ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.79 by root, Sat Jan 1 20:01:07 2011 UTC vs.
Revision 1.140 by root, Wed Mar 6 19:29:18 2024 UTC

…		…
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use AnyEvent::HTTP;	7	use AnyEvent::HTTP;
8		8
9	http_get "http://www.nethype.de/", sub { print $_[1] };	9	http_get "http://www.nethype.de/", sub {
		10	my ($body, $hdr) = @_;
		11	print "$hdr->{URL} Status: $hdr->{Status}\n";
		12	print $body;
		13	};
10		14
11	# ... do something else here	15	# ... do something else here
12		16
13	=head1 DESCRIPTION	17	=head1 DESCRIPTION
14		18
15	This module is an L<AnyEvent> user, you need to make sure that you use and	19	This module is an L<AnyEvent> user, you need to make sure that you use and
16	run a supported event loop.	20	run a supported event loop.
17		21
18	This module implements a simple, stateless and non-blocking HTTP	22	This module implements a simple, stateless and non-blocking HTTP
19	client. It supports GET, POST and other request methods, cookies and more,	23	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	24	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	25	automatically limits the number of connections to the values specified in
22	the RFC.	26	the RFC.
23		27
24	It should generally be a "good client" that is enough for most HTTP	28	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	29	tasks. Simple tasks should be simple, but complex tasks should still be
…		…
36		40
37	=cut	41	=cut
38		42
39	package AnyEvent::HTTP;	43	package AnyEvent::HTTP;
40		44
41	use strict;	45	use common::sense;
42	no warnings;
43		46
44	use Errno ();	47	use Errno ();
45		48
46	use AnyEvent 5.0 ();	49	use AnyEvent 5.0 ();
47	use AnyEvent::Util ();	50	use AnyEvent::Util ();
48	use AnyEvent::Handle ();	51	use AnyEvent::Handle ();
49		52
50	use base Exporter::;	53	use base Exporter::;
51		54
52	our $VERSION = '1.5';	55	our $VERSION = 2.25;
53		56
54	our @EXPORT = qw(http_get http_post http_head http_request);	57	our @EXPORT = qw(http_get http_post http_head http_request);
55		58
56	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";	59	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
57	our $MAX_RECURSE = 10;	60	our $MAX_RECURSE = 10;
58	our $MAX_PERSISTENT = 8;
59	our $PERSISTENT_TIMEOUT = 2;	61	our $PERSISTENT_TIMEOUT = 3;
60	our $TIMEOUT = 300;	62	our $TIMEOUT = 300;
61		63	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		64
66	our $PROXY;	65	our $PROXY;
67	our $ACTIVE = 0;	66	our $ACTIVE = 0;
68		67
69	my %KA_COUNT; # number of open keep-alive connections per host	68	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
70	my %CO_SLOT; # number of open connections, and wait queue, per host	69	my %CO_SLOT; # number of open connections, and wait queue, per host
71		70
72	=item http_get $url, key => value..., $cb->($data, $headers)	71	=item http_get $url, key => value..., $cb->($data, $headers)
73		72
74	Executes an HTTP-GET request. See the http_request function for details on	73	Executes an HTTP-GET request. See the http_request function for details on
…		…
94	C<http_request> returns a "cancellation guard" - you have to keep the	93	C<http_request> returns a "cancellation guard" - you have to keep the
95	object at least alive until the callback get called. If the object gets	94	object at least alive until the callback get called. If the object gets
96	destroyed before the callback is called, the request will be cancelled.	95	destroyed before the callback is called, the request will be cancelled.
97		96
98	The callback will be called with the response body data as first argument	97	The callback will be called with the response body data as first argument
99	(or C<undef> if an error occured), and a hash-ref with response headers	98	(or C<undef> if an error occurred), and a hash-ref with response headers
100	(and trailers) as second argument.	99	(and trailers) as second argument.
101		100
102	All the headers in that hash are lowercased. In addition to the response	101	All the headers in that hash are lowercased. In addition to the response
103	headers, the "pseudo-headers" (uppercase to avoid clashing with possible	102	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
104	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the	103	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
…		…
128	C<590>-C<599> and the C<Reason> pseudo-header will contain an error	127	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
129	message. Currently the following status codes are used:	128	message. Currently the following status codes are used:
130		129
131	=over 4	130	=over 4
132		131
133	=item 595 - errors during connection etsbalishment, proxy handshake.	132	=item 595 - errors during connection establishment, proxy handshake.
134		133
135	=item 596 - errors during TLS negotiation, request sending and header processing.	134	=item 596 - errors during TLS negotiation, request sending and header processing.
136		135
137	=item 597 - errors during body receiving or processing.	136	=item 597 - errors during body receiving or processing.
138		137
…		…
159		158
160	=over 4	159	=over 4
161		160
162	=item recurse => $count (default: $MAX_RECURSE)	161	=item recurse => $count (default: $MAX_RECURSE)
163		162
164	Whether to recurse requests or not, e.g. on redirects, authentication	163	Whether to recurse requests or not, e.g. on redirects, authentication and
165	retries and so on, and how often to do so.	164	other retries and so on, and how often to do so.
		165
		166	Only redirects to http and https URLs are supported. While most common
		167	redirection forms are handled entirely within this module, some require
		168	the use of the optional L<URI> module. If it is required but missing, then
		169	the request will fail with an error.
166		170
167	=item headers => hashref	171	=item headers => hashref
168		172
169	The request headers to use. Currently, C<http_request> may provide its own	173	The request headers to use. Currently, C<http_request> may provide its own
170	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and	174	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
171	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>	175	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
172	(this can be suppressed by using C<undef> for these headers in which case	176	(this can be suppressed by using C<undef> for these headers in which case
173	they won't be sent at all).	177	they won't be sent at all).
174		178
		179	You really should provide your own C<User-Agent:> header value that is
		180	appropriate for your program - I wouldn't be surprised if the default
		181	AnyEvent string gets blocked by webservers sooner or later.
		182
		183	Also, make sure that your headers names and values do not contain any
		184	embedded newlines.
		185
175	=item timeout => $seconds	186	=item timeout => $seconds
176		187
177	The time-out to use for various stages - each connect attempt will reset	188	The time-out to use for various stages - each connect attempt will reset
178	the timeout, as will read or write activity, i.e. this is not an overall	189	the timeout, as will read or write activity, i.e. this is not an overall
179	timeout.	190	timeout.
180		191
181	Default timeout is 5 minutes.	192	Default timeout is 5 minutes.
182		193
183	=item proxy => [$host, $port[, $scheme]] or undef	194	=item proxy => [$host, $port[, $scheme]] or undef
184		195
185	Use the given http proxy for all requests. If not specified, then the	196	Use the given http proxy for all requests, or no proxy if C<undef> is
186	default proxy (as specified by C<$ENV{http_proxy}>) is used.	197	used.
187		198
188	C<$scheme> must be either missing, C<http> for HTTP or C<https> for	199	C<$scheme> must be either missing or must be C<http> for HTTP.
189	HTTPS.	200
		201	If not specified, then the default proxy is used (see
		202	C<AnyEvent::HTTP::set_proxy>).
		203
		204	Currently, if your proxy requires authorization, you have to specify an
		205	appropriate "Proxy-Authorization" header in every request.
