ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.34 by root, Wed Oct 29 14:51:07 2008 UTC vs.
Revision 1.114 by root, Mon Jan 14 21:36:26 2013 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 = '1.05';	51	our $VERSION = '2.15';
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
…		…
92	must be an absolute http or https URL.	86	must be an absolute http or https URL.
93		87
94	When called in void context, nothing is returned. In other contexts,	88	When called in void context, nothing is returned. In other contexts,
95	C<http_request> returns a "cancellation guard" - you have to keep the	89	C<http_request> returns a "cancellation guard" - you have to keep the
96	object at least alive until the callback get called. If the object gets	90	object at least alive until the callback get called. If the object gets
97	destroyed before the callbakc is called, the request will be cancelled.	91	destroyed before the callback is called, the request will be cancelled.
98		92
99	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
100	(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
101	response headers as second argument.	95	(and trailers) as second argument.
102		96
103	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
104	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
105	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
106	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
107	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).
		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	$headers->{Redirect}[1]{Redirect} >> will then contain the original
		116	response, and so on.
108		117
109	If the server sends a header multiple times, then their contents will be	118	If the server sends a header multiple times, then their contents will be
110	joined together with a comma (C<,>), as per the HTTP spec.	119	joined together with a comma (C<,>), as per the HTTP spec.
111		120
112	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
113	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<599>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
114	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection establishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
115		139
116	A typical callback might look like this:	140	A typical callback might look like this:
117		141
118	sub {	142	sub {
119	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
130		154
131	=over 4	155	=over 4
132		156
133	=item recurse => $count (default: $MAX_RECURSE)	157	=item recurse => $count (default: $MAX_RECURSE)
134		158
135	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
136	retries and so on, and how often to do so.	160	other retries and so on, and how often to do so.
137		161
138	=item headers => hashref	162	=item headers => hashref
139		163
140	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
141	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
142	and will provide defaults for C<User-Agent:> and C<Referer:>.	166	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
		167	(this can be suppressed by using C<undef> for these headers in which case
		168	they won't be sent at all).
		169
		170	You really should provide your own C<User-Agent:> header value that is
		171	appropriate for your program - I wouldn't be surprised if the default
		172	AnyEvent string gets blocked by webservers sooner or later.
		173
		174	Also, make sure that your headers names and values do not contain any
		175	embedded newlines.
143		176
144	=item timeout => $seconds	177	=item timeout => $seconds
145		178
146	The time-out to use for various stages - each connect attempt will reset	179	The time-out to use for various stages - each connect attempt will reset
147	the timeout, as will read or write activity. Default timeout is 5 minutes.	180	the timeout, as will read or write activity, i.e. this is not an overall
		181	timeout.
		182
		183	Default timeout is 5 minutes.
148		184
149	=item proxy => [$host, $port[, $scheme]] or undef	185	=item proxy => [$host, $port[, $scheme]] or undef
150		186
151	Use the given http proxy for all requests. If not specified, then the	187	Use the given http proxy for all requests, or no proxy if C<undef> is
152	default proxy (as specified by C<$ENV{http_proxy}>) is used.	188	used.
153		189
154	C<$scheme> must be either missing or C<http> for HTTP, or C<https> for	190	C<$scheme> must be either missing or must be C<http> for HTTP.
155	HTTPS.	191
		192	If not specified, then the default proxy is used (see
		193	C<AnyEvent::HTTP::set_proxy>).
156		194
157	=item body => $string	195	=item body => $string
158		196
159	The request body, usually empty. Will be-sent as-is (future versions of	197	The request body, usually empty. Will be sent as-is (future versions of
160	this module might offer more options).	198	this module might offer more options).
161		199
162	=item cookie_jar => $hash_ref	200	=item cookie_jar => $hash_ref
163		201
164	Passing this parameter enables (simplified) cookie-processing, loosely	202	Passing this parameter enables (simplified) cookie-processing, loosely
165	based on the original netscape specification.	203	based on the original netscape specification.
166		204
167	The C<$hash_ref> must be an (initially empty) hash reference which will	205	The C<$hash_ref> must be an (initially empty) hash reference which
168	get updated automatically. It is possible to save the cookie_jar to	206	will get updated automatically. It is possible to save the cookie jar
169	persistent storage with something like JSON or Storable, but this is not	207	to persistent storage with something like JSON or Storable - see the
170	recommended, as expire times are currently being ignored.	208	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		209	expired or session-only cookies, and also for documentation on the format
		210	of the cookie jar.
171		211
172	Note that this cookie implementation is not of very high quality, nor	212	Note that this cookie implementation is not meant to be complete. If
173	meant to be complete. If you want complete cookie management you have to	213	you want complete cookie management you have to do that on your
174	do that on your own. C<cookie_jar> is meant as a quick fix to get some	214	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
175	cookie-using sites working. Cookies are a privacy disaster, do not use	215	working. Cookies are a privacy disaster, do not use them unless required
176	them unless required to.	216	to.
		217
		218	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		219	headers will be set and handled by this module, otherwise they will be
		220	left untouched.
		221
		222	=item tls_ctx => $scheme | $tls_ctx
		223
		224	Specifies the AnyEvent::TLS context to be used for https connections. This
		225	parameter follows the same rules as the C<tls_ctx> parameter to
		226	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		227	C<high> can be specified, which give you a predefined low-security (no
		228	verification, highest compatibility) and high-security (CA and common-name
		229	verification) TLS context.
		230
		231	The default for this option is C<low>, which could be interpreted as "give
		232	me the page, no matter what".
		233
		234	See also the C<sessionid> parameter.
		235
		236	=item session => $string
		237
		238	The module might reuse connections to the same host internally. Sometimes
		239	(e.g. when using TLS), you do not want to reuse connections from other
		240	sessions. This can be achieved by setting this parameter to some unique
		241	ID (such as the address of an object storing your state data, or the TLS
		242	context) - only connections using the same unique ID will be reused.
		243
		244	=item on_prepare => $callback->($fh)
		245
		246	In rare cases you need to "tune" the socket before it is used to
		247	connect (for exmaple, to bind it on a given IP address). This parameter
		248	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		249	and behaves exactly the same way (e.g. it has to provide a
		250	timeout). See the description for the C<$prepare_cb> argument of
		251	C<AnyEvent::Socket::tcp_connect> for details.
		252
		253	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		254
		255	In even rarer cases you want total control over how AnyEvent::HTTP
		256	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		257	to do this, but you can provide your own C<tcp_connect> function -
		258	obviously, it has to follow the same calling conventions, except that it
		259	may always return a connection guard object.
		260
		261	There are probably lots of weird uses for this function, starting from
		262	tracing the hosts C<http_request> actually tries to connect, to (inexact
		263	but fast) host => IP address caching or even socks protocol support.
		264
		265	=item on_header => $callback->($headers)
		266
		267	When specified, this callback will be called with the header hash as soon
		268	as headers have been successfully received from the remote server (not on
		269	locally-generated errors).
