ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.24 by root, Wed Jul 2 01:30:33 2008 UTC vs.
Revision 1.95 by root, Wed Jan 12 03:30:05 2011 UTC

…		…
15	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
16	run a supported event loop.	16	run a supported event loop.
17		17
18	This module implements a simple, stateless and non-blocking HTTP	18	This module implements a simple, stateless and non-blocking HTTP
19	client. It supports GET, POST and other request methods, cookies and more,	19	client. It supports GET, POST and other request methods, cookies and more,
20	all on a very low level. It can follow redirects supports proxies and	20	all on a very low level. It can follow redirects, supports proxies, and
21	automatically limits the number of connections to the values specified in	21	automatically limits the number of connections to the values specified in
22	the RFC.	22	the RFC.
23		23
24	It should generally be a "good client" that is enough for most HTTP	24	It should generally be a "good client" that is enough for most HTTP
25	tasks. Simple tasks should be simple, but complex tasks should still be	25	tasks. Simple tasks should be simple, but complex tasks should still be
…		…
36		36
37	=cut	37	=cut
38		38
39	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
40		40
41	use strict;	41	use common::sense;
42	no warnings;
43		42
44	use Carp;	43	use Errno ();
45		44
46	use AnyEvent ();	45	use AnyEvent 5.0 ();
47	use AnyEvent::Util ();	46	use AnyEvent::Util ();
48	use AnyEvent::Socket ();
49	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
50		48
51	use base Exporter::;	49	use base Exporter::;
52		50
53	our $VERSION = '1.03';	51	our $VERSION = '2.02';
54		52
55	our @EXPORT = qw(http_get http_post http_head http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
56		54
57	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)";
58	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
59	our $MAX_PERSISTENT = 8;
60	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
61	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
62		59	our $MAX_PER_HOST = 4; # changing this is evil
63	# changing these is evil
64	our $MAX_PERSISTENT_PER_HOST = 2;
65	our $MAX_PER_HOST = 4;
66		60
67	our $PROXY;	61	our $PROXY;
68	our $ACTIVE = 0;	62	our $ACTIVE = 0;
69		63
70	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
71	my %CO_SLOT; # number of open connections, and wait queue, per host	65	my %CO_SLOT; # number of open connections, and wait queue, per host
72		66
73	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
74		68
75	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
76	additional parameters.	70	additional parameters and the return value.
77		71
78	=item http_head $url, key => value..., $cb->($data, $headers)	72	=item http_head $url, key => value..., $cb->($data, $headers)
79		73
80	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
81	additional parameters.	75	on additional parameters and the return value.
82		76
83	=item http_post $url, $body, key => value..., $cb->($data, $headers)	77	=item http_post $url, $body, key => value..., $cb->($data, $headers)
84		78
85	Executes an HTTP-POST request with a request body of C<$bod>. See the	79	Executes an HTTP-POST request with a request body of C<$body>. See the
86	http_request function for details on additional parameters.	80	http_request function for details on additional parameters and the return
		81	value.
87		82
88	=item http_request $method => $url, key => value..., $cb->($data, $headers)	83	=item http_request $method => $url, key => value..., $cb->($data, $headers)
89		84
90	Executes a HTTP request of type C<$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
91	must be an absolute http or https URL.	86	must be an absolute http or https URL.
92		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
93	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
94	(or C<undef> if it wasn't available due to errors), and a hash-ref with	94	(or C<undef> if an error occured), and a hash-ref with response headers
95	response headers as second argument.	95	(and trailers) as second argument.
96		96
97	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
98	headers, the "pseudo-headers" C<HTTPVersion>, C<Status> and C<Reason>	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
		99	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
99	contain the three parts of the HTTP Status-Line of the same name. The	100	three parts of the HTTP Status-Line of the same name. If an error occurs
		101	during the body phase of a request, then the original C<Status> and
		102	C<Reason> values from the header are available as C<OrigStatus> and
		103	C<OrigReason>.
		104
100	pseudo-header C<URL> contains the original URL (which can differ from the	105	The pseudo-header C<URL> contains the actual URL (which can differ from
101	requested URL when following redirects).	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).
102		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	$headers->{Redirect}[1]{Redirect} >> will then contain the original
		116	response, and so on.
		117
103	If the server sends a header multiple lines, then their contents will be	118	If the server sends a header multiple times, then their contents will be
104	joined together with C<\x00>.	119	joined together with a comma (C<,>), as per the HTTP spec.
105		120
106	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
107	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<599>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
108	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection etsbalishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
109		139
110	A typical callback might look like this:	140	A typical callback might look like this:
111		141
112	sub {	142	sub {
113	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
129	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication
130	retries and so on, and how often to do so.	160	retries and so on, and how often to do so.
131		161
132	=item headers => hashref	162	=item headers => hashref
133		163
134	The request headers to use. Currently, C<http_request> may provide its	164	The request headers to use. Currently, C<http_request> may provide its own
135	own C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers	165	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
136	and will provide defaults for C<User-Agent:> and C<Referer:>.	166	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
		167	(this can be suppressed by using C<undef> for these headers in which case
		168	they won't be sent at all).
		169
		170	You really should provide your own C<User-Agent:> header value that is
		171	appropriate for your program - I wouldn't be surprised if the default
		172	AnyEvent string gets blocked by webservers sooner or later.
137		173
138	=item timeout => $seconds	174	=item timeout => $seconds
139		175
140	The time-out to use for various stages - each connect attempt will reset	176	The time-out to use for various stages - each connect attempt will reset
141	the timeout, as will read or write activity. Default timeout is 5 minutes.	177	the timeout, as will read or write activity, i.e. this is not an overall
		178	timeout.
		179
		180	Default timeout is 5 minutes.
142		181
143	=item proxy => [$host, $port[, $scheme]] or undef	182	=item proxy => [$host, $port[, $scheme]] or undef
144		183
145	Use the given http proxy for all requests. If not specified, then the	184	Use the given http proxy for all requests. If not specified, then the
146	default proxy (as specified by C<$ENV{http_proxy}>) is used.	185	default proxy (as specified by C<$ENV{http_proxy}>) is used.
147		186
148	C<$scheme> must be either missing or C<http> for HTTP, or C<https> for	187	C<$scheme> must be either missing or must be C<http> for HTTP.
149	HTTPS.
150		188
151	=item body => $string	189	=item body => $string
152		190
153	The request body, usually empty. Will be-sent as-is (future versions of	191	The request body, usually empty. Will be sent as-is (future versions of
154	this module might offer more options).	192	this module might offer more options).
155		193
156	=item cookie_jar => $hash_ref	194	=item cookie_jar => $hash_ref
157		195
158	Passing this parameter enables (simplified) cookie-processing, loosely	196	Passing this parameter enables (simplified) cookie-processing, loosely
159	based on the original netscape specification.	197	based on the original netscape specification.