		206
		207	Note that this module will prefer an existing persistent connection,
		208	even if that connection was made using another proxy. If you need to
		209	ensure that a new connection is made in this case, you can either force
		210	C<persistent> to false or e.g. use the proxy address in your C<sessionid>.
190		211
191	=item body => $string	212	=item body => $string
192		213
193	The request body, usually empty. Will be sent as-is (future versions of	214	The request body, usually empty. Will be sent as-is (future versions of
194	this module might offer more options).	215	this module might offer more options).
…		…
196	=item cookie_jar => $hash_ref	217	=item cookie_jar => $hash_ref
197		218
198	Passing this parameter enables (simplified) cookie-processing, loosely	219	Passing this parameter enables (simplified) cookie-processing, loosely
199	based on the original netscape specification.	220	based on the original netscape specification.
200		221
201	The C<$hash_ref> must be an (initially empty) hash reference which will	222	The C<$hash_ref> must be an (initially empty) hash reference which
202	get updated automatically. It is possible to save the cookie jar to	223	will get updated automatically. It is possible to save the cookie jar
203	persistent storage with something like JSON or Storable, but this is not	224	to persistent storage with something like JSON or Storable - see the
204	recommended, as session-only cookies might survive longer than expected.	225	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		226	expired or session-only cookies, and also for documentation on the format
		227	of the cookie jar.
205		228
206	Note that this cookie implementation is not meant to be complete. If	229	Note that this cookie implementation is not meant to be complete. If
207	you want complete cookie management you have to do that on your	230	you want complete cookie management you have to do that on your
208	own. C<cookie_jar> is meant as a quick fix to get some cookie-using sites	231	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
209	working. Cookies are a privacy disaster, do not use them unless required	232	working. Cookies are a privacy disaster, do not use them unless required
210	to.	233	to.
211		234
212	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>	235	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
213	headers will be set and handled by this module, otherwise they will be	236	headers will be set and handled by this module, otherwise they will be
…		…
223	verification) TLS context.	246	verification) TLS context.
224		247
225	The default for this option is C<low>, which could be interpreted as "give	248	The default for this option is C<low>, which could be interpreted as "give
226	me the page, no matter what".	249	me the page, no matter what".
227		250
		251	See also the C<sessionid> parameter.
		252
		253	=item sessionid => $string
		254
		255	The module might reuse connections to the same host internally (regardless
		256	of other settings, such as C<tcp_connect> or C<proxy>). Sometimes (e.g.
		257	when using TLS or a specfic proxy), you do not want to reuse connections
		258	from other sessions. This can be achieved by setting this parameter to
		259	some unique ID (such as the address of an object storing your state data
		260	or the TLS context, or the proxy IP) - only connections using the same
		261	unique ID will be reused.
		262
228	=item on_prepare => $callback->($fh)	263	=item on_prepare => $callback->($fh)
229		264
230	In rare cases you need to "tune" the socket before it is used to	265	In rare cases you need to "tune" the socket before it is used to
231	connect (for exmaple, to bind it on a given IP address). This parameter	266	connect (for example, to bind it on a given IP address). This parameter
232	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>	267	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
233	and behaves exactly the same way (e.g. it has to provide a	268	and behaves exactly the same way (e.g. it has to provide a
234	timeout). See the description for the C<$prepare_cb> argument of	269	timeout). See the description for the C<$prepare_cb> argument of
235	C<AnyEvent::Socket::tcp_connect> for details.	270	C<AnyEvent::Socket::tcp_connect> for details.
236		271
…		…
239	In even rarer cases you want total control over how AnyEvent::HTTP	274	In even rarer cases you want total control over how AnyEvent::HTTP
240	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>	275	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
241	to do this, but you can provide your own C<tcp_connect> function -	276	to do this, but you can provide your own C<tcp_connect> function -
242	obviously, it has to follow the same calling conventions, except that it	277	obviously, it has to follow the same calling conventions, except that it
243	may always return a connection guard object.	278	may always return a connection guard object.
		279
		280	The connections made by this hook will be treated as equivalent to
		281	connections made the built-in way, specifically, they will be put into
		282	and taken from the persistent connection cache. If your C<$tcp_connect>
		283	function is incompatible with this kind of re-use, consider switching off
		284	C<persistent> connections and/or providing a C<sessionid> identifier.
244		285
245	There are probably lots of weird uses for this function, starting from	286	There are probably lots of weird uses for this function, starting from
246	tracing the hosts C<http_request> actually tries to connect, to (inexact	287	tracing the hosts C<http_request> actually tries to connect, to (inexact
247	but fast) host => IP address caching or even socks protocol support.	288	but fast) host => IP address caching or even socks protocol support.
248		289
…		…
301	called. Instead of the C<$body> argument containing the body data, the	342	called. Instead of the C<$body> argument containing the body data, the
302	callback will receive the L<AnyEvent::Handle> object associated with the	343	callback will receive the L<AnyEvent::Handle> object associated with the
303	connection. In error cases, C<undef> will be passed. When there is no body	344	connection. In error cases, C<undef> will be passed. When there is no body
304	(e.g. status C<304>), the empty string will be passed.	345	(e.g. status C<304>), the empty string will be passed.
305		346
306	The handle object might or might not be in TLS mode, might be connected to	347	The handle object might or might not be in TLS mode, might be connected
307	a proxy, be a persistent connection etc., and configured in unspecified	348	to a proxy, be a persistent connection, use chunked transfer encoding
308	ways. The user is responsible for this handle (it will not be used by this	349	etc., and configured in unspecified ways. The user is responsible for this
309	module anymore).	350	handle (it will not be used by this module anymore).
310		351
311	This is useful with some push-type services, where, after the initial	352	This is useful with some push-type services, where, after the initial
312	headers, an interactive protocol is used (typical example would be the	353	headers, an interactive protocol is used (typical example would be the
313	push-style twitter API which starts a JSON/XML stream).	354	push-style twitter API which starts a JSON/XML stream).
314		355
315	If you think you need this, first have a look at C<on_body>, to see if	356	If you think you need this, first have a look at C<on_body>, to see if
316	that doesn't solve your problem in a better way.	357	that doesn't solve your problem in a better way.
		358
		359	=item persistent => $boolean
		360
		361	Try to create/reuse a persistent connection. When this flag is set
		362	(default: true for idempotent requests, false for all others), then
		363	C<http_request> tries to re-use an existing (previously-created)
		364	persistent connection to same host (i.e. identical URL scheme, hostname,
		365	port and sessionid) and, failing that, tries to create a new one.
		366
		367	Requests failing in certain ways will be automatically retried once, which
		368	is dangerous for non-idempotent requests, which is why it defaults to off
		369	for them. The reason for this is because the bozos who designed HTTP/1.1
		370	made it impossible to distinguish between a fatal error and a normal
		371	connection timeout, so you never know whether there was a problem with
		372	your request or not.
		373
		374	When reusing an existent connection, many parameters (such as TLS context)
		375	will be ignored. See the C<sessionid> parameter for a workaround.
		376
		377	=item keepalive => $boolean
		378
		379	Only used when C<persistent> is also true. This parameter decides whether
		380	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		381	(as opposed to only a HTTP/1.1 persistent connection).
		382
		383	The default is true, except when using a proxy, in which case it defaults
		384	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		385
		386	=item handle_params => { key => value ... }
		387
		388	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		389	constructor that is called - not all requests will create a handle, and
		390	sometimes more than one is created, so this parameter is only good for
		391	setting hints.
		392
		393	Example: set the maximum read size to 4096, to potentially conserve memory
		394	at the cost of speed.
		395
		396	handle_params => {
		397	max_read_size => 4096,
		398	},
317		399
318	=back	400	=back
319		401
320	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	402	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
321	the response body.	403	the response body.
