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 1.131 by root, Thu Aug 30 17:04:28 2018 UTC

…		…
15	This module is an L<AnyEvent> user, you need to make sure that you use and	15	This module is an L<AnyEvent> user, you need to make sure that you use and
16	run a supported event loop.	16	run a supported event loop.
17		17
18	This module implements a simple, stateless and non-blocking HTTP	18	This module implements a simple, stateless and non-blocking HTTP
19	client. It supports GET, POST and other request methods, cookies and more,	19	client. It supports GET, POST and other request methods, cookies and more,
20	all on a very low level. It can follow redirects supports proxies and	20	all on a very low level. It can follow redirects, supports proxies, and
21	automatically limits the number of connections to the values specified in	21	automatically limits the number of connections to the values specified in
22	the RFC.	22	the RFC.
23		23
24	It should generally be a "good client" that is enough for most HTTP	24	It should generally be a "good client" that is enough for most HTTP
25	tasks. Simple tasks should be simple, but complex tasks should still be	25	tasks. Simple tasks should be simple, but complex tasks should still be
…		…
36		36
37	=cut	37	=cut
38		38
39	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
40		40
41	use strict;	41	use common::sense;
42	no warnings;
43		42
44	use Errno ();	43	use Errno ();
45		44
46	use AnyEvent 5.0 ();	45	use AnyEvent 5.0 ();
47	use AnyEvent::Util ();	46	use AnyEvent::Util ();
48	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
49		48
50	use base Exporter::;	49	use base Exporter::;
51		50
52	our $VERSION = '1.5';	51	our $VERSION = 2.24;
53		52
54	our @EXPORT = qw(http_get http_post http_head http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
55		54
56	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";	55	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
57	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
58	our $MAX_PERSISTENT = 8;
59	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
60	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
61		59	our $MAX_PER_HOST = 4; # changing this is evil
62	# changing these is evil
63	our $MAX_PERSISTENT_PER_HOST = 0;
64	our $MAX_PER_HOST = 4;
65		60
66	our $PROXY;	61	our $PROXY;
67	our $ACTIVE = 0;	62	our $ACTIVE = 0;
68		63
69	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
70	my %CO_SLOT; # number of open connections, and wait queue, per host	65	my %CO_SLOT; # number of open connections, and wait queue, per host
71		66
72	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
73		68
74	Executes an HTTP-GET request. See the http_request function for details on	69	Executes an HTTP-GET request. See the http_request function for details on
…		…
94	C<http_request> returns a "cancellation guard" - you have to keep the	89	C<http_request> returns a "cancellation guard" - you have to keep the
95	object at least alive until the callback get called. If the object gets	90	object at least alive until the callback get called. If the object gets
96	destroyed before the callback is called, the request will be cancelled.	91	destroyed before the callback is called, the request will be cancelled.
97		92
98	The callback will be called with the response body data as first argument	93	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	94	(or C<undef> if an error occurred), and a hash-ref with response headers
100	(and trailers) as second argument.	95	(and trailers) as second argument.
101		96
102	All the headers in that hash are lowercased. In addition to the response	97	All the headers in that hash are lowercased. In addition to the response
103	headers, the "pseudo-headers" (uppercase to avoid clashing with possible	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
104	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the	99	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	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
129	message. Currently the following status codes are used:	124	message. Currently the following status codes are used:
130		125
131	=over 4	126	=over 4
132		127
133	=item 595 - errors during connection etsbalishment, proxy handshake.	128	=item 595 - errors during connection establishment, proxy handshake.
134		129
135	=item 596 - errors during TLS negotiation, request sending and header processing.	130	=item 596 - errors during TLS negotiation, request sending and header processing.
136		131
137	=item 597 - errors during body receiving or processing.	132	=item 597 - errors during body receiving or processing.
138		133
…		…
159		154
160	=over 4	155	=over 4
161		156
162	=item recurse => $count (default: $MAX_RECURSE)	157	=item recurse => $count (default: $MAX_RECURSE)
163		158
164	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
165	retries and so on, and how often to do so.	160	other retries and so on, and how often to do so.