160		198
161	The C<$hash_ref> must be an (initially empty) hash reference which will	199	The C<$hash_ref> must be an (initially empty) hash reference which
162	get updated automatically. It is possible to save the cookie_jar to	200	will get updated automatically. It is possible to save the cookie jar
163	persistent storage with something like JSON or Storable, but this is not	201	to persistent storage with something like JSON or Storable - see the
164	recommended, as expire times are currently being ignored.	202	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		203	expired or session-only cookies, and also for documentation on the format
		204	of the cookie jar.
165		205
166	Note that this cookie implementation is not of very high quality, nor	206	Note that this cookie implementation is not meant to be complete. If
167	meant to be complete. If you want complete cookie management you have to	207	you want complete cookie management you have to do that on your
168	do that on your own. C<cookie_jar> is meant as a quick fix to get some	208	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
169	cookie-using sites working. Cookies are a privacy disaster, do not use	209	working. Cookies are a privacy disaster, do not use them unless required
170	them unless required to.	210	to.
		211
		212	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		213	headers will be set and handled by this module, otherwise they will be
		214	left untouched.
		215
		216	=item tls_ctx => $scheme | $tls_ctx
		217
		218	Specifies the AnyEvent::TLS context to be used for https connections. This
		219	parameter follows the same rules as the C<tls_ctx> parameter to
		220	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		221	C<high> can be specified, which give you a predefined low-security (no
		222	verification, highest compatibility) and high-security (CA and common-name
		223	verification) TLS context.
		224
		225	The default for this option is C<low>, which could be interpreted as "give
		226	me the page, no matter what".
		227
		228	See also the C<sessionid> parameter.
		229
		230	=item session => $string
		231
		232	The module might reuse connections to the same host internally. Sometimes
		233	(e.g. when using TLS), you do not want to reuse connections from other
		234	sessions. This can be achieved by setting this parameter to some unique
		235	ID (such as the address of an object storing your state data, or the TLS
		236	context) - only connections using the same unique ID will be reused.
		237
		238	=item on_prepare => $callback->($fh)
		239
		240	In rare cases you need to "tune" the socket before it is used to
		241	connect (for exmaple, to bind it on a given IP address). This parameter
		242	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		243	and behaves exactly the same way (e.g. it has to provide a
		244	timeout). See the description for the C<$prepare_cb> argument of
		245	C<AnyEvent::Socket::tcp_connect> for details.
		246
		247	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		248
		249	In even rarer cases you want total control over how AnyEvent::HTTP
		250	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		251	to do this, but you can provide your own C<tcp_connect> function -
		252	obviously, it has to follow the same calling conventions, except that it
		253	may always return a connection guard object.
		254
		255	There are probably lots of weird uses for this function, starting from
		256	tracing the hosts C<http_request> actually tries to connect, to (inexact
		257	but fast) host => IP address caching or even socks protocol support.
		258
		259	=item on_header => $callback->($headers)
		260
		261	When specified, this callback will be called with the header hash as soon
		262	as headers have been successfully received from the remote server (not on
		263	locally-generated errors).
		264
		265	It has to return either true (in which case AnyEvent::HTTP will continue),
		266	or false, in which case AnyEvent::HTTP will cancel the download (and call
		267	the finish callback with an error code of C<598>).
		268
		269	This callback is useful, among other things, to quickly reject unwanted
		270	content, which, if it is supposed to be rare, can be faster than first
		271	doing a C<HEAD> request.
		272
		273	The downside is that cancelling the request makes it impossible to re-use
		274	the connection. Also, the C<on_header> callback will not receive any
		275	trailer (headers sent after the response body).
		276
		277	Example: cancel the request unless the content-type is "text/html".
		278
		279	on_header => sub {
		280	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		281	},
		282
		283	=item on_body => $callback->($partial_body, $headers)
		284
		285	When specified, all body data will be passed to this callback instead of
		286	to the completion callback. The completion callback will get the empty
		287	string instead of the body data.
		288
		289	It has to return either true (in which case AnyEvent::HTTP will continue),
		290	or false, in which case AnyEvent::HTTP will cancel the download (and call
		291	the completion callback with an error code of C<598>).
		292
		293	The downside to cancelling the request is that it makes it impossible to
		294	re-use the connection.
		295
		296	This callback is useful when the data is too large to be held in memory
		297	(so the callback writes it to a file) or when only some information should
		298	be extracted, or when the body should be processed incrementally.
		299
		300	It is usually preferred over doing your own body handling via
		301	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		302	only used to create a connection, C<want_body_handle> is the better
		303	alternative, as it allows you to install your own event handler, reducing
		304	resource usage.
		305
		306	=item want_body_handle => $enable
		307
		308	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		309	changes considerably: after parsing the headers, and instead of
		310	downloading the body (if any), the completion callback will be
		311	called. Instead of the C<$body> argument containing the body data, the
		312	callback will receive the L<AnyEvent::Handle> object associated with the
		313	connection. In error cases, C<undef> will be passed. When there is no body
		314	(e.g. status C<304>), the empty string will be passed.
		315
		316	The handle object might or might not be in TLS mode, might be connected
		317	to a proxy, be a persistent connection, use chunked transfer encoding
		318	etc., and configured in unspecified ways. The user is responsible for this
		319	handle (it will not be used by this module anymore).
		320
		321	This is useful with some push-type services, where, after the initial
		322	headers, an interactive protocol is used (typical example would be the
		323	push-style twitter API which starts a JSON/XML stream).
		324
		325	If you think you need this, first have a look at C<on_body>, to see if
		326	that doesn't solve your problem in a better way.
		327
		328	=item persistent => $boolean
		329
		330	Try to create/reuse a persistent connection. When this flag is set
		331	(default: true for idempotent requests, false for all others), then
		332	C<http_request> tries to re-use an existing (previously-created)
		333	persistent connection to the host and, failing that, tries to create a new
		334	one.
		335
		336	Requests failing in certain ways will be automatically retried once, which
		337	is dangerous for non-idempotent requests, which is why it defaults to off
		338	for them. The reason for this is because the bozos who designed HTTP/1.1
		339	made it impossible to distinguish between a fatal error and a normal
		340	connection timeout, so you never know whether there was a problem with
		341	your request or not.
		342
		343	When reusing an existent connection, many parameters (such as TLS context)
		344	will be ignored. See the C<session> parameter for a workaround.
		345
		346	=item keepalive => $boolean
		347
		348	Only used when C<persistent> is also true. This parameter decides whether
		349	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		350	(as opposed to only a HTTP/1.1 persistent connection).
		351
		352	The default is true, except when using a proxy, in which case it defaults
		353	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		354
		355	=item handle_params => { key => value ... }
		356
		357	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		358	constructor that is called - not all requests will create a handle, and
		359	sometimes more than one is created, so this parameter is only good for
		360	setting hints.
		361
		362	Example: set the maximum read size to 4096, to potentially conserve memory
		363	at the cost of speed.