…		…
327		409
328	Example: do a HTTP HEAD request on https://www.google.com/, use a	410	Example: do a HTTP HEAD request on https://www.google.com/, use a
329	timeout of 30 seconds.	411	timeout of 30 seconds.
330		412
331	http_request	413	http_request
332	GET => "https://www.google.com",	414	HEAD => "https://www.google.com",
		415	headers => { "user-agent" => "MySearchClient 1.0" },
333	timeout => 30,	416	timeout => 30,
334	sub {	417	sub {
335	my ($body, $hdr) = @_;	418	my ($body, $hdr) = @_;
336	use Data::Dumper;	419	use Data::Dumper;
337	print Dumper $hdr;	420	print Dumper $hdr;
…		…
347	};	430	};
348		431
349	undef $request;	432	undef $request;
350		433
351	=cut	434	=cut
		435
		436	#############################################################################
		437	# wait queue/slots
352		438
353	sub _slot_schedule;	439	sub _slot_schedule;
354	sub _slot_schedule($) {	440	sub _slot_schedule($) {
355	my $host = shift;	441	my $host = shift;
356		442
…		…
378	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	464	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
379		465
380	_slot_schedule $_[0];	466	_slot_schedule $_[0];
381	}	467	}
382		468
		469	#############################################################################
		470	# cookie handling
		471
		472	# expire cookies
		473	sub cookie_jar_expire($;$) {
		474	my ($jar, $session_end) = @_;
		475
		476	%$jar = () if $jar->{version} != 2;
		477
		478	my $anow = AE::now;
		479
		480	while (my ($chost, $paths) = each %$jar) {
		481	next unless ref $paths;
		482
		483	while (my ($cpath, $cookies) = each %$paths) {
		484	while (my ($cookie, $kv) = each %$cookies) {
		485	if (exists $kv->{_expires}) {
		486	delete $cookies->{$cookie}
		487	if $anow > $kv->{_expires};
		488	} elsif ($session_end) {
		489	delete $cookies->{$cookie};
		490	}
		491	}
		492
		493	delete $paths->{$cpath}
		494	unless %$cookies;
		495	}
		496
		497	delete $jar->{$chost}
		498	unless %$paths;
		499	}
		500	}
		501
383	# extract cookies from jar	502	# extract cookies from jar
384	sub cookie_jar_extract($$$$) {	503	sub cookie_jar_extract($$$$) {
385	my ($jar, $uscheme, $uhost, $upath) = @_;	504	my ($jar, $scheme, $host, $path) = @_;
386		505
387	%$jar = () if $jar->{version} != 1;	506	%$jar = () if $jar->{version} != 2;
		507
		508	$host = AnyEvent::Util::idn_to_ascii $host
		509	if $host =~ /[^\x00-\x7f]/;
388		510
389	my @cookies;	511	my @cookies;
390		512
391	while (my ($chost, $paths) = each %$jar) {	513	while (my ($chost, $paths) = each %$jar) {
392	next unless ref $paths;	514	next unless ref $paths;
393		515
394	if ($chost =~ /^\./) {	516	# exact match or suffix including . match
395	next unless $chost eq substr $uhost, -length $chost;	517	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
396	} elsif ($chost =~ /\./) {
397	next unless $chost eq $uhost;
398	} else {
399	next;	518	or next;
400	}
401		519
402	while (my ($cpath, $cookies) = each %$paths) {	520	while (my ($cpath, $cookies) = each %$paths) {
403	next unless $cpath eq substr $upath, 0, length $cpath;	521	next unless $cpath eq substr $path, 0, length $cpath;
404		522
405	while (my ($cookie, $kv) = each %$cookies) {	523	while (my ($cookie, $kv) = each %$cookies) {
406	next if $uscheme ne "https" && exists $kv->{secure};	524	next if $scheme ne "https" && exists $kv->{secure};
407		525
408	if (exists $kv->{expires}) {	526	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
409	if (AE::now > parse_date ($kv->{expires})) {
410	delete $cookies->{$cookie};	527	delete $cookies->{$cookie};
411	next;	528	next;
412	}
413	}	529	}
414		530
415	my $value = $kv->{value};	531	my $value = $kv->{value};
416		532
417	if ($value =~ /[=;,[:space:]]/) {	533	if ($value =~ /[=;,[:space:]]/) {
…		…
426		542
427	\@cookies	543	\@cookies
428	}	544	}
429		545
430	# parse set_cookie header into jar	546	# parse set_cookie header into jar
431	sub cookie_jar_set_cookie($$$) {	547	sub cookie_jar_set_cookie($$$$) {
432	my ($jar, $set_cookie, $uhost) = @_;	548	my ($jar, $set_cookie, $host, $date) = @_;
		549
		550	%$jar = () if $jar->{version} != 2;
		551
		552	my $anow = int AE::now;
		553	my $snow; # server-now
433		554
434	for ($set_cookie) {	555	for ($set_cookie) {
435	# parse NAME=VALUE	556	# parse NAME=VALUE
436	my @kv;	557	my @kv;
437		558
…		…
440	while (	561	while (
441	m{	562	m{
442	\G\s*	563	\G\s*
443	(?:	564	(?:
444	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)	565	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
445	| ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )	566	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
446	)	567	)
447	}gcxsi	568	}gcxsi
448	) {	569	) {
449	my $name = $2;	570	my $name = $2;
450	my $value = $4;	571	my $value = $4;
451		572
452	unless (defined $name) {	573	if (defined $1) {
453	# expires	574	# expires
454	$name = "expires";	575	$name = "expires";
455	$value = $1;	576	$value = $1;
456	} elsif (!defined $value) {	577	} elsif (defined $3) {
457	# quoted	578	# quoted
458	$value = $3;	579	$value = $3;
459	$value =~ s/\\(.)/$1/gs;	580	$value =~ s/\\(.)/$1/gs;
460	}	581	}
461		582
462	push @kv, lc $name, $value;	583	push @kv, @kv ? lc $name : $name, $value;
463		584
464	last unless /\G\s*;/gc;	585	last unless /\G\s*;/gc;
465	}	586	}
466		587
467	last unless @kv;	588	last unless @kv;
468		589
469	my $name = shift @kv;	590	my $name = shift @kv;
470	my %kv = (value => shift @kv, @kv);	591	my %kv = (value => shift @kv, @kv);
471		592
472	$kv{expires} ||= format_date (AE::now + $kv{"max-age"})
473	if exists $kv{"max-age"};	593	if (exists $kv{"max-age"}) {
		594	$kv{_expires} = $anow + delete $kv{"max-age"};
		595	} elsif (exists $kv{expires}) {
		596	$snow ||= parse_date ($date) || $anow;
		597	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		598	} else {
		599	delete $kv{_expires};
		600	}
474		601
475	my $cdom;	602	my $cdom;
476	my $cpath = (delete $kv{path}) || "/";	603	my $cpath = (delete $kv{path}) || "/";
477		604
478	if (exists $kv{domain}) {	605	if (exists $kv{domain}) {
479	$cdom = delete $kv{domain};	606	$cdom = $kv{domain};
480		607
481	$cdom =~ s/^\.?/./; # make sure it starts with a "."	608	$cdom =~ s/^\.?/./; # make sure it starts with a "."
482		609
483	next if $cdom =~ /\.$/;	610	next if $cdom =~ /\.$/;
484		611
485	# this is not rfc-like and not netscape-like. go figure.	612	# this is not rfc-like and not netscape-like. go figure.