		161
		162	Only redirects to http and https URLs are supported. While most common
		163	redirection forms are handled entirely within this module, some require
		164	the use of the optional L<URI> module. If it is required but missing, then
		165	the request will fail with an error.
166		166
167	=item headers => hashref	167	=item headers => hashref
168		168
169	The request headers to use. Currently, C<http_request> may provide its own	169	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	170	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:>	171	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	172	(this can be suppressed by using C<undef> for these headers in which case
173	they won't be sent at all).	173	they won't be sent at all).
174		174
		175	You really should provide your own C<User-Agent:> header value that is
		176	appropriate for your program - I wouldn't be surprised if the default
		177	AnyEvent string gets blocked by webservers sooner or later.
		178
		179	Also, make sure that your headers names and values do not contain any
		180	embedded newlines.
		181
175	=item timeout => $seconds	182	=item timeout => $seconds
176		183
177	The time-out to use for various stages - each connect attempt will reset	184	The time-out to use for various stages - each connect attempt will reset
178	the timeout, as will read or write activity, i.e. this is not an overall	185	the timeout, as will read or write activity, i.e. this is not an overall
179	timeout.	186	timeout.
180		187
181	Default timeout is 5 minutes.	188	Default timeout is 5 minutes.
182		189
183	=item proxy => [$host, $port[, $scheme]] or undef	190	=item proxy => [$host, $port[, $scheme]] or undef
184		191
185	Use the given http proxy for all requests. If not specified, then the	192	Use the given http proxy for all requests, or no proxy if C<undef> is
186	default proxy (as specified by C<$ENV{http_proxy}>) is used.	193	used.
187		194
188	C<$scheme> must be either missing, C<http> for HTTP or C<https> for	195	C<$scheme> must be either missing or must be C<http> for HTTP.
189	HTTPS.	196
		197	If not specified, then the default proxy is used (see
		198	C<AnyEvent::HTTP::set_proxy>).
		199
		200	Currently, if your proxy requires authorization, you have to specify an
		201	appropriate "Proxy-Authorization" header in every request.
190		202
191	=item body => $string	203	=item body => $string
192		204
193	The request body, usually empty. Will be sent as-is (future versions of	205	The request body, usually empty. Will be sent as-is (future versions of
194	this module might offer more options).	206	this module might offer more options).
…		…
196	=item cookie_jar => $hash_ref	208	=item cookie_jar => $hash_ref
197		209
198	Passing this parameter enables (simplified) cookie-processing, loosely	210	Passing this parameter enables (simplified) cookie-processing, loosely
199	based on the original netscape specification.	211	based on the original netscape specification.
200		212
201	The C<$hash_ref> must be an (initially empty) hash reference which will	213	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	214	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	215	to persistent storage with something like JSON or Storable - see the
204	recommended, as session-only cookies might survive longer than expected.	216	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		217	expired or session-only cookies, and also for documentation on the format
		218	of the cookie jar.
205		219
206	Note that this cookie implementation is not meant to be complete. If	220	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	221	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	222	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	223	working. Cookies are a privacy disaster, do not use them unless required
210	to.	224	to.
211		225
212	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>	226	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	227	headers will be set and handled by this module, otherwise they will be
…		…
223	verification) TLS context.	237	verification) TLS context.
224		238
225	The default for this option is C<low>, which could be interpreted as "give	239	The default for this option is C<low>, which could be interpreted as "give
226	me the page, no matter what".	240	me the page, no matter what".
227		241
		242	See also the C<sessionid> parameter.
		243
		244	=item session => $string
		245
		246	The module might reuse connections to the same host internally. Sometimes
		247	(e.g. when using TLS), you do not want to reuse connections from other
		248	sessions. This can be achieved by setting this parameter to some unique
		249	ID (such as the address of an object storing your state data, or the TLS
		250	context) - only connections using the same unique ID will be reused.