		270
		271	It has to return either true (in which case AnyEvent::HTTP will continue),
		272	or false, in which case AnyEvent::HTTP will cancel the download (and call
		273	the finish callback with an error code of C<598>).
		274
		275	This callback is useful, among other things, to quickly reject unwanted
		276	content, which, if it is supposed to be rare, can be faster than first
		277	doing a C<HEAD> request.
		278
		279	The downside is that cancelling the request makes it impossible to re-use
		280	the connection. Also, the C<on_header> callback will not receive any
		281	trailer (headers sent after the response body).
		282
		283	Example: cancel the request unless the content-type is "text/html".
		284
		285	on_header => sub {
		286	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		287	},
		288
		289	=item on_body => $callback->($partial_body, $headers)
		290
		291	When specified, all body data will be passed to this callback instead of
		292	to the completion callback. The completion callback will get the empty
		293	string instead of the body data.
		294
		295	It has to return either true (in which case AnyEvent::HTTP will continue),
		296	or false, in which case AnyEvent::HTTP will cancel the download (and call
		297	the completion callback with an error code of C<598>).
		298
		299	The downside to cancelling the request is that it makes it impossible to
		300	re-use the connection.
		301
		302	This callback is useful when the data is too large to be held in memory
		303	(so the callback writes it to a file) or when only some information should
		304	be extracted, or when the body should be processed incrementally.
		305
		306	It is usually preferred over doing your own body handling via
		307	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		308	only used to create a connection, C<want_body_handle> is the better
		309	alternative, as it allows you to install your own event handler, reducing
		310	resource usage.
		311
		312	=item want_body_handle => $enable
		313
		314	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		315	changes considerably: after parsing the headers, and instead of
		316	downloading the body (if any), the completion callback will be
		317	called. Instead of the C<$body> argument containing the body data, the
		318	callback will receive the L<AnyEvent::Handle> object associated with the
		319	connection. In error cases, C<undef> will be passed. When there is no body
		320	(e.g. status C<304>), the empty string will be passed.
		321
		322	The handle object might or might not be in TLS mode, might be connected
		323	to a proxy, be a persistent connection, use chunked transfer encoding
		324	etc., and configured in unspecified ways. The user is responsible for this
		325	handle (it will not be used by this module anymore).
		326
		327	This is useful with some push-type services, where, after the initial
		328	headers, an interactive protocol is used (typical example would be the
		329	push-style twitter API which starts a JSON/XML stream).
		330
		331	If you think you need this, first have a look at C<on_body>, to see if
		332	that doesn't solve your problem in a better way.
		333
		334	=item persistent => $boolean
		335
		336	Try to create/reuse a persistent connection. When this flag is set
		337	(default: true for idempotent requests, false for all others), then
		338	C<http_request> tries to re-use an existing (previously-created)
		339	persistent connection to the host and, failing that, tries to create a new
		340	one.
		341
		342	Requests failing in certain ways will be automatically retried once, which
		343	is dangerous for non-idempotent requests, which is why it defaults to off
		344	for them. The reason for this is because the bozos who designed HTTP/1.1
		345	made it impossible to distinguish between a fatal error and a normal
		346	connection timeout, so you never know whether there was a problem with
		347	your request or not.
		348
		349	When reusing an existent connection, many parameters (such as TLS context)
		350	will be ignored. See the C<session> parameter for a workaround.
		351
		352	=item keepalive => $boolean
		353
		354	Only used when C<persistent> is also true. This parameter decides whether
		355	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		356	(as opposed to only a HTTP/1.1 persistent connection).
		357
		358	The default is true, except when using a proxy, in which case it defaults
		359	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		360
		361	=item handle_params => { key => value ... }
		362
		363	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		364	constructor that is called - not all requests will create a handle, and
		365	sometimes more than one is created, so this parameter is only good for
		366	setting hints.
		367
		368	Example: set the maximum read size to 4096, to potentially conserve memory
		369	at the cost of speed.
		370
		371	handle_params => {
		372	max_read_size => 4096,
		373	},
177		374
178	=back	375	=back
179		376
180	Example: make a simple HTTP GET request for http://www.nethype.de/	377	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		378	the response body.
181		379
182	http_request GET => "http://www.nethype.de/", sub {	380	http_request GET => "http://www.nethype.de/", sub {
183	my ($body, $hdr) = @_;	381	my ($body, $hdr) = @_;
184	print "$body\n";	382	print "$body\n";
185	};	383	};
186		384
187	Example: make a HTTP HEAD request on https://www.google.com/, use a	385	Example: do a HTTP HEAD request on https://www.google.com/, use a
188	timeout of 30 seconds.	386	timeout of 30 seconds.
189		387
190	http_request	388	http_request
191	GET => "https://www.google.com",	389	HEAD => "https://www.google.com",
		390	headers => { "user-agent" => "MySearchClient 1.0" },
192	timeout => 30,	391	timeout => 30,
193	sub {	392	sub {
194	my ($body, $hdr) = @_;	393	my ($body, $hdr) = @_;
195	use Data::Dumper;	394	use Data::Dumper;
196	print Dumper $hdr;	395	print Dumper $hdr;
197	}	396	}
198	;	397	;
199		398
200	Example: make another simple HTTP GET request, but immediately try to	399	Example: do another simple HTTP GET request, but immediately try to
201	cancel it.	400	cancel it.
