ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.82 by root, Sun Jan 2 04:50:40 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).
…		…
225	verification) TLS context.	246	verification) TLS context.
226		247
227	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
228	me the page, no matter what".	249	me the page, no matter what".
229		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
230	=item on_prepare => $callback->($fh)	263	=item on_prepare => $callback->($fh)
231		264
232	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
233	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
234	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>	267	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
235	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
236	timeout). See the description for the C<$prepare_cb> argument of	269	timeout). See the description for the C<$prepare_cb> argument of
237	C<AnyEvent::Socket::tcp_connect> for details.	270	C<AnyEvent::Socket::tcp_connect> for details.
238		271
…		…
241	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
242	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>	275	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
243	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 -
244	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
245	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<$tcp_connect>
		283	function is incompatible with this kind of re-use, consider switching off
		284	C<persistent> connections and/or providing a C<sessionid> identifier.
246		285
247	There are probably lots of weird uses for this function, starting from	286	There are probably lots of weird uses for this function, starting from
248	tracing the hosts C<http_request> actually tries to connect, to (inexact	287	tracing the hosts C<http_request> actually tries to connect, to (inexact
249	but fast) host => IP address caching or even socks protocol support.	288	but fast) host => IP address caching or even socks protocol support.
250		289
…		…
303	called. Instead of the C<$body> argument containing the body data, the	342	called. Instead of the C<$body> argument containing the body data, the
304	callback will receive the L<AnyEvent::Handle> object associated with the	343	callback will receive the L<AnyEvent::Handle> object associated with the
305	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
306	(e.g. status C<304>), the empty string will be passed.	345	(e.g. status C<304>), the empty string will be passed.
307		346
308	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
309	a proxy, be a persistent connection etc., and configured in unspecified	348	to a proxy, be a persistent connection, use chunked transfer encoding
310	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
311	module anymore).	350	handle (it will not be used by this module anymore).
312		351
313	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
314	headers, an interactive protocol is used (typical example would be the	353	headers, an interactive protocol is used (typical example would be the
315	push-style twitter API which starts a JSON/XML stream).	354	push-style twitter API which starts a JSON/XML stream).
316		355
317	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
318	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 => $boolean
		378
		379	Only used when C<persistent> is also true. This parameter decides whether
		380	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		381	(as opposed to only a HTTP/1.1 persistent connection).
		382
		383	The default is true, except when using a proxy, in which case it defaults
		384	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		385
		386	=item handle_params => { key => value ... }
		387
		388	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		389	constructor that is called - not all requests will create a handle, and
		390	sometimes more than one is created, so this parameter is only good for
		391	setting hints.
		392
		393	Example: set the maximum read size to 4096, to potentially conserve memory
		394	at the cost of speed.
		395
		396	handle_params => {
		397	max_read_size => 4096,
		398	},
319		399
320	=back	400	=back
321		401
322	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	402	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
323	the response body.	403	the response body.
…		…
329		409
330	Example: do a HTTP HEAD request on https://www.google.com/, use a	410	Example: do a HTTP HEAD request on https://www.google.com/, use a
331	timeout of 30 seconds.	411	timeout of 30 seconds.