486	my $ndots = $cdom =~ y/.//;	613	my $ndots = $cdom =~ y/.//;
487	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);	614	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		615
		616	$cdom = substr $cdom, 1; # remove initial .
488	} else {	617	} else {
489	$cdom = $uhost;	618	$cdom = $host;
490	}	619	}
491		620
492	# store it	621	# store it
493	$jar->{version} = 1;	622	$jar->{version} = 2;
494	$jar->{$cdom}{$cpath}{$name} = \%kv;	623	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
495		624
496	redo if /\G\s*,/gc;	625	redo if /\G\s*,/gc;
497	}	626	}
498	}	627	}
499		628
		629	#############################################################################
		630	# keepalive/persistent connection cache
		631
		632	# fetch a connection from the keepalive cache
		633	sub ka_fetch($) {
		634	my $ka_key = shift;
		635
		636	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		637	delete $KA_CACHE{$ka_key}
		638	unless @{ $KA_CACHE{$ka_key} };
		639
		640	$hdl
		641	}
		642
		643	sub ka_store($$) {
		644	my ($ka_key, $hdl) = @_;
		645
		646	my $kaa = $KA_CACHE{$ka_key} ||= [];
		647
		648	my $destroy = sub {
		649	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		650
		651	$hdl->destroy;
		652
		653	@ka
		654	? $KA_CACHE{$ka_key} = \@ka
		655	: delete $KA_CACHE{$ka_key};
		656	};
		657
		658	# on error etc., destroy
		659	$hdl->on_error ($destroy);
		660	$hdl->on_eof ($destroy);
		661	$hdl->on_read ($destroy);
		662	$hdl->timeout ($PERSISTENT_TIMEOUT);
		663
		664	push @$kaa, $hdl;
		665	shift @$kaa while @$kaa > $MAX_PER_HOST;
		666	}
		667
		668	#############################################################################
		669	# utilities
		670
500	# continue to parse $_ for headers and place them into the arg	671	# continue to parse $_ for headers and place them into the arg
501	sub parse_hdr() {	672	sub _parse_hdr() {
502	my %hdr;	673	my %hdr;
503		674
504	# things seen, not parsed:	675	# things seen, not parsed:
505	# p3pP="NON CUR OTPi OUR NOR UNI"	676	# p3pP="NON CUR OTPi OUR NOR UNI"
506		677
…		…
520	for values %hdr;	691	for values %hdr;
521		692
522	\%hdr	693	\%hdr
523	}	694	}
524		695
		696	#############################################################################
		697	# http_get
		698
525	our $qr_nlnl = qr{(?<![^\012])\015?\012};	699	our $qr_nlnl = qr{(?<![^\012])\015?\012};
526		700
527	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	701	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
528	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };	702	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		703
		704	# maybe it should just become a normal object :/
		705
		706	sub _destroy_state(\%) {
		707	my ($state) = @_;
		708
		709	$state->{handle}->destroy if $state->{handle};
		710	%$state = ();
		711	}
		712
		713	sub _error(\%$$) {
		714	my ($state, $cb, $hdr) = @_;
		715
		716	&_destroy_state ($state);
		717
		718	$cb->(undef, $hdr);
		719	()
		720	}
		721
		722	our %IDEMPOTENT = (
		723	DELETE => 1,
		724	GET => 1,
		725	QUERY => 1,
		726	HEAD => 1,
		727	OPTIONS => 1,
		728	PUT => 1,
		729	TRACE => 1,
		730
		731	ACL => 1,
		732	"BASELINE-CONTROL" => 1,
		733	BIND => 1,
		734	CHECKIN => 1,
		735	CHECKOUT => 1,
		736	COPY => 1,
		737	LABEL => 1,
		738	LINK => 1,
		739	MERGE => 1,
		740	MKACTIVITY => 1,
		741	MKCALENDAR => 1,
		742	MKCOL => 1,
		743	MKREDIRECTREF => 1,
		744	MKWORKSPACE => 1,
		745	MOVE => 1,
		746	ORDERPATCH => 1,
		747	PRI => 1,
		748	PROPFIND => 1,
		749	PROPPATCH => 1,
		750	REBIND => 1,
		751	REPORT => 1,
		752	SEARCH => 1,
		753	UNBIND => 1,
		754	UNCHECKOUT => 1,
		755	UNLINK => 1,
		756	UNLOCK => 1,
		757	UPDATE => 1,
		758	UPDATEREDIRECTREF => 1,
		759	"VERSION-CONTROL" => 1,
		760	);
529		761
530	sub http_request($$@) {	762	sub http_request($$@) {
531	my $cb = pop;	763	my $cb = pop;
532	my ($method, $url, %arg) = @_;	764	my ($method, $url, %arg) = @_;
533		765
…		…
551	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	783	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
552		784
553	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	785	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
554	if $recurse < 0;	786	if $recurse < 0;
555		787
556	my $proxy = $arg{proxy} || $PROXY;	788	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
557	my $timeout = $arg{timeout} || $TIMEOUT;	789	my $timeout = $arg{timeout} || $TIMEOUT;
558		790
559	my ($uscheme, $uauthority, $upath, $query, $fragment) =	791	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
560	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	792	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
561		793
562	$uscheme = lc $uscheme;	794	$uscheme = lc $uscheme;
563		795
564	my $uport = $uscheme eq "http" ? 80	796	my $uport = $uscheme eq "http" ? 80
565	: $uscheme eq "https" ? 443	797	: $uscheme eq "https" ? 443
566	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });	798	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
567		799
568	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	800	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
569	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	801	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
570		802
571	my $uhost = $1;	803	my $uhost = lc $1;
572	$uport = $2 if defined $2;	804	$uport = $2 if defined $2;
573		805
574	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	806	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
575	unless exists $hdr{host};	807	unless exists $hdr{host};
576		808
…		…
595	$rscheme = "http" unless defined $rscheme;	827	$rscheme = "http" unless defined $rscheme;
596		828
597	# don't support https requests over https-proxy transport,	829	# don't support https requests over https-proxy transport,
598	# can't be done with tls as spec'ed, unless you double-encrypt.	830	# can't be done with tls as spec'ed, unless you double-encrypt.
599	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";	831	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		832
		833	$rhost = lc $rhost;
		834	$rscheme = lc $rscheme;
600	} else {	835	} else {
601	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	836	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
602	}	837	}
603		838
604	# leave out fragment and query string, just a heuristic	839	# leave out fragment and query string, just a heuristic
…		…
606	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	841	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
607		842
608	$hdr{"content-length"} = length $arg{body}	843	$hdr{"content-length"} = length $arg{body}
609	if length $arg{body} || $method ne "GET";	844	if length $arg{body} || $method ne "GET";
610		845
611	$hdr{connection} = "close TE"; #1.1	846	my $idempotent = $IDEMPOTENT{$method};
		847
		848	# default value for keepalive is true iff the request is for an idempotent method
		849	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		850	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		851	my $was_persistent; # true if this is actually a recycled connection
		852
		853	# the key to use in the keepalive cache
		854	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		855
		856	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
612	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	857	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
613		858
614	my %state = (connect_guard => 1);	859	my %state = (connect_guard => 1);
		860
		861	my $ae_error = 595; # connecting
		862
		863	# handle actual, non-tunneled, request
		864	my $handle_actual_request = sub {
		865	$ae_error = 596; # request phase
		866
		867	my $hdl = $state{handle};
		868
		869	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		870
		871	# send request
		872	$hdl->push_write (
		873	"$method $rpath HTTP/1.1\015\012"
		874	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		875	. "\015\012"
		876	. $arg{body}
		877	);
		878
		879	# return if error occurred during push_write()
		880	return unless %state;
		881
		882	# reduce memory usage, save a kitten, also re-use it for the response headers.
