915	if ($loc =~ m%^//%) { # //
		916	$loc = "$uscheme:$loc";
		917
		918	} elsif ($loc eq "") {
		919	$loc = $url;
		920
		921	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		922	$loc =~ s/^\.\/+//;
		923
		924	if ($loc !~ m%^[.?#]%) {
		925	my $prefix = "$uscheme://$uauthority";
		926
		927	unless ($loc =~ s/^\///) {
		928	$prefix .= $upath;
		929	$prefix =~ s/\/[^\/]*$//;
		930	}
		931
		932	$loc = "$prefix/$loc";
		933
		934	} elsif (eval { require URI }) { # uri
		935	$loc = URI->new_abs ($loc, $url)->as_string;
		936
		937	} else {
		938	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		939	#$hdr{Status} = 599;
		940	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		941	#$recurse = 0;
		942	}
		943	}
		944
		945	$hdr{location} = $loc;
		946	}
		947
		948	my $redirect;
		949
		950	if ($recurse) {
		951	my $status = $hdr{Status};
		952
		953	# industry standard is to redirect POST as GET for
		954	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		955	# also, the UA should ask the user for 301 and 307 and POST,
		956	# industry standard seems to be to simply follow.
		957	# we go with the industry standard. 308 is defined
		958	# by rfc7538
		959	if ($status == 301 or $status == 302 or $status == 303) {
		960	$redirect = 1;
		961	# HTTP/1.1 is unclear on how to mutate the method
		962	unless ($method eq "HEAD") {
		963	$method = "GET";
		964	delete $arg{body};
		965	}
		966	} elsif ($status == 307 or $status == 308) {
		967	$redirect = 1;
		968	}
		969	}
		970
		971	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		972	if ($state{handle}) {
		973	# handle keepalive
		974	if (
		975	$persistent
		976	&& $_[3]
		977	&& ($hdr{HTTPVersion} < 1.1
		978	? $hdr{connection} =~ /\bkeep-?alive\b/i
		979	: $hdr{connection} !~ /\bclose\b/i)
		980	) {
		981	ka_store $ka_key, delete $state{handle};
		982	} else {
		983	# no keepalive, destroy the handle
		984	$state{handle}->destroy;
		985	}
		986	}
		987
		988	%state = ();
		989
		990	if (defined $_[1]) {
		991	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		992	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		993	}
		994
		995	# set-cookie processing
		996	if ($arg{cookie_jar}) {
		997	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		998	}
		999
		1000	if ($redirect && exists $hdr{location}) {
		1001	# we ignore any errors, as it is very common to receive
		1002	# Content-Length != 0 but no actual body
		1003	# we also access %hdr, as $_[1] might be an erro
		1004	$state{recurse} =
		1005	http_request (
		1006	$method => $hdr{location},
		1007	%arg,
		1008	recurse => $recurse - 1,
		1009	Redirect => [$_[0], \%hdr],
		1010	sub {
		1011	%state = ();
		1012	&$cb
		1013	},
		1014	);
		1015	} else {
		1016	$cb->($_[0], \%hdr);
		1017	}
		1018	};
		1019
		1020	$ae_error = 597; # body phase
		1021
		1022	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1023
		1024	my $len = $chunked ? undef : $hdr{"content-length"};
		1025
		1026	# body handling, many different code paths
		1027	# - no body expected
		1028	# - want_body_handle
		1029	# - te chunked
		1030	# - 2x length known (with or without on_body)
		1031	# - 2x length not known (with or without on_body)
		1032	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1033	$finish->(undef, 598 => "Request cancelled by on_header");
		1034	} elsif (
		1035	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1036	or $method eq "HEAD"
		1037	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1038	) {
		1039	# no body
		1040	$finish->("", undef, undef, 1);
		1041
		1042	} elsif (!$redirect && $arg{want_body_handle}) {
		1043	$_[0]->on_eof (undef);
		1044	$_[0]->on_error (undef);
		1045	$_[0]->on_read (undef);
		1046
		1047	$finish->(delete $state{handle});
		1048
		1049	} elsif ($chunked) {
		1050	my $cl = 0;
		1051	my $body = "";
		1052	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1053
