ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.71 by root, Fri Dec 31 20:50:58 2010 UTC vs.
Revision 1.112 by root, Wed Nov 14 22:22:24 2012 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 Errno ();	43	use Errno ();
45		44
46	use AnyEvent 5.0 ();	45	use AnyEvent 5.0 ();
47	use AnyEvent::Util ();	46	use AnyEvent::Util ();
48	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
49		48
50	use base Exporter::;	49	use base Exporter::;
51		50
52	our $VERSION = '1.5';	51	our $VERSION = '2.15';
53		52
54	our @EXPORT = qw(http_get http_post http_head http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
55		54
56	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";	55	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
57	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
58	our $MAX_PERSISTENT = 8;
59	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
60	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
61		59	our $MAX_PER_HOST = 4; # changing this is evil
62	# changing these is evil
63	our $MAX_PERSISTENT_PER_HOST = 0;
64	our $MAX_PER_HOST = 4;
65		60
66	our $PROXY;	61	our $PROXY;
67	our $ACTIVE = 0;	62	our $ACTIVE = 0;
68		63
69	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
70	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
71		66
72	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
73		68
74	Executes an HTTP-GET request. See the http_request function for details on	69	Executes an HTTP-GET request. See the http_request function for details on
…		…
122		117
123	If the server sends a header multiple times, then their contents will be	118	If the server sends a header multiple times, then their contents will be
124	joined together with a comma (C<,>), as per the HTTP spec.	119	joined together with a comma (C<,>), as per the HTTP spec.
125		120
126	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
127	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<59x>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
128	(usually C<599>) and the C<Reason> pseudo-header will contain an error	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
129	message.	124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection etsbalishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
130		139
131	A typical callback might look like this:	140	A typical callback might look like this:
132		141
133	sub {	142	sub {
134	my ($body, $hdr) = @_;	143	my ($body, $hdr) = @_;
…		…
145		154
146	=over 4	155	=over 4
147		156
148	=item recurse => $count (default: $MAX_RECURSE)	157	=item recurse => $count (default: $MAX_RECURSE)
149		158
150	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
151	retries and so on, and how often to do so.	160	other retries and so on, and how often to do so.
152		161
153	=item headers => hashref	162	=item headers => hashref
154		163
155	The request headers to use. Currently, C<http_request> may provide its own	164	The request headers to use. Currently, C<http_request> may provide its own
156	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and	165	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
157	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>	166	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
158	(this can be suppressed by using C<undef> for these headers in which case	167	(this can be suppressed by using C<undef> for these headers in which case
159	they won't be sent at all).	168	they won't be sent at all).
160		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.
		176
161	=item timeout => $seconds	177	=item timeout => $seconds
162		178
163	The time-out to use for various stages - each connect attempt will reset	179	The time-out to use for various stages - each connect attempt will reset
164	the timeout, as will read or write activity, i.e. this is not an overall	180	the timeout, as will read or write activity, i.e. this is not an overall
165	timeout.	181	timeout.
166		182
167	Default timeout is 5 minutes.	183	Default timeout is 5 minutes.
168		184
169	=item proxy => [$host, $port[, $scheme]] or undef	185	=item proxy => [$host, $port[, $scheme]] or undef
170		186
171	Use the given http proxy for all requests. If not specified, then the	187	Use the given http proxy for all requests, or no proxy if C<undef> is
172	default proxy (as specified by C<$ENV{http_proxy}>) is used.	188	used.
173		189
174	C<$scheme> must be either missing, C<http> for HTTP or C<https> for	190	C<$scheme> must be either missing or must be C<http> for HTTP.
175	HTTPS.	191
		192	If not specified, then the default proxy is used (see
		193	C<AnyEvent::HTTP::set_proxy>).
176		194
177	=item body => $string	195	=item body => $string
178		196
179	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
180	this module might offer more options).	198	this module might offer more options).
…		…
182	=item cookie_jar => $hash_ref	200	=item cookie_jar => $hash_ref
183		201
184	Passing this parameter enables (simplified) cookie-processing, loosely	202	Passing this parameter enables (simplified) cookie-processing, loosely
185	based on the original netscape specification.	203	based on the original netscape specification.