		883	%hdr = ();
		884
		885	# status line and headers
		886	$state{read_response} = sub {
		887	return unless %state;
		888
		889	for ("$_[1]") {
		890	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		891
		892	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		893	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		894
		895	# 100 Continue handling
		896	# should not happen as we don't send expect: 100-continue,
		897	# but we handle it just in case.
		898	# since we send the request body regardless, if we get an error
		899	# we are out of-sync, which we currently do NOT handle correctly.
		900	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		901	if $2 eq 100;
		902
		903	push @pseudo,
		904	HTTPVersion => $1,
		905	Status => $2,
		906	Reason => $3,
		907	;
		908
		909	my $hdr = _parse_hdr
		910	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		911
		912	%hdr = (%$hdr, @pseudo);
		913	}
		914
		915	# redirect handling
		916	# relative uri handling forced by microsoft and other shitheads.
		917	# we give our best and fall back to URI if available.
		918	if (exists $hdr{location}) {
		919	my $loc = $hdr{location};
		920
		921	if ($loc =~ m%^//%) { # //
		922	$loc = "$uscheme:$loc";
		923
		924	} elsif ($loc eq "") {
		925	$loc = $url;
		926
		927	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		928	$loc =~ s/^\.\/+//;
		929
		930	if ($loc !~ m%^[.?#]%) {
		931	my $prefix = "$uscheme://$uauthority";
		932
		933	unless ($loc =~ s/^\///) {
		934	$prefix .= $upath;
		935	$prefix =~ s/\/[^\/]*$//;
		936	}
		937
		938	$loc = "$prefix/$loc";
		939
		940	} elsif (eval { require URI }) { # uri
		941	$loc = URI->new_abs ($loc, $url)->as_string;
		942
		943	} else {
		944	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		945	#$hdr{Status} = 599;
		946	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		947	#$recurse = 0;
		948	}
		949	}
		950
		951	$hdr{location} = $loc;
		952	}
		953
		954	my $redirect;
		955
		956	if ($recurse) {
		957	my $status = $hdr{Status};
		958
		959	# industry standard is to redirect POST as GET for
		960	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		961	# also, the UA should ask the user for 301 and 307 and POST,
		962	# industry standard seems to be to simply follow.
		963	# we go with the industry standard. 308 is defined
		964	# by rfc7538
		965	if ($status == 301 or $status == 302 or $status == 303) {
		966	$redirect = 1;
		967	# HTTP/1.1 is unclear on how to mutate the method
		968	unless ($method eq "HEAD") {
		969	$method = "GET";
		970	delete $arg{body};
		971	}
		972	} elsif ($status == 307 or $status == 308) {
		973	$redirect = 1;
		974	}
		975	}
		976
		977	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		978	if ($state{handle}) {
		979	# handle keepalive
		980	if (
		981	$persistent
		982	&& $_[3]
		983	&& ($hdr{HTTPVersion} < 1.1
		984	? $hdr{connection} =~ /\bkeep-?alive\b/i
		985	: $hdr{connection} !~ /\bclose\b/i)
		986	) {
		987	ka_store $ka_key, delete $state{handle};
		988	} else {
		989	# no keepalive, destroy the handle
		990	$state{handle}->destroy;
		991	}
		992	}
		993
		994	%state = ();
		995
		996	if (defined $_[1]) {
		997	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		998	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		999	}
		1000
		1001	# set-cookie processing
		1002	if ($arg{cookie_jar}) {
		1003	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		1004	}
		1005
		1006	if ($redirect && exists $hdr{location}) {
		1007	# we ignore any errors, as it is very common to receive
		1008	# Content-Length != 0 but no actual body
		1009	# we also access %hdr, as $_[1] might be an erro
		1010	$state{recurse} =
		1011	http_request (
		1012	$method => $hdr{location},
		1013	%arg,
		1014	recurse => $recurse - 1,
		1015	Redirect => [$_[0], \%hdr],
		1016	sub {
		1017	%state = ();
		1018	&$cb
		1019	},
		1020	);
		1021	} else {
		1022	$cb->($_[0], \%hdr);
		1023	}
		1024	};
		1025
		1026	$ae_error = 597; # body phase
		1027
		1028	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1029
		1030	my $len = $chunked ? undef : $hdr{"content-length"};
		1031
		1032	# body handling, many different code paths
		1033	# - no body expected
		1034	# - want_body_handle
		1035	# - te chunked
		1036	# - 2x length known (with or without on_body)
		1037	# - 2x length not known (with or without on_body)
		1038	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1039	$finish->(undef, 598 => "Request cancelled by on_header");
		1040	} elsif (
		1041	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1042	or $method eq "HEAD"
		1043	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1044	) {
		1045	# no body
		1046	$finish->("", undef, undef, 1);
		1047
		1048	} elsif (!$redirect && $arg{want_body_handle}) {
		1049	$_[0]->on_eof (undef);
		1050	$_[0]->on_error (undef);
		1051	$_[0]->on_read (undef);
		1052
		1053	$finish->(delete $state{handle});
		1054
		1055	} elsif ($chunked) {
		1056	my $cl = 0;
		1057	my $body = "";
		1058	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1059
		1060	$state{read_chunk} = sub {
		1061	$_[1] =~ /^([0-9a-fA-F]+)/
		1062	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1063
		1064	my $len = hex $1;
		1065
		1066	if ($len) {
		1067	$cl += $len;
		1068
		1069	$_[0]->push_read (chunk => $len, sub {
		1070	$on_body->($_[1], \%hdr)
		1071	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1072
		1073	$_[0]->push_read (line => sub {
		1074	length $_[1]
		1075	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1076	$_[0]->push_read (line => $state{read_chunk});
		1077	});
		1078	});
		1079	} else {
		1080	$hdr{"content-length"} ||= $cl;
		1081
		1082	$_[0]->push_read (line => $qr_nlnl, sub {
		1083	if (length $_[1]) {
		1084	for ("$_[1]") {
		1085	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1086
		1087	my $hdr = _parse_hdr
		1088	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1089
		1090	%hdr = (%hdr, %$hdr);
		1091	}
		1092	}
		1093
		1094	$finish->($body, undef, undef, 1);
		1095	});
		1096	}
		1097	};
		1098
		1099	$_[0]->push_read (line => $state{read_chunk});
		1100
		1101	} elsif (!$redirect && $arg{on_body}) {
		1102	if (defined $len) {
		1103	$_[0]->on_read (sub {
		1104	$len -= length $_[0]{rbuf};
		1105
		1106	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1107	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1108
		1109	$len > 0
		1110	or $finish->("", undef, undef, 1);
		1111	});
		1112	} else {
		1113	$_[0]->on_eof (sub {
		1114	$finish->("");
		1115	});
		1116	$_[0]->on_read (sub {
		1117	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1118	or $finish->(undef, 598 => "Request cancelled by on_body");
		1119	});
		1120	}
		1121	} else {
		1122	$_[0]->on_eof (undef);
		1123
		1124	if (defined $len) {
		1125	$_[0]->on_read (sub {
		1126	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1127	if $len <= length $_[0]{rbuf};
		1128	});
		1129	} else {
		1130	$_[0]->on_error (sub {
		1131	($! == Errno::EPIPE || !$!)