202		401
203	my $request = http_request GET => "http://www.nethype.de/", sub {	402	my $request = http_request GET => "http://www.nethype.de/", sub {
204	my ($body, $hdr) = @_;	403	my ($body, $hdr) = @_;
205	print "$body\n";	404	print "$body\n";
206	};	405	};
207		406
208	undef $request;	407	undef $request;
209		408
210	=cut	409	=cut
		410
		411	#############################################################################
		412	# wait queue/slots
211		413
212	sub _slot_schedule;	414	sub _slot_schedule;
213	sub _slot_schedule($) {	415	sub _slot_schedule($) {
214	my $host = shift;	416	my $host = shift;
215		417
…		…
237	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	439	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
238		440
239	_slot_schedule $_[0];	441	_slot_schedule $_[0];
240	}	442	}
241		443
242	our $qr_nl = qr<\015?\012>;	444	#############################################################################
243	our $qr_nlnl = qr<\015?\012\015?\012>;	445	# cookie handling
		446
		447	# expire cookies
		448	sub cookie_jar_expire($;$) {
		449	my ($jar, $session_end) = @_;
		450
		451	%$jar = () if $jar->{version} != 1;
		452
		453	my $anow = AE::now;
		454
		455	while (my ($chost, $paths) = each %$jar) {
		456	next unless ref $paths;
		457
		458	while (my ($cpath, $cookies) = each %$paths) {
		459	while (my ($cookie, $kv) = each %$cookies) {
		460	if (exists $kv->{_expires}) {
		461	delete $cookies->{$cookie}
		462	if $anow > $kv->{_expires};
		463	} elsif ($session_end) {
		464	delete $cookies->{$cookie};
		465	}
		466	}
		467
		468	delete $paths->{$cpath}
		469	unless %$cookies;
		470	}
		471
		472	delete $jar->{$chost}
		473	unless %$paths;
		474	}
		475	}
		476
		477	# extract cookies from jar
		478	sub cookie_jar_extract($$$$) {
		479	my ($jar, $scheme, $host, $path) = @_;
		480
		481	%$jar = () if $jar->{version} != 1;
		482
		483	my @cookies;
		484
		485	while (my ($chost, $paths) = each %$jar) {
		486	next unless ref $paths;
		487
		488	if ($chost =~ /^\./) {
		489	next unless $chost eq substr $host, -length $chost;
		490	} elsif ($chost =~ /\./) {
		491	next unless $chost eq $host;
		492	} else {
		493	next;
		494	}
		495
		496	while (my ($cpath, $cookies) = each %$paths) {
		497	next unless $cpath eq substr $path, 0, length $cpath;
		498
		499	while (my ($cookie, $kv) = each %$cookies) {
		500	next if $scheme ne "https" && exists $kv->{secure};
		501
		502	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		503	delete $cookies->{$cookie};
		504	next;
		505	}
		506
		507	my $value = $kv->{value};
		508
		509	if ($value =~ /[=;,[:space:]]/) {
		510	$value =~ s/([\\"])/\\$1/g;
		511	$value = "\"$value\"";
		512	}
		513
		514	push @cookies, "$cookie=$value";
		515	}
		516	}
		517	}
		518
		519	\@cookies
		520	}
		521
		522	# parse set_cookie header into jar
		523	sub cookie_jar_set_cookie($$$$) {
		524	my ($jar, $set_cookie, $host, $date) = @_;
		525
		526	my $anow = int AE::now;
		527	my $snow; # server-now
		528
		529	for ($set_cookie) {
		530	# parse NAME=VALUE
		531	my @kv;
		532
		533	# expires is not http-compliant in the original cookie-spec,
		534	# we support the official date format and some extensions
		535	while (
		536	m{
		537	\G\s*
		538	(?:
		539	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		540	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
		541	)
		542	}gcxsi
		543	) {
		544	my $name = $2;
		545	my $value = $4;
		546
		547	if (defined $1) {
		548	# expires
		549	$name = "expires";
		550	$value = $1;
		551	} elsif (defined $3) {
		552	# quoted
		553	$value = $3;
		554	$value =~ s/\\(.)/$1/gs;
		555	}
		556
		557	push @kv, @kv ? lc $name : $name, $value;
		558
		559	last unless /\G\s*;/gc;
		560	}
		561
		562	last unless @kv;
		563
		564	my $name = shift @kv;
		565	my %kv = (value => shift @kv, @kv);
		566
		567	if (exists $kv{"max-age"}) {
		568	$kv{_expires} = $anow + delete $kv{"max-age"};
		569	} elsif (exists $kv{expires}) {
		570	$snow ||= parse_date ($date) || $anow;
		571	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		572	} else {
		573	delete $kv{_expires};
		574	}
		575
		576	my $cdom;
		577	my $cpath = (delete $kv{path}) || "/";
		578
		579	if (exists $kv{domain}) {
		580	$cdom = delete $kv{domain};
		581
		582	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		583
		584	next if $cdom =~ /\.$/;
		585
		586	# this is not rfc-like and not netscape-like. go figure.
		587	my $ndots = $cdom =~ y/.//;
		588	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		589	} else {
		590	$cdom = $host;
		591	}
		592
		593	# store it
		594	$jar->{version} = 1;
		595	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		596
		597	redo if /\G\s*,/gc;
		598	}
		599	}
		600
		601	#############################################################################
		602	# keepalive/persistent connection cache
		603
		604	# fetch a connection from the keepalive cache
		605	sub ka_fetch($) {
		606	my $ka_key = shift;
		607
		608	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		609	delete $KA_CACHE{$ka_key}
		610	unless @{ $KA_CACHE{$ka_key} };
		611
		612	$hdl
		613	}
		614
		615	sub ka_store($$) {
		616	my ($ka_key, $hdl) = @_;
		617
		618	my $kaa = $KA_CACHE{$ka_key} ||= [];
		619
		620	my $destroy = sub {
		621	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		622
		623	$hdl->destroy;
		624
		625	@ka
		626	? $KA_CACHE{$ka_key} = \@ka
		627	: delete $KA_CACHE{$ka_key};
		628	};
		629
		630	# on error etc., destroy
		631	$hdl->on_error ($destroy);
		632	$hdl->on_eof ($destroy);
		633	$hdl->on_read ($destroy);
		634	$hdl->timeout ($PERSISTENT_TIMEOUT);
		635
		636	push @$kaa, $hdl;
		637	shift @$kaa while @$kaa > $MAX_PER_HOST;
		638	}
		639
		640	#############################################################################
		641	# utilities
		642
		643	# continue to parse $_ for headers and place them into the arg
		644	sub _parse_hdr() {
		645	my %hdr;
		646
		647	# things seen, not parsed:
		648	# p3pP="NON CUR OTPi OUR NOR UNI"
		649
		650	$hdr{lc $1} .= ",$2"
		651	while /\G
		652	([^:\000-\037]*):
		653	[\011\040]*
		654	((?: [^\012]+ | \012[\011\040] )*)
		655	\012
		656	/gxc;
		657
		658	/\G$/
		659	or return;
		660
		661	# remove the "," prefix we added to all headers above
		662	substr $_, 0, 1, ""
		663	for values %hdr;
		664
		665	\%hdr
		666	}
		667
		668	#############################################################################
		669	# http_get
		670
		671	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		672
		673	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		674	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		675
		676	# maybe it should just become a normal object :/
		677
		678	sub _destroy_state(\%) {
		679	my ($state) = @_;
		680
		681	$state->{handle}->destroy if $state->{handle};
		682	%$state = ();
		683	}
		684
		685	sub _error(\%$$) {
		686	my ($state, $cb, $hdr) = @_;
		687
		688	&_destroy_state ($state);
		689
		690	$cb->(undef, $hdr);
		691	()
		692	}
244		693
245	sub http_request($$@) {	694	sub http_request($$@) {
246	my $cb = pop;	695	my $cb = pop;
247	my ($method, $url, %arg) = @_;	696	my ($method, $url, %arg) = @_;
248		697
249	my %hdr;	698	my %hdr;
		699
		700	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		701	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
250		702
251	$method = uc $method;	703	$method = uc $method;
252		704
253	if (my $hdr = $arg{headers}) {	705	if (my $hdr = $arg{headers}) {
254	while (my ($k, $v) = each %$hdr) {	706	while (my ($k, $v) = each %$hdr) {
255	$hdr{lc $k} = $v;	707	$hdr{lc $k} = $v;