		364
		365	handle_params => {
		366	max_read_size => 4096,
		367	},
171		368
172	=back	369	=back
173		370
174	Example: make a simple HTTP GET request for http://www.nethype.de/	371	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		372	the response body.
175		373
176	http_request GET => "http://www.nethype.de/", sub {	374	http_request GET => "http://www.nethype.de/", sub {
177	my ($body, $hdr) = @_;	375	my ($body, $hdr) = @_;
178	print "$body\n";	376	print "$body\n";
179	};	377	};
180		378
181	Example: make a HTTP HEAD request on https://www.google.com/, use a	379	Example: do a HTTP HEAD request on https://www.google.com/, use a
182	timeout of 30 seconds.	380	timeout of 30 seconds.
183		381
184	http_request	382	http_request
185	GET => "https://www.google.com",	383	GET => "https://www.google.com",
		384	headers => { "user-agent" => "MySearchClient 1.0" },
186	timeout => 30,	385	timeout => 30,
187	sub {	386	sub {
188	my ($body, $hdr) = @_;	387	my ($body, $hdr) = @_;
189	use Data::Dumper;	388	use Data::Dumper;
190	print Dumper $hdr;	389	print Dumper $hdr;
191	}	390	}
192	;	391	;
193		392
		393	Example: do another simple HTTP GET request, but immediately try to
		394	cancel it.
		395
		396	my $request = http_request GET => "http://www.nethype.de/", sub {
		397	my ($body, $hdr) = @_;
		398	print "$body\n";
		399	};
		400
		401	undef $request;
		402
194	=cut	403	=cut
		404
		405	#############################################################################
		406	# wait queue/slots
195		407
196	sub _slot_schedule;	408	sub _slot_schedule;
197	sub _slot_schedule($) {	409	sub _slot_schedule($) {
198	my $host = shift;	410	my $host = shift;
199		411
…		…
221	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	433	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
222		434
223	_slot_schedule $_[0];	435	_slot_schedule $_[0];
224	}	436	}
225		437
		438	#############################################################################
		439	# cookie handling
		440
		441	# expire cookies
		442	sub cookie_jar_expire($;$) {
		443	my ($jar, $session_end) = @_;
		444
		445	%$jar = () if $jar->{version} != 1;
		446
		447	my $anow = AE::now;
		448
		449	while (my ($chost, $paths) = each %$jar) {
		450	next unless ref $paths;
		451
		452	while (my ($cpath, $cookies) = each %$paths) {
		453	while (my ($cookie, $kv) = each %$cookies) {
		454	if (exists $kv->{_expires}) {
		455	delete $cookies->{$cookie}
		456	if $anow > $kv->{_expires};
		457	} elsif ($session_end) {
		458	delete $cookies->{$cookie};
		459	}
		460	}
		461
		462	delete $paths->{$cpath}
		463	unless %$cookies;
		464	}
		465
		466	delete $jar->{$chost}
		467	unless %$paths;
		468	}
		469	}
		470
		471	# extract cookies from jar
		472	sub cookie_jar_extract($$$$) {
		473	my ($jar, $scheme, $host, $path) = @_;
		474
		475	%$jar = () if $jar->{version} != 1;
		476
		477	my @cookies;
		478
		479	while (my ($chost, $paths) = each %$jar) {
		480	next unless ref $paths;
		481
		482	if ($chost =~ /^\./) {
		483	next unless $chost eq substr $host, -length $chost;
		484	} elsif ($chost =~ /\./) {
		485	next unless $chost eq $host;
		486	} else {
		487	next;
		488	}
		489
		490	while (my ($cpath, $cookies) = each %$paths) {
		491	next unless $cpath eq substr $path, 0, length $cpath;
		492
		493	while (my ($cookie, $kv) = each %$cookies) {
		494	next if $scheme ne "https" && exists $kv->{secure};
		495
		496	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		497	delete $cookies->{$cookie};
		498	next;
		499	}
		500
		501	my $value = $kv->{value};
		502
		503	if ($value =~ /[=;,[:space:]]/) {
		504	$value =~ s/([\\"])/\\$1/g;
		505	$value = "\"$value\"";
		506	}
		507
		508	push @cookies, "$cookie=$value";
		509	}
		510	}
		511	}
		512
		513	\@cookies
		514	}
		515
		516	# parse set_cookie header into jar
		517	sub cookie_jar_set_cookie($$$$) {
		518	my ($jar, $set_cookie, $host, $date) = @_;
		519
		520	my $anow = int AE::now;
		521	my $snow; # server-now
		522
		523	for ($set_cookie) {
		524	# parse NAME=VALUE
		525	my @kv;
		526
		527	# expires is not http-compliant in the original cookie-spec,
		528	# we support the official date format and some extensions
		529	while (
		530	m{
		531	\G\s*
		532	(?:
		533	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		534	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) ) )?
		535	)
		536	}gcxsi
		537	) {
		538	my $name = $2;
		539	my $value = $4;
		540
		541	if (defined $1) {
		542	# expires
		543	$name = "expires";
		544	$value = $1;
		545	} elsif (defined $3) {
		546	# quoted
		547	$value = $3;
		548	$value =~ s/\\(.)/$1/gs;
		549	}
		550
		551	push @kv, @kv ? lc $name : $name, $value;
		552
		553	last unless /\G\s*;/gc;
		554	}
		555
		556	last unless @kv;
		557
		558	my $name = shift @kv;
		559	my %kv = (value => shift @kv, @kv);
		560
		561	if (exists $kv{"max-age"}) {
		562	$kv{_expires} = $anow + delete $kv{"max-age"};
		563	} elsif (exists $kv{expires}) {
		564	$snow ||= parse_date ($date) || $anow;
		565	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		566	} else {
		567	delete $kv{_expires};
		568	}
		569
		570	my $cdom;
		571	my $cpath = (delete $kv{path}) || "/";
		572
		573	if (exists $kv{domain}) {
		574	$cdom = delete $kv{domain};
		575
		576	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		577
		578	next if $cdom =~ /\.$/;
		579
		580	# this is not rfc-like and not netscape-like. go figure.