332		412
333	http_request	413	http_request
334	GET => "https://www.google.com",	414	HEAD => "https://www.google.com",
		415	headers => { "user-agent" => "MySearchClient 1.0" },
335	timeout => 30,	416	timeout => 30,
336	sub {	417	sub {
337	my ($body, $hdr) = @_;	418	my ($body, $hdr) = @_;
338	use Data::Dumper;	419	use Data::Dumper;
339	print Dumper $hdr;	420	print Dumper $hdr;
…		…
349	};	430	};
350		431
351	undef $request;	432	undef $request;
352		433
353	=cut	434	=cut
		435
		436	#############################################################################
		437	# wait queue/slots
354		438
355	sub _slot_schedule;	439	sub _slot_schedule;
356	sub _slot_schedule($) {	440	sub _slot_schedule($) {
357	my $host = shift;	441	my $host = shift;
358		442
…		…
381		465
382	_slot_schedule $_[0];	466	_slot_schedule $_[0];
383	}	467	}
384		468
385	#############################################################################	469	#############################################################################
		470	# cookie handling
386		471
387	# expire cookies	472	# expire cookies
388	sub cookie_jar_expire($;$) {	473	sub cookie_jar_expire($;$) {
389	my ($jar, $session_end) = @_;	474	my ($jar, $session_end) = @_;
390		475
391	%$jar = () if $jar->{version} != 1;	476	%$jar = () if $jar->{version} != 2;
392		477
393	my $anow = AE::now;	478	my $anow = AE::now;
394		479
395	while (my ($chost, $paths) = each %$jar) {	480	while (my ($chost, $paths) = each %$jar) {
396	next unless ref $paths;	481	next unless ref $paths;
…		…
414	}	499	}
415	}	500	}
416		501
417	# extract cookies from jar	502	# extract cookies from jar
418	sub cookie_jar_extract($$$$) {	503	sub cookie_jar_extract($$$$) {
419	my ($jar, $uscheme, $uhost, $upath) = @_;	504	my ($jar, $scheme, $host, $path) = @_;
420		505
421	%$jar = () if $jar->{version} != 1;	506	%$jar = () if $jar->{version} != 2;
		507
		508	$host = AnyEvent::Util::idn_to_ascii $host
		509	if $host =~ /[^\x00-\x7f]/;
422		510
423	my @cookies;	511	my @cookies;
424		512
425	while (my ($chost, $paths) = each %$jar) {	513	while (my ($chost, $paths) = each %$jar) {
426	next unless ref $paths;	514	next unless ref $paths;
427		515
428	if ($chost =~ /^\./) {	516	# exact match or suffix including . match
429	next unless $chost eq substr $uhost, -length $chost;	517	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
430	} elsif ($chost =~ /\./) {
431	next unless $chost eq $uhost;
432	} else {
433	next;	518	or next;
434	}
435		519
436	while (my ($cpath, $cookies) = each %$paths) {	520	while (my ($cpath, $cookies) = each %$paths) {
437	next unless $cpath eq substr $upath, 0, length $cpath;	521	next unless $cpath eq substr $path, 0, length $cpath;
438		522
439	while (my ($cookie, $kv) = each %$cookies) {	523	while (my ($cookie, $kv) = each %$cookies) {
440	next if $uscheme ne "https" && exists $kv->{secure};	524	next if $scheme ne "https" && exists $kv->{secure};
441		525
442	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {	526	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
443	delete $cookies->{$cookie};	527	delete $cookies->{$cookie};
444	next;	528	next;
445	}	529	}
…		…
459	\@cookies	543	\@cookies
460	}	544	}
461		545
462	# parse set_cookie header into jar	546	# parse set_cookie header into jar
463	sub cookie_jar_set_cookie($$$$) {	547	sub cookie_jar_set_cookie($$$$) {
464	my ($jar, $set_cookie, $uhost, $date) = @_;	548	my ($jar, $set_cookie, $host, $date) = @_;
		549
		550	%$jar = () if $jar->{version} != 2;
465		551
466	my $anow = int AE::now;	552	my $anow = int AE::now;
467	my $snow; # server-now	553	my $snow; # server-now
468		554
469	for ($set_cookie) {	555	for ($set_cookie) {
…		…
475	while (	561	while (
476	m{	562	m{
477	\G\s*	563	\G\s*
478	(?:	564	(?:
479	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)	565	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
480	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) ) )?	566	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
481	)	567	)
482	}gcxsi	568	}gcxsi
483	) {	569	) {
484	my $name = $2;	570	my $name = $2;
485	my $value = $4;	571	my $value = $4;
…		…
492	# quoted	578	# quoted
493	$value = $3;	579	$value = $3;
494	$value =~ s/\\(.)/$1/gs;	580	$value =~ s/\\(.)/$1/gs;
495	}	581	}
496		582
497	push @kv, lc $name, $value;	583	push @kv, @kv ? lc $name : $name, $value;
498		584
499	last unless /\G\s*;/gc;	585	last unless /\G\s*;/gc;
500	}	586	}
501		587
502	last unless @kv;	588	last unless @kv;
…		…
515		601
516	my $cdom;	602	my $cdom;
517	my $cpath = (delete $kv{path}) || "/";	603	my $cpath = (delete $kv{path}) || "/";
518		604
519	if (exists $kv{domain}) {	605	if (exists $kv{domain}) {
520	$cdom = delete $kv{domain};	606	$cdom = $kv{domain};
521		607
522	$cdom =~ s/^\.?/./; # make sure it starts with a "."	608	$cdom =~ s/^\.?/./; # make sure it starts with a "."