256	}	708	}
257	}	709	}
258		710
		711	# pseudo headers for all subsequent responses
		712	my @pseudo = (URL => $url);
		713	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		714
259	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	715	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
260		716
261	return $cb->(undef, { Status => 599, Reason => "recursion limit reached", URL => $url })	717	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
262	if $recurse < 0;	718	if $recurse < 0;
263		719
264	my $proxy = $arg{proxy} || $PROXY;	720	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
265	my $timeout = $arg{timeout} || $TIMEOUT;	721	my $timeout = $arg{timeout} || $TIMEOUT;
266		722
267	$hdr{"user-agent"} ||= $USERAGENT;
268
269	my ($uscheme, $uauthority, $upath, $query, $fragment) =	723	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
270	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;	724	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
271		725
272	$uscheme = lc $uscheme;	726	$uscheme = lc $uscheme;
273		727
274	my $uport = $uscheme eq "http" ? 80	728	my $uport = $uscheme eq "http" ? 80
275	: $uscheme eq "https" ? 443	729	: $uscheme eq "https" ? 443
276	: return $cb->(undef, { Status => 599, Reason => "only http and https URL schemes supported", URL => $url });	730	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
277
278	$hdr{referer} ||= "$uscheme://$uauthority$upath"; # leave out fragment and query string, just a heuristic
279		731
280	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	732	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
281	or return $cb->(undef, { Status => 599, Reason => "unparsable URL", URL => $url });	733	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
282		734
283	my $uhost = $1;	735	my $uhost = lc $1;
284	$uport = $2 if defined $2;	736	$uport = $2 if defined $2;
285		737
		738	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		739	unless exists $hdr{host};
		740
286	$uhost =~ s/^\[(.*)\]$/$1/;	741	$uhost =~ s/^\[(.*)\]$/$1/;
287	$upath .= "?$query" if length $query;	742	$upath .= $query if length $query;
288		743
289	$upath =~ s%^/?%/%;	744	$upath =~ s%^/?%/%;
290		745
291	# cookie processing	746	# cookie processing
292	if (my $jar = $arg{cookie_jar}) {	747	if (my $jar = $arg{cookie_jar}) {
293	%$jar = () if $jar->{version} != 1;	748	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
294		749
295	my @cookie;
296
297	while (my ($chost, $v) = each %$jar) {
298	if ($chost =~ /^\./) {
299	next unless $chost eq substr $uhost, -length $chost;
300	} elsif ($chost =~ /\./) {
301	next unless $chost eq $uhost;
302	} else {
303	next;
304	}
305
306	while (my ($cpath, $v) = each %$v) {
307	next unless $cpath eq substr $upath, 0, length $cpath;
308
309	while (my ($k, $v) = each %$v) {
310	next if $uscheme ne "https" && exists $v->{secure};
311	my $value = $v->{value};
312	$value =~ s/([\\"])/\\$1/g;
313	push @cookie, "$k=\"$value\"";
314	}
315	}
316	}
317
318	$hdr{cookie} = join "; ", @cookie	750	$hdr{cookie} = join "; ", @$cookies
319	if @cookie;	751	if @$cookies;
320	}	752	}
321		753
322	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path	754	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
323		755
324	if ($proxy) {	756	if ($proxy) {
325	($rhost, $rport, $rscheme, $rpath) = (@$proxy, $url);	757	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
		758
		759	$rscheme = "http" unless defined $rscheme;
326		760
327	# don't support https requests over https-proxy transport,	761	# don't support https requests over https-proxy transport,
328	# can't be done with tls as spec'ed.	762	# can't be done with tls as spec'ed, unless you double-encrypt.
329	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";	763	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		764
		765	$rhost = lc $rhost;
		766	$rscheme = lc $rscheme;
330	} else {	767	} else {
331	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	768	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
332	}	769	}
333		770
334	$hdr{host} = $uhost;	771	# leave out fragment and query string, just a heuristic
		772	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		773	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
		774
335	$hdr{"content-length"} = length $arg{body};	775	$hdr{"content-length"} = length $arg{body}
		776	if length $arg{body} || $method ne "GET";
		777
		778	my $idempotent = $method =~ /^(?:GET|HEAD|PUT|DELETE|OPTIONS|TRACE)$/;
		779
		780	# default value for keepalive is true iff the request is for an idempotent method
		781	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		782	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		783	my $was_persistent; # true if this is actually a recycled connection
		784
		785	# the key to use in the keepalive cache
		786	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		787
		788	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive " : "" : "close ") . "Te"; #1.1
		789	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
336		790
337	my %state = (connect_guard => 1);	791	my %state = (connect_guard => 1);
		792
		793	my $ae_error = 595; # connecting
		794
		795	# handle actual, non-tunneled, request
		796	my $handle_actual_request = sub {
		797	$ae_error = 596; # request phase
		798
		799	my $hdl = $state{handle};
		800
		801	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		802
		803	# send request
		804	$hdl->push_write (
		805	"$method $rpath HTTP/1.1\015\012"
		806	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		807	. "\015\012"
		808	. (delete $arg{body})
		809	);
		810
		811	# return if error occured during push_write()
		812	return unless %state;
		813
		814	# reduce memory usage, save a kitten, also re-use it for the response headers.
		815	%hdr = ();
		816
		817	# status line and headers
		818	$state{read_response} = sub {
		819	return unless %state;
		820
		821	for ("$_[1]") {
		822	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		823
		824	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		825	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		826
		827	# 100 Continue handling
		828	# should not happen as we don't send expect: 100-continue,
		829	# but we handle it just in case.
		830	# since we send the request body regardless, if we get an error
		831	# we are out of-sync, which we currently do NOT handle correctly.
		832	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		833	if $2 eq 100;
		834
		835	push @pseudo,
		836	HTTPVersion => $1,
		837	Status => $2,
		838	Reason => $3,
		839	;
		840
		841	my $hdr = _parse_hdr
		842	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		843
		844	%hdr = (%$hdr, @pseudo);
		845	}
		846
		847	# redirect handling
		848	# microsoft and other shitheads don't give a shit for following standards,
		849	# try to support some common forms of broken Location headers.
		850	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
		851	$hdr{location} =~ s/^\.\/+//;
		852
		853	my $url = "$rscheme://$uhost:$uport";
		854
		855	unless ($hdr{location} =~ s/^\///) {
		856	$url .= $upath;
		857	$url =~ s/\/[^\/]*$//;
		858	}
		859
		860	$hdr{location} = "$url/$hdr{location}";
		861	}
		862
		863	my $redirect;
		864
		865	if ($recurse) {
		866	my $status = $hdr{Status};
		867
		868	# industry standard is to redirect POST as GET for
		869	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		870	# also, the UA should ask the user for 301 and 307 and POST,
		871	# industry standard seems to be to simply follow.
		872	# we go with the industry standard.