		581	my $ndots = $cdom =~ y/.//;
		582	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		583	} else {
		584	$cdom = $host;
		585	}
		586
		587	# store it
		588	$jar->{version} = 1;
		589	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		590
		591	redo if /\G\s*,/gc;
		592	}
		593	}
		594
		595	#############################################################################
		596	# keepalive/persistent connection cache
		597
		598	# fetch a connection from the keepalive cache
		599	sub ka_fetch($) {
		600	my $ka_key = shift;
		601
		602	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		603	delete $KA_CACHE{$ka_key}
		604	unless @{ $KA_CACHE{$ka_key} };
		605
		606	$hdl
		607	}
		608
		609	sub ka_store($$) {
		610	my ($ka_key, $hdl) = @_;
		611
		612	my $kaa = $KA_CACHE{$ka_key} ||= [];
		613
		614	my $destroy = sub {
		615	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		616
		617	$hdl->destroy;
		618
		619	@ka
		620	? $KA_CACHE{$ka_key} = \@ka
		621	: delete $KA_CACHE{$ka_key};
		622	};
		623
		624	# on error etc., destroy
		625	$hdl->on_error ($destroy);
		626	$hdl->on_eof ($destroy);
		627	$hdl->on_read ($destroy);
		628	$hdl->timeout ($PERSISTENT_TIMEOUT);
		629
		630	push @$kaa, $hdl;
		631	shift @$kaa while @$kaa > $MAX_PER_HOST;
		632	}
		633
		634	#############################################################################
		635	# utilities
		636
		637	# continue to parse $_ for headers and place them into the arg
		638	sub _parse_hdr() {
		639	my %hdr;
		640
		641	# things seen, not parsed:
		642	# p3pP="NON CUR OTPi OUR NOR UNI"
		643
		644	$hdr{lc $1} .= ",$2"
		645	while /\G
		646	([^:\000-\037]*):
		647	[\011\040]*
		648	((?: [^\012]+ | \012[\011\040] )*)
		649	\012
		650	/gxc;
		651
		652	/\G$/
		653	or return;
		654
		655	# remove the "," prefix we added to all headers above
		656	substr $_, 0, 1, ""
		657	for values %hdr;
		658
		659	\%hdr
		660	}
		661
		662	#############################################################################
		663	# http_get
		664
		665	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		666
		667	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		668	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		669
		670	# maybe it should just become a normal object :/
		671
		672	sub _destroy_state(\%) {
		673	my ($state) = @_;
		674
		675	$state->{handle}->destroy if $state->{handle};
		676	%$state = ();
		677	}
		678
		679	sub _error(\%$$) {
		680	my ($state, $cb, $hdr) = @_;
		681
		682	&_destroy_state ($state);
		683
		684	$cb->(undef, $hdr);
		685	()
		686	}
		687
226	sub http_request($$@) {	688	sub http_request($$@) {
227	my $cb = pop;	689	my $cb = pop;
228	my ($method, $url, %arg) = @_;	690	my ($method, $url, %arg) = @_;
229		691
230	my %hdr;	692	my %hdr;
		693
		694	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		695	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
231		696
232	$method = uc $method;	697	$method = uc $method;
233		698
234	if (my $hdr = $arg{headers}) {	699	if (my $hdr = $arg{headers}) {
235	while (my ($k, $v) = each %$hdr) {	700	while (my ($k, $v) = each %$hdr) {
236	$hdr{lc $k} = $v;	701	$hdr{lc $k} = $v;
237	}	702	}
238	}	703	}
239		704
		705	# pseudo headers for all subsequent responses
		706	my @pseudo = (URL => $url);
		707	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		708
240	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	709	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
241		710
242	return $cb->(undef, { Status => 599, Reason => "recursion limit reached", URL => $url })	711	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
243	if $recurse < 0;	712	if $recurse < 0;
244		713
245	my $proxy = $arg{proxy} || $PROXY;	714	my $proxy = $arg{proxy} || $PROXY;
246	my $timeout = $arg{timeout} || $TIMEOUT;	715	my $timeout = $arg{timeout} || $TIMEOUT;
247		716
248	$hdr{"user-agent"} ||= $USERAGENT;
249
250	my ($scheme, $authority, $upath, $query, $fragment) =	717	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
251	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;	718	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;
252		719
253	$scheme = lc $scheme;	720	$uscheme = lc $uscheme;
254		721
255	my $uport = $scheme eq "http" ? 80	722	my $uport = $uscheme eq "http" ? 80
256	: $scheme eq "https" ? 443	723	: $uscheme eq "https" ? 443
257	: return $cb->(undef, { Status => 599, Reason => "only http and https URL schemes supported", URL => $url });	724	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
258		725
259	$hdr{referer} ||= "$scheme://$authority$upath"; # leave out fragment and query string, just a heuristic
260
261	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	726	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
262	or return $cb->(undef, { Status => 599, Reason => "unparsable URL", URL => $url });	727	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
263		728
264	my $uhost = $1;	729	my $uhost = lc $1;
265	$uport = $2 if defined $2;	730	$uport = $2 if defined $2;
266		731
		732	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		733	unless exists $hdr{host};
		734
267	$uhost =~ s/^\[(.*)\]$/$1/;	735	$uhost =~ s/^\[(.*)\]$/$1/;
268	$upath .= "?$query" if length $query;	736	$upath .= $query if length $query;
269		737
270	$upath =~ s%^/?%/%;	738	$upath =~ s%^/?%/%;
271		739
272	# cookie processing	740	# cookie processing
273	if (my $jar = $arg{cookie_jar}) {	741	if (my $jar = $arg{cookie_jar}) {
274	%$jar = () if $jar->{version} < 1;	742	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
275		743
		744	$hdr{cookie} = join "; ", @$cookies
276	my @cookie;	745	if @$cookies;
277		746	}
278	while (my ($chost, $v) = each %$jar) {	747
279	next unless $chost eq substr $uhost, -length $chost;	748	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
280	next unless $chost =~ /^\./;	749
281		750	if ($proxy) {
282	while (my ($cpath, $v) = each %$v) {	751	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
283	next unless $cpath eq substr $upath, 0, length $cpath;	752
284		753	$rscheme = "http" unless defined $rscheme;
285	while (my ($k, $v) = each %$v) {	754
286	next if $scheme ne "https" && exists $v->{secure};	755	# don't support https requests over https-proxy transport,
287	push @cookie, "$k=$v->{value}";	756	# can't be done with tls as spec'ed, unless you double-encrypt.
		757	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		758
		759	$rhost = lc $rhost;
		760	$rscheme = lc $rscheme;
		761	} else {
		762	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
		763	}
		764
		765	# leave out fragment and query string, just a heuristic
		766	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		767	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
		768
		769	$hdr{"content-length"} = length $arg{body}
		770	if length $arg{body} || $method ne "GET";
		771
		772	my $idempotent = $method =~ /^(?:GET|HEAD|PUT|DELETE|OPTIONS|TRACE)$/;
		773
		774	# default value for keepalive is true iff the request is for an idempotent method
		775	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : $idempotent;
		776	my $keepalive10 = exists $arg{keepalive10} ? $arg{keepalive10} : !$proxy;
		777	my $keptalive; # true if this is actually a recycled connection
		778
		779	# the key to use in the keepalive cache
		780	my $ka_key = "$uhost\x00$arg{sessionid}";
		781
		782	$hdr{connection} = ($keepalive ? $keepalive10 ? "keep-alive " : "" : "close ") . "Te"; #1.1
		783	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		784
		785	my %state = (connect_guard => 1);
		786
		787	my $ae_error = 595; # connecting
		788
		789	# handle actual, non-tunneled, request
		790	my $handle_actual_request = sub {
		791	$ae_error = 596; # request phase
		792
		793	my $hdl = $state{handle};
		794
		795	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		796
		797	# send request
		798	$hdl->push_write (
		799	"$method $rpath HTTP/1.1\015\012"
		800	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		801	. "\015\012"
		802	. (delete $arg{body})
		803	);
		804
		805	# return if error occured during push_write()
		806	return unless %state;
		807
		808	# reduce memory usage, save a kitten, also re-use it for the response headers.