186		204
187	The C<$hash_ref> must be an (initially empty) hash reference which will	205	The C<$hash_ref> must be an (initially empty) hash reference which
188	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
189	persistent storage with something like JSON or Storable, but this is not	207	to persistent storage with something like JSON or Storable - see the
190	recommended, as session-only cookies might survive longer than expected.	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.
191		211
192	Note that this cookie implementation is not meant to be complete. If	212	Note that this cookie implementation is not meant to be complete. If
193	you want complete cookie management you have to do that on your	213	you want complete cookie management you have to do that on your
194	own. C<cookie_jar> is meant as a quick fix to get some cookie-using sites	214	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
195	working. Cookies are a privacy disaster, do not use them unless required	215	working. Cookies are a privacy disaster, do not use them unless required
196	to.	216	to.
197		217
198	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>	218	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
199	headers will be set and handled by this module, otherwise they will be	219	headers will be set and handled by this module, otherwise they will be
…		…
208	verification, highest compatibility) and high-security (CA and common-name	228	verification, highest compatibility) and high-security (CA and common-name
209	verification) TLS context.	229	verification) TLS context.
210		230
211	The default for this option is C<low>, which could be interpreted as "give	231	The default for this option is C<low>, which could be interpreted as "give
212	me the page, no matter what".	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.
213		243
214	=item on_prepare => $callback->($fh)	244	=item on_prepare => $callback->($fh)
215		245
216	In rare cases you need to "tune" the socket before it is used to	246	In rare cases you need to "tune" the socket before it is used to
217	connect (for exmaple, to bind it on a given IP address). This parameter	247	connect (for exmaple, to bind it on a given IP address). This parameter
…		…
287	called. Instead of the C<$body> argument containing the body data, the	317	called. Instead of the C<$body> argument containing the body data, the
288	callback will receive the L<AnyEvent::Handle> object associated with the	318	callback will receive the L<AnyEvent::Handle> object associated with the
289	connection. In error cases, C<undef> will be passed. When there is no body	319	connection. In error cases, C<undef> will be passed. When there is no body
290	(e.g. status C<304>), the empty string will be passed.	320	(e.g. status C<304>), the empty string will be passed.
291		321
292	The handle object might or might not be in TLS mode, might be connected to	322	The handle object might or might not be in TLS mode, might be connected
293	a proxy, be a persistent connection etc., and configured in unspecified	323	to a proxy, be a persistent connection, use chunked transfer encoding
294	ways. The user is responsible for this handle (it will not be used by this	324	etc., and configured in unspecified ways. The user is responsible for this
295	module anymore).	325	handle (it will not be used by this module anymore).
296		326
297	This is useful with some push-type services, where, after the initial	327	This is useful with some push-type services, where, after the initial
298	headers, an interactive protocol is used (typical example would be the	328	headers, an interactive protocol is used (typical example would be the
299	push-style twitter API which starts a JSON/XML stream).	329	push-style twitter API which starts a JSON/XML stream).
300		330
301	If you think you need this, first have a look at C<on_body>, to see if	331	If you think you need this, first have a look at C<on_body>, to see if
302	that doesn't solve your problem in a better way.	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 => $boolean
		353
		354	Only used when C<persistent> is also true. This parameter decides whether
		355	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		356	(as opposed to only a HTTP/1.1 persistent connection).
		357
		358	The default is true, except when using a proxy, in which case it defaults
		359	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		360
		361	=item handle_params => { key => value ... }
		362
		363	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		364	constructor that is called - not all requests will create a handle, and
		365	sometimes more than one is created, so this parameter is only good for
		366	setting hints.
		367
		368	Example: set the maximum read size to 4096, to potentially conserve memory
		369	at the cost of speed.
		370
		371	handle_params => {
		372	max_read_size => 4096,
		373	},
303		374
304	=back	375	=back
305		376
306	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	377	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
307	the response body.	378	the response body.
…		…
313		384
314	Example: do a HTTP HEAD request on https://www.google.com/, use a	385	Example: do a HTTP HEAD request on https://www.google.com/, use a
315	timeout of 30 seconds.	386	timeout of 30 seconds.