		873	if ($status == 301 or $status == 302 or $status == 303) {
		874	# HTTP/1.1 is unclear on how to mutate the method
		875	$method = "GET" unless $method eq "HEAD";
		876	$redirect = 1;
		877	} elsif ($status == 307) {
		878	$redirect = 1;
		879	}
		880	}
		881
		882	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		883	if ($state{handle}) {
		884	# handle keepalive
		885	if (
		886	$persistent
		887	&& $_[3]
		888	&& ($hdr{HTTPVersion} < 1.1
		889	? $hdr{connection} =~ /\bkeep-?alive\b/i
		890	: $hdr{connection} !~ /\bclose\b/i)
		891	) {
		892	ka_store $ka_key, delete $state{handle};
		893	} else {
		894	# no keepalive, destroy the handle
		895	$state{handle}->destroy;
		896	}
		897	}
		898
		899	%state = ();
		900
		901	if (defined $_[1]) {
		902	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		903	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		904	}
		905
		906	# set-cookie processing
		907	if ($arg{cookie_jar}) {
		908	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		909	}
		910
		911	if ($redirect && exists $hdr{location}) {
		912	# we ignore any errors, as it is very common to receive
		913	# Content-Length != 0 but no actual body
		914	# we also access %hdr, as $_[1] might be an erro
		915	$state{recurse} =
		916	http_request (
		917	$method => $hdr{location},
		918	%arg,
		919	recurse => $recurse - 1,
		920	Redirect => [$_[0], \%hdr],
		921	sub {
		922	%state = ();
		923	&$cb
		924	},
		925	);
		926	} else {
		927	$cb->($_[0], \%hdr);
		928	}
		929	};
		930
		931	$ae_error = 597; # body phase
		932
		933	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		934
		935	my $len = $chunked ? undef : $hdr{"content-length"};
		936
		937	# body handling, many different code paths
		938	# - no body expected
		939	# - want_body_handle
		940	# - te chunked
		941	# - 2x length known (with or without on_body)
		942	# - 2x length not known (with or without on_body)
		943	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		944	$finish->(undef, 598 => "Request cancelled by on_header");
		945	} elsif (
		946	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		947	or $method eq "HEAD"
		948	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		949	) {
		950	# no body
		951	$finish->("", undef, undef, 1);
		952
		953	} elsif (!$redirect && $arg{want_body_handle}) {
		954	$_[0]->on_eof (undef);
		955	$_[0]->on_error (undef);
		956	$_[0]->on_read (undef);
		957
		958	$finish->(delete $state{handle});
		959
		960	} elsif ($chunked) {
		961	my $cl = 0;
		962	my $body = "";
		963	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
		964
		965	$state{read_chunk} = sub {
		966	$_[1] =~ /^([0-9a-fA-F]+)/
		967	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		968
		969	my $len = hex $1;
		970
		971	if ($len) {
		972	$cl += $len;
		973
		974	$_[0]->push_read (chunk => $len, sub {
		975	$on_body->($_[1], \%hdr)
		976	or return $finish->(undef, 598 => "Request cancelled by on_body");
		977
		978	$_[0]->push_read (line => sub {
		979	length $_[1]
		980	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		981	$_[0]->push_read (line => $state{read_chunk});
		982	});
		983	});
		984	} else {
		985	$hdr{"content-length"} ||= $cl;
		986
		987	$_[0]->push_read (line => $qr_nlnl, sub {
		988	if (length $_[1]) {
		989	for ("$_[1]") {
		990	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		991
		992	my $hdr = _parse_hdr
		993	or return $finish->(undef, $ae_error => "Garbled response trailers");
		994
		995	%hdr = (%hdr, %$hdr);
		996	}
		997	}
		998
		999	$finish->($body, undef, undef, 1);
		1000	});
		1001	}
		1002	};
		1003
		1004	$_[0]->push_read (line => $state{read_chunk});
		1005
		1006	} elsif ($arg{on_body}) {
		1007	if (defined $len) {
		1008	$_[0]->on_read (sub {
		1009	$len -= length $_[0]{rbuf};
		1010
		1011	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1012	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1013
		1014	$len > 0
		1015	or $finish->("", undef, undef, 1);
		1016	});
		1017	} else {
		1018	$_[0]->on_eof (sub {
		1019	$finish->("");
		1020	});
		1021	$_[0]->on_read (sub {
		1022	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1023	or $finish->(undef, 598 => "Request cancelled by on_body");
		1024	});
		1025	}
		1026	} else {
		1027	$_[0]->on_eof (undef);
		1028
		1029	if (defined $len) {
		1030	$_[0]->on_read (sub {
		1031	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1032	if $len <= length $_[0]{rbuf};
		1033	});
		1034	} else {
		1035	$_[0]->on_error (sub {
		1036	($! == Errno::EPIPE || !$!)
		1037	? $finish->(delete $_[0]{rbuf})
		1038	: $finish->(undef, $ae_error => $_[2]);
		1039	});
		1040	$_[0]->on_read (sub { });
		1041	}
		1042	}
		1043	};
		1044
		1045	# if keepalive is enabled, then the server closing the connection
		1046	# before a response can happen legally - we retry on idempotent methods.
		1047	if ($was_persistent && $idempotent) {
		1048	my $old_eof = $hdl->{on_eof};
		1049	$hdl->{on_eof} = sub {
		1050	_destroy_state %state;
		1051
		1052	%state = ();
		1053	$state{recurse} =
		1054	http_request (
		1055	$method => $url,
		1056	%arg,
		1057	recurse => $recurse - 1,
		1058	keepalive => 0,
		1059	sub {
		1060	%state = ();
		1061	&$cb
		1062	}
		1063	);
		1064	};
		1065	$hdl->on_read (sub {
		1066	return unless %state;
		1067
		1068	# as soon as we receive something, a connection close
		1069	# once more becomes a hard error
		1070	$hdl->{on_eof} = $old_eof;
		1071	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1072	});
		1073	} else {
		1074	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1075	}
		1076	};
		1077
		1078	my $prepare_handle = sub {
		1079	my ($hdl) = $state{handle};
		1080
		1081	$hdl->on_error (sub {
		1082	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1083	});
		1084	$hdl->on_eof (sub {
		1085	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1086	});
		1087	$hdl->timeout_reset;
		1088	$hdl->timeout ($timeout);
		1089	};
		1090
		1091	# connected to proxy (or origin server)
		1092	my $connect_cb = sub {
		1093	my $fh = shift
		1094	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1095
		1096	return unless delete $state{connect_guard};
		1097
		1098	# get handle
		1099	$state{handle} = new AnyEvent::Handle
		1100	%{ $arg{handle_params} },
		1101	fh => $fh,
		1102	peername => $uhost,
		1103	tls_ctx => $arg{tls_ctx},
		1104	;
		1105
		1106	$prepare_handle->();
		1107
		1108	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1109
		1110	# now handle proxy-CONNECT method
		1111	if ($proxy && $uscheme eq "https") {
		1112	# oh dear, we have to wrap it into a connect request
		1113
		1114	# maybe re-use $uauthority with patched port?