		809	%hdr = ();
		810
		811	# status line and headers
		812	$state{read_response} = sub {
		813	return unless %state;
		814
		815	for ("$_[1]") {
		816	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		817
		818	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		819	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		820
		821	# 100 Continue handling
		822	# should not happen as we don't send expect: 100-continue,
		823	# but we handle it just in case.
		824	# since we send the request body regardless, if we get an error
		825	# we are out of-sync, which we currently do NOT handle correctly.
		826	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		827	if $2 eq 100;
		828
		829	push @pseudo,
		830	HTTPVersion => $1,
		831	Status => $2,
		832	Reason => $3,
		833	;
		834
		835	my $hdr = _parse_hdr
		836	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		837
		838	%hdr = (%$hdr, @pseudo);
		839	}
		840
		841	# redirect handling
		842	# microsoft and other shitheads don't give a shit for following standards,
		843	# try to support some common forms of broken Location headers.
		844	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
		845	$hdr{location} =~ s/^\.\/+//;
		846
		847	my $url = "$rscheme://$uhost:$uport";
		848
		849	unless ($hdr{location} =~ s/^\///) {
		850	$url .= $upath;
		851	$url =~ s/\/[^\/]*$//;
		852	}
		853
		854	$hdr{location} = "$url/$hdr{location}";
		855	}
		856
		857	my $redirect;
		858
		859	if ($recurse) {
		860	my $status = $hdr{Status};
		861
		862	# industry standard is to redirect POST as GET for
		863	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		864	# also, the UA should ask the user for 301 and 307 and POST,
		865	# industry standard seems to be to simply follow.
		866	# we go with the industry standard.
		867	if ($status == 301 or $status == 302 or $status == 303) {
		868	# HTTP/1.1 is unclear on how to mutate the method
		869	$method = "GET" unless $method eq "HEAD";
		870	$redirect = 1;
		871	} elsif ($status == 307) {
		872	$redirect = 1;
288	}	873	}
289	}	874	}
		875
		876	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
		877	if ($state{handle}) {
		878	# handle keepalive
		879	if (
		880	$keepalive
		881	&& $_[3]
		882	&& ($hdr{HTTPVersion} < 1.1
		883	? $hdr{connection} =~ /\bkeep-?alive\b/i
		884	: $hdr{connection} !~ /\bclose\b/i)
		885	) {
		886	ka_store $ka_key, delete $state{handle};
		887	} else {
		888	# no keepalive, destroy the handle
		889	$state{handle}->destroy;
		890	}
		891	}
		892
		893	%state = ();
		894
		895	if (defined $_[1]) {
		896	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		897	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		898	}
		899
		900	# set-cookie processing
		901	if ($arg{cookie_jar}) {
		902	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		903	}
		904
		905	if ($redirect && exists $hdr{location}) {
		906	# we ignore any errors, as it is very common to receive
		907	# Content-Length != 0 but no actual body
		908	# we also access %hdr, as $_[1] might be an erro
		909	http_request (
		910	$method => $hdr{location},
		911	%arg,
		912	recurse => $recurse - 1,
		913	Redirect => [$_[0], \%hdr],
		914	$cb
		915	);
		916	} else {
		917	$cb->($_[0], \%hdr);
		918	}
		919	};
		920
		921	$ae_error = 597; # body phase
		922
		923	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		924
		925	my $len = $chunked ? undef : $hdr{"content-length"};
		926
		927	# body handling, many different code paths
		928	# - no body expected
		929	# - want_body_handle
		930	# - te chunked
		931	# - 2x length known (with or without on_body)
		932	# - 2x length not known (with or without on_body)
		933	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		934	$finish->(undef, 598 => "Request cancelled by on_header");
		935	} elsif (
		936	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		937	or $method eq "HEAD"
		938	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		939	) {
		940	# no body
		941	$finish->("", undef, undef, 1);
		942
		943	} elsif (!$redirect && $arg{want_body_handle}) {
		944	$_[0]->on_eof (undef);
		945	$_[0]->on_error (undef);
		946	$_[0]->on_read (undef);
		947
		948	$finish->(delete $state{handle});
		949
		950	} elsif ($chunked) {
		951	my $cl = 0;
		952	my $body = "";
		953	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		954
		955	$state{read_chunk} = sub {
		956	$_[1] =~ /^([0-9a-fA-F]+)/
		957	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		958
		959	my $len = hex $1;
		960
		961	if ($len) {
		962	$cl += $len;
		963
		964	$_[0]->push_read (chunk => $len, sub {
		965	$on_body->($_[1], \%hdr)
		966	or return $finish->(undef, 598 => "Request cancelled by on_body");
		967
		968	$_[0]->push_read (line => sub {
		969	length $_[1]
		970	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		971	$_[0]->push_read (line => $state{read_chunk});
		972	});
		973	});
		974	} else {
		975	$hdr{"content-length"} ||= $cl;
		976
		977	$_[0]->push_read (line => $qr_nlnl, sub {
		978	if (length $_[1]) {
		979	for ("$_[1]") {
		980	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		981
		982	my $hdr = _parse_hdr
		983	or return $finish->(undef, $ae_error => "Garbled response trailers");
		984
		985	%hdr = (%hdr, %$hdr);
		986	}
		987	}
		988
		989	$finish->($body, undef, undef, 1);
		990	});
		991	}
		992	};
		993
		994	$_[0]->push_read (line => $state{read_chunk});
		995
		996	} elsif ($arg{on_body}) {
		997	if (defined $len) {
		998	$_[0]->on_read (sub {
		999	$len -= length $_[0]{rbuf};
		1000
		1001	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1002	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1003
		1004	$len > 0
		1005	or $finish->("", undef, undef, 1);
		1006	});
		1007	} else {
		1008	$_[0]->on_eof (sub {
		1009	$finish->("");
		1010	});
		1011	$_[0]->on_read (sub {
		1012	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1013	or $finish->(undef, 598 => "Request cancelled by on_body");
		1014	});
		1015	}
		1016	} else {
		1017	$_[0]->on_eof (undef);
		1018
		1019	if (defined $len) {
		1020	$_[0]->on_read (sub {
		1021	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1022	if $len <= length $_[0]{rbuf};
		1023	});
		1024	} else {
		1025	$_[0]->on_error (sub {
		1026	($! == Errno::EPIPE || !$!)
		1027	? $finish->(delete $_[0]{rbuf})
		1028	: $finish->(undef, $ae_error => $_[2]);
		1029	});
		1030	$_[0]->on_read (sub { });
		1031	}
		1032	}
		1033	};
		1034
		1035	# if keepalive is enabled, then the server closing the connection
		1036	# before a response can happen legally - we retry on idempotent methods.