		251
228	=item on_prepare => $callback->($fh)	252	=item on_prepare => $callback->($fh)
229		253
230	In rare cases you need to "tune" the socket before it is used to	254	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	255	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>	256	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	257	and behaves exactly the same way (e.g. it has to provide a
234	timeout). See the description for the C<$prepare_cb> argument of	258	timeout). See the description for the C<$prepare_cb> argument of
235	C<AnyEvent::Socket::tcp_connect> for details.	259	C<AnyEvent::Socket::tcp_connect> for details.
236		260
…		…
301	called. Instead of the C<$body> argument containing the body data, the	325	called. Instead of the C<$body> argument containing the body data, the
302	callback will receive the L<AnyEvent::Handle> object associated with the	326	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	327	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.	328	(e.g. status C<304>), the empty string will be passed.
305		329
306	The handle object might or might not be in TLS mode, might be connected to	330	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	331	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	332	etc., and configured in unspecified ways. The user is responsible for this
309	module anymore).	333	handle (it will not be used by this module anymore).
310		334
311	This is useful with some push-type services, where, after the initial	335	This is useful with some push-type services, where, after the initial
312	headers, an interactive protocol is used (typical example would be the	336	headers, an interactive protocol is used (typical example would be the
313	push-style twitter API which starts a JSON/XML stream).	337	push-style twitter API which starts a JSON/XML stream).
314		338
315	If you think you need this, first have a look at C<on_body>, to see if	339	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.	340	that doesn't solve your problem in a better way.
		341
		342	=item persistent => $boolean
		343
		344	Try to create/reuse a persistent connection. When this flag is set
		345	(default: true for idempotent requests, false for all others), then
		346	C<http_request> tries to re-use an existing (previously-created)
		347	persistent connection to the host and, failing that, tries to create a new
		348	one.
		349
		350	Requests failing in certain ways will be automatically retried once, which
		351	is dangerous for non-idempotent requests, which is why it defaults to off
		352	for them. The reason for this is because the bozos who designed HTTP/1.1
		353	made it impossible to distinguish between a fatal error and a normal
		354	connection timeout, so you never know whether there was a problem with
		355	your request or not.
		356
		357	When reusing an existent connection, many parameters (such as TLS context)
		358	will be ignored. See the C<session> parameter for a workaround.
		359
		360	=item keepalive => $boolean
		361
		362	Only used when C<persistent> is also true. This parameter decides whether
		363	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		364	(as opposed to only a HTTP/1.1 persistent connection).
		365
		366	The default is true, except when using a proxy, in which case it defaults
		367	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		368
		369	=item handle_params => { key => value ... }
		370
		371	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		372	constructor that is called - not all requests will create a handle, and
		373	sometimes more than one is created, so this parameter is only good for
		374	setting hints.
		375
		376	Example: set the maximum read size to 4096, to potentially conserve memory
		377	at the cost of speed.
		378
		379	handle_params => {
		380	max_read_size => 4096,
		381	},
317		382
318	=back	383	=back
319		384
320	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	385	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
321	the response body.	386	the response body.
…		…
327		392
328	Example: do a HTTP HEAD request on https://www.google.com/, use a	393	Example: do a HTTP HEAD request on https://www.google.com/, use a
329	timeout of 30 seconds.	394	timeout of 30 seconds.