		1115	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1116	$state{handle}->push_read (line => $qr_nlnl, sub {
		1117	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1118	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1119
		1120	if ($2 == 200) {
		1121	$rpath = $upath;
		1122	$handle_actual_request->();
		1123	} else {
		1124	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1125	}
		1126	});
		1127	} else {
		1128	$handle_actual_request->();
		1129	}
		1130	};
338		1131
339	_get_slot $uhost, sub {	1132	_get_slot $uhost, sub {
340	$state{slot_guard} = shift;	1133	$state{slot_guard} = shift;
341		1134
342	return unless $state{connect_guard};	1135	return unless $state{connect_guard};
343		1136
344	$state{connect_guard} = AnyEvent::Socket::tcp_connect $rhost, $rport, sub {	1137	# try to use an existing keepalive connection, but only if we, ourselves, plan
345	$state{fh} = shift	1138	# on a keepalive request (in theory, this should be a separate config option).
346	or return $cb->(undef, { Status => 599, Reason => "$!", URL => $url });	1139	if ($persistent && $KA_CACHE{$ka_key}) {
347	pop; # free memory, save a tree	1140	$was_persistent = 1;
348		1141
349	return unless delete $state{connect_guard};	1142	$state{handle} = ka_fetch $ka_key;
		1143	$state{handle}->destroyed
		1144	and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1145	$prepare_handle->();
		1146	$state{handle}->destroyed
		1147	and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1148	$handle_actual_request->();
350		1149
351	# get handle
352	$state{handle} = new AnyEvent::Handle
353	fh => $state{fh},
354	timeout => $timeout;
355
356	# limit the number of persistent connections
357	# keepalive not yet supported
358	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
359	++$KA_COUNT{$_[1]};
360	$state{handle}{ka_count_guard} = AnyEvent::Util::guard {
361	--$KA_COUNT{$_[1]}
362	};
363	$hdr{connection} = "keep-alive";
364	} else {	1150	} else {
365	delete $hdr{connection};	1151	my $tcp_connect = $arg{tcp_connect}
366	}	1152	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
367		1153
368	# (re-)configure handle	1154	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
369	$state{handle}->on_error (sub {
370	my $errno = "$!";
371	%state = ();
372	$cb->(undef, { Status => 599, Reason => $errno, URL => $url });
373	});
374	$state{handle}->on_eof (sub {
375	%state = ();
376	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file", URL => $url });
377	});
378
379	$state{handle}->starttls ("connect") if $rscheme eq "https";
380
381	# handle actual, non-tunneled, request
382	my $handle_actual_request = sub {
383	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};
384
385	# send request
386	$state{handle}->push_write (
387	"$method $rpath HTTP/1.0\015\012"
388	. (join "", map "\u$_: $hdr{$_}\015\012", keys %hdr)
389	. "\015\012"
390	. (delete $arg{body})
391	);
392
393	%hdr = (); # reduce memory usage, save a kitten
394
395	# status line
396	$state{handle}->push_read (line => $qr_nl, sub {
397	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
398	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])", URL => $url }));
399
400	my %hdr = ( # response headers
401	HTTPVersion => ",$1",
402	Status => ",$2",
403	Reason => ",$3",
404	URL => ",$url"
405	);
406
407	# headers, could be optimized a bit
408	$state{handle}->unshift_read (line => $qr_nlnl, sub {
409	for ("$_[1]\012") {
410	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
411
412	# we support spaces in field names, as lotus domino
413	# creates them (actually spaces around seperators
414	# are strictly allowed in http, they are a security issue).
415	$hdr{lc $1} .= ",$2"
416	while /\G
417	([^:\000-\037]+):
418	[\011\040]*
419	((?: [^\012]+ | \012[\011\040] )*)
420	\012
421	/gxc;
422
423	/\G$/
424	or return (%state = (), $cb->(undef, { Status => 599, Reason => "garbled response headers", URL => $url }));
425	}
426
427	substr $_, 0, 1, ""
428	for values %hdr;
429
430	my $finish = sub {
431	# TODO: use destroy method, when/if available
432	#$state{handle}->destroy;
433	$state{handle}->on_eof (undef);
434	$state{handle}->on_error (undef);
435	%state = ();
436
437	# set-cookie processing
438	if ($arg{cookie_jar}) {
439	for ($hdr{"set-cookie"}) {
440	# parse NAME=VALUE
441	my @kv;
442
443	while (/\G\s* ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )/gcxs) {
444	my $name = $1;
445	my $value = $3;
446
447	unless ($value) {
448	$value = $2;
449	$value =~ s/\\(.)/$1/gs;
450	}
451
452	push @kv, $name => $value;
453
454	last unless /\G\s*;/gc;
455	}
456
457	last unless @kv;
458
459	my $name = shift @kv;
460	my %kv = (value => shift @kv, @kv);
461
462	my $cdom;
463	my $cpath = (delete $kv{path}) || "/";
464
465	if (exists $kv{domain}) {
466	$cdom = delete $kv{domain};
467
468	$cdom =~ s/^\.?/./; # make sure it starts with a "."
469
470	next if $cdom =~ /\.$/;
471
472	# this is not rfc-like and not netscape-like. go figure.
473	my $ndots = $cdom =~ y/.//;
474	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
475	} else {
476	$cdom = $uhost;
477	}
478
479	# store it
480	$arg{cookie_jar}{version} = 1;
481	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
482
483	redo if /\G\s*,/gc;
484	}
485	}
486
487	# microsoft and other shitheads don't give a shit for following standards,
488	# try to support some common forms of broken Location headers.
489	if ($_[1]{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
490	$_[1]{location} =~ s/^\.\/+//;
491
492	my $url = "$rscheme://$uhost:$uport";
493
494	unless ($_[1]{location} =~ s/^\///) {
495	$url .= $upath;
496	$url =~ s/\/[^\/]*$//;
497	}
498
499	$_[1]{location} = "$url/$_[1]{location}";
500	}
501
502	if ($_[1]{Status} =~ /^30[12]$/ && $recurse && $method ne "POST") {
503	# apparently, mozilla et al. just change POST to GET here
504	# more research is needed before we do the same
505	http_request ($method, $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
506	} elsif ($_[1]{Status} == 303 && $recurse) {
507	# even http/1.1 is unclear on how to mutate the method
508	$method = "GET" unless $method eq "HEAD";
509	http_request ($method => $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
510	} elsif ($_[1]{Status} == 307 && $recurse && $method =~ /^(?:GET|HEAD)$/) {
511	http_request ($method => $_[1]{location}, %arg, recurse => $recurse - 1, $cb);
512	} else {
513	$cb->($_[0], $_[1]);
514	}
515	};
516
517	if ($hdr{Status} =~ /^(?:1..|204|304)$/ or $method eq "HEAD") {
518	$finish->(undef, \%hdr);
519	} else {
520	if (exists $hdr{"content-length"}) {
521	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
522	# could cache persistent connection now
523	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
524	# but we don't, due to misdesigns, this is annoyingly complex
525	};
526
527	$finish->($_[1], \%hdr);
528	});
529	} else {
530	# too bad, need to read until we get an error or EOF,
531	# no way to detect winged data.