523		609
524	next if $cdom =~ /\.$/;	610	next if $cdom =~ /\.$/;
525		611
526	# this is not rfc-like and not netscape-like. go figure.	612	# this is not rfc-like and not netscape-like. go figure.
527	my $ndots = $cdom =~ y/.//;	613	my $ndots = $cdom =~ y/.//;
528	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);	614	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		615
		616	$cdom = substr $cdom, 1; # remove initial .
529	} else {	617	} else {
530	$cdom = $uhost;	618	$cdom = $host;
531	}	619	}
532		620
533	# store it	621	# store it
534	$jar->{version} = 1;	622	$jar->{version} = 2;
535	$jar->{$cdom}{$cpath}{$name} = \%kv;	623	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
536		624
537	redo if /\G\s*,/gc;	625	redo if /\G\s*,/gc;
538	}	626	}
539	}	627	}
540		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
541	# continue to parse $_ for headers and place them into the arg	671	# continue to parse $_ for headers and place them into the arg
542	sub parse_hdr() {	672	sub _parse_hdr() {
543	my %hdr;	673	my %hdr;
544		674
545	# things seen, not parsed:	675	# things seen, not parsed:
546	# p3pP="NON CUR OTPi OUR NOR UNI"	676	# p3pP="NON CUR OTPi OUR NOR UNI"
547		677
…		…
561	for values %hdr;	691	for values %hdr;
562		692
563	\%hdr	693	\%hdr
564	}	694	}
565		695
		696	#############################################################################
		697	# http_get
		698
566	our $qr_nlnl = qr{(?<![^\012])\015?\012};	699	our $qr_nlnl = qr{(?<![^\012])\015?\012};
567		700
568	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	701	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
569	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(\%$$) {
		714	my ($state, $cb, $hdr) = @_;
		715
		716	&_destroy_state ($state);
		717
		718	$cb->(undef, $hdr);
		719	()
		720	}
		721
		722	our %IDEMPOTENT = (
		723	DELETE => 1,
		724	GET => 1,
		725	QUERY => 1,
		726	HEAD => 1,
		727	OPTIONS => 1,
		728	PUT => 1,
		729	TRACE => 1,
		730
		731	ACL => 1,
		732	"BASELINE-CONTROL" => 1,
		733	BIND => 1,
		734	CHECKIN => 1,
		735	CHECKOUT => 1,
		736	COPY => 1,
		737	LABEL => 1,
		738	LINK => 1,
		739	MERGE => 1,
		740	MKACTIVITY => 1,
		741	MKCALENDAR => 1,
		742	MKCOL => 1,
		743	MKREDIRECTREF => 1,
		744	MKWORKSPACE => 1,
		745	MOVE => 1,
		746	ORDERPATCH => 1,
		747	PRI => 1,
		748	PROPFIND => 1,
		749	PROPPATCH => 1,
		750	REBIND => 1,
		751	REPORT => 1,
		752	SEARCH => 1,
		753	UNBIND => 1,
		754	UNCHECKOUT => 1,
		755	UNLINK => 1,
		756	UNLOCK => 1,
		757	UPDATE => 1,
		758	UPDATEREDIRECTREF => 1,
		759	"VERSION-CONTROL" => 1,
		760	);
570		761
571	sub http_request($$@) {	762	sub http_request($$@) {
572	my $cb = pop;	763	my $cb = pop;
573	my ($method, $url, %arg) = @_;	764	my ($method, $url, %arg) = @_;
574		765
…		…
592	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	783	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
593		784
594	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	785	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
595	if $recurse < 0;	786	if $recurse < 0;
596		787
597	my $proxy = $arg{proxy} || $PROXY;	788	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
598	my $timeout = $arg{timeout} || $TIMEOUT;	789	my $timeout = $arg{timeout} || $TIMEOUT;
599		790
600	my ($uscheme, $uauthority, $upath, $query, $fragment) =	791	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
601	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	792	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
602		793
603	$uscheme = lc $uscheme;	794	$uscheme = lc $uscheme;
604		795
605	my $uport = $uscheme eq "http" ? 80	796	my $uport = $uscheme eq "http" ? 80
606	: $uscheme eq "https" ? 443	797	: $uscheme eq "https" ? 443
607	: 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" });
608		799
609	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	800	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
610	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	801	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
611		802
612	my $uhost = $1;	803	my $uhost = lc $1;
613	$uport = $2 if defined $2;	804	$uport = $2 if defined $2;
614		805
615	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	806	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
616	unless exists $hdr{host};	807	unless exists $hdr{host};
617		808
…		…
636	$rscheme = "http" unless defined $rscheme;	827	$rscheme = "http" unless defined $rscheme;
637		828
638	# don't support https requests over https-proxy transport,	829	# don't support https requests over https-proxy transport,
639	# 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.