330		395
331	http_request	396	http_request
332	GET => "https://www.google.com",	397	HEAD => "https://www.google.com",
		398	headers => { "user-agent" => "MySearchClient 1.0" },
333	timeout => 30,	399	timeout => 30,
334	sub {	400	sub {
335	my ($body, $hdr) = @_;	401	my ($body, $hdr) = @_;
336	use Data::Dumper;	402	use Data::Dumper;
337	print Dumper $hdr;	403	print Dumper $hdr;
…		…
347	};	413	};
348		414
349	undef $request;	415	undef $request;
350		416
351	=cut	417	=cut
		418
		419	#############################################################################
		420	# wait queue/slots
352		421
353	sub _slot_schedule;	422	sub _slot_schedule;
354	sub _slot_schedule($) {	423	sub _slot_schedule($) {
355	my $host = shift;	424	my $host = shift;
356		425
…		…
378	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	447	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
379		448
380	_slot_schedule $_[0];	449	_slot_schedule $_[0];
381	}	450	}
382		451
		452	#############################################################################
		453	# cookie handling
		454
		455	# expire cookies
		456	sub cookie_jar_expire($;$) {
		457	my ($jar, $session_end) = @_;
		458
		459	%$jar = () if $jar->{version} != 2;
		460
		461	my $anow = AE::now;
		462
		463	while (my ($chost, $paths) = each %$jar) {
		464	next unless ref $paths;
		465
		466	while (my ($cpath, $cookies) = each %$paths) {
		467	while (my ($cookie, $kv) = each %$cookies) {
		468	if (exists $kv->{_expires}) {
		469	delete $cookies->{$cookie}
		470	if $anow > $kv->{_expires};
		471	} elsif ($session_end) {
		472	delete $cookies->{$cookie};
		473	}
		474	}
		475
		476	delete $paths->{$cpath}
		477	unless %$cookies;
		478	}
		479
		480	delete $jar->{$chost}
		481	unless %$paths;
		482	}
		483	}
		484
383	# extract cookies from jar	485	# extract cookies from jar
384	sub cookie_jar_extract($$$$) {	486	sub cookie_jar_extract($$$$) {
385	my ($jar, $uscheme, $uhost, $upath) = @_;	487	my ($jar, $scheme, $host, $path) = @_;
386		488
387	%$jar = () if $jar->{version} != 1;	489	%$jar = () if $jar->{version} != 2;
		490
		491	$host = AnyEvent::Util::idn_to_ascii $host
		492	if $host =~ /[^\x00-\x7f]/;
388		493
389	my @cookies;	494	my @cookies;
390		495
391	while (my ($chost, $paths) = each %$jar) {	496	while (my ($chost, $paths) = each %$jar) {
392	next unless ref $paths;	497	next unless ref $paths;
393		498
394	if ($chost =~ /^\./) {	499	# exact match or suffix including . match
395	next unless $chost eq substr $uhost, -length $chost;	500	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
396	} elsif ($chost =~ /\./) {
397	next unless $chost eq $uhost;
398	} else {
399	next;	501	or next;
400	}
401		502
402	while (my ($cpath, $cookies) = each %$paths) {	503	while (my ($cpath, $cookies) = each %$paths) {
403	next unless $cpath eq substr $upath, 0, length $cpath;	504	next unless $cpath eq substr $path, 0, length $cpath;
404		505
405	while (my ($cookie, $kv) = each %$cookies) {	506	while (my ($cookie, $kv) = each %$cookies) {
406	next if $uscheme ne "https" && exists $kv->{secure};	507	next if $scheme ne "https" && exists $kv->{secure};
407		508
408	if (exists $kv->{expires}) {	509	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
409	if (AE::now > parse_date ($kv->{expires})) {
410	delete $cookies->{$cookie};	510	delete $cookies->{$cookie};
411	next;	511	next;
412	}
413	}	512	}
414		513
415	my $value = $kv->{value};	514	my $value = $kv->{value};
416		515
417	if ($value =~ /[=;,[:space:]]/) {	516	if ($value =~ /[=;,[:space:]]/) {
…		…
426		525
427	\@cookies	526	\@cookies
428	}	527	}
429		528
430	# parse set_cookie header into jar	529	# parse set_cookie header into jar
431	sub cookie_jar_set_cookie($$$) {	530	sub cookie_jar_set_cookie($$$$) {
432	my ($jar, $set_cookie, $uhost) = @_;	531	my ($jar, $set_cookie, $host, $date) = @_;
		532
		533	%$jar = () if $jar->{version} != 2;
		534
		535	my $anow = int AE::now;
		536	my $snow; # server-now
433		537
434	for ($set_cookie) {	538	for ($set_cookie) {