316		387
317	http_request	388	http_request
318	GET => "https://www.google.com",	389	HEAD => "https://www.google.com",
		390	headers => { "user-agent" => "MySearchClient 1.0" },
319	timeout => 30,	391	timeout => 30,
320	sub {	392	sub {
321	my ($body, $hdr) = @_;	393	my ($body, $hdr) = @_;
322	use Data::Dumper;	394	use Data::Dumper;
323	print Dumper $hdr;	395	print Dumper $hdr;
…		…
333	};	405	};
334		406
335	undef $request;	407	undef $request;
336		408
337	=cut	409	=cut
		410
		411	#############################################################################
		412	# wait queue/slots
338		413
339	sub _slot_schedule;	414	sub _slot_schedule;
340	sub _slot_schedule($) {	415	sub _slot_schedule($) {
341	my $host = shift;	416	my $host = shift;
342		417
…		…
364	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	439	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
365		440
366	_slot_schedule $_[0];	441	_slot_schedule $_[0];
367	}	442	}
368		443
		444	#############################################################################
		445	# cookie handling
		446
		447	# expire cookies
369	sub cookie_jar_extract($$$$) {	448	sub cookie_jar_expire($;$) {
370	my ($jar, $uscheme, $uhost, $upath) = @_;	449	my ($jar, $session_end) = @_;
371		450
372	%$jar = () if $jar->{version} != 1;	451	%$jar = () if $jar->{version} != 1;
373		452
374	my @cookies;	453	my $anow = AE::now;
375		454
376	while (my ($chost, $paths) = each %$jar) {	455	while (my ($chost, $paths) = each %$jar) {
377	next unless ref $paths;	456	next unless ref $paths;
378		457
		458	while (my ($cpath, $cookies) = each %$paths) {
		459	while (my ($cookie, $kv) = each %$cookies) {
		460	if (exists $kv->{_expires}) {
		461	delete $cookies->{$cookie}
		462	if $anow > $kv->{_expires};
		463	} elsif ($session_end) {
		464	delete $cookies->{$cookie};
		465	}
		466	}
		467
		468	delete $paths->{$cpath}
		469	unless %$cookies;
		470	}
		471
		472	delete $jar->{$chost}
		473	unless %$paths;
		474	}
		475	}
		476
		477	# extract cookies from jar
		478	sub cookie_jar_extract($$$$) {
		479	my ($jar, $scheme, $host, $path) = @_;
		480
		481	%$jar = () if $jar->{version} != 1;
		482
		483	my @cookies;
		484
		485	while (my ($chost, $paths) = each %$jar) {
		486	next unless ref $paths;
		487
379	if ($chost =~ /^\./) {	488	if ($chost =~ /^\./) {
380	next unless $chost eq substr $uhost, -length $chost;	489	next unless $chost eq substr $host, -length $chost;
381	} elsif ($chost =~ /\./) {	490	} elsif ($chost =~ /\./) {
382	next unless $chost eq $uhost;	491	next unless $chost eq $host;
383	} else {	492	} else {
384	next;	493	next;
385	}	494	}
386		495
387	while (my ($cpath, $cookies) = each %$paths) {	496	while (my ($cpath, $cookies) = each %$paths) {
388	next unless $cpath eq substr $upath, 0, length $cpath;	497	next unless $cpath eq substr $path, 0, length $cpath;
389		498
390	while (my ($cookie, $kv) = each %$cookies) {	499	while (my ($cookie, $kv) = each %$cookies) {
391	next if $uscheme ne "https" && exists $kv->{secure};	500	next if $scheme ne "https" && exists $kv->{secure};
392		501
393	if (exists $kv->{expires}) {	502	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
394	if (AE::now > parse_date ($kv->{expires})) {
395	delete $cookies->{$cookie};	503	delete $cookies->{$cookie};
396	next;	504	next;
397	}
398	}	505	}
399		506
400	my $value = $kv->{value};	507	my $value = $kv->{value};
401		508
402	if ($value =~ /[=;,[:space:]]/) {	509	if ($value =~ /[=;,[:space:]]/) {
…		…
410	}	517	}
411		518
412	\@cookies	519	\@cookies
413	}	520	}
414		521
		522	# parse set_cookie header into jar
		523	sub cookie_jar_set_cookie($$$$) {
		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
415	# continue to parse $_ for headers and place them into the arg	643	# continue to parse $_ for headers and place them into the arg
416	sub parse_hdr() {	644	sub _parse_hdr() {
417	my %hdr;	645	my %hdr;
418		646
419	# things seen, not parsed:	647	# things seen, not parsed:
420	# p3pP="NON CUR OTPi OUR NOR UNI"	648	# p3pP="NON CUR OTPi OUR NOR UNI"
421		649
…		…
435	for values %hdr;	663	for values %hdr;
436		664
437	\%hdr	665	\%hdr
438	}	666	}
439		667
