ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.70 by root, Fri Dec 31 20:31:47 2010 UTC vs.
Revision 1.105 by root, Tue May 10 12:33:51 2011 UTC

…		…
15	This module is an L<AnyEvent> user, you need to make sure that you use and	15	This module is an L<AnyEvent> user, you need to make sure that you use and
16	run a supported event loop.	16	run a supported event loop.
17		17
18	This module implements a simple, stateless and non-blocking HTTP	18	This module implements a simple, stateless and non-blocking HTTP
19	client. It supports GET, POST and other request methods, cookies and more,	19	client. It supports GET, POST and other request methods, cookies and more,
20	all on a very low level. It can follow redirects supports proxies and	20	all on a very low level. It can follow redirects, supports proxies, and
21	automatically limits the number of connections to the values specified in	21	automatically limits the number of connections to the values specified in
22	the RFC.	22	the RFC.
23		23
24	It should generally be a "good client" that is enough for most HTTP	24	It should generally be a "good client" that is enough for most HTTP
25	tasks. Simple tasks should be simple, but complex tasks should still be	25	tasks. Simple tasks should be simple, but complex tasks should still be
…		…
36		36
37	=cut	37	=cut
38		38
39	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
40		40
41	use strict;	41	use common::sense;
42	no warnings;
43		42
44	use 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 = '2.11';
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) = @_;
…		…
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 => $boolean
		353
		354	Only used when C<persistent> is also true. This parameter decides whether
		355	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		356	(as opposed to only a HTTP/1.1 persistent connection).
		357
		358	The default is true, except when using a proxy, in which case it defaults
		359	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		360
		361	=item handle_params => { key => value ... }
		362
		363	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		364	constructor that is called - not all requests will create a handle, and
		365	sometimes more than one is created, so this parameter is only good for
		366	setting hints.
		367
		368	Example: set the maximum read size to 4096, to potentially conserve memory
		369	at the cost of speed.
		370
		371	handle_params => {
		372	max_read_size => 4096,
		373	},
303		374
304	=back	375	=back
305		376
306	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	377	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
307	the response body.	378	the response body.
…		…
314	Example: do a HTTP HEAD request on https://www.google.com/, use a	385	Example: do a HTTP HEAD request on https://www.google.com/, use a
315	timeout of 30 seconds.	386	timeout of 30 seconds.
316		387
317	http_request	388	http_request
318	GET => "https://www.google.com",	389	GET => "https://www.google.com",
		390	headers => { "user-agent" => "MySearchClient 1.0" },
319	timeout => 30,	391	timeout => 30,
320	sub {	392	sub {
321	my ($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 $_[0];
367	}	442	}
368		443
		444	#############################################################################
		445	# cookie handling
		446
		447	# expire cookies
		448	sub cookie_jar_expire($;$) {
		449	my ($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
369	# continue to parse $_ for headers and place them into the arg	643	# continue to parse $_ for headers and place them into the arg
370	sub parse_hdr() {	644	sub _parse_hdr() {
371	my %hdr;	645	my %hdr;
372		646
373	# things seen, not parsed:	647	# things seen, not parsed:
374	# p3pP="NON CUR OTPi OUR NOR UNI"	648	# p3pP="NON CUR OTPi OUR NOR UNI"
375		649
…		…
389	for values %hdr;	663	for values %hdr;
390		664
391	\%hdr	665	\%hdr
392	}	666	}
393		667
		668	#############################################################################
		669	# http_get
		670
394	our $qr_nlnl = qr{(?<![^\012])\015?\012};	671	our $qr_nlnl = qr{(?<![^\012])\015?\012};
395		672
396	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	673	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
397	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(\%$$) {
		686	my ($state, $cb, $hdr) = @_;
		687
		688	&_destroy_state ($state);
		689
		690	$cb->(undef, $hdr);
		691	()
		692	}
398		693
399	sub http_request($$@) {	694	sub http_request($$@) {
400	my $cb = pop;	695	my $cb = pop;
401	my ($method, $url, %arg) = @_;	696	my ($method, $url, %arg) = @_;
402		697
…		…
420	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	715	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
421		716
422	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	717	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
423	if $recurse < 0;	718	if $recurse < 0;
424		719
425	my $proxy = $arg{proxy} || $PROXY;	720	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
426	my $timeout = $arg{timeout} || $TIMEOUT;	721	my $timeout = $arg{timeout} || $TIMEOUT;
427		722