		1054	$state{read_chunk} = sub {
		1055	$_[1] =~ /^([0-9a-fA-F]+)/
		1056	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1057
		1058	my $len = hex $1;
		1059
		1060	if ($len) {
		1061	$cl += $len;
		1062
		1063	$_[0]->push_read (chunk => $len, sub {
		1064	$on_body->($_[1], \%hdr)
		1065	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1066
		1067	$_[0]->push_read (line => sub {
		1068	length $_[1]
		1069	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1070	$_[0]->push_read (line => $state{read_chunk});
		1071	});
		1072	});
		1073	} else {
		1074	$hdr{"content-length"} ||= $cl;
		1075
		1076	$_[0]->push_read (line => $qr_nlnl, sub {
		1077	if (length $_[1]) {
		1078	for ("$_[1]") {
		1079	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1080
		1081	my $hdr = _parse_hdr
		1082	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1083
		1084	%hdr = (%hdr, %$hdr);
		1085	}
		1086	}
		1087
		1088	$finish->($body, undef, undef, 1);
		1089	});
		1090	}
		1091	};
		1092
		1093	$_[0]->push_read (line => $state{read_chunk});
		1094
		1095	} elsif (!$redirect && $arg{on_body}) {
		1096	if (defined $len) {
		1097	$_[0]->on_read (sub {
		1098	$len -= length $_[0]{rbuf};
		1099
		1100	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1101	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1102
		1103	$len > 0
		1104	or $finish->("", undef, undef, 1);
		1105	});
		1106	} else {
		1107	$_[0]->on_eof (sub {
		1108	$finish->("");
		1109	});
		1110	$_[0]->on_read (sub {
		1111	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1112	or $finish->(undef, 598 => "Request cancelled by on_body");
		1113	});
		1114	}
		1115	} else {
		1116	$_[0]->on_eof (undef);
		1117
		1118	if (defined $len) {
		1119	$_[0]->on_read (sub {
		1120	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1121	if $len <= length $_[0]{rbuf};
		1122	});
		1123	} else {
		1124	$_[0]->on_error (sub {
		1125	($! == Errno::EPIPE || !$!)
		1126	? $finish->(delete $_[0]{rbuf})
		1127	: $finish->(undef, $ae_error => $_[2]);
		1128	});
		1129	$_[0]->on_read (sub { });
		1130	}
		1131	}
		1132	};
		1133
		1134	# if keepalive is enabled, then the server closing the connection
		1135	# before a response can happen legally - we retry on idempotent methods.
		1136	if ($was_persistent && $idempotent) {
		1137	my $old_eof = $hdl->{on_eof};
		1138	$hdl->{on_eof} = sub {
		1139	_destroy_state %state;
		1140
		1141	%state = ();
		1142	$state{recurse} =
		1143	http_request (
		1144	$method => $url,
		1145	%arg,
		1146	recurse => $recurse - 1,
		1147	persistent => 0,
		1148	sub {
		1149	%state = ();
		1150	&$cb
		1151	}
		1152	);
		1153	};
		1154	$hdl->on_read (sub {
		1155	return unless %state;
		1156
		1157	# as soon as we receive something, a connection close
		1158	# once more becomes a hard error
		1159	$hdl->{on_eof} = $old_eof;
		1160	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1161	});
		1162	} else {
		1163	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1164	}
		1165	};
		1166
		1167	my $prepare_handle = sub {
		1168	my ($hdl) = $state{handle};
		1169
		1170	$hdl->on_error (sub {
		1171	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1172	});
		1173	$hdl->on_eof (sub {
		1174	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1175	});
		1176	$hdl->timeout_reset;
		1177	$hdl->timeout ($timeout);
		1178	};
		1179
		1180	# connected to proxy (or origin server)
		1181	my $connect_cb = sub {
179	$state{fh} = shift	1182	my $fh = shift
180	or return $cb->(undef, { Status => 599, Reason => "$!" });	1183	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
181		1184
182	delete $state{connect_guard}; # reduce memory usage, save a tree	1185	return unless delete $state{connect_guard};
183		1186
184	# get handle	1187	# get handle
185	$state{handle} = new AnyEvent::Handle	1188	$state{handle} = new AnyEvent::Handle