		668	#############################################################################
		669	# http_get
		670
440	our $qr_nlnl = qr{(?<![^\012])\015?\012};	671	our $qr_nlnl = qr{(?<![^\012])\015?\012};
441		672
442	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	673	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
443	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };	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(\%$$) {
		686	my ($state, $cb, $hdr) = @_;
		687
		688	&_destroy_state ($state);
		689
		690	$cb->(undef, $hdr);
		691	()
		692	}
444		693
445	sub http_request($$@) {	694	sub http_request($$@) {
446	my $cb = pop;	695	my $cb = pop;
447	my ($method, $url, %arg) = @_;	696	my ($method, $url, %arg) = @_;
448		697
…		…
466	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	715	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
467		716
468	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	717	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
469	if $recurse < 0;	718	if $recurse < 0;
470		719
471	my $proxy = $arg{proxy} || $PROXY;	720	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
472	my $timeout = $arg{timeout} || $TIMEOUT;	721	my $timeout = $arg{timeout} || $TIMEOUT;
473		722
474	my ($uscheme, $uauthority, $upath, $query, $fragment) =	723	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
475	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	724	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
476		725
477	$uscheme = lc $uscheme;	726	$uscheme = lc $uscheme;
478		727
479	my $uport = $uscheme eq "http" ? 80	728	my $uport = $uscheme eq "http" ? 80
480	: $uscheme eq "https" ? 443	729	: $uscheme eq "https" ? 443
481	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });	730	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
482		731
483	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	732	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
484	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	733	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
485		734
486	my $uhost = $1;	735	my $uhost = lc $1;
487	$uport = $2 if defined $2;	736	$uport = $2 if defined $2;
488		737
489	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	738	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
490	unless exists $hdr{host};	739	unless exists $hdr{host};
491		740
…		…
510	$rscheme = "http" unless defined $rscheme;	759	$rscheme = "http" unless defined $rscheme;
511		760
512	# don't support https requests over https-proxy transport,	761	# don't support https requests over https-proxy transport,
513	# can't be done with tls as spec'ed, unless you double-encrypt.	762	# can't be done with tls as spec'ed, unless you double-encrypt.
514	$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;
515	} else {	767	} else {
516	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	768	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
517	}	769	}
518		770
519	# leave out fragment and query string, just a heuristic	771	# leave out fragment and query string, just a heuristic
…		…
521	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	773	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
522		774
523	$hdr{"content-length"} = length $arg{body}	775	$hdr{"content-length"} = length $arg{body}
524	if length $arg{body} || $method ne "GET";	776	if length $arg{body} || $method ne "GET";
525		777
526	$hdr{connection} = "close TE"; #1.1	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
527	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	789	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
528		790
529	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	};
530		1131
531	_get_slot $uhost, sub {	1132	_get_slot $uhost, sub {
532	$state{slot_guard} = shift;	1133	$state{slot_guard} = shift;
533		1134
534	return unless $state{connect_guard};	1135	return unless $state{connect_guard};
535		1136
536	my $connect_cb = sub {	1137	# try to use an existing keepalive connection, but only if we, ourselves, plan
537	$state{fh} = shift	1138	# on a keepalive request (in theory, this should be a separate config option).