640	$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;
641	} else {	835	} else {
642	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	836	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
643	}	837	}
644		838
645	# leave out fragment and query string, just a heuristic	839	# leave out fragment and query string, just a heuristic
…		…
647	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	841	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
648		842
649	$hdr{"content-length"} = length $arg{body}	843	$hdr{"content-length"} = length $arg{body}
650	if length $arg{body} || $method ne "GET";	844	if length $arg{body} || $method ne "GET";
651		845
652	$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
653	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	857	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
654		858
655	my %state = (connect_guard => 1);	859	my %state = (connect_guard => 1);
656		860
657	_get_slot $uhost, sub {
658	$state{slot_guard} = shift;
659
660	return unless $state{connect_guard};
661
662	my $ae_error = 595; # connecting	861	my $ae_error = 595; # connecting
663		862
664	# handle actual, non-tunneled, request	863	# handle actual, non-tunneled, request
665	my $handle_actual_request = sub {	864	my $handle_actual_request = sub {
666	$ae_error = 596; # request phase	865	$ae_error = 596; # request phase
667		866
		867	my $hdl = $state{handle};
		868
668	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};	869	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
669		870
670	# send request	871	# send request
671	$state{handle}->push_write (	872	$hdl->push_write (
672	"$method $rpath HTTP/1.1\015\012"	873	"$method $rpath HTTP/1.1\015\012"
673	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)	874	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
674	. "\015\012"	875	. "\015\012"
675	. (delete $arg{body})	876	. $arg{body}
676	);	877	);
677		878
678	# return if error occured during push_write()	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 {
679	return unless %state;	887	return unless %state;
680		888
681	%hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
682
683	# status line and headers
684	$state{read_response} = sub {
685	for ("$_[1]") {	889	for ("$_[1]") {
686	y/\015//d; # weed out any \015, as they show up in the weirdest of places.	890	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
687		891
688	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci	892	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
689	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid server response" }));	893	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
690		894
691	# 100 Continue handling	895	# 100 Continue handling
692	# should not happen as we don't send expect: 100-continue,	896	# should not happen as we don't send expect: 100-continue,
693	# but we handle it just in case.	897	# but we handle it just in case.
694	# since we send the request body regardless, if we get an error	898	# since we send the request body regardless, if we get an error
695	# we are out of-sync, which we currently do NOT handle correctly.	899	# we are out of-sync, which we currently do NOT handle correctly.
696	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})	900	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
697	if $2 eq 100;	901	if $2 eq 100;
698		902
699	push @pseudo,	903	push @pseudo,
700	HTTPVersion => $1,	904	HTTPVersion => $1,
701	Status => $2,	905	Status => $2,
702	Reason => $3,	906	Reason => $3,
703	;	907	;
704		908
705	my $hdr = parse_hdr	909	my $hdr = _parse_hdr
706	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Garbled response headers" }));	910	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
707		911
708	%hdr = (%$hdr, @pseudo);	912	%hdr = (%$hdr, @pseudo);
709	}	913	}
710		914
711	# redirect handling	915	# redirect handling
712	# microsoft and other shitheads don't give a shit for following standards,	916	# relative uri handling forced by microsoft and other shitheads.