186	fh => $state{fh},	1189	%{ $arg{handle_params} },
187	($scheme eq "https" ? (tls => "connect") : ());	1190	fh => $fh,
188		1191	peername => $uhost,
189	# limit the number of persistent connections	1192	tls_ctx => $arg{tls_ctx},
190	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
191	++$KA_COUNT{$_[1]};
192	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
193	$hdr{connection} = "keep-alive";
194	delete $hdr{connection}; # keep-alive not yet supported
195	} else {
196	delete $hdr{connection};
197	}
198
199	# (re-)configure handle
200	$state{handle}->timeout ($timeout);
201	$state{handle}->on_error (sub {
202	%state = ();
203	$cb->(undef, { Status => 599, Reason => "$!" });
204	});
205	$state{handle}->on_eof (sub {
206	%state = ();
207	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file" });
208	});
209
210	# send request
211	$state{handle}->push_write (
212	"$method $path HTTP/1.0\015\012"
213	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
214	. "\015\012"
215	. (delete $state{body})
216	);	1193	;
217		1194
218	%hdr = (); # reduce memory usage, save a kitten	1195	$prepare_handle->();
219		1196
220	# status line	1197	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1198
		1199	# now handle proxy-CONNECT method
		1200	if ($proxy && $uscheme eq "https") {
		1201	# oh dear, we have to wrap it into a connect request
		1202
		1203	my $auth = exists $hdr{"proxy-authorization"}
		1204	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1205	: "";
		1206
		1207	# maybe re-use $uauthority with patched port?
		1208	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
221	$state{handle}->push_read (line => qr/\015?\012/, sub {	1209	$state{handle}->push_read (line => $qr_nlnl, sub {
222	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix	1210	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
223	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])" }));	1211	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
224		1212
225	my %hdr = ( # response headers	1213	if ($2 == 200) {
226	HTTPVersion => ",$1",	1214	$rpath = $upath;
227	Status => ",$2",	1215	$handle_actual_request->();
228	Reason => ",$3",
229	);
230
231	# headers, could be optimized a bit
232	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
233	for ("$_[1]\012") {
234	# we support spaces in field names, as lotus domino
235	# creates them.
236	$hdr{lc $1} .= ",$2"
237	while /\G
238	([^:\000-\037]+):
239	[\011\040]*
240	((?: [^\015\012]+ | \015?\012[\011\040] )*)
241	\015?\012
242	/gxc;
243
244	/\G$/
245	or return $cb->(undef, { Status => 599, Reason => "garbled response headers" });
246	}
247
248	substr $_, 0, 1, ""
249	for values %hdr;
250
251	if ($method eq "HEAD") {
252	%state = ();
253	$cb->(undef, \%hdr);
254	} else {	1216	} else {
255	if (exists $hdr{"content-length"}) {	1217	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
256	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
257	# could cache persistent connection now
258	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
259	# but we don't, due to misdesigns, this is annoyingly complex
260	};
261
262	%state = ();
263	$cb->($_[1], \%hdr);
264	});
265	} else {
266	# too bad, need to read until we get an error or EOF,
267	# no way to detect winged data.
268	$_[0]->on_error (sub {
269	%state = ();
270	$cb->($_[0]{rbuf}, \%hdr);
271	});
272	$_[0]->on_eof (undef);
273	$_[0]->on_read (sub { });
274	}
275	}	1218	}
276	});	1219	});
		1220	} else {
		1221	delete $hdr{"proxy-authorization"} unless $proxy;
		1222
		1223	$handle_actual_request->();
277	});	1224	}
278	}, sub {
279	$timeout
280	};	1225	};
281		1226
		1227	_get_slot $uhost, sub {
		1228	$state{slot_guard} = shift;
		1229
		1230	return unless $state{connect_guard};
		1231
		1232	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1233	# on a keepalive request (in theory, this should be a separate config option).