538	or do {	1139	if ($persistent && $KA_CACHE{$ka_key}) {
539	my $err = "$!";	1140	$was_persistent = 1;
540	%state = ();
541	return $cb->(undef, { @pseudo, Status => 599, Reason => $err });
542	};
543		1141
544	pop; # free memory, save a tree	1142	$state{handle} = ka_fetch $ka_key;
545
546	return unless delete $state{connect_guard};
547
548	# get handle
549	$state{handle} = new AnyEvent::Handle
550	fh => $state{fh},
551	peername => $rhost,
552	tls_ctx => $arg{tls_ctx},
553	# these need to be reconfigured on keepalive handles
554	timeout => $timeout,
555	on_error => sub {
556	%state = ();
557	$cb->(undef, { @pseudo, Status => 599, Reason => $_[2] });
558	},
559	on_eof => sub {
560	%state = ();
561	$cb->(undef, { @pseudo, Status => 599, Reason => "Unexpected end-of-file" });
562	},
563	;
564
565	# limit the number of persistent connections
566	# keepalive not yet supported
567	# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
568	# ++$KA_COUNT{$_[1]};
569	# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
570	# --$KA_COUNT{$_[1]}
571	# };
572	# $hdr{connection} = "keep-alive";
573	# }
574
575	$state{handle}->starttls ("connect") if $rscheme eq "https";
576
577	# handle actual, non-tunneled, request
578	my $handle_actual_request = sub {
579	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};
580
581	# send request
582	$state{handle}->push_write (	1143	$state{handle}->destroyed
583	"$method $rpath HTTP/1.1\015\012"	1144	and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
584	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)	1145	$prepare_handle->();
585	. "\015\012"	1146	$state{handle}->destroyed
586	. (delete $arg{body})	1147	and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
587	);
588
589	# return if error occured during push_write()
590	return unless %state;
591
592	%hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
593
594	# status line and headers
595	$state{read_response} = sub {
596	for ("$_[1]") {
597	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
598
599	/^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/igxc
600	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid server response" }));
601
602	# 100 Continue handling
603	# should not happen as we don't send expect: 100-continue,
604	# but we handle it just in case.
605	# since we send the request body regardless, if we get an error
606	# we are out of-sync, which we currently do NOT handle correctly.
607	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
608	if $2 eq 100;
609
610	push @pseudo,
611	HTTPVersion => $1,
612	Status => $2,
613	Reason => $3,
614	;
615
616	my $hdr = parse_hdr
617	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Garbled response headers" }));
618
619	%hdr = (%$hdr, @pseudo);
620	}
621
622	# redirect handling
623	# microsoft and other shitheads don't give a shit for following standards,
624	# try to support some common forms of broken Location headers.
625	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
626	$hdr{location} =~ s/^\.\/+//;
627
628	my $url = "$rscheme://$uhost:$uport";
629
630	unless ($hdr{location} =~ s/^\///) {
631	$url .= $upath;
632	$url =~ s/\/[^\/]*$//;
633	}
634
635	$hdr{location} = "$url/$hdr{location}";
636	}
637
638	my $redirect;
639
640	if ($recurse) {
641	my $status = $hdr{Status};
642
643	# industry standard is to redirect POST as GET for
644	# 301, 302 and 303, in contrast to http/1.0 and 1.1.
645	# also, the UA should ask the user for 301 and 307 and POST,
646	# industry standard seems to be to simply follow.
647	# we go with the industry standard.
648	if ($status == 301 or $status == 302 or $status == 303) {
649	# HTTP/1.1 is unclear on how to mutate the method
650	$method = "GET" unless $method eq "HEAD";
651	$redirect = 1;
652	} elsif ($status == 307) {
653	$redirect = 1;
654	}
655	}
656
657	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
658	my $keepalive = pop;
659
660	$state{handle}->destroy if $state{handle};
661	%state = ();
662
663	if (defined $_[1]) {
664	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
665	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
666	}
667
668	# set-cookie processing
669	if ($arg{cookie_jar}) {
670	for ($hdr{"set-cookie"}) {
671	# parse NAME=VALUE
672	my @kv;
673
674	while (
675	m{
676	\G\s*
677	(?:
678	expires \s*=\s* ([A-Z][a-z][a-z],\ [^,;]+)
679	| ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )
680	)
681	}gcxsi
682	) {
683	my $name = $2;
684	my $value = $4;
685
686	unless (defined $name) {
687	# expires
688	$name = "expires";
689	$value = $1;
690	} elsif (!defined $value) {
691	# quoted
692	$value = $3;
693	$value =~ s/\\(.)/$1/gs;
694	}
695
696	push @kv, lc $name, $value;
697
698	last unless /\G\s*;/gc;
699	}
700
701	last unless @kv;
702
703	my $name = shift @kv;
704	my %kv = (value => shift @kv, @kv);
705
706	$kv{expires} ||= format_date (AE::now + $kv{"max-age"})
707	if exists $kv{"max-age"};
708
709	my $cdom;
710	my $cpath = (delete $kv{path}) || "/";
711
712	if (exists $kv{domain}) {
713	$cdom = delete $kv{domain};
714
715	$cdom =~ s/^\.?/./; # make sure it starts with a "."