		1037	if ($keptalive && $idempotent) {
		1038	my $old_eof = $hdl->{on_eof};
		1039	$hdl->{on_eof} = sub {
		1040	_destroy_state %state;
		1041
		1042	http_request (
		1043	$method => $url,
		1044	%arg,
		1045	keepalive => 0,
		1046	$cb
		1047	);
		1048	};
		1049	$hdl->on_read (sub {
		1050	return unless %state;
		1051
		1052	# as soon as we receive something, a connection close
		1053	# once more becomes a hard error
		1054	$hdl->{on_eof} = $old_eof;
		1055	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1056	});
		1057	} else {
		1058	$hdl->push_read (line => $qr_nlnl, $state{read_response});
290	}	1059	}
291
292	$hdr{cookie} = join "; ", @cookie
293	if @cookie;
294	}	1060	};
295		1061
296	my ($rhost, $rport, $rpath); # request host, port, path	1062	my $prepare_handle = sub {
		1063	my ($hdl) = $state{handle};
297		1064
298	if ($proxy) {	1065	$hdl->timeout ($timeout);
299	($rhost, $rport, $scheme) = @$proxy;	1066	$hdl->on_error (sub {
300	$rpath = $url;	1067	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
301	} else {	1068	});
302	($rhost, $rport, $rpath) = ($uhost, $uport, $upath);	1069	$hdl->on_eof (sub {
303	$hdr{host} = $uhost;	1070	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1071	});
304	}	1072	};
305		1073
306	$hdr{"content-length"} = length $arg{body};	1074	# connected to proxy (or origin server)
		1075	my $connect_cb = sub {
		1076	my $fh = shift
		1077	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
307		1078
308	my %state = (connect_guard => 1);	1079	return unless delete $state{connect_guard};
		1080
		1081	# get handle
		1082	$state{handle} = new AnyEvent::Handle
		1083	%{ $arg{handle_params} },
		1084	fh => $fh,
		1085	peername => $uhost,
		1086	tls_ctx => $arg{tls_ctx},
		1087	;
		1088
		1089	$prepare_handle->();
		1090
		1091	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1092
		1093	# now handle proxy-CONNECT method
		1094	if ($proxy && $uscheme eq "https") {
		1095	# oh dear, we have to wrap it into a connect request
		1096
		1097	# maybe re-use $uauthority with patched port?
		1098	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1099	$state{handle}->push_read (line => $qr_nlnl, sub {
		1100	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1101	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1102
		1103	if ($2 == 200) {
		1104	$rpath = $upath;
		1105	$handle_actual_request->();
		1106	} else {
		1107	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1108	}
		1109	});
		1110	} else {
		1111	$handle_actual_request->();
		1112	}
		1113	};
309		1114
310	_get_slot $uhost, sub {	1115	_get_slot $uhost, sub {
311	$state{slot_guard} = shift;	1116	$state{slot_guard} = shift;
312		1117
313	return unless $state{connect_guard};	1118	return unless $state{connect_guard};
314		1119
315	$state{connect_guard} = AnyEvent::Socket::tcp_connect $rhost, $rport, sub {	1120	# try to use an existing keepalive connection, but only if we, ourselves, plan
316	$state{fh} = shift	1121	# on a keepalive request (in theory, this should be a separate config option).
317	or return $cb->(undef, { Status => 599, Reason => "$!", URL => $url });	1122	if ($keepalive && $KA_CACHE{$ka_key}) {
		1123	$keptalive = 1;
		1124	$state{handle} = ka_fetch $ka_key;
		1125	$prepare_handle->();
		1126	$handle_actual_request->();
318		1127
319	delete $state{connect_guard}; # reduce memory usage, save a tree
320
321	# get handle
322	$state{handle} = new AnyEvent::Handle
323	fh => $state{fh},
324	($scheme eq "https" ? (tls => "connect") : ());
325
326	# limit the number of persistent connections
327	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
328	++$KA_COUNT{$_[1]};
329	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
330	$hdr{connection} = "keep-alive";
331	delete $hdr{connection}; # keep-alive not yet supported
332	} else {	1128	} else {
333	delete $hdr{connection};	1129	my $tcp_connect = $arg{tcp_connect}
334	}	1130	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
335		1131
336	# (re-)configure handle	1132	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
337	$state{handle}->timeout ($timeout);
338	$state{handle}->on_error (sub {
339	my $errno = "$!";
340	%state = ();
341	$cb->(undef, { Status => 599, Reason => $errno, URL => $url });
342	});
343	$state{handle}->on_eof (sub {
344	%state = ();
345	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file", URL => $url });
346	});
347
348	# send request
349	$state{handle}->push_write (
350	"$method $rpath HTTP/1.0\015\012"
351	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
352	. "\015\012"
353	. (delete $arg{body})
354	);
355
356	%hdr = (); # reduce memory usage, save a kitten
357
358	# status line
359	$state{handle}->push_read (line => qr/\015?\012/, sub {
360	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix
361	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])", URL => $url }));
362
363	my %hdr = ( # response headers
364	HTTPVersion => "\x00$1",
365	Status => "\x00$2",
366	Reason => "\x00$3",
367	URL => "\x00$url"
368	);
369
370	# headers, could be optimized a bit
371	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
372	for ("$_[1]\012") {
373	# we support spaces in field names, as lotus domino
374	# creates them.
375	$hdr{lc $1} .= "\x00$2"
376	while /\G
377	([^:\000-\037]+):
378	[\011\040]*
379	((?: [^\015\012]+ | \015?\012[\011\040] )*)
380	\015?\012
381	/gxc;
382
383	/\G$/
384	or return (%state = (), $cb->(undef, { Status => 599, Reason => "garbled response headers", URL => $url }));
385	}
386
387	substr $_, 0, 1, ""
388	for values %hdr;
389
390	my $finish = sub {
391	%state = ();
392
393	# set-cookie processing
394	if ($arg{cookie_jar} && exists $hdr{"set-cookie"}) {
395	for (split /\x00/, $hdr{"set-cookie"}) {
396	my ($cookie, @arg) = split /;\s*/;
397	my ($name, $value) = split /=/, $cookie, 2;
398	my %kv = (value => $value, map { split /=/, $_, 2 } @arg);
399
400	my $cdom = (delete $kv{domain}) || $uhost;
401	my $cpath = (delete $kv{path}) || "/";
402
403	$cdom =~ s/^.?/./; # make sure it starts with a "."
404
405	next if $cdom =~ /\.$/;
406
407	# this is not rfc-like and not netscape-like. go figure.
408	my $ndots = $cdom =~ y/.//;
409	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
410
411	# store it
412	$arg{cookie_jar}{version} = 1;
413	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
414	}
415	}
416
417	# microsoft and other assholes don't give a shit for following standards,
418	# try to support a common form of broken Location header.