435	# parse NAME=VALUE	539	# parse NAME=VALUE
436	my @kv;	540	my @kv;
437		541
…		…
440	while (	544	while (
441	m{	545	m{
442	\G\s*	546	\G\s*
443	(?:	547	(?:
444	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)	548	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
445	| ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )	549	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
446	)	550	)
447	}gcxsi	551	}gcxsi
448	) {	552	) {
449	my $name = $2;	553	my $name = $2;
450	my $value = $4;	554	my $value = $4;
451		555
452	unless (defined $name) {	556	if (defined $1) {
453	# expires	557	# expires
454	$name = "expires";	558	$name = "expires";
455	$value = $1;	559	$value = $1;
456	} elsif (!defined $value) {	560	} elsif (defined $3) {
457	# quoted	561	# quoted
458	$value = $3;	562	$value = $3;
459	$value =~ s/\\(.)/$1/gs;	563	$value =~ s/\\(.)/$1/gs;
460	}	564	}
461		565
462	push @kv, lc $name, $value;	566	push @kv, @kv ? lc $name : $name, $value;
463		567
464	last unless /\G\s*;/gc;	568	last unless /\G\s*;/gc;
465	}	569	}
466		570
467	last unless @kv;	571	last unless @kv;
468		572
469	my $name = shift @kv;	573	my $name = shift @kv;
470	my %kv = (value => shift @kv, @kv);	574	my %kv = (value => shift @kv, @kv);
471		575
472	$kv{expires} ||= format_date (AE::now + $kv{"max-age"})
473	if exists $kv{"max-age"};	576	if (exists $kv{"max-age"}) {
		577	$kv{_expires} = $anow + delete $kv{"max-age"};
		578	} elsif (exists $kv{expires}) {
		579	$snow ||= parse_date ($date) || $anow;
		580	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		581	} else {
		582	delete $kv{_expires};
		583	}
474		584
475	my $cdom;	585	my $cdom;
476	my $cpath = (delete $kv{path}) || "/";	586	my $cpath = (delete $kv{path}) || "/";
477		587
478	if (exists $kv{domain}) {	588	if (exists $kv{domain}) {
479	$cdom = delete $kv{domain};	589	$cdom = $kv{domain};
480		590
481	$cdom =~ s/^\.?/./; # make sure it starts with a "."	591	$cdom =~ s/^\.?/./; # make sure it starts with a "."
482		592
483	next if $cdom =~ /\.$/;	593	next if $cdom =~ /\.$/;
484		594
485	# this is not rfc-like and not netscape-like. go figure.	595	# this is not rfc-like and not netscape-like. go figure.
486	my $ndots = $cdom =~ y/.//;	596	my $ndots = $cdom =~ y/.//;
487	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);	597	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		598
		599	$cdom = substr $cdom, 1; # remove initial .
488	} else {	600	} else {
489	$cdom = $uhost;	601	$cdom = $host;
490	}	602	}
491		603
492	# store it	604	# store it
493	$jar->{version} = 1;	605	$jar->{version} = 2;
494	$jar->{$cdom}{$cpath}{$name} = \%kv;	606	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
495		607
496	redo if /\G\s*,/gc;	608	redo if /\G\s*,/gc;
497	}	609	}
498	}	610	}
499		611
		612	#############################################################################
		613	# keepalive/persistent connection cache
		614
		615	# fetch a connection from the keepalive cache
		616	sub ka_fetch($) {
		617	my $ka_key = shift;
		618
		619	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		620	delete $KA_CACHE{$ka_key}
		621	unless @{ $KA_CACHE{$ka_key} };
		622
		623	$hdl
		624	}
		625
		626	sub ka_store($$) {
		627	my ($ka_key, $hdl) = @_;
		628
		629	my $kaa = $KA_CACHE{$ka_key} ||= [];
		630
		631	my $destroy = sub {
		632	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		633
		634	$hdl->destroy;
		635
		636	@ka
		637	? $KA_CACHE{$ka_key} = \@ka
		638	: delete $KA_CACHE{$ka_key};
		639	};
		640
		641	# on error etc., destroy
		642	$hdl->on_error ($destroy);
		643	$hdl->on_eof ($destroy);
		644	$hdl->on_read ($destroy);
		645	$hdl->timeout ($PERSISTENT_TIMEOUT);
		646
		647	push @$kaa, $hdl;
		648	shift @$kaa while @$kaa > $MAX_PER_HOST;
		649	}
		650
		651	#############################################################################
		652	# utilities
		653
500	# continue to parse $_ for headers and place them into the arg	654	# continue to parse $_ for headers and place them into the arg