		1234	if ($persistent && $KA_CACHE{$ka_key}) {
		1235	$was_persistent = 1;
		1236
		1237	$state{handle} = ka_fetch $ka_key;
		1238	# $state{handle}->destroyed
		1239	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1240	$prepare_handle->();
		1241	# $state{handle}->destroyed
		1242	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1243	$handle_actual_request->();
		1244
		1245	} else {
		1246	my $tcp_connect = $arg{tcp_connect}
		1247	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1248
		1249	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
		1250	}
		1251	};
		1252
282	defined wantarray && AnyEvent::Util::guard { %state = () }	1253	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
283	}	1254	}
284		1255
285	sub http_get($$;@) {	1256	sub http_get($@) {
286	unshift @_, "GET";	1257	unshift @_, "GET";
287	&http_request	1258	&http_request
288	}	1259	}
289		1260
290	sub http_head($$;@) {	1261	sub http_head($@) {
291	unshift @_, "HEAD";	1262	unshift @_, "HEAD";
292	&http_request	1263	&http_request
293	}	1264	}
294		1265
295	sub http_post($$$;@) {	1266	sub http_post($$@) {
		1267	my $url = shift;
296	unshift @_, "POST", "body";	1268	unshift @_, "POST", $url, "body";
297	&http_request	1269	&http_request
298	}	1270	}
299		1271
		1272	=back
		1273
		1274	=head2 DNS CACHING
		1275
		1276	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1277	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1278	hostnames. The latter is a simple stub resolver and does no caching
		1279	on its own. If you want DNS caching, you currently have to provide
		1280	your own default resolver (by storing a suitable resolver object in
		1281	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1282
300	=head2 GLOBAL FUNCTIONS AND VARIABLES	1283	=head2 GLOBAL FUNCTIONS AND VARIABLES
301		1284
302	=over 4	1285	=over 4
303		1286
304	=item AnyEvent::HTTP::set_proxy "proxy-url"	1287	=item AnyEvent::HTTP::set_proxy "proxy-url"
305		1288
306	Sets the default proxy server to use. The proxy-url must begin with a	1289	Sets the default proxy server to use. The proxy-url must begin with a
307	string of the form C<http://host:port> (optionally C<https:...>).	1290	string of the form C<http://host:port>, croaks otherwise.
		1291
		1292	To clear an already-set proxy, use C<undef>.
		1293
		1294	When AnyEvent::HTTP is loaded for the first time it will query the
		1295	default proxy from the operating system, currently by looking at
		1296	C<$ENV{http_proxy>}.
		1297
		1298	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1299
		1300	Remove all cookies from the cookie jar that have been expired. If
		1301	C<$session_end> is given and true, then additionally remove all session
		1302	cookies.
		1303
		1304	You should call this function (with a true C<$session_end>) before you
		1305	save cookies to disk, and you should call this function after loading them
		1306	again. If you have a long-running program you can additionally call this
		1307	function from time to time.
		1308
		1309	A cookie jar is initially an empty hash-reference that is managed by this
		1310	module. Its format is subject to change, but currently it is as follows:
		1311
		1312	The key C<version> has to contain C<2>, otherwise the hash gets
		1313	cleared. All other keys are hostnames or IP addresses pointing to
		1314	hash-references. The key for these inner hash references is the
		1315	server path for which this cookie is meant, and the values are again
		1316	hash-references. Each key of those hash-references is a cookie name, and
		1317	the value, you guessed it, is another hash-reference, this time with the
		1318	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1319	which have been replaced by a C<_expires> key that contains the cookie
		1320	expiry timestamp. Session cookies are indicated by not having an
		1321	C<_expires> key.
		1322
		1323	Here is an example of a cookie jar with a single cookie, so you have a
		1324	chance of understanding the above paragraph:
		1325
		1326	{
		1327	version => 2,
		1328	"10.0.0.1" => {
		1329	"/" => {
		1330	"mythweb_id" => {
		1331	_expires => 1293917923,
		1332	value => "ooRung9dThee3ooyXooM1Ohm",
		1333	},
		1334	},
		1335	},
		1336	}
		1337
		1338	=item $date = AnyEvent::HTTP::format_date $timestamp
		1339
		1340	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1341	Date (RFC 2616).
		1342
		1343	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1344
		1345	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1346	bunch of minor variations of those, and returns the corresponding POSIX
		1347	timestamp, or C<undef> if the date cannot be parsed.
308		1348
309	=item $AnyEvent::HTTP::MAX_RECURSE	1349	=item $AnyEvent::HTTP::MAX_RECURSE
310		1350
311	The default value for the C<recurse> request parameter (default: C<10>).	1351	The default value for the C<recurse> request parameter (default: C<10>).