713	# try to support some common forms of broken Location headers.	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
714	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {	927	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
715	$hdr{location} =~ s/^\.\/+//;	928	$loc =~ s/^\.\/+//;
716		929
717	my $url = "$rscheme://$uhost:$uport";	930	if ($loc !~ m%^[.?#]%) {
		931	my $prefix = "$uscheme://$uauthority";
718		932
719	unless ($hdr{location} =~ s/^\///) {	933	unless ($loc =~ s/^\///) {
720	$url .= $upath;	934	$prefix .= $upath;
721	$url =~ s/\/[^\/]*$//;	935	$prefix =~ s/\/[^\/]*$//;
722	}	936	}
723		937
724	$hdr{location} = "$url/$hdr{location}";	938	$loc = "$prefix/$loc";
725	}
726		939
727	my $redirect;	940	} elsif (eval { require URI }) { # uri
		941	$loc = URI->new_abs ($loc, $url)->as_string;
728		942
729	if ($recurse) {	943	} else {
730	my $status = $hdr{Status};	944	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
731		945	#$hdr{Status} = 599;
732	# industry standard is to redirect POST as GET for	946	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
733	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
734	# also, the UA should ask the user for 301 and 307 and POST,
735	# industry standard seems to be to simply follow.
736	# we go with the industry standard.
737	if ($status == 301 or $status == 302 or $status == 303) {
738	# HTTP/1.1 is unclear on how to mutate the method
739	$method = "GET" unless $method eq "HEAD";
740	$redirect = 1;	947	#$recurse = 0;
741	} elsif ($status == 307) {
742	$redirect = 1;
743	}	948	}
744	}	949	}
745		950
746	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])	951	$hdr{location} = $loc;
747	my $may_keep_alive = $_[3];	952	}
748		953
749	$state{handle}->destroy if $state{handle};	954	my $redirect;
750	%state = ();
751		955
752	if (defined $_[1]) {	956	if ($recurse) {
753	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];	957	my $status = $hdr{Status};
754	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];	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};
755	}	971	}
		972	} elsif ($status == 307 or $status == 308) {
		973	$redirect = 1;
		974	}
		975	}
756		976
757	# set-cookie processing	977	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
758	if ($arg{cookie_jar}) {	978	if ($state{handle}) {
759	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};	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;
760	}	991	}
		992	}
761		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
762	if ($redirect && exists $hdr{location}) {	1006	if ($redirect && exists $hdr{location}) {
763	# we ignore any errors, as it is very common to receive	1007	# we ignore any errors, as it is very common to receive
764	# Content-Length != 0 but no actual body	1008	# Content-Length != 0 but no actual body
765	# we also access %hdr, as $_[1] might be an erro	1009	# we also access %hdr, as $_[1] might be an erro
		1010	$state{recurse} =
766	http_request (	1011	http_request (
767	$method => $hdr{location},	1012	$method => $hdr{location},
768	%arg,	1013	%arg,
769	recurse => $recurse - 1,	1014	recurse => $recurse - 1,
770	Redirect => [$_[0], \%hdr],	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});
771	$cb);	1077	});
		1078	});
772	} else {	1079	} else {
773	$cb->($_[0], \%hdr);	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	});
774	}	1096	}
775	};	1097	};
776		1098
777	$ae_error = 597; # body phase	1099	$_[0]->push_read (line => $state{read_chunk});
778		1100
779	my $len = $hdr{"content-length"};	1101	} elsif (!$redirect && $arg{on_body}) {
		1102	if (defined $len) {
		1103	$_[0]->on_read (sub {
		1104	$len -= length $_[0]{rbuf};
780		1105
781	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {	1106	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
782	$finish->(undef, 598 => "Request cancelled by on_header");	1107	or return $finish->(undef, 598 => "Request cancelled by on_body");
783	} elsif (	1108
784	$hdr{Status} =~ /^(?:1..|204|205|304)$/	1109	$len > 0
785	or $method eq "HEAD"
786	or (defined $len && !$len)
787	) {
788	# no body