419	$_[1]{location} =~ s%^/%$scheme://$uhost:$uport/%
420	if exists $_[1]{location};
421
422	if ($_[1]{Status} =~ /^30[12]$/ && $recurse && $method ne "POST") {
423	# apparently, mozilla et al. just change POST to GET here
424	# more research is needed before we do the same
425	http_request ($method, $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
426	} elsif ($_[1]{Status} == 303 && $recurse) {
427	# even http/1.1 is unlear on how to mutate the method
428	$method = "GET" unless $method eq "HEAD";
429	http_request ($method => $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
430	} elsif ($_[1]{Status} == 307 && $recurse && $method =~ /^(?:GET|HEAD)$/) {
431	http_request ($method => $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
432	} else {
433	$cb->($_[0], $_[1]);
434	}
435	};
436
437	if ($hdr{Status} =~ /^(?:1..|204|304)$/ or $method eq "HEAD") {
438	$finish->(undef, \%hdr);
439	} else {
440	if (exists $hdr{"content-length"}) {
441	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
442	# could cache persistent connection now
443	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
444	# but we don't, due to misdesigns, this is annoyingly complex
445	};
446
447	$finish->($_[1], \%hdr);
448	});
449	} else {
450	# too bad, need to read until we get an error or EOF,
451	# no way to detect winged data.
452	$_[0]->on_error (sub {
453	$finish->($_[0]{rbuf}, \%hdr);
454	});
455	$_[0]->on_eof (undef);
456	$_[0]->on_read (sub { });
457	}
458	}
459	});
460	});
461	}, sub {
462	$timeout
463	};	1133	}
464	};	1134	};
465		1135
466	defined wantarray && AnyEvent::Util::guard { %state = () }	1136	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
467	}	1137	}
468		1138
469	sub http_get($@) {	1139	sub http_get($@) {
470	unshift @_, "GET";	1140	unshift @_, "GET";
471	&http_request	1141	&http_request
…		…
482	&http_request	1152	&http_request
483	}	1153	}
484		1154
485	=back	1155	=back
486		1156
		1157	=head2 DNS CACHING
		1158
		1159	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1160	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1161	hostnames. The latter is a simple stub resolver and does no caching
		1162	on its own. If you want DNS caching, you currently have to provide
		1163	your own default resolver (by storing a suitable resolver object in
		1164	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1165
487	=head2 GLOBAL FUNCTIONS AND VARIABLES	1166	=head2 GLOBAL FUNCTIONS AND VARIABLES
488		1167
489	=over 4	1168	=over 4
490		1169
491	=item AnyEvent::HTTP::set_proxy "proxy-url"	1170	=item AnyEvent::HTTP::set_proxy "proxy-url"
492		1171
493	Sets the default proxy server to use. The proxy-url must begin with a	1172	Sets the default proxy server to use. The proxy-url must begin with a
494	string of the form C<http://host:port> (optionally C<https:...>).	1173	string of the form C<http://host:port>, croaks otherwise.
		1174
		1175	To clear an already-set proxy, use C<undef>.
		1176
		1177	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1178
		1179	Remove all cookies from the cookie jar that have been expired. If
		1180	C<$session_end> is given and true, then additionally remove all session
		1181	cookies.
		1182
		1183	You should call this function (with a true C<$session_end>) before you
		1184	save cookies to disk, and you should call this function after loading them
		1185	again. If you have a long-running program you can additonally call this
		1186	function from time to time.
		1187
		1188	A cookie jar is initially an empty hash-reference that is managed by this
		1189	module. It's format is subject to change, but currently it is like this:
		1190
		1191	The key C<version> has to contain C<1>, otherwise the hash gets
		1192	emptied. All other keys are hostnames or IP addresses pointing to
		1193	hash-references. The key for these inner hash references is the
		1194	server path for which this cookie is meant, and the values are again
		1195	hash-references. The keys of those hash-references is the cookie name, and
		1196	the value, you guessed it, is another hash-reference, this time with the
		1197	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1198	which have been replaced by a C<_expires> key that contains the cookie
		1199	expiry timestamp.
		1200
		1201	Here is an example of a cookie jar with a single cookie, so you have a
		1202	chance of understanding the above paragraph:
		1203
		1204	{
		1205	version => 1,
		1206	"10.0.0.1" => {
		1207	"/" => {
		1208	"mythweb_id" => {
		1209	_expires => 1293917923,
		1210	value => "ooRung9dThee3ooyXooM1Ohm",
		1211	},
		1212	},
		1213	},
		1214	}
		1215
		1216	=item $date = AnyEvent::HTTP::format_date $timestamp
		1217
		1218	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1219	Date (RFC 2616).
		1220
		1221	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1222
		1223	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1224	bunch of minor variations of those, and returns the corresponding POSIX
		1225	timestamp, or C<undef> if the date cannot be parsed.
495		1226
496	=item $AnyEvent::HTTP::MAX_RECURSE	1227	=item $AnyEvent::HTTP::MAX_RECURSE
497		1228
498	The default value for the C<recurse> request parameter (default: C<10>).	1229	The default value for the C<recurse> request parameter (default: C<10>).
499		1230
		1231	=item $AnyEvent::HTTP::TIMEOUT
		1232
		1233	The default timeout for conenction operations (default: C<300>).
		1234
500	=item $AnyEvent::HTTP::USERAGENT	1235	=item $AnyEvent::HTTP::USERAGENT
501		1236
502	The default value for the C<User-Agent> header (the default is	1237	The default value for the C<User-Agent> header (the default is
503	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1238	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
504		1239
505	=item $AnyEvent::HTTP::MAX_PERSISTENT	1240	=item $AnyEvent::HTTP::MAX_PER_HOST
506		1241
507	The maximum number of persistent connections to keep open (default: 8).	1242	The maximum number of concurrent connections to the same host (identified
		1243	by the hostname). If the limit is exceeded, then the additional requests
		1244	are queued until previous connections are closed. Both persistent and
		1245	non-persistent connections are counted in this limit.
508		1246
509	Not implemented currently.	1247	The default value for this is C<4>, and it is highly advisable to not
		1248	increase it much.
		1249
		1250	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1251	connections, older browsers used 2, newers (such as firefox 3) typically
		1252	use 6, and Opera uses 8 because like, they have the fastest browser and
		1253	give a shit for everybody else on the planet.
510		1254
511	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1255	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
512		1256
513	The maximum time to cache a persistent connection, in seconds (default: 2).	1257	The time after which idle persistent conenctions get closed by
514		1258	AnyEvent::HTTP (default: C<3>).
515	Not implemented currently.
516		1259
517	=item $AnyEvent::HTTP::ACTIVE	1260	=item $AnyEvent::HTTP::ACTIVE
518		1261
519	The number of active connections. This is not the number of currently	1262	The number of active connections. This is not the number of currently
520	running requests, but the number of currently open and non-idle TCP	1263	running requests, but the number of currently open and non-idle TCP
521	connections. This number of can be useful for load-leveling.	1264	connections. This number can be useful for load-leveling.