532	$_[0]->on_error (sub {
533	$finish->($_[0]{rbuf}, \%hdr);
534	});
535	$_[0]->on_eof (undef);
536	$_[0]->on_read (sub { });
537	}
538	}
539	});
540	});
541	};
542
543	# now handle proxy-CONNECT method
544	if ($proxy && $uscheme eq "https") {
545	# oh dear, we have to wrap it into a connect request
546
547	# maybe re-use $uauthority with patched port?
548	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
549	$state{handle}->push_read (line => $qr_nlnl, sub {
550	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
551	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid proxy connect response ($_[1])", URL => $url }));
552
553	if ($2 == 200) {
554	$rpath = $upath;
555	&$handle_actual_request;
556	} else {
557	%state = ();
558	$cb->(undef, { Status => $2, Reason => $3, URL => $url });
559	}
560	});
561	} else {
562	&$handle_actual_request;
563	}
564
565	}, sub {
566	$timeout
567	};	1155	}
568	};	1156	};
569		1157
570	defined wantarray && AnyEvent::Util::guard { %state = () }	1158	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
571	}	1159	}
572		1160
573	sub http_get($@) {	1161	sub http_get($@) {
574	unshift @_, "GET";	1162	unshift @_, "GET";
575	&http_request	1163	&http_request
…		…
586	&http_request	1174	&http_request
587	}	1175	}
588		1176
589	=back	1177	=back
590		1178
		1179	=head2 DNS CACHING
		1180
		1181	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1182	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1183	hostnames. The latter is a simple stub resolver and does no caching
		1184	on its own. If you want DNS caching, you currently have to provide
		1185	your own default resolver (by storing a suitable resolver object in
		1186	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1187
591	=head2 GLOBAL FUNCTIONS AND VARIABLES	1188	=head2 GLOBAL FUNCTIONS AND VARIABLES
592		1189
593	=over 4	1190	=over 4
594		1191
595	=item AnyEvent::HTTP::set_proxy "proxy-url"	1192	=item AnyEvent::HTTP::set_proxy "proxy-url"
596		1193
597	Sets the default proxy server to use. The proxy-url must begin with a	1194	Sets the default proxy server to use. The proxy-url must begin with a
598	string of the form C<http://host:port> (optionally C<https:...>).	1195	string of the form C<http://host:port>, croaks otherwise.
		1196
		1197	To clear an already-set proxy, use C<undef>.
		1198
		1199	When AnyEvent::HTTP is loaded for the first time it will query the
		1200	default proxy from the operating system, currently by looking at
		1201	C<$ENV{http_proxy>}.
		1202
		1203	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1204
		1205	Remove all cookies from the cookie jar that have been expired. If
		1206	C<$session_end> is given and true, then additionally remove all session
		1207	cookies.
		1208
		1209	You should call this function (with a true C<$session_end>) before you
		1210	save cookies to disk, and you should call this function after loading them
		1211	again. If you have a long-running program you can additonally call this
		1212	function from time to time.
		1213
		1214	A cookie jar is initially an empty hash-reference that is managed by this
		1215	module. It's format is subject to change, but currently it is like this:
		1216
		1217	The key C<version> has to contain C<1>, otherwise the hash gets
		1218	emptied. All other keys are hostnames or IP addresses pointing to
		1219	hash-references. The key for these inner hash references is the
		1220	server path for which this cookie is meant, and the values are again
		1221	hash-references. The keys of those hash-references is the cookie name, and
		1222	the value, you guessed it, is another hash-reference, this time with the
		1223	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1224	which have been replaced by a C<_expires> key that contains the cookie
		1225	expiry timestamp.
		1226
		1227	Here is an example of a cookie jar with a single cookie, so you have a
		1228	chance of understanding the above paragraph:
		1229
		1230	{
		1231	version => 1,
		1232	"10.0.0.1" => {
		1233	"/" => {
		1234	"mythweb_id" => {
		1235	_expires => 1293917923,
		1236	value => "ooRung9dThee3ooyXooM1Ohm",
		1237	},
		1238	},
		1239	},
		1240	}
		1241
		1242	=item $date = AnyEvent::HTTP::format_date $timestamp
		1243
		1244	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1245	Date (RFC 2616).
		1246
		1247	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1248
		1249	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1250	bunch of minor variations of those, and returns the corresponding POSIX
		1251	timestamp, or C<undef> if the date cannot be parsed.
599		1252
600	=item $AnyEvent::HTTP::MAX_RECURSE	1253	=item $AnyEvent::HTTP::MAX_RECURSE
601		1254
602	The default value for the C<recurse> request parameter (default: C<10>).	1255	The default value for the C<recurse> request parameter (default: C<10>).
603		1256
		1257	=item $AnyEvent::HTTP::TIMEOUT
		1258
		1259	The default timeout for connection operations (default: C<300>).
		1260
604	=item $AnyEvent::HTTP::USERAGENT	1261	=item $AnyEvent::HTTP::USERAGENT
605		1262
606	The default value for the C<User-Agent> header (the default is	1263	The default value for the C<User-Agent> header (the default is
607	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1264	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
608		1265
609	=item $AnyEvent::HTTP::MAX_PERSISTENT	1266	=item $AnyEvent::HTTP::MAX_PER_HOST
610		1267
611	The maximum number of persistent connections to keep open (default: 8).	1268	The maximum number of concurrent connections to the same host (identified
		1269	by the hostname). If the limit is exceeded, then the additional requests
		1270	are queued until previous connections are closed. Both persistent and
		1271	non-persistent connections are counted in this limit.
612		1272
613	Not implemented currently.	1273	The default value for this is C<4>, and it is highly advisable to not
		1274	increase it much.
		1275
		1276	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1277	connections, older browsers used 2, newers (such as firefox 3) typically
		1278	use 6, and Opera uses 8 because like, they have the fastest browser and
		1279	give a shit for everybody else on the planet.
614		1280
615	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1281	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
616		1282
617	The maximum time to cache a persistent connection, in seconds (default: 2).	1283	The time after which idle persistent conenctions get closed by
618		1284	AnyEvent::HTTP (default: C<3>).
619	Not implemented currently.
620		1285
621	=item $AnyEvent::HTTP::ACTIVE	1286	=item $AnyEvent::HTTP::ACTIVE
622		1287
623	The number of active connections. This is not the number of currently	1288	The number of active connections. This is not the number of currently
624	running requests, but the number of currently open and non-idle TCP	1289	running requests, but the number of currently open and non-idle TCP
625	connections. This number of can be useful for load-leveling.	1290	connections. This number can be useful for load-leveling.