		1132	? $finish->(delete $_[0]{rbuf})
		1133	: $finish->(undef, $ae_error => $_[2]);
		1134	});
		1135	$_[0]->on_read (sub { });
		1136	}
		1137	}
		1138	};
		1139
		1140	# if keepalive is enabled, then the server closing the connection
		1141	# before a response can happen legally - we retry on idempotent methods.
		1142	if ($was_persistent && $idempotent) {
		1143	my $old_eof = $hdl->{on_eof};
		1144	$hdl->{on_eof} = sub {
		1145	_destroy_state %state;
		1146
		1147	%state = ();
		1148	$state{recurse} =
		1149	http_request (
		1150	$method => $url,
		1151	%arg,
		1152	recurse => $recurse - 1,
		1153	persistent => 0,
		1154	sub {
		1155	%state = ();
		1156	&$cb
		1157	}
		1158	);
		1159	};
		1160	$hdl->on_read (sub {
		1161	return unless %state;
		1162
		1163	# as soon as we receive something, a connection close
		1164	# once more becomes a hard error
		1165	$hdl->{on_eof} = $old_eof;
		1166	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1167	});
		1168	} else {
		1169	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1170	}
		1171	};
		1172
		1173	my $prepare_handle = sub {
		1174	my ($hdl) = $state{handle};
		1175
		1176	$hdl->on_error (sub {
		1177	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1178	});
		1179	$hdl->on_eof (sub {
		1180	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1181	});
		1182	$hdl->timeout_reset;
		1183	$hdl->timeout ($timeout);
		1184	};
		1185
		1186	# connected to proxy (or origin server)
		1187	my $connect_cb = sub {
		1188	my $fh = shift
		1189	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1190
		1191	return unless delete $state{connect_guard};
		1192
		1193	# get handle
		1194	$state{handle} = new AnyEvent::Handle
		1195	%{ $arg{handle_params} },
		1196	fh => $fh,
		1197	peername => $uhost,
		1198	tls_ctx => $arg{tls_ctx},
		1199	;
		1200
		1201	$prepare_handle->();
		1202
		1203	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1204
		1205	# now handle proxy-CONNECT method
		1206	if ($proxy && $uscheme eq "https") {
		1207	# oh dear, we have to wrap it into a connect request
		1208
		1209	my $auth = exists $hdr{"proxy-authorization"}
		1210	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1211	: "";
		1212
		1213	# maybe re-use $uauthority with patched port?
		1214	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
		1215	$state{handle}->push_read (line => $qr_nlnl, sub {
		1216	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1217	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1218
		1219	if ($2 == 200) {
		1220	$rpath = $upath;
		1221	$handle_actual_request->();
		1222	} else {
		1223	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1224	}
		1225	});
		1226	} else {
		1227	delete $hdr{"proxy-authorization"} unless $proxy;
		1228
		1229	$handle_actual_request->();
		1230	}
		1231	};
615		1232
616	_get_slot $uhost, sub {	1233	_get_slot $uhost, sub {
617	$state{slot_guard} = shift;	1234	$state{slot_guard} = shift;
618		1235
619	return unless $state{connect_guard};	1236	return unless $state{connect_guard};
620		1237
621	my $ae_error = 595; # connecting	1238	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1239	# on a keepalive request (in theory, this should be a separate config option).
		1240	if ($persistent && $KA_CACHE{$ka_key}) {
		1241	$was_persistent = 1;
622		1242
623	my $connect_cb = sub {	1243	$state{handle} = ka_fetch $ka_key;
624	$state{fh} = shift	1244	# $state{handle}->destroyed
625	or do {	1245	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
626	my $err = "$!";	1246	$prepare_handle->();
627	%state = ();	1247	# $state{handle}->destroyed
628	return $cb->(undef, { @pseudo, Status => $ae_error, Reason => $err });	1248	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
629	};
630
631	return unless delete $state{connect_guard};
632
633	# get handle
634	$state{handle} = new AnyEvent::Handle
635	fh => $state{fh},
636	peername => $rhost,
637	tls_ctx => $arg{tls_ctx},
638	# these need to be reconfigured on keepalive handles
639	timeout => $timeout,
640	on_error => sub {
641	%state = ();
642	$cb->(undef, { @pseudo, Status => $ae_error, Reason => $_[2] });
643	},
644	on_eof => sub {
645	%state = ();
646	$cb->(undef, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" });
647	},
648	;
649
650	# limit the number of persistent connections
651	# keepalive not yet supported
652	# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
653	# ++$KA_COUNT{$_[1]};
654	# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
655	# --$KA_COUNT{$_[1]}
656	# };
657	# $hdr{connection} = "keep-alive";
658	# }
659
660	$state{handle}->starttls ("connect") if $rscheme eq "https";
661
662	# handle actual, non-tunneled, request
663	my $handle_actual_request = sub {
664	$ae_error = 596; # request phase
665
666	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};
667
668	# send request
669	$state{handle}->push_write (
670	"$method $rpath HTTP/1.1\015\012"
671	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
672	. "\015\012"
673	. (delete $arg{body})
674	);
675
676	# return if error occured during push_write()
677	return unless %state;
678
679	%hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
680
681	# status line and headers
682	$state{read_response} = sub {
683	for ("$_[1]") {
684	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
685
686	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
687	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid server response" }));
688
689	# 100 Continue handling
690	# should not happen as we don't send expect: 100-continue,
691	# but we handle it just in case.
692	# since we send the request body regardless, if we get an error
693	# we are out of-sync, which we currently do NOT handle correctly.
694	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
695	if $2 eq 100;
696
697	push @pseudo,
698	HTTPVersion => $1,
699	Status => $2,
700	Reason => $3,
701	;
702
703	my $hdr = parse_hdr
704	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Garbled response headers" }));
705
706	%hdr = (%$hdr, @pseudo);
707	}
708
709	# redirect handling
710	# microsoft and other shitheads don't give a shit for following standards,
711	# try to support some common forms of broken Location headers.
712	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
713	$hdr{location} =~ s/^\.\/+//;
714
715	my $url = "$rscheme://$uhost:$uport";
716
717	unless ($hdr{location} =~ s/^\///) {
718	$url .= $upath;
719	$url =~ s/\/[^\/]*$//;
720	}
721
722	$hdr{location} = "$url/$hdr{location}";
723	}
724
725	my $redirect;
726
727	if ($recurse) {
728	my $status = $hdr{Status};
729
730	# industry standard is to redirect POST as GET for
731	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
732	# also, the UA should ask the user for 301 and 307 and POST,
733	# industry standard seems to be to simply follow.
734	# we go with the industry standard.