789	$finish->("", undef, undef, 1);	1110	or $finish->("", undef, undef, 1);
		1111	});
790	} else {	1112	} else {
791	# body handling, many different code paths
792	# - no body expected
793	# - want_body_handle
794	# - te chunked
795	# - 2x length known (with or without on_body)
796	# - 2x length not known (with or without on_body)
797	if (!$redirect && $arg{want_body_handle}) {
798	$_[0]->on_eof (undef);	1113	$_[0]->on_eof (sub {
799	$_[0]->on_error (undef);	1114	$finish->("");
800	$_[0]->on_read (undef);
801
802	$finish->(delete $state{handle});
803
804	} elsif ($hdr{"transfer-encoding"} =~ /\bchunked\b/i) {
805	my $cl = 0;
806	my $body = undef;
807	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
808
809	$state{read_chunk} = sub {
810	$_[1] =~ /^([0-9a-fA-F]+)/
811	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
812
813	my $len = hex $1;
814
815	if ($len) {
816	$cl += $len;
817
818	$_[0]->push_read (chunk => $len, sub {
819	$on_body->($_[1], \%hdr)
820	or return $finish->(undef, 598 => "Request cancelled by on_body");
821
822	$_[0]->push_read (line => sub {
823	length $_[1]
824	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
825	$_[0]->push_read (line => $state{read_chunk});
826	});
827	});
828	} else {
829	$hdr{"content-length"} ||= $cl;
830
831	$_[0]->push_read (line => $qr_nlnl, sub {
832	if (length $_[1]) {
833	for ("$_[1]") {
834	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
835
836	my $hdr = parse_hdr
837	or return $finish->(undef, $ae_error => "Garbled response trailers");
838
839	%hdr = (%hdr, %$hdr);
840	}
841	}
842
843	$finish->($body, undef, undef, 1);
844	});
845	}
846	};	1115	});
847
848	$_[0]->push_read (line => $state{read_chunk});
849
850	} elsif ($arg{on_body}) {
851	if ($len) {
852	$_[0]->on_read (sub {	1116	$_[0]->on_read (sub {
853	$len -= length $_[0]{rbuf};
854
855	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)	1117	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
856	or return $finish->(undef, 598 => "Request cancelled by on_body");
857
858	$len > 0
859	or $finish->("", undef, undef, 1);
860	});
861	} else {
862	$_[0]->on_eof (sub {
863	$finish->("");
864	});
865	$_[0]->on_read (sub {
866	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
867	or $finish->(undef, 598 => "Request cancelled by on_body");	1118	or $finish->(undef, 598 => "Request cancelled by on_body");
868	});
869	}
870	} else {
871	$_[0]->on_eof (undef);
872
873	if ($len) {
874	$_[0]->on_read (sub {
875	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
876	if $len <= length $_[0]{rbuf};
877	});
878	} else {
879	$_[0]->on_error (sub {
880	($! == Errno::EPIPE || !$!)
881	? $finish->(delete $_[0]{rbuf})
882	: $finish->(undef, $ae_error => $_[2]);
883	});
884	$_[0]->on_read (sub { });
885	}
886	}	1119	});
887	}	1120	}
888	};
889
890	$state{handle}->push_read (line => $qr_nlnl, $state{read_response});
891	};
892
893	my $connect_cb = sub {
894	$state{fh} = shift
895	or do {
896	my $err = "$!";
897	%state = ();
898	return $cb->(undef, { @pseudo, Status => $ae_error, Reason => $err });
899	};
900
901	return unless delete $state{connect_guard};
902
903	# get handle
904	$state{handle} = new AnyEvent::Handle
905	fh => $state{fh},
906	peername => $rhost,
907	tls_ctx => $arg{tls_ctx},
908	# these need to be reconfigured on keepalive handles
909	timeout => $timeout,
910	on_error => sub {
911	%state = ();
912	$cb->(undef, { @pseudo, Status => $ae_error, Reason => $_[2] });
913	},
914	on_eof => sub {
915	%state = ();
916	$cb->(undef, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" });
917	},
918	;
919
920	# limit the number of persistent connections
921	# keepalive not yet supported
922	# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
923	# ++$KA_COUNT{$_[1]};
924	# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
925	# --$KA_COUNT{$_[1]}
926	# };
927	# $hdr{connection} = "keep-alive";
928	# }
929
930	$state{handle}->starttls ("connect") if $rscheme eq "https";
931
932	# now handle proxy-CONNECT method
933	if ($proxy && $uscheme eq "https") {
934	# oh dear, we have to wrap it into a connect request
935
936	# maybe re-use $uauthority with patched port?