428	my ($uscheme, $uauthority, $upath, $query, $fragment) =	723	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
429	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	724	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
430		725
431	$uscheme = lc $uscheme;	726	$uscheme = lc $uscheme;
432		727
433	my $uport = $uscheme eq "http" ? 80	728	my $uport = $uscheme eq "http" ? 80
434	: $uscheme eq "https" ? 443	729	: $uscheme eq "https" ? 443
435	: 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" });
436		731
437	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	732	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
438	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	733	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
439		734
440	my $uhost = $1;	735	my $uhost = lc $1;
441	$uport = $2 if defined $2;	736	$uport = $2 if defined $2;
442		737
443	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	738	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
444	unless exists $hdr{host};	739	unless exists $hdr{host};
445		740
…		…
448		743
449	$upath =~ s%^/?%/%;	744	$upath =~ s%^/?%/%;
450		745
451	# cookie processing	746	# cookie processing
452	if (my $jar = $arg{cookie_jar}) {	747	if (my $jar = $arg{cookie_jar}) {
453	%$jar = () if $jar->{version} != 1;	748	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
454
455	my @cookie;
456
457	while (my ($chost, $paths) = each %$jar) {
458	if ($chost =~ /^\./) {
459	next unless $chost eq substr $uhost, -length $chost;
460	} elsif ($chost =~ /\./) {
461	next unless $chost eq $uhost;
462	} else {
463	next;
464	}
465		749
466	while (my ($cpath, $cookies) = each %$paths) {
467	next unless $cpath eq substr $upath, 0, length $cpath;
468
469	while (my ($cookie, $kv) = each %$cookies) {
470	next if $uscheme ne "https" && exists $kv->{secure};
471
472	if (exists $kv->{expires}) {
473	if (AE::now > parse_date ($kv->{expires})) {
474	delete $cookies->{$cookie};
475	next;
476	}
477	}
478
479	my $value = $kv->{value};
480	$value =~ s/([\\"])/\\$1/g;
481	push @cookie, "$cookie=\"$value\"";
482	}
483	}
484	}
485
486	$hdr{cookie} = join "; ", @cookie	750	$hdr{cookie} = join "; ", @$cookies
487	if @cookie;	751	if @$cookies;
488	}	752	}
489		753
490	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path	754	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
491		755
492	if ($proxy) {	756	if ($proxy) {
…		…
495	$rscheme = "http" unless defined $rscheme;	759	$rscheme = "http" unless defined $rscheme;
496		760
497	# don't support https requests over https-proxy transport,	761	# don't support https requests over https-proxy transport,
498	# 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.
499	$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;
500	} else {	767	} else {
501	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	768	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
502	}	769	}
503		770
504	# leave out fragment and query string, just a heuristic	771	# leave out fragment and query string, just a heuristic
…		…
506	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	773	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
507		774
508	$hdr{"content-length"} = length $arg{body}	775	$hdr{"content-length"} = length $arg{body}
509	if length $arg{body} || $method ne "GET";	776	if length $arg{body} || $method ne "GET";
510		777
511	$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
512	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	789	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
513		790
514	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 $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	keepalive => 0,
		1058	sub {
		1059	%state = ();
		1060	&$cb
		1061	}
		1062	);
		1063	};
		1064	$hdl->on_read (sub {
		1065	return unless %state;
		1066
		1067	# as soon as we receive something, a connection close
		1068	# once more becomes a hard error
		1069	$hdl->{on_eof} = $old_eof;
		1070	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1071	});
		1072	} else {
		1073	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1074	}
		1075	};
		1076
		1077	my $prepare_handle = sub {
		1078	my ($hdl) = $state{handle};
		1079
		1080	$hdl->on_error (sub {
		1081	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1082	});
		1083	$hdl->on_eof (sub {
		1084	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1085	});
		1086	$hdl->timeout_reset;
		1087	$hdl->timeout ($timeout);
		1088	};
		1089
		1090	# connected to proxy (or origin server)
		1091	my $connect_cb = sub {
		1092	my $fh = shift
		1093	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1094
		1095	return unless delete $state{connect_guard};
		1096
		1097	# get handle
		1098	$state{handle} = new AnyEvent::Handle
		1099	%{ $arg{handle_params} },
		1100	fh => $fh,
		1101	peername => $uhost,
		1102	tls_ctx => $arg{tls_ctx},
		1103	;
		1104
		1105	$prepare_handle->();
		1106
		1107	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1108
		1109	# now handle proxy-CONNECT method
		1110	if ($proxy && $uscheme eq "https") {
		1111	# oh dear, we have to wrap it into a connect request
		1112
		1113	# maybe re-use $uauthority with patched port?