716
717	next if $cdom =~ /\.$/;
718
719	# this is not rfc-like and not netscape-like. go figure.
720	my $ndots = $cdom =~ y/.//;
721	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
722	} else {
723	$cdom = $uhost;
724	}
725
726	# store it
727	$arg{cookie_jar}{version} = 1;
728	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
729
730	redo if /\G\s*,/gc;
731	}
732	}
733
734	if ($redirect && exists $hdr{location}) {
735	# we ignore any errors, as it is very common to receive
736	# Content-Length != 0 but no actual body
737	# we also access %hdr, as $_[1] might be an erro
738	http_request (
739	$method => $hdr{location},
740	%arg,
741	recurse => $recurse - 1,
742	Redirect => [$_[0], \%hdr],
743	$cb);
744	} else {
745	$cb->($_[0], \%hdr);
746	}
747	};
748
749	my $len = $hdr{"content-length"};
750
751	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
752	$finish->(undef, 598 => "Request cancelled by on_header");
753	} elsif (
754	$hdr{Status} =~ /^(?:1..|204|205|304)$/
755	or $method eq "HEAD"
756	or (defined $len && !$len)
757	) {
758	# no body
759	$finish->("", undef, undef, 1);
760	} else {
761	# body handling, many different code paths
762	# - no body expected
763	# - want_body_handle
764	# - te chunked
765	# - 2x length known (with or without on_body)
766	# - 2x length not known (with or without on_body)
767	if (!$redirect && $arg{want_body_handle}) {
768	$_[0]->on_eof (undef);
769	$_[0]->on_error (undef);
770	$_[0]->on_read (undef);
771
772	$finish->(delete $state{handle});
773
774	} elsif ($hdr{"transfer-encoding"} =~ /\bchunked\b/i) {
775	my $cl = 0;
776	my $body = undef;
777	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
778
779	$_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
780
781	my $read_chunk; $read_chunk = sub {
782	$_[1] =~ /^([0-9a-fA-F]+)/
783	or $finish->(undef, 599 => "Garbled chunked transfer encoding");
784
785	my $len = hex $1;
786
787	if ($len) {
788	$cl += $len;
789
790	$_[0]->push_read (chunk => $len, sub {
791	$on_body->($_[1], \%hdr)
792	or return $finish->(undef, 598 => "Request cancelled by on_body");
793
794	$_[0]->push_read (line => sub {
795	length $_[1]
796	and return $finish->(undef, 599 => "Garbled chunked transfer encoding");
797	$_[0]->push_read (line => $read_chunk);
798	});
799	});
800	} else {
801	$hdr{"content-length"} ||= $cl;
802
803	$_[0]->push_read (line => $qr_nlnl, sub {
804	if (length $_[1]) {
805	for ("$_[1]") {
806	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
807
808	my $hdr = parse_hdr
809	or return $finish->(undef, 599 => "Garbled response trailers");
810
811	%hdr = (%hdr, %$hdr);
812	}
813	}
814
815	$finish->($body, undef, undef, 1);
816	});
817	}
818	};
819
820	$_[0]->push_read (line => $read_chunk);
821
822	} elsif ($arg{on_body}) {
823	$_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
824
825	if ($len) {
826	$_[0]->on_read (sub {
827	$len -= length $_[0]{rbuf};
828
829	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
830	or return $finish->(undef, 598 => "Request cancelled by on_body");
831
832	$len > 0
833	or $finish->("", undef, undef, 1);
834	});
835	} else {
836	$_[0]->on_eof (sub {
837	$finish->("");
838	});
839	$_[0]->on_read (sub {
840	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
841	or $finish->(undef, 598 => "Request cancelled by on_body");
842	});
843	}
844	} else {
845	$_[0]->on_eof (undef);
846
847	if ($len) {
848	$_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
849	$_[0]->on_read (sub {
850	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
851	if $len <= length $_[0]{rbuf};
852	});
853	} else {
854	$_[0]->on_error (sub {
855	($! == Errno::EPIPE || !$!)