937	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
938	$state{handle}->push_read (line => $qr_nlnl, sub {
939	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
940	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" }));
941
942	if ($2 == 200) {
943	$rpath = $upath;
944	$handle_actual_request->();
945	} else {
946	%state = ();
947	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });
948	}
949	});
950	} else {	1121	} else {
951	$handle_actual_request->();	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	}
952	}	1137	}
953	};	1138	};
954		1139
955	my $tcp_connect = $arg{tcp_connect}	1140	# if keepalive is enabled, then the server closing the connection
956	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	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;
957		1146
958	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	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	}
959	};	1171	};
960		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	};
		1232
		1233	_get_slot $uhost, sub {
		1234	$state{slot_guard} = shift;
		1235
		1236	return unless $state{connect_guard};
		1237
		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;
		1242
		1243	$state{handle} = ka_fetch $ka_key;
		1244	# $state{handle}->destroyed
		1245	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1246	$prepare_handle->();
		1247	# $state{handle}->destroyed
		1248	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1249	$rpath = $upath;
		1250	$handle_actual_request->();
		1251
		1252	} else {
		1253	my $tcp_connect = $arg{tcp_connect}
		1254	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1255
		1256	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
		1257	}
		1258	};
		1259
961	defined wantarray && AnyEvent::Util::guard { %state = () }	1260	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
962	}	1261	}
963		1262
964	sub http_get($@) {	1263	sub http_get($@) {
965	unshift @_, "GET";	1264	unshift @_, "GET";
966	&http_request	1265	&http_request
…		…
984	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1283	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
985	the actual connection, which in turn uses AnyEvent::DNS to resolve	1284	the actual connection, which in turn uses AnyEvent::DNS to resolve
986	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
987	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
988	your own default resolver (by storing a suitable resolver object in	1287	your own default resolver (by storing a suitable resolver object in
989	C<$AnyEvent::DNS::RESOLVER>).	1288	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
990		1289
991	=head2 GLOBAL FUNCTIONS AND VARIABLES	1290	=head2 GLOBAL FUNCTIONS AND VARIABLES
992		1291
993	=over 4	1292	=over 4
994		1293
995	=item AnyEvent::HTTP::set_proxy "proxy-url"	1294	=item AnyEvent::HTTP::set_proxy "proxy-url"
996		1295
997	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
998	string of the form C<http://host:port> (optionally C<https:...>), croaks	1297	string of the form C<http://host:port>, croaks otherwise.
999	otherwise.
1000		1298
1001	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>}.
1002		1304
1003	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]	1305	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
1004		1306
1005	Remove all cookies from the cookie jar that have been expired. If	1307	Remove all cookies from the cookie jar that have been expired. If
1006	C<$session_end> is given and true, then additionally remove all session	1308	C<$session_end> is given and true, then additionally remove all session
1007	cookies.	1309	cookies.
1008		1310
1009	You should call this function (with a true C<$session_end>) before you	1311	You should call this function (with a true C<$session_end>) before you
1010	save cookies to disk, and you should call this function after loading them	1312	save cookies to disk, and you should call this function after loading them
1011	again. If you have a long-running program you can additonally call this	1313	again. If you have a long-running program you can additionally call this
1012	function from time to time.	1314	function from time to time.
1013		1315
1014	A cookie jar is initially an empty hash-reference that is managed by this	1316	A cookie jar is initially an empty hash-reference that is managed by this
1015	module. It's format is subject to change, but currently it is like this:	1317	module. Its format is subject to change, but currently it is as follows:
1016		1318
1017	The key C<version> has to contain C<1>, otherwise the hash gets	1319	The key C<version> has to contain C<2>, otherwise the hash gets
1018	emptied. All other keys are hostnames or IP addresses pointing to	1320	cleared. All other keys are hostnames or IP addresses pointing to
1019	hash-references. The key for these inner hash references is the	1321	hash-references. The key for these inner hash references is the
1020	server path for which this cookie is meant, and the values are again	1322	server path for which this cookie is meant, and the values are again
1021	hash-references. The keys of those hash-references is the cookie name, and	1323	hash-references. Each key of those hash-references is a cookie name, and
1022	the value, you guessed it, is another hash-reference, this time with the	1324	the value, you guessed it, is another hash-reference, this time with the
1023	key-value pairs from the cookie, except for C<expires> and C<max-age>,	1325	key-value pairs from the cookie, except for C<expires> and C<max-age>,
1024	which have been replaced by a C<_expires> key that contains the cookie	1326	which have been replaced by a C<_expires> key that contains the cookie
1025	expiry timestamp.	1327	expiry timestamp. Session cookies are indicated by not having an