		1114	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012\015\012");
		1115	$state{handle}->push_read (line => $qr_nlnl, sub {
		1116	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1117	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1118
		1119	if ($2 == 200) {
		1120	$rpath = $upath;
		1121	$handle_actual_request->();
		1122	} else {
		1123	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1124	}
		1125	});
		1126	} else {
		1127	$handle_actual_request->();
		1128	}
		1129	};
515		1130
516	_get_slot $uhost, sub {	1131	_get_slot $uhost, sub {
517	$state{slot_guard} = shift;	1132	$state{slot_guard} = shift;
518		1133
519	return unless $state{connect_guard};	1134	return unless $state{connect_guard};
520		1135
521	my $connect_cb = sub {	1136	# try to use an existing keepalive connection, but only if we, ourselves, plan
522	$state{fh} = shift	1137	# on a keepalive request (in theory, this should be a separate config option).
523	or do {	1138	if ($persistent && $KA_CACHE{$ka_key}) {
524	my $err = "$!";	1139	$was_persistent = 1;
525	%state = ();
526	return $cb->(undef, { @pseudo, Status => 599, Reason => $err });
527	};
528		1140
529	pop; # free memory, save a tree	1141	$state{handle} = ka_fetch $ka_key;
530
531	return unless delete $state{connect_guard};
532
533	# get handle
534	$state{handle} = new AnyEvent::Handle
535	fh => $state{fh},
536	peername => $rhost,
537	tls_ctx => $arg{tls_ctx},
538	# these need to be reconfigured on keepalive handles
539	timeout => $timeout,
540	on_error => sub {
541	%state = ();
542	$cb->(undef, { @pseudo, Status => 599, Reason => $_[2] });
543	},
544	on_eof => sub {
545	%state = ();
546	$cb->(undef, { @pseudo, Status => 599, Reason => "Unexpected end-of-file" });
547	},
548	;
549
550	# limit the number of persistent connections
551	# keepalive not yet supported
552	# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
553	# ++$KA_COUNT{$_[1]};
554	# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
555	# --$KA_COUNT{$_[1]}
556	# };
557	# $hdr{connection} = "keep-alive";
558	# }
559
560	$state{handle}->starttls ("connect") if $rscheme eq "https";
561
562	# handle actual, non-tunneled, request
563	my $handle_actual_request = sub {
564	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};
565
566	# send request
567	$state{handle}->push_write (	1142	$state{handle}->destroyed
568	"$method $rpath HTTP/1.1\015\012"	1143	and die "got a destructed habndle. pah\n";#d#
569	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)	1144	$prepare_handle->();
570	. "\015\012"	1145	$state{handle}->destroyed
571	. (delete $arg{body})	1146	and die "got a destructed habndle. pa2\n";#d#
572	);
573
574	# return if error occured during push_write()
575	return unless %state;
576
577	%hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
578
579	# status line and headers
580	$state{read_response} = sub {
581	for ("$_[1]") {
582	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
583
584	/^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/igxc
585	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid server response" }));
586
587	# 100 Continue handling
588	# should not happen as we don't send expect: 100-continue,
589	# but we handle it just in case.