856	? $finish->(delete $_[0]{rbuf})
857	: $finish->(undef, 599 => $_[2]);
858	});
859	$_[0]->on_read (sub { });
860	}
861	}
862	}
863	};
864
865	$state{handle}->push_read (line => $qr_nlnl, $state{read_response});
866	};
867
868	# now handle proxy-CONNECT method
869	if ($proxy && $uscheme eq "https") {
870	# oh dear, we have to wrap it into a connect request
871
872	# maybe re-use $uauthority with patched port?
873	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
874	$state{handle}->push_read (line => $qr_nlnl, sub {
875	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
876	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" }));
877
878	if ($2 == 200) {
879	$rpath = $upath;
880	&$handle_actual_request;	1148	$handle_actual_request->();
881	} else {	1149
882	%state = ();
883	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });
884	}
885	});
886	} else {	1150	} else {
887	&$handle_actual_request;
888	}
889	};
890
891	my $tcp_connect = $arg{tcp_connect}	1151	my $tcp_connect = $arg{tcp_connect}
892	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	1152	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
893		1153
894	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	1154	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
895		1155	}
896	};	1156	};
897		1157
898	defined wantarray && AnyEvent::Util::guard { %state = () }	1158	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
899	}	1159	}
900		1160
901	sub http_get($@) {	1161	sub http_get($@) {
902	unshift @_, "GET";	1162	unshift @_, "GET";
903	&http_request	1163	&http_request
…		…
921	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1181	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
922	the actual connection, which in turn uses AnyEvent::DNS to resolve	1182	the actual connection, which in turn uses AnyEvent::DNS to resolve
923	hostnames. The latter is a simple stub resolver and does no caching	1183	hostnames. The latter is a simple stub resolver and does no caching
924	on its own. If you want DNS caching, you currently have to provide	1184	on its own. If you want DNS caching, you currently have to provide
925	your own default resolver (by storing a suitable resolver object in	1185	your own default resolver (by storing a suitable resolver object in
926	C<$AnyEvent::DNS::RESOLVER>).	1186	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
927		1187
928	=head2 GLOBAL FUNCTIONS AND VARIABLES	1188	=head2 GLOBAL FUNCTIONS AND VARIABLES
929		1189
930	=over 4	1190	=over 4
931		1191
932	=item AnyEvent::HTTP::set_proxy "proxy-url"	1192	=item AnyEvent::HTTP::set_proxy "proxy-url"
933		1193
934	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
935	string of the form C<http://host:port> (optionally C<https:...>), croaks	1195	string of the form C<http://host:port>, croaks otherwise.
936	otherwise.
937		1196
938	To clear an already-set proxy, use C<undef>.	1197	To clear an already-set proxy, use C<undef>.
		1198
		1199	When AnyEvent::HTTP is laoded 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	}
939		1241
940	=item $date = AnyEvent::HTTP::format_date $timestamp	1242	=item $date = AnyEvent::HTTP::format_date $timestamp
941		1243
942	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP	1244	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
943	Date (RFC 2616).	1245	Date (RFC 2616).
944		1246
945	=item $timestamp = AnyEvent::HTTP::parse_date $date	1247	=item $timestamp = AnyEvent::HTTP::parse_date $date
946		1248
947	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) and	1249	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
948	returns the corresponding POSIX timestamp, or C<undef> if the date cannot	1250	bunch of minor variations of those, and returns the corresponding POSIX
949	be parsed.	1251	timestamp, or C<undef> if the date cannot be parsed.
950		1252
951	=item $AnyEvent::HTTP::MAX_RECURSE	1253	=item $AnyEvent::HTTP::MAX_RECURSE
952		1254
953	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>).