312		1352
		1353	=item $AnyEvent::HTTP::TIMEOUT
		1354
		1355	The default timeout for connection operations (default: C<300>).
		1356
313	=item $AnyEvent::HTTP::USERAGENT	1357	=item $AnyEvent::HTTP::USERAGENT
314		1358
315	The default value for the C<User-Agent> header (the default is	1359	The default value for the C<User-Agent> header (the default is
316	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1360	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
317		1361
318	=item $AnyEvent::HTTP::MAX_PERSISTENT	1362	=item $AnyEvent::HTTP::MAX_PER_HOST
319		1363
320	The maximum number of persistent connections to keep open (default: 8).	1364	The maximum number of concurrent connections to the same host (identified
		1365	by the hostname). If the limit is exceeded, then additional requests
		1366	are queued until previous connections are closed. Both persistent and
		1367	non-persistent connections are counted in this limit.
321		1368
322	Not implemented currently.	1369	The default value for this is C<4>, and it is highly advisable to not
		1370	increase it much.
		1371
		1372	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1373	connections, older browsers used 2, newer ones (such as firefox 3)
		1374	typically use 6, and Opera uses 8 because like, they have the fastest
		1375	browser and give a shit for everybody else on the planet.
323		1376
324	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1377	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
325		1378
326	The maximum time to cache a persistent connection, in seconds (default: 2).	1379	The time after which idle persistent connections get closed by
		1380	AnyEvent::HTTP (default: C<3>).
327		1381
328	Not implemented currently.	1382	=item $AnyEvent::HTTP::ACTIVE
		1383
		1384	The number of active connections. This is not the number of currently
		1385	running requests, but the number of currently open and non-idle TCP
		1386	connections. This number can be useful for load-leveling.
329		1387
330	=back	1388	=back
331		1389
332	=cut	1390	=cut
333		1391
		1392	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1393	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1394
		1395	sub format_date($) {
		1396	my ($time) = @_;
		1397
		1398	# RFC 822/1123 format
		1399	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1400
		1401	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1402	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1403	$H, $M, $S;
		1404	}
		1405
		1406	sub parse_date($) {
		1407	my ($date) = @_;
		1408
		1409	my ($d, $m, $y, $H, $M, $S);
		1410
		1411	if ($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$/) {
		1412	# RFC 822/1123, required by RFC 2616 (with " ")
		1413	# cookie dates (with "-")
		1414
		1415	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1416
		1417	} elsif ($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$/) {
		1418	# RFC 850
		1419	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1420
		1421	} elsif ($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])$/) {
		1422	# ISO C's asctime
		1423	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1424	}
		1425	# other formats fail in the loop below
		1426
		1427	for (0..11) {
		1428	if ($m eq $month[$_]) {
		1429	require Time::Local;
		1430	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
		1431	}
		1432	}
		1433
		1434	undef
		1435	}
		1436
334	sub set_proxy($) {	1437	sub set_proxy($) {
335	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1438	if (length $_[0]) {
		1439	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1440	or Carp::croak "$_[0]: invalid proxy URL";
		1441	$PROXY = [$2, $3 || 3128, $1]
		1442	} else {
		1443	undef $PROXY;
		1444	}
336	}	1445	}
337		1446
338	# initialise proxy from environment	1447	# initialise proxy from environment
		1448	eval {
339	set_proxy $ENV{http_proxy};	1449	set_proxy $ENV{http_proxy};
		1450	};
		1451
		1452	=head2 SHOWCASE
		1453
		1454	This section contains some more elaborate "real-world" examples or code
		1455	snippets.
		1456
		1457	=head2 HTTP/1.1 FILE DOWNLOAD
		1458
		1459	Downloading files with HTTP can be quite tricky, especially when something
		1460	goes wrong and you want to resume.
		1461
		1462	Here is a function that initiates and resumes a download. It uses the
		1463	last modified time to check for file content changes, and works with many
		1464	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1465	on older servers.
		1466
		1467	It calls the completion callback with either C<undef>, which means a
		1468	nonretryable error occurred, C<0> when the download was partial and should
		1469	be retried, and C<1> if it was successful.