590	# since we send the request body regardless, if we get an error
591	# we are out of-sync, which we currently do NOT handle correctly.
592	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
593	if $2 eq 100;
594
595	push @pseudo,
596	HTTPVersion => $1,
597	Status => $2,
598	Reason => $3,
599	;
600
601	my $hdr = parse_hdr
602	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Garbled response headers" }));
603
604	%hdr = (%$hdr, @pseudo);
605	}
606
607	# redirect handling
608	# microsoft and other shitheads don't give a shit for following standards,
609	# try to support some common forms of broken Location headers.
610	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
611	$hdr{location} =~ s/^\.\/+//;
612
613	my $url = "$rscheme://$uhost:$uport";
614
615	unless ($hdr{location} =~ s/^\///) {
616	$url .= $upath;
617	$url =~ s/\/[^\/]*$//;
618	}
619
620	$hdr{location} = "$url/$hdr{location}";
621	}
622
623	my $redirect;
624
625	if ($recurse) {
626	my $status = $hdr{Status};
627
628	# industry standard is to redirect POST as GET for
629	# 301, 302 and 303, in contrast to http/1.0 and 1.1.
630	# also, the UA should ask the user for 301 and 307 and POST,
631	# industry standard seems to be to simply follow.
632	# we go with the industry standard.
633	if ($status == 301 or $status == 302 or $status == 303) {
634	# HTTP/1.1 is unclear on how to mutate the method
635	$method = "GET" unless $method eq "HEAD";
636	$redirect = 1;
637	} elsif ($status == 307) {
638	$redirect = 1;
639	}
640	}
641
642	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
643	my $keepalive = pop;
644
645	$state{handle}->destroy if $state{handle};
646	%state = ();
647
648	if (defined $_[1]) {
649	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
650	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
651	}
652
653	# set-cookie processing
654	if ($arg{cookie_jar}) {
655	for ($hdr{"set-cookie"}) {
656	# parse NAME=VALUE
657	my @kv;
658
659	while (
660	m{
661	\G\s*
662	(?:
663	expires \s*=\s* ([A-Z][a-z][a-z],\ [^,;]+)
664	| ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )
665	)
666	}gcxsi
667	) {
668	my $name = $2;
669	my $value = $4;
670
671	unless (defined $name) {
672	# expires
673	$name = "expires";
674	$value = $1;
675	} elsif (!defined $value) {
676	# quoted
677	$value = $3;
678	$value =~ s/\\(.)/$1/gs;
679	}
680
681	push @kv, lc $name, $value;
682
683	last unless /\G\s*;/gc;
684	}
685
686	last unless @kv;
687
688	my $name = shift @kv;
689	my %kv = (value => shift @kv, @kv);
690
691	$kv{expires} ||= format_date (AE::now + $kv{"max-age"})
692	if exists $kv{"max-age"};
693
694	my $cdom;
695	my $cpath = (delete $kv{path}) || "/";
696
697	if (exists $kv{domain}) {
698	$cdom = delete $kv{domain};
699
700	$cdom =~ s/^\.?/./; # make sure it starts with a "."
701
702	next if $cdom =~ /\.$/;
703
704	# this is not rfc-like and not netscape-like. go figure.