501	sub parse_hdr() {	655	sub _parse_hdr() {
502	my %hdr;	656	my %hdr;
503		657
504	# things seen, not parsed:	658	# things seen, not parsed:
505	# p3pP="NON CUR OTPi OUR NOR UNI"	659	# p3pP="NON CUR OTPi OUR NOR UNI"
506		660
…		…
520	for values %hdr;	674	for values %hdr;
521		675
522	\%hdr	676	\%hdr
523	}	677	}
524		678
		679	#############################################################################
		680	# http_get
		681
525	our $qr_nlnl = qr{(?<![^\012])\015?\012};	682	our $qr_nlnl = qr{(?<![^\012])\015?\012};
526		683
527	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	684	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
528	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };	685	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		686
		687	# maybe it should just become a normal object :/
		688
		689	sub _destroy_state(\%) {
		690	my ($state) = @_;
		691
		692	$state->{handle}->destroy if $state->{handle};
		693	%$state = ();
		694	}
		695
		696	sub _error(\%$$) {
		697	my ($state, $cb, $hdr) = @_;
		698
		699	&_destroy_state ($state);
		700
		701	$cb->(undef, $hdr);
		702	()
		703	}
		704
		705	our %IDEMPOTENT = (
		706	DELETE => 1,
		707	GET => 1,
		708	HEAD => 1,
		709	OPTIONS => 1,
		710	PUT => 1,
		711	TRACE => 1,
		712
		713	ACL => 1,
		714	"BASELINE-CONTROL" => 1,
		715	BIND => 1,
		716	CHECKIN => 1,
		717	CHECKOUT => 1,
		718	COPY => 1,
		719	LABEL => 1,
		720	LINK => 1,
		721	MERGE => 1,
		722	MKACTIVITY => 1,
		723	MKCALENDAR => 1,
		724	MKCOL => 1,
		725	MKREDIRECTREF => 1,
		726	MKWORKSPACE => 1,
		727	MOVE => 1,
		728	ORDERPATCH => 1,
		729	PROPFIND => 1,
		730	PROPPATCH => 1,
		731	REBIND => 1,
		732	REPORT => 1,
		733	SEARCH => 1,
		734	UNBIND => 1,
		735	UNCHECKOUT => 1,
		736	UNLINK => 1,
		737	UNLOCK => 1,
		738	UPDATE => 1,
		739	UPDATEREDIRECTREF => 1,
		740	"VERSION-CONTROL" => 1,
		741	);
529		742
530	sub http_request($$@) {	743	sub http_request($$@) {
531	my $cb = pop;	744	my $cb = pop;
532	my ($method, $url, %arg) = @_;	745	my ($method, $url, %arg) = @_;
533		746
…		…
551	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	764	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
552		765
553	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	766	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
554	if $recurse < 0;	767	if $recurse < 0;
555		768
556	my $proxy = $arg{proxy} || $PROXY;	769	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
557	my $timeout = $arg{timeout} || $TIMEOUT;	770	my $timeout = $arg{timeout} || $TIMEOUT;
558		771
559	my ($uscheme, $uauthority, $upath, $query, $fragment) =	772	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
560	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	773	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
561		774
562	$uscheme = lc $uscheme;	775	$uscheme = lc $uscheme;
563		776
564	my $uport = $uscheme eq "http" ? 80	777	my $uport = $uscheme eq "http" ? 80
565	: $uscheme eq "https" ? 443	778	: $uscheme eq "https" ? 443
566	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });	779	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
567		780
568	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	781	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
569	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	782	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
570		783
571	my $uhost = $1;	784	my $uhost = lc $1;
572	$uport = $2 if defined $2;	785	$uport = $2 if defined $2;
573		786
574	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	787	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
575	unless exists $hdr{host};	788	unless exists $hdr{host};
576		789
…		…
595	$rscheme = "http" unless defined $rscheme;	808	$rscheme = "http" unless defined $rscheme;
596		809
597	# don't support https requests over https-proxy transport,	810	# don't support https requests over https-proxy transport,
598	# can't be done with tls as spec'ed, unless you double-encrypt.	811	# can't be done with tls as spec'ed, unless you double-encrypt.