		1328	C<_expires> key.
1026		1329
1027	Here is an example of a cookie jar with a single cookie, so you have a	1330	Here is an example of a cookie jar with a single cookie, so you have a
1028	chance of understanding the above paragraph:	1331	chance of understanding the above paragraph:
1029		1332
1030	{	1333	{
1031	version => 1,	1334	version => 2,
1032	"10.0.0.1" => {	1335	"10.0.0.1" => {
1033	"/" => {	1336	"/" => {
1034	"mythweb_id" => {	1337	"mythweb_id" => {
1035	_expires => 1293917923,	1338	_expires => 1293917923,
1036	value => "ooRung9dThee3ooyXooM1Ohm",	1339	value => "ooRung9dThee3ooyXooM1Ohm",
…		…
1052		1355
1053	=item $AnyEvent::HTTP::MAX_RECURSE	1356	=item $AnyEvent::HTTP::MAX_RECURSE
1054		1357
1055	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>).
1056		1359
		1360	=item $AnyEvent::HTTP::TIMEOUT
		1361
		1362	The default timeout for connection operations (default: C<300>).
		1363
1057	=item $AnyEvent::HTTP::USERAGENT	1364	=item $AnyEvent::HTTP::USERAGENT
1058		1365
1059	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
1060	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)>).
1061		1368
1062	=item $AnyEvent::HTTP::MAX_PER_HOST	1369	=item $AnyEvent::HTTP::MAX_PER_HOST
1063		1370
1064	The maximum number of concurrent connections to the same host (identified	1371	The maximum number of concurrent connections to the same host (identified
1065	by the hostname). If the limit is exceeded, then the additional requests	1372	by the hostname). If the limit is exceeded, then additional requests
1066	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.
1067		1375
1068	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
1069	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>).
1070		1388
1071	=item $AnyEvent::HTTP::ACTIVE	1389	=item $AnyEvent::HTTP::ACTIVE
1072		1390
1073	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
1074	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
1075	connections. This number of can be useful for load-leveling.	1393	connections. This number can be useful for load-leveling.
1076		1394
1077	=back	1395	=back
1078		1396
1079	=cut	1397	=cut
1080		1398
…		…
1114	# other formats fail in the loop below	1432	# other formats fail in the loop below
1115		1433
1116	for (0..11) {	1434	for (0..11) {
1117	if ($m eq $month[$_]) {	1435	if ($m eq $month[$_]) {
1118	require Time::Local;	1436	require Time::Local;
1119	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);	1437	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
1120	}	1438	}
1121	}	1439	}
1122		1440
1123	undef	1441	undef
1124	}	1442	}
1125		1443
1126	sub set_proxy($) {	1444	sub set_proxy($) {
1127	if (length $_[0]) {	1445	if (length $_[0]) {
1128	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1446	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1129	or Carp::croak "$_[0]: invalid proxy URL";	1447	or Carp::croak "$_[0]: invalid proxy URL";
1130	$PROXY = [$2, $3 || 3128, $1]	1448	$PROXY = [$2, $3 || 3128, $1]
1131	} else {	1449	} else {
1132	undef $PROXY;	1450	undef $PROXY;
1133	}	1451	}
…		…
1136	# initialise proxy from environment	1454	# initialise proxy from environment
1137	eval {	1455	eval {
1138	set_proxy $ENV{http_proxy};	1456	set_proxy $ENV{http_proxy};
1139	};	1457	};
1140		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
1141	=head2 SOCKS PROXIES	1559	=head3 SOCKS PROXIES
1142		1560
1143	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1561	Socks proxies are not directly supported by AnyEvent::HTTP. You can
1144	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
1145	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
1146	transparently.	1564	transparently.

Diff Legend

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