705	my $ndots = $cdom =~ y/.//;
706	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
707	} else {
708	$cdom = $uhost;
709	}
710
711	# store it
712	$arg{cookie_jar}{version} = 1;
713	$arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
714
715	redo if /\G\s*,/gc;
716	}
717	}
718
719	if ($redirect && exists $hdr{location}) {
720	# we ignore any errors, as it is very common to receive
721	# Content-Length != 0 but no actual body
722	# we also access %hdr, as $_[1] might be an erro
723	http_request (
724	$method => $hdr{location},
725	%arg,
726	recurse => $recurse - 1,
727	Redirect => [$_[0], \%hdr],
728	$cb);
729	} else {
730	$cb->($_[0], \%hdr);
731	}
732	};
733
734	my $len = $hdr{"content-length"};
735
736	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
737	$finish->(undef, 598 => "Request cancelled by on_header");
738	} elsif (
739	$hdr{Status} =~ /^(?:1..|204|205|304)$/
740	or $method eq "HEAD"
741	or (defined $len && !$len)
742	) {
743	# no body
744	$finish->("", undef, undef, 1);
745	} else {
746	# body handling, many different code paths
747	# - no body expected
748	# - want_body_handle
749	# - te chunked
750	# - 2x length known (with or without on_body)
751	# - 2x length not known (with or without on_body)
752	if (!$redirect && $arg{want_body_handle}) {
753	$_[0]->on_eof (undef);
754	$_[0]->on_error (undef);
755	$_[0]->on_read (undef);
756
757	$finish->(delete $state{handle});
758
759	} elsif ($hdr{"transfer-encoding"} =~ /\bchunked\b/i) {
760	my $cl = 0;
761	my $body = undef;
762	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
763
764	$_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
765
766	my $read_chunk; $read_chunk = sub {
767	$_[1] =~ /^([0-9a-fA-F]+)/
768	or $finish->(undef, 599 => "Garbled chunked transfer encoding");
769
770	my $len = hex $1;
771
772	if ($len) {
773	$cl += $len;
774
775	$_[0]->push_read (chunk => $len, sub {
776	$on_body->($_[1], \%hdr)
777	or return $finish->(undef, 598 => "Request cancelled by on_body");
778
779	$_[0]->push_read (line => sub {
780	length $_[1]
781	and return $finish->(undef, 599 => "Garbled chunked transfer encoding");
782	$_[0]->push_read (line => $read_chunk);
783	});
784	});
785	} else {
786	$hdr{"content-length"} ||= $cl;
787
788	$_[0]->push_read (line => $qr_nlnl, sub {
789	if (length $_[1]) {
790	for ("$_[1]") {
791	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
792
793	my $hdr = parse_hdr
794	or return $finish->(undef, 599 => "Garbled response trailers");
795
796	%hdr = (%hdr, %$hdr);
797	}
798	}
799
800	$finish->($body, undef, undef, 1);
801	});
802	}
803	};
804
805	$_[0]->push_read (line => $read_chunk);
806
807	} elsif ($arg{on_body}) {
808	$_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
809
810	if ($len) {
811	$_[0]->on_read (sub {
812	$len -= length $_[0]{rbuf};
813
814	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
815	or return $finish->(undef, 598 => "Request cancelled by on_body");
816
817	$len > 0
818	or $finish->("", undef, undef, 1);
819	});
820	} else {
821	$_[0]->on_eof (sub {
822	$finish->("");
823	});
824	$_[0]->on_read (sub {
825	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
826	or $finish->(undef, 598 => "Request cancelled by on_body");
827	});
828	}
829	} else {
830	$_[0]->on_eof (undef);
831
832	if ($len) {
833	$_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
834	$_[0]->on_read (sub {
835	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
836	if $len <= length $_[0]{rbuf};
837	});
838	} else {
839	$_[0]->on_error (sub {
840	($! == Errno::EPIPE || !$!)
841	? $finish->(delete $_[0]{rbuf})
842	: $finish->(undef, 599 => $_[2]);
843	});
844	$_[0]->on_read (sub { });
845	}
846	}
847	}
848	};
849
850	$state{handle}->push_read (line => $qr_nlnl, $state{read_response});
851	};
852
853	# now handle proxy-CONNECT method
854	if ($proxy && $uscheme eq "https") {
855	# oh dear, we have to wrap it into a connect request
856
857	# maybe re-use $uauthority with patched port?