599	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";	812	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		813
		814	$rhost = lc $rhost;
		815	$rscheme = lc $rscheme;
600	} else {	816	} else {
601	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	817	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
602	}	818	}
603		819
604	# leave out fragment and query string, just a heuristic	820	# leave out fragment and query string, just a heuristic
…		…
606	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	822	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
607		823
608	$hdr{"content-length"} = length $arg{body}	824	$hdr{"content-length"} = length $arg{body}
609	if length $arg{body} || $method ne "GET";	825	if length $arg{body} || $method ne "GET";
610		826
611	$hdr{connection} = "close TE"; #1.1	827	my $idempotent = $IDEMPOTENT{$method};
		828
		829	# default value for keepalive is true iff the request is for an idempotent method
		830	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		831	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		832	my $was_persistent; # true if this is actually a recycled connection
		833
		834	# the key to use in the keepalive cache
		835	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		836
		837	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
612	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	838	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
613		839
614	my %state = (connect_guard => 1);	840	my %state = (connect_guard => 1);
		841
		842	my $ae_error = 595; # connecting
		843
		844	# handle actual, non-tunneled, request
		845	my $handle_actual_request = sub {
		846	$ae_error = 596; # request phase
		847
		848	my $hdl = $state{handle};
		849
		850	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		851
		852	# send request
		853	$hdl->push_write (
		854	"$method $rpath HTTP/1.1\015\012"
		855	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		856	. "\015\012"
		857	. $arg{body}
		858	);
		859
		860	# return if error occurred during push_write()
		861	return unless %state;
		862
		863	# reduce memory usage, save a kitten, also re-use it for the response headers.
		864	%hdr = ();
		865
		866	# status line and headers
		867	$state{read_response} = sub {
		868	return unless %state;
		869
		870	for ("$_[1]") {
		871	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		872
		873	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		874	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		875
		876	# 100 Continue handling
		877	# should not happen as we don't send expect: 100-continue,
		878	# but we handle it just in case.
		879	# since we send the request body regardless, if we get an error
		880	# we are out of-sync, which we currently do NOT handle correctly.
		881	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		882	if $2 eq 100;
		883
		884	push @pseudo,
		885	HTTPVersion => $1,
		886	Status => $2,
		887	Reason => $3,
		888	;
		889
		890	my $hdr = _parse_hdr
		891	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		892
		893	%hdr = (%$hdr, @pseudo);
		894	}
		895
		896	# redirect handling
		897	# relative uri handling forced by microsoft and other shitheads.
		898	# we give our best and fall back to URI if available.
		899	if (exists $hdr{location}) {
		900	my $loc = $hdr{location};
		901
		902	if ($loc =~ m%^//%) { # //
		903	$loc = "$uscheme:$loc";
		904
		905	} elsif ($loc eq "") {
		906	$loc = $url;
		907
		908	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		909	$loc =~ s/^\.\/+//;
		910
		911	if ($loc !~ m%^[.?#]%) {
		912	my $prefix = "$uscheme://$uauthority";
		913
		914	unless ($loc =~ s/^\///) {
		915	$prefix .= $upath;
		916	$prefix =~ s/\/[^\/]*$//;
		917	}
		918
		919	$loc = "$prefix/$loc";
		920
		921	} elsif (eval { require URI }) { # uri
		922	$loc = URI->new_abs ($loc, $url)->as_string;
		923
		924	} else {
		925	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		926	#$hdr{Status} = 599;
		927	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		928	#$recurse = 0;
		929	}
		930	}
		931
		932	$hdr{location} = $loc;
		933	}
		934
		935	my $redirect;
		936
		937	if ($recurse) {
		938	my $status = $hdr{Status};
		939
		940	# industry standard is to redirect POST as GET for
		941	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		942	# also, the UA should ask the user for 301 and 307 and POST,
		943	# industry standard seems to be to simply follow.