626		1291
627	=back	1292	=back
628		1293
629	=cut	1294	=cut
630		1295
		1296	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1297	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1298
		1299	sub format_date($) {
		1300	my ($time) = @_;
		1301
		1302	# RFC 822/1123 format
		1303	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1304
		1305	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1306	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1307	$H, $M, $S;
		1308	}
		1309
		1310	sub parse_date($) {
		1311	my ($date) = @_;
		1312
		1313	my ($d, $m, $y, $H, $M, $S);
		1314
		1315	if ($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$/) {
		1316	# RFC 822/1123, required by RFC 2616 (with " ")
		1317	# cookie dates (with "-")
		1318
		1319	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1320
		1321	} elsif ($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$/) {
		1322	# RFC 850
		1323	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1324
		1325	} elsif ($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])$/) {
		1326	# ISO C's asctime
		1327	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1328	}
		1329	# other formats fail in the loop below
		1330
		1331	for (0..11) {
		1332	if ($m eq $month[$_]) {
		1333	require Time::Local;
		1334	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
		1335	}
		1336	}
		1337
		1338	undef
		1339	}
		1340
631	sub set_proxy($) {	1341	sub set_proxy($) {
632	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1342	if (length $_[0]) {
		1343	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1344	or Carp::croak "$_[0]: invalid proxy URL";
		1345	$PROXY = [$2, $3 || 3128, $1]
		1346	} else {
		1347	undef $PROXY;
		1348	}
633	}	1349	}
634		1350
635	# initialise proxy from environment	1351	# initialise proxy from environment
		1352	eval {
636	set_proxy $ENV{http_proxy};	1353	set_proxy $ENV{http_proxy};
		1354	};
		1355
		1356	=head2 SHOWCASE
		1357
		1358	This section contaisn some more elaborate "real-world" examples or code
		1359	snippets.
		1360
		1361	=head2 HTTP/1.1 FILE DOWNLOAD
		1362
		1363	Downloading files with HTTP can be quite tricky, especially when something
		1364	goes wrong and you want to resume.
		1365
		1366	Here is a function that initiates and resumes a download. It uses the
		1367	last modified time to check for file content changes, and works with many
		1368	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1369	on older servers.
		1370
		1371	It calls the completion callback with either C<undef>, which means a
		1372	nonretryable error occured, C<0> when the download was partial and should
		1373	be retried, and C<1> if it was successful.
		1374
		1375	use AnyEvent::HTTP;
		1376
		1377	sub download($$$) {
		1378	my ($url, $file, $cb) = @_;
		1379
		1380	open my $fh, "+<", $file
		1381	or die "$file: $!";
		1382
		1383	my %hdr;
		1384	my $ofs = 0;
		1385
		1386	warn stat $fh;
		1387	warn -s _;
		1388	if (stat $fh and -s _) {
		1389	$ofs = -s _;
		1390	warn "-s is ", $ofs;
		1391	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1392	$hdr{"range"} = "bytes=$ofs-";
		1393	}
		1394
		1395	http_get $url,
		1396	headers => \%hdr,
		1397	on_header => sub {
		1398	my ($hdr) = @_;
		1399
		1400	if ($hdr->{Status} == 200 && $ofs) {
		1401	# resume failed
		1402	truncate $fh, $ofs = 0;
		1403	}
		1404
		1405	sysseek $fh, $ofs, 0;
		1406
		1407	1
		1408	},
		1409	on_body => sub {
		1410	my ($data, $hdr) = @_;
		1411
		1412	if ($hdr->{Status} =~ /^2/) {
		1413	length $data == syswrite $fh, $data
		1414	or return; # abort on write errors
		1415	}
		1416
		1417	1
		1418	},
		1419	sub {
		1420	my (undef, $hdr) = @_;
		1421
		1422	my $status = $hdr->{Status};
		1423
		1424	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1425	utime $fh, $time, $time;
		1426	}
		1427
		1428	if ($status == 200 || $status == 206 || $status == 416) {
		1429	# download ok || resume ok || file already fully downloaded
		1430	$cb->(1, $hdr);
		1431
		1432	} elsif ($status == 412) {
		1433	# file has changed while resuming, delete and retry
		1434	unlink $file;
		1435	$cb->(0, $hdr);
		1436
		1437	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1438	# retry later
		1439	$cb->(0, $hdr);
		1440
		1441	} else {
		1442	$cb->(undef, $hdr);
		1443	}
		1444	}
		1445	;
		1446	}
		1447
		1448	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1449	if ($_[0]) {
		1450	print "OK!\n";
		1451	} elsif (defined $_[0]) {
		1452	print "please retry later\n";
		1453	} else {
		1454	print "ERROR\n";
		1455	}
		1456	};
		1457
		1458	=head3 SOCKS PROXIES
		1459
		1460	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1461	compile your perl to support socks, or use an external program such as
		1462	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1463	transparently.
		1464
		1465	Alternatively, for AnyEvent::HTTP only, you can use your own
		1466	C<tcp_connect> function that does the proxy handshake - here is an example
		1467	that works with socks4a proxies:
		1468
		1469	use Errno;
		1470	use AnyEvent::Util;
		1471	use AnyEvent::Socket;
		1472	use AnyEvent::Handle;
		1473
		1474	# host, port and username of/for your socks4a proxy
		1475	my $socks_host = "10.0.0.23";
		1476	my $socks_port = 9050;
		1477	my $socks_user = "";
		1478
		1479	sub socks4a_connect {
		1480	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1481
		1482	my $hdl = new AnyEvent::Handle
		1483	connect => [$socks_host, $socks_port],
		1484	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1485	on_error => sub { $connect_cb->() },
		1486	;
		1487
		1488	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1489
		1490	$hdl->push_read (chunk => 8, sub {
		1491	my ($hdl, $chunk) = @_;
		1492	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1493
		1494	if ($status == 0x5a) {
		1495	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1496	} else {
		1497	$! = Errno::ENXIO; $connect_cb->();
		1498	}
		1499	});
		1500
		1501	$hdl
		1502	}
		1503
		1504	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1505	possibly after switching off other proxy types:
		1506
		1507	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1508
		1509	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1510	my ($data, $headers) = @_;
		1511	...
		1512	};
637		1513
638	=head1 SEE ALSO	1514	=head1 SEE ALSO
639		1515
640	L<AnyEvent>.	1516	L<AnyEvent>.
641		1517
642	=head1 AUTHOR	1518	=head1 AUTHOR
643		1519
644	Marc Lehmann <schmorp@schmorp.de>	1520	Marc Lehmann <schmorp@schmorp.de>
645	http://home.schmorp.de/	1521	http://home.schmorp.de/
646		1522
		1523	With many thanks to Дмитрий Шалашов, who provided countless
		1524	testcases and bugreports.
		1525
647	=cut	1526	=cut
648		1527
649	1	1528	1
650		1529

Diff Legend

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