858	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
859	$state{handle}->push_read (line => $qr_nlnl, sub {
860	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
861	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" }));
862
863	if ($2 == 200) {
864	$rpath = $upath;
865	&$handle_actual_request;	1147	$handle_actual_request->();
866	} else {	1148	$state{handle}->destroyed
867	%state = ();	1149	and die "got a destructed habndle. pa3\n";#d#
868	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });	1150
869	}
870	});
871	} else {	1151	} else {
872	&$handle_actual_request;
873	}
874	};
875
876	my $tcp_connect = $arg{tcp_connect}	1152	my $tcp_connect = $arg{tcp_connect}
877	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	1153	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
878		1154
879	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	1155	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
880		1156	}
881	};	1157	};
882		1158
883	defined wantarray && AnyEvent::Util::guard { %state = () }	1159	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
884	}	1160	}
885		1161
886	sub http_get($@) {	1162	sub http_get($@) {
887	unshift @_, "GET";	1163	unshift @_, "GET";
888	&http_request	1164	&http_request
…		…
906	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1182	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
907	the actual connection, which in turn uses AnyEvent::DNS to resolve	1183	the actual connection, which in turn uses AnyEvent::DNS to resolve
908	hostnames. The latter is a simple stub resolver and does no caching	1184	hostnames. The latter is a simple stub resolver and does no caching
909	on its own. If you want DNS caching, you currently have to provide	1185	on its own. If you want DNS caching, you currently have to provide
910	your own default resolver (by storing a suitable resolver object in	1186	your own default resolver (by storing a suitable resolver object in
911	C<$AnyEvent::DNS::RESOLVER>).	1187	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
912		1188
913	=head2 GLOBAL FUNCTIONS AND VARIABLES	1189	=head2 GLOBAL FUNCTIONS AND VARIABLES
914		1190
915	=over 4	1191	=over 4
916		1192
917	=item AnyEvent::HTTP::set_proxy "proxy-url"	1193	=item AnyEvent::HTTP::set_proxy "proxy-url"
918		1194
919	Sets the default proxy server to use. The proxy-url must begin with a	1195	Sets the default proxy server to use. The proxy-url must begin with a
920	string of the form C<http://host:port> (optionally C<https:...>), croaks	1196	string of the form C<http://host:port>, croaks otherwise.
921	otherwise.
922		1197
923	To clear an already-set proxy, use C<undef>.	1198	To clear an already-set proxy, use C<undef>.
		1199
		1200	When AnyEvent::HTTP is laoded for the first time it will query the
		1201	default proxy from the operating system, currently by looking at
		1202	C<$ENV{http_proxy>}.
		1203
		1204	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1205
		1206	Remove all cookies from the cookie jar that have been expired. If
		1207	C<$session_end> is given and true, then additionally remove all session
		1208	cookies.
		1209
		1210	You should call this function (with a true C<$session_end>) before you
		1211	save cookies to disk, and you should call this function after loading them
		1212	again. If you have a long-running program you can additonally call this
		1213	function from time to time.
		1214
		1215	A cookie jar is initially an empty hash-reference that is managed by this
		1216	module. It's format is subject to change, but currently it is like this:
		1217
		1218	The key C<version> has to contain C<1>, otherwise the hash gets
		1219	emptied. All other keys are hostnames or IP addresses pointing to
		1220	hash-references. The key for these inner hash references is the
		1221	server path for which this cookie is meant, and the values are again
		1222	hash-references. The keys of those hash-references is the cookie name, and
		1223	the value, you guessed it, is another hash-reference, this time with the
		1224	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1225	which have been replaced by a C<_expires> key that contains the cookie
		1226	expiry timestamp.
		1227
		1228	Here is an example of a cookie jar with a single cookie, so you have a
		1229	chance of understanding the above paragraph:
		1230
		1231	{
		1232	version => 1,
		1233	"10.0.0.1" => {
		1234	"/" => {
		1235	"mythweb_id" => {
		1236	_expires => 1293917923,
		1237	value => "ooRung9dThee3ooyXooM1Ohm",
		1238	},
		1239	},
		1240	},
		1241	}
924		1242
925	=item $date = AnyEvent::HTTP::format_date $timestamp	1243	=item $date = AnyEvent::HTTP::format_date $timestamp
926		1244
927	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP	1245	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
928	Date (RFC 2616).	1246	Date (RFC 2616).
929		1247
930	=item $timestamp = AnyEvent::HTTP::parse_date $date	1248	=item $timestamp = AnyEvent::HTTP::parse_date $date
931		1249
932	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) and	1250	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
933	returns the corresponding POSIX timestamp, or C<undef> if the date cannot	1251	bunch of minor variations of those, and returns the corresponding POSIX
934	be parsed.	1252	timestamp, or C<undef> if the date cannot be parsed.