522		1265
523	=back	1266	=back
524		1267
525	=cut	1268	=cut
526		1269
		1270	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1271	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1272
		1273	sub format_date($) {
		1274	my ($time) = @_;
		1275
		1276	# RFC 822/1123 format
		1277	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1278
		1279	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1280	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1281	$H, $M, $S;
		1282	}
		1283
		1284	sub parse_date($) {
		1285	my ($date) = @_;
		1286
		1287	my ($d, $m, $y, $H, $M, $S);
		1288
		1289	if ($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$/) {
		1290	# RFC 822/1123, required by RFC 2616 (with " ")
		1291	# cookie dates (with "-")
		1292
		1293	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1294
		1295	} elsif ($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$/) {
		1296	# RFC 850
		1297	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1298
		1299	} elsif ($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])$/) {
		1300	# ISO C's asctime
		1301	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1302	}
		1303	# other formats fail in the loop below
		1304
		1305	for (0..11) {
		1306	if ($m eq $month[$_]) {
		1307	require Time::Local;
		1308	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);
		1309	}
		1310	}
		1311
		1312	undef
		1313	}
		1314
527	sub set_proxy($) {	1315	sub set_proxy($) {
528	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1316	if (length $_[0]) {
		1317	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1318	or Carp::croak "$_[0]: invalid proxy URL";
		1319	$PROXY = [$2, $3 || 3128, $1]
		1320	} else {
		1321	undef $PROXY;
		1322	}
529	}	1323	}
530		1324
531	# initialise proxy from environment	1325	# initialise proxy from environment
		1326	eval {
532	set_proxy $ENV{http_proxy};	1327	set_proxy $ENV{http_proxy};
		1328	};
		1329
		1330	=head2 SHOWCASE
		1331
		1332	This section contaisn some more elaborate "real-world" examples or code
		1333	snippets.
		1334
		1335	=head2 HTTP/1.1 FILE DOWNLOAD
		1336
		1337	Downloading files with HTTP cna be quite tricky, especially when something
		1338	goes wrong and you want tor esume.
		1339
		1340	Here is a function that initiates and resumes a download. It uses the
		1341	last modified time to check for file content changes, and works with many
		1342	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1343	on older servers.
		1344
		1345	It calls the completion callback with either C<undef>, which means a
		1346	nonretryable error occured, C<0> when the download was partial and should
		1347	be retried, and C<1> if it was successful.
		1348
		1349	use AnyEvent::HTTP;
		1350
		1351	sub download($$$) {
		1352	my ($url, $file, $cb) = @_;
		1353
		1354	open my $fh, "+<", $file
		1355	or die "$file: $!";
		1356
		1357	my %hdr;
		1358	my $ofs = 0;
		1359
		1360	warn stat $fh;
		1361	warn -s _;
		1362	if (stat $fh and -s _) {
		1363	$ofs = -s _;
		1364	warn "-s is ", $ofs;#d#
		1365	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1366	$hdr{"range"} = "bytes=$ofs-";
		1367	}
		1368
		1369	http_get $url,
		1370	headers => \%hdr,
		1371	on_header => sub {
		1372	my ($hdr) = @_;
		1373
		1374	if ($hdr->{Status} == 200 && $ofs) {
		1375	# resume failed
		1376	truncate $fh, $ofs = 0;
		1377	}
		1378
		1379	sysseek $fh, $ofs, 0;
		1380
		1381	1
		1382	},
		1383	on_body => sub {
		1384	my ($data, $hdr) = @_;
		1385
		1386	if ($hdr->{Status} =~ /^2/) {
		1387	length $data == syswrite $fh, $data
		1388	or return; # abort on write errors
		1389	}
		1390
		1391	1
		1392	},
		1393	sub {
		1394	my (undef, $hdr) = @_;
		1395
		1396	my $status = $hdr->{Status};
		1397
		1398	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1399	utime $fh, $time, $time;
		1400	}
		1401
		1402	if ($status == 200 || $status == 206 || $status == 416) {
		1403	# download ok || resume ok || file already fully downloaded
		1404	$cb->(1, $hdr);
		1405
		1406	} elsif ($status == 412) {
		1407	# file has changed while resuming, delete and retry
		1408	unlink $file;
		1409	$cb->(0, $hdr);
		1410
		1411	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1412	# retry later
		1413	$cb->(0, $hdr);
		1414
		1415	} else {
		1416	$cb->(undef, $hdr);
		1417	}
		1418	}
		1419	;
		1420	}
		1421
		1422	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1423	if ($_[0]) {
		1424	print "OK!\n";
		1425	} elsif (defined $_[0]) {
		1426	print "please retry later\n";
		1427	} else {
		1428	print "ERROR\n";
		1429	}
		1430	};
		1431
		1432	=head3 SOCKS PROXIES
		1433
		1434	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1435	compile your perl to support socks, or use an external program such as
		1436	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1437	transparently.
		1438
		1439	Alternatively, for AnyEvent::HTTP only, you can use your own
		1440	C<tcp_connect> function that does the proxy handshake - here is an example
		1441	that works with socks4a proxies:
		1442
		1443	use Errno;
		1444	use AnyEvent::Util;
		1445	use AnyEvent::Socket;
		1446	use AnyEvent::Handle;
		1447
		1448	# host, port and username of/for your socks4a proxy
		1449	my $socks_host = "10.0.0.23";
		1450	my $socks_port = 9050;
		1451	my $socks_user = "";
		1452
		1453	sub socks4a_connect {
		1454	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1455
		1456	my $hdl = new AnyEvent::Handle
		1457	connect => [$socks_host, $socks_port],
		1458	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1459	on_error => sub { $connect_cb->() },
		1460	;
		1461
		1462	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1463
		1464	$hdl->push_read (chunk => 8, sub {
		1465	my ($hdl, $chunk) = @_;
		1466	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1467
		1468	if ($status == 0x5a) {
		1469	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1470	} else {
		1471	$! = Errno::ENXIO; $connect_cb->();
		1472	}
		1473	});
		1474
		1475	$hdl
		1476	}
		1477
		1478	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1479	possibly after switching off other proxy types:
		1480
		1481	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1482
		1483	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1484	my ($data, $headers) = @_;
		1485	...
		1486	};
533		1487
534	=head1 SEE ALSO	1488	=head1 SEE ALSO
535		1489
536	L<AnyEvent>.	1490	L<AnyEvent>.
537		1491
538	=head1 AUTHOR	1492	=head1 AUTHOR
539		1493
540	Marc Lehmann <schmorp@schmorp.de>	1494	Marc Lehmann <schmorp@schmorp.de>
541	http://home.schmorp.de/	1495	http://home.schmorp.de/
542		1496
		1497	With many thanks to Дмитрий Шалашов, who provided countless
		1498	testcases and bugreports.
		1499
543	=cut	1500	=cut
544		1501
545	1	1502	1
546		1503

Diff Legend

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