735	if ($status == 301 or $status == 302 or $status == 303) {
736	# HTTP/1.1 is unclear on how to mutate the method
737	$method = "GET" unless $method eq "HEAD";
738	$redirect = 1;
739	} elsif ($status == 307) {
740	$redirect = 1;
741	}
742	}
743
744	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
745	my $may_keep_alive = $_[3];
746
747	$state{handle}->destroy if $state{handle};
748	%state = ();
749
750	if (defined $_[1]) {
751	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
752	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
753	}
754
755	# set-cookie processing
756	if ($arg{cookie_jar}) {
757	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost;
758	}
759
760	if ($redirect && exists $hdr{location}) {
761	# we ignore any errors, as it is very common to receive
762	# Content-Length != 0 but no actual body
763	# we also access %hdr, as $_[1] might be an erro
764	http_request (
765	$method => $hdr{location},
766	%arg,
767	recurse => $recurse - 1,
768	Redirect => [$_[0], \%hdr],
769	$cb);
770	} else {
771	$cb->($_[0], \%hdr);
772	}
773	};
774
775	$ae_error = 597; # body phase
776
777	my $len = $hdr{"content-length"};
778
779	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
780	$finish->(undef, 598 => "Request cancelled by on_header");
781	} elsif (
782	$hdr{Status} =~ /^(?:1..|204|205|304)$/
783	or $method eq "HEAD"
784	or (defined $len && !$len)
785	) {
786	# no body
787	$finish->("", undef, undef, 1);
788	} else {
789	# body handling, many different code paths
790	# - no body expected
791	# - want_body_handle
792	# - te chunked
793	# - 2x length known (with or without on_body)
794	# - 2x length not known (with or without on_body)
795	if (!$redirect && $arg{want_body_handle}) {
796	$_[0]->on_eof (undef);
797	$_[0]->on_error (undef);
798	$_[0]->on_read (undef);
799
800	$finish->(delete $state{handle});
801
802	} elsif ($hdr{"transfer-encoding"} =~ /\bchunked\b/i) {
803	my $cl = 0;
804	my $body = undef;
805	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
806
807	my $read_chunk; $read_chunk = sub {
808	$_[1] =~ /^([0-9a-fA-F]+)/
809	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
810
811	my $len = hex $1;
812
813	if ($len) {
814	$cl += $len;
815
816	$_[0]->push_read (chunk => $len, sub {
817	$on_body->($_[1], \%hdr)
818	or return $finish->(undef, 598 => "Request cancelled by on_body");
819
820	$_[0]->push_read (line => sub {
821	length $_[1]
822	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
823	$_[0]->push_read (line => $read_chunk);
824	});
825	});
826	} else {
827	$hdr{"content-length"} ||= $cl;
828
829	$_[0]->push_read (line => $qr_nlnl, sub {
830	if (length $_[1]) {
831	for ("$_[1]") {
832	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
833
834	my $hdr = parse_hdr
835	or return $finish->(undef, $ae_error => "Garbled response trailers");
836
837	%hdr = (%hdr, %$hdr);
838	}
839	}
840
841	$finish->($body, undef, undef, 1);
842	});
843	}
844	};
845
846	$_[0]->push_read (line => $read_chunk);
847
848	} elsif ($arg{on_body}) {
849	if ($len) {
850	$_[0]->on_read (sub {
851	$len -= length $_[0]{rbuf};
852
853	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
854	or return $finish->(undef, 598 => "Request cancelled by on_body");
855
856	$len > 0
857	or $finish->("", undef, undef, 1);
858	});
859	} else {
860	$_[0]->on_eof (sub {
861	$finish->("");
862	});
863	$_[0]->on_read (sub {
864	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
865	or $finish->(undef, 598 => "Request cancelled by on_body");
866	});
867	}
868	} else {
869	$_[0]->on_eof (undef);
870
871	if ($len) {
872	$_[0]->on_read (sub {
873	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
874	if $len <= length $_[0]{rbuf};
875	});
876	} else {
877	$_[0]->on_error (sub {
878	($! == Errno::EPIPE || !$!)
879	? $finish->(delete $_[0]{rbuf})
880	: $finish->(undef, $ae_error => $_[2]);
881	});
882	$_[0]->on_read (sub { });
883	}
884	}
885	}
886	};
887
888	$state{handle}->push_read (line => $qr_nlnl, $state{read_response});
889	};
890
891	# now handle proxy-CONNECT method
892	if ($proxy && $uscheme eq "https") {
893	# oh dear, we have to wrap it into a connect request
894
895	# maybe re-use $uauthority with patched port?
896	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
897	$state{handle}->push_read (line => $qr_nlnl, sub {
898	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
899	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" }));
900
901	if ($2 == 200) {
902	$rpath = $upath;	1249	$rpath = $upath;
903	&$handle_actual_request;	1250	$handle_actual_request->();
904	} else {	1251
905	%state = ();
906	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });
907	}
908	});
909	} else {	1252	} else {
910	&$handle_actual_request;
911	}
912	};
913
914	my $tcp_connect = $arg{tcp_connect}	1253	my $tcp_connect = $arg{tcp_connect}
915	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	1254	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
916		1255
917	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	1256	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
918		1257	}
919	};	1258	};
920		1259
921	defined wantarray && AnyEvent::Util::guard { %state = () }	1260	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
922	}	1261	}
923		1262
924	sub http_get($@) {	1263	sub http_get($@) {
925	unshift @_, "GET";	1264	unshift @_, "GET";
926	&http_request	1265	&http_request
…		…
944	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1283	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
945	the actual connection, which in turn uses AnyEvent::DNS to resolve	1284	the actual connection, which in turn uses AnyEvent::DNS to resolve
946	hostnames. The latter is a simple stub resolver and does no caching	1285	hostnames. The latter is a simple stub resolver and does no caching
947	on its own. If you want DNS caching, you currently have to provide	1286	on its own. If you want DNS caching, you currently have to provide
948	your own default resolver (by storing a suitable resolver object in	1287	your own default resolver (by storing a suitable resolver object in
949	C<$AnyEvent::DNS::RESOLVER>).	1288	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
950		1289
951	=head2 GLOBAL FUNCTIONS AND VARIABLES	1290	=head2 GLOBAL FUNCTIONS AND VARIABLES
952		1291
953	=over 4	1292	=over 4
954		1293
955	=item AnyEvent::HTTP::set_proxy "proxy-url"	1294	=item AnyEvent::HTTP::set_proxy "proxy-url"
956		1295
957	Sets the default proxy server to use. The proxy-url must begin with a	1296	Sets the default proxy server to use. The proxy-url must begin with a
958	string of the form C<http://host:port> (optionally C<https:...>), croaks	1297	string of the form C<http://host:port>, croaks otherwise.
959	otherwise.
960		1298
961	To clear an already-set proxy, use C<undef>.	1299	To clear an already-set proxy, use C<undef>.
		1300
		1301	When AnyEvent::HTTP is loaded for the first time it will query the
		1302	default proxy from the operating system, currently by looking at
		1303	C<$ENV{http_proxy>}.
		1304
		1305	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1306
		1307	Remove all cookies from the cookie jar that have been expired. If
		1308	C<$session_end> is given and true, then additionally remove all session
		1309	cookies.
		1310
		1311	You should call this function (with a true C<$session_end>) before you
		1312	save cookies to disk, and you should call this function after loading them
		1313	again. If you have a long-running program you can additionally call this
		1314	function from time to time.