935		1253
936	=item $AnyEvent::HTTP::MAX_RECURSE	1254	=item $AnyEvent::HTTP::MAX_RECURSE
937		1255
938	The default value for the C<recurse> request parameter (default: C<10>).	1256	The default value for the C<recurse> request parameter (default: C<10>).
		1257
		1258	=item $AnyEvent::HTTP::TIMEOUT
		1259
		1260	The default timeout for conenction operations (default: C<300>).
939		1261
940	=item $AnyEvent::HTTP::USERAGENT	1262	=item $AnyEvent::HTTP::USERAGENT
941		1263
942	The default value for the C<User-Agent> header (the default is	1264	The default value for the C<User-Agent> header (the default is
943	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1265	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
944		1266
945	=item $AnyEvent::HTTP::MAX_PER_HOST	1267	=item $AnyEvent::HTTP::MAX_PER_HOST
946		1268
947	The maximum number of concurrent connections to the same host (identified	1269	The maximum number of concurrent connections to the same host (identified
948	by the hostname). If the limit is exceeded, then the additional requests	1270	by the hostname). If the limit is exceeded, then the additional requests
949	are queued until previous connections are closed.	1271	are queued until previous connections are closed. Both persistent and
		1272	non-persistent connections are counted in this limit.
950		1273
951	The default value for this is C<4>, and it is highly advisable to not	1274	The default value for this is C<4>, and it is highly advisable to not
952	increase it.	1275	increase it much.
		1276
		1277	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1278	connections, older browsers used 2, newers (such as firefox 3) typically
		1279	use 6, and Opera uses 8 because like, they have the fastest browser and
		1280	give a shit for everybody else on the planet.
		1281
		1282	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1283
		1284	The time after which idle persistent conenctions get closed by
		1285	AnyEvent::HTTP (default: C<3>).
953		1286
954	=item $AnyEvent::HTTP::ACTIVE	1287	=item $AnyEvent::HTTP::ACTIVE
955		1288
956	The number of active connections. This is not the number of currently	1289	The number of active connections. This is not the number of currently
957	running requests, but the number of currently open and non-idle TCP	1290	running requests, but the number of currently open and non-idle TCP
958	connections. This number of can be useful for load-leveling.	1291	connections. This number can be useful for load-leveling.
959		1292
960	=back	1293	=back
961		1294
962	=cut	1295	=cut
963		1296
…		…
978	sub parse_date($) {	1311	sub parse_date($) {
979	my ($date) = @_;	1312	my ($date) = @_;
980		1313
981	my ($d, $m, $y, $H, $M, $S);	1314	my ($d, $m, $y, $H, $M, $S);
982		1315
983	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	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$/) {
984	# RFC 822/1123, required by RFC 2616 (with " ")	1317	# RFC 822/1123, required by RFC 2616 (with " ")
985	# cookie dates (with "-")	1318	# cookie dates (with "-")
986		1319
987	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);	1320	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
988		1321
989	} 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$/) {	1322	} 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$/) {
990	# RFC 850	1323	# RFC 850
991	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);	1324	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
992		1325
993	} 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	} 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])$/) {
994	# ISO C's asctime	1327	# ISO C's asctime
995	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);	1328	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
996	}	1329	}
997	# other formats fail in the loop below	1330	# other formats fail in the loop below
998		1331
…		…
1006	undef	1339	undef
1007	}	1340	}
1008		1341
1009	sub set_proxy($) {	1342	sub set_proxy($) {
1010	if (length $_[0]) {	1343	if (length $_[0]) {
1011	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1344	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1012	or Carp::croak "$_[0]: invalid proxy URL";	1345	or Carp::croak "$_[0]: invalid proxy URL";
1013	$PROXY = [$2, $3 || 3128, $1]	1346	$PROXY = [$2, $3 || 3128, $1]
1014	} else {	1347	} else {
1015	undef $PROXY;	1348	undef $PROXY;
1016	}	1349	}
…		…
1019	# initialise proxy from environment	1352	# initialise proxy from environment
1020	eval {	1353	eval {
1021	set_proxy $ENV{http_proxy};	1354	set_proxy $ENV{http_proxy};
1022	};	1355	};
1023		1356
		1357	=head2 SHOWCASE
		1358
		1359	This section contaisn some more elaborate "real-world" examples or code
		1360	snippets.