		1470
		1471	use AnyEvent::HTTP;
		1472
		1473	sub download($$$) {
		1474	my ($url, $file, $cb) = @_;
		1475
		1476	open my $fh, "+<", $file
		1477	or die "$file: $!";
		1478
		1479	my %hdr;
		1480	my $ofs = 0;
		1481
		1482	if (stat $fh and -s _) {
		1483	$ofs = -s _;
		1484	warn "-s is ", $ofs;
		1485	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1486	$hdr{"range"} = "bytes=$ofs-";
		1487	}
		1488
		1489	http_get $url,
		1490	headers => \%hdr,
		1491	on_header => sub {
		1492	my ($hdr) = @_;
		1493
		1494	if ($hdr->{Status} == 200 && $ofs) {
		1495	# resume failed
		1496	truncate $fh, $ofs = 0;
		1497	}
		1498
		1499	sysseek $fh, $ofs, 0;
		1500
		1501	1
		1502	},
		1503	on_body => sub {
		1504	my ($data, $hdr) = @_;
		1505
		1506	if ($hdr->{Status} =~ /^2/) {
		1507	length $data == syswrite $fh, $data
		1508	or return; # abort on write errors
		1509	}
		1510
		1511	1
		1512	},
		1513	sub {
		1514	my (undef, $hdr) = @_;
		1515
		1516	my $status = $hdr->{Status};
		1517
		1518	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1519	utime $time, $time, $fh;
		1520	}
		1521
		1522	if ($status == 200 || $status == 206 || $status == 416) {
		1523	# download ok || resume ok || file already fully downloaded
		1524	$cb->(1, $hdr);
		1525
		1526	} elsif ($status == 412) {
		1527	# file has changed while resuming, delete and retry
		1528	unlink $file;
		1529	$cb->(0, $hdr);
		1530
		1531	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1532	# retry later
		1533	$cb->(0, $hdr);
		1534
		1535	} else {
		1536	$cb->(undef, $hdr);
		1537	}
		1538	}
		1539	;
		1540	}
		1541
		1542	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1543	if ($_[0]) {
		1544	print "OK!\n";
		1545	} elsif (defined $_[0]) {
		1546	print "please retry later\n";
		1547	} else {
		1548	print "ERROR\n";
		1549	}
		1550	};
		1551
		1552	=head3 SOCKS PROXIES
		1553
		1554	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1555	compile your perl to support socks, or use an external program such as
		1556	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1557	transparently.
		1558
		1559	Alternatively, for AnyEvent::HTTP only, you can use your own
		1560	C<tcp_connect> function that does the proxy handshake - here is an example
		1561	that works with socks4a proxies:
		1562
		1563	use Errno;
		1564	use AnyEvent::Util;
		1565	use AnyEvent::Socket;
		1566	use AnyEvent::Handle;
		1567
		1568	# host, port and username of/for your socks4a proxy
		1569	my $socks_host = "10.0.0.23";
		1570	my $socks_port = 9050;
		1571	my $socks_user = "";
		1572
		1573	sub socks4a_connect {
		1574	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1575
		1576	my $hdl = new AnyEvent::Handle
		1577	connect => [$socks_host, $socks_port],
		1578	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1579	on_error => sub { $connect_cb->() },
		1580	;
		1581
		1582	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1583
		1584	$hdl->push_read (chunk => 8, sub {
		1585	my ($hdl, $chunk) = @_;
		1586	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1587
		1588	if ($status == 0x5a) {
		1589	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1590	} else {
		1591	$! = Errno::ENXIO; $connect_cb->();
		1592	}
		1593	});
		1594
		1595	$hdl
		1596	}
		1597
		1598	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1599	possibly after switching off other proxy types:
		1600
		1601	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1602
		1603	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1604	my ($data, $headers) = @_;
		1605	...
		1606	};
340		1607
341	=head1 SEE ALSO	1608	=head1 SEE ALSO
342		1609
343	L<AnyEvent>.	1610	L<AnyEvent>.
344		1611
345	=head1 AUTHOR	1612	=head1 AUTHOR
346		1613
347	Marc Lehmann <schmorp@schmorp.de>	1614	Marc Lehmann <schmorp@schmorp.de>
348	http://home.schmorp.de/	1615	http://home.schmorp.de/
		1616
		1617	With many thanks to Дмитрий Шалашов, who provided countless
		1618	testcases and bugreports.
349		1619
350	=cut	1620	=cut
351		1621
352	1	1622	1
353		1623

Diff Legend

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