		1315
		1316	A cookie jar is initially an empty hash-reference that is managed by this
		1317	module. Its format is subject to change, but currently it is as follows:
		1318
		1319	The key C<version> has to contain C<2>, otherwise the hash gets
		1320	cleared. All other keys are hostnames or IP addresses pointing to
		1321	hash-references. The key for these inner hash references is the
		1322	server path for which this cookie is meant, and the values are again
		1323	hash-references. Each key of those hash-references is a cookie name, and
		1324	the value, you guessed it, is another hash-reference, this time with the
		1325	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1326	which have been replaced by a C<_expires> key that contains the cookie
		1327	expiry timestamp. Session cookies are indicated by not having an
		1328	C<_expires> key.
		1329
		1330	Here is an example of a cookie jar with a single cookie, so you have a
		1331	chance of understanding the above paragraph:
		1332
		1333	{
		1334	version => 2,
		1335	"10.0.0.1" => {
		1336	"/" => {
		1337	"mythweb_id" => {
		1338	_expires => 1293917923,
		1339	value => "ooRung9dThee3ooyXooM1Ohm",
		1340	},
		1341	},
		1342	},
		1343	}
962		1344
963	=item $date = AnyEvent::HTTP::format_date $timestamp	1345	=item $date = AnyEvent::HTTP::format_date $timestamp
964		1346
965	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP	1347	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
966	Date (RFC 2616).	1348	Date (RFC 2616).
…		…
973		1355
974	=item $AnyEvent::HTTP::MAX_RECURSE	1356	=item $AnyEvent::HTTP::MAX_RECURSE
975		1357
976	The default value for the C<recurse> request parameter (default: C<10>).	1358	The default value for the C<recurse> request parameter (default: C<10>).
977		1359
		1360	=item $AnyEvent::HTTP::TIMEOUT
		1361
		1362	The default timeout for connection operations (default: C<300>).
		1363
978	=item $AnyEvent::HTTP::USERAGENT	1364	=item $AnyEvent::HTTP::USERAGENT
979		1365
980	The default value for the C<User-Agent> header (the default is	1366	The default value for the C<User-Agent> header (the default is
981	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1367	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
982		1368
983	=item $AnyEvent::HTTP::MAX_PER_HOST	1369	=item $AnyEvent::HTTP::MAX_PER_HOST
984		1370
985	The maximum number of concurrent connections to the same host (identified	1371	The maximum number of concurrent connections to the same host (identified
986	by the hostname). If the limit is exceeded, then the additional requests	1372	by the hostname). If the limit is exceeded, then additional requests
987	are queued until previous connections are closed.	1373	are queued until previous connections are closed. Both persistent and
		1374	non-persistent connections are counted in this limit.
988		1375
989	The default value for this is C<4>, and it is highly advisable to not	1376	The default value for this is C<4>, and it is highly advisable to not
990	increase it.	1377	increase it much.
		1378
		1379	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1380	connections, older browsers used 2, newer ones (such as firefox 3)
		1381	typically use 6, and Opera uses 8 because like, they have the fastest
		1382	browser and give a shit for everybody else on the planet.
		1383
		1384	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1385
		1386	The time after which idle persistent connections get closed by
		1387	AnyEvent::HTTP (default: C<3>).
991		1388
992	=item $AnyEvent::HTTP::ACTIVE	1389	=item $AnyEvent::HTTP::ACTIVE
993		1390
994	The number of active connections. This is not the number of currently	1391	The number of active connections. This is not the number of currently
995	running requests, but the number of currently open and non-idle TCP	1392	running requests, but the number of currently open and non-idle TCP
996	connections. This number of can be useful for load-leveling.	1393	connections. This number can be useful for load-leveling.
997		1394
998	=back	1395	=back
999		1396
1000	=cut	1397	=cut
1001		1398
…		…
1035	# other formats fail in the loop below	1432	# other formats fail in the loop below
1036		1433
1037	for (0..11) {	1434	for (0..11) {
1038	if ($m eq $month[$_]) {	1435	if ($m eq $month[$_]) {
1039	require Time::Local;	1436	require Time::Local;
1040	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);	1437	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
1041	}	1438	}
1042	}	1439	}
1043		1440
1044	undef	1441	undef
1045	}	1442	}
1046		1443
1047	sub set_proxy($) {	1444	sub set_proxy($) {
1048	if (length $_[0]) {	1445	if (length $_[0]) {
1049	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1446	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1050	or Carp::croak "$_[0]: invalid proxy URL";	1447	or Carp::croak "$_[0]: invalid proxy URL";
1051	$PROXY = [$2, $3 || 3128, $1]	1448	$PROXY = [$2, $3 || 3128, $1]
1052	} else {	1449	} else {
1053	undef $PROXY;	1450	undef $PROXY;
1054	}	1451	}
…		…
1057	# initialise proxy from environment	1454	# initialise proxy from environment
1058	eval {	1455	eval {
1059	set_proxy $ENV{http_proxy};	1456	set_proxy $ENV{http_proxy};
1060	};	1457	};
1061		1458
		1459	=head2 SHOWCASE
		1460
		1461	This section contains some more elaborate "real-world" examples or code
		1462	snippets.
		1463
		1464	=head2 HTTP/1.1 FILE DOWNLOAD
		1465
		1466	Downloading files with HTTP can be quite tricky, especially when something
		1467	goes wrong and you want to resume.
		1468
		1469	Here is a function that initiates and resumes a download. It uses the
		1470	last modified time to check for file content changes, and works with many
		1471	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1472	on older servers.
		1473
		1474	It calls the completion callback with either C<undef>, which means a
		1475	nonretryable error occurred, C<0> when the download was partial and should
		1476	be retried, and C<1> if it was successful.
		1477
		1478	use AnyEvent::HTTP;
		1479
		1480	sub download($$$) {
		1481	my ($url, $file, $cb) = @_;
		1482
		1483	open my $fh, "+<", $file
		1484	or die "$file: $!";
		1485
		1486	my %hdr;
		1487	my $ofs = 0;
		1488
		1489	if (stat $fh and -s _) {
		1490	$ofs = -s _;
		1491	warn "-s is ", $ofs;
		1492	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1493	$hdr{"range"} = "bytes=$ofs-";
		1494	}
		1495
		1496	http_get $url,
		1497	headers => \%hdr,
		1498	on_header => sub {
		1499	my ($hdr) = @_;
		1500
		1501	if ($hdr->{Status} == 200 && $ofs) {
		1502	# resume failed
		1503	truncate $fh, $ofs = 0;
		1504	}
		1505
		1506	sysseek $fh, $ofs, 0;
		1507
		1508	1
		1509	},
		1510	on_body => sub {
		1511	my ($data, $hdr) = @_;
		1512
		1513	if ($hdr->{Status} =~ /^2/) {
		1514	length $data == syswrite $fh, $data
		1515	or return; # abort on write errors
		1516	}
		1517
		1518	1
		1519	},
		1520	sub {
		1521	my (undef, $hdr) = @_;
		1522
		1523	my $status = $hdr->{Status};
		1524
		1525	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1526	utime $time, $time, $fh;
		1527	}
		1528
		1529	if ($status == 200 || $status == 206 || $status == 416) {
		1530	# download ok || resume ok || file already fully downloaded
		1531	$cb->(1, $hdr);
		1532
		1533	} elsif ($status == 412) {
		1534	# file has changed while resuming, delete and retry
		1535	unlink $file;
		1536	$cb->(0, $hdr);
		1537
		1538	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1539	# retry later
		1540	$cb->(0, $hdr);
		1541
		1542	} else {
		1543	$cb->(undef, $hdr);
		1544	}
		1545	}
		1546	;
		1547	}
		1548
		1549	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1550	if ($_[0]) {
		1551	print "OK!\n";
		1552	} elsif (defined $_[0]) {
		1553	print "please retry later\n";
		1554	} else {
		1555	print "ERROR\n";
		1556	}
		1557	};
		1558
1062	=head2 SOCKS PROXIES	1559	=head3 SOCKS PROXIES
1063		1560
1064	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1561	Socks proxies are not directly supported by AnyEvent::HTTP. You can
1065	compile your perl to support socks, or use an external program such as	1562	compile your perl to support socks, or use an external program such as
1066	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy	1563	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1067	transparently.	1564	transparently.

Diff Legend

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