		1256
		1257	=item $AnyEvent::HTTP::TIMEOUT
		1258
		1259	The default timeout for connection operations (default: C<300>).
954		1260
955	=item $AnyEvent::HTTP::USERAGENT	1261	=item $AnyEvent::HTTP::USERAGENT
956		1262
957	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
958	C<Mozilla/5.0 (compatible; U; 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)>).
959		1265
960	=item $AnyEvent::HTTP::MAX_PER_HOST	1266	=item $AnyEvent::HTTP::MAX_PER_HOST
961		1267
962	The maximum number of concurrent connections to the same host (identified	1268	The maximum number of concurrent connections to the same host (identified
963	by the hostname). If the limit is exceeded, then the additional requests	1269	by the hostname). If the limit is exceeded, then the additional requests
964	are queued until previous connections are closed.	1270	are queued until previous connections are closed. Both persistent and
		1271	non-persistent connections are counted in this limit.
965		1272
966	The default value for this is C<4>, and it is highly advisable to not	1273	The default value for this is C<4>, and it is highly advisable to not
967	increase it.	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.
		1280
		1281	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1282
		1283	The time after which idle persistent conenctions get closed by
		1284	AnyEvent::HTTP (default: C<3>).
968		1285
969	=item $AnyEvent::HTTP::ACTIVE	1286	=item $AnyEvent::HTTP::ACTIVE
970		1287
971	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
972	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
973	connections. This number of can be useful for load-leveling.	1290	connections. This number can be useful for load-leveling.
974		1291
975	=back	1292	=back
976		1293
977	=cut	1294	=cut
978		1295
…		…
993	sub parse_date($) {	1310	sub parse_date($) {
994	my ($date) = @_;	1311	my ($date) = @_;
995		1312
996	my ($d, $m, $y, $H, $M, $S);	1313	my ($d, $m, $y, $H, $M, $S);
997		1314
998	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$/) {	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$/) {
999	# RFC 822/1123, required by RFC 2616 (with " ")	1316	# RFC 822/1123, required by RFC 2616 (with " ")
1000	# cookie dates (with "-")	1317	# cookie dates (with "-")
1001		1318
1002	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);	1319	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
1003		1320
1004	} elsif ($date =~ /^[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$/) {	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$/) {
1005	# RFC 850	1322	# RFC 850
1006	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);	1323	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
1007		1324
1008	} 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])$/) {	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])$/) {
1009	# ISO C's asctime	1326	# ISO C's asctime
1010	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);	1327	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
1011	}	1328	}
1012	# other formats fail in the loop below	1329	# other formats fail in the loop below
1013		1330
1014	for (0..11) {	1331	for (0..11) {
1015	if ($m eq $month[$_]) {	1332	if ($m eq $month[$_]) {
1016	require Time::Local;	1333	require Time::Local;
1017	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);	1334	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
1018	}	1335	}
1019	}	1336	}
1020		1337
1021	undef	1338	undef
1022	}	1339	}
1023		1340
1024	sub set_proxy($) {	1341	sub set_proxy($) {
1025	if (length $_[0]) {	1342	if (length $_[0]) {
1026	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1343	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1027	or Carp::croak "$_[0]: invalid proxy URL";	1344	or Carp::croak "$_[0]: invalid proxy URL";
1028	$PROXY = [$2, $3 || 3128, $1]	1345	$PROXY = [$2, $3 || 3128, $1]
1029	} else {	1346	} else {
1030	undef $PROXY;	1347	undef $PROXY;
1031	}	1348	}
…		…
1034	# initialise proxy from environment	1351	# initialise proxy from environment
1035	eval {	1352	eval {
1036	set_proxy $ENV{http_proxy};	1353	set_proxy $ENV{http_proxy};
1037	};	1354	};
1038		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
1039	=head2 SOCKS PROXIES	1458	=head3 SOCKS PROXIES
1040		1459
1041	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1460	Socks proxies are not directly supported by AnyEvent::HTTP. You can
1042	compile your perl to support socks, or use an external program such as	1461	compile your perl to support socks, or use an external program such as
1043	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy	1462	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1044	transparently.	1463	transparently.

Diff Legend

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