		1361
		1362	=head2 HTTP/1.1 FILE DOWNLOAD
		1363
		1364	Downloading files with HTTP can be quite tricky, especially when something
		1365	goes wrong and you want to resume.
		1366
		1367	Here is a function that initiates and resumes a download. It uses the
		1368	last modified time to check for file content changes, and works with many
		1369	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1370	on older servers.
		1371
		1372	It calls the completion callback with either C<undef>, which means a
		1373	nonretryable error occured, C<0> when the download was partial and should
		1374	be retried, and C<1> if it was successful.
		1375
		1376	use AnyEvent::HTTP;
		1377
		1378	sub download($$$) {
		1379	my ($url, $file, $cb) = @_;
		1380
		1381	open my $fh, "+<", $file
		1382	or die "$file: $!";
		1383
		1384	my %hdr;
		1385	my $ofs = 0;
		1386
		1387	warn stat $fh;
		1388	warn -s _;
		1389	if (stat $fh and -s _) {
		1390	$ofs = -s _;
		1391	warn "-s is ", $ofs;#d#
		1392	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1393	$hdr{"range"} = "bytes=$ofs-";
		1394	}
		1395
		1396	http_get $url,
		1397	headers => \%hdr,
		1398	on_header => sub {
		1399	my ($hdr) = @_;
		1400
		1401	if ($hdr->{Status} == 200 && $ofs) {
		1402	# resume failed
		1403	truncate $fh, $ofs = 0;
		1404	}
		1405
		1406	sysseek $fh, $ofs, 0;
		1407
		1408	1
		1409	},
		1410	on_body => sub {
		1411	my ($data, $hdr) = @_;
		1412
		1413	if ($hdr->{Status} =~ /^2/) {
		1414	length $data == syswrite $fh, $data
		1415	or return; # abort on write errors
		1416	}
		1417
		1418	1
		1419	},
		1420	sub {
		1421	my (undef, $hdr) = @_;
		1422
		1423	my $status = $hdr->{Status};
		1424
		1425	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1426	utime $fh, $time, $time;
		1427	}
		1428
		1429	if ($status == 200 || $status == 206 || $status == 416) {
		1430	# download ok || resume ok || file already fully downloaded
		1431	$cb->(1, $hdr);
		1432
		1433	} elsif ($status == 412) {
		1434	# file has changed while resuming, delete and retry
		1435	unlink $file;
		1436	$cb->(0, $hdr);
		1437
		1438	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1439	# retry later
		1440	$cb->(0, $hdr);
		1441
		1442	} else {
		1443	$cb->(undef, $hdr);
		1444	}
		1445	}
		1446	;
		1447	}
		1448
		1449	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1450	if ($_[0]) {
		1451	print "OK!\n";
		1452	} elsif (defined $_[0]) {
		1453	print "please retry later\n";
		1454	} else {
		1455	print "ERROR\n";
		1456	}
		1457	};
		1458
1024	=head2 SOCKS PROXIES	1459	=head3 SOCKS PROXIES
1025		1460
1026	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1461	Socks proxies are not directly supported by AnyEvent::HTTP. You can
1027	compile your perl to support socks, or use an external program such as	1462	compile your perl to support socks, or use an external program such as
1028	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy	1463	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1029	transparently.	1464	transparently.

Diff Legend

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