		944	# we go with the industry standard. 308 is defined
		945	# by rfc7538
		946	if ($status == 301 or $status == 302 or $status == 303) {
		947	$redirect = 1;
		948	# HTTP/1.1 is unclear on how to mutate the method
		949	unless ($method eq "HEAD") {
		950	$method = "GET";
		951	delete $arg{body};
		952	}
		953	} elsif ($status == 307 or $status == 308) {
		954	$redirect = 1;
		955	}
		956	}
		957
		958	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		959	if ($state{handle}) {
		960	# handle keepalive
		961	if (
		962	$persistent
		963	&& $_[3]
		964	&& ($hdr{HTTPVersion} < 1.1
		965	? $hdr{connection} =~ /\bkeep-?alive\b/i
		966	: $hdr{connection} !~ /\bclose\b/i)
		967	) {
		968	ka_store $ka_key, delete $state{handle};
		969	} else {
		970	# no keepalive, destroy the handle
		971	$state{handle}->destroy;
		972	}
		973	}
		974
		975	%state = ();
		976
		977	if (defined $_[1]) {
		978	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		979	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		980	}
		981
		982	# set-cookie processing
		983	if ($arg{cookie_jar}) {
		984	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		985	}
		986
		987	if ($redirect && exists $hdr{location}) {
		988	# we ignore any errors, as it is very common to receive
		989	# Content-Length != 0 but no actual body
		990	# we also access %hdr, as $_[1] might be an erro
		991	$state{recurse} =
		992	http_request (
		993	$method => $hdr{location},
		994	%arg,
		995	recurse => $recurse - 1,
		996	Redirect => [$_[0], \%hdr],
		997	sub {
		998	%state = ();
		999	&$cb
		1000	},
		1001	);
		1002	} else {
		1003	$cb->($_[0], \%hdr);
		1004	}
		1005	};
		1006
		1007	$ae_error = 597; # body phase
		1008
		1009	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1010
		1011	my $len = $chunked ? undef : $hdr{"content-length"};
		1012
		1013	# body handling, many different code paths
		1014	# - no body expected
		1015	# - want_body_handle
		1016	# - te chunked
		1017	# - 2x length known (with or without on_body)
		1018	# - 2x length not known (with or without on_body)
		1019	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1020	$finish->(undef, 598 => "Request cancelled by on_header");
		1021	} elsif (
		1022	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1023	or $method eq "HEAD"
		1024	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1025	) {
		1026	# no body
		1027	$finish->("", undef, undef, 1);
		1028
		1029	} elsif (!$redirect && $arg{want_body_handle}) {
		1030	$_[0]->on_eof (undef);
		1031	$_[0]->on_error (undef);
		1032	$_[0]->on_read (undef);
		1033
		1034	$finish->(delete $state{handle});
		1035
		1036	} elsif ($chunked) {
		1037	my $cl = 0;
		1038	my $body = "";
		1039	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1040
		1041	$state{read_chunk} = sub {
		1042	$_[1] =~ /^([0-9a-fA-F]+)/
		1043	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1044
		1045	my $len = hex $1;
		1046
		1047	if ($len) {
		1048	$cl += $len;
		1049
		1050	$_[0]->push_read (chunk => $len, sub {
		1051	$on_body->($_[1], \%hdr)
		1052	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1053
		1054	$_[0]->push_read (line => sub {
		1055	length $_[1]
		1056	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1057	$_[0]->push_read (line => $state{read_chunk});
		1058	});
		1059	});
		1060	} else {
		1061	$hdr{"content-length"} ||= $cl;
		1062
		1063	$_[0]->push_read (line => $qr_nlnl, sub {
		1064	if (length $_[1]) {
		1065	for ("$_[1]") {
		1066	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1067
		1068	my $hdr = _parse_hdr
		1069	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1070
		1071	%hdr = (%hdr, %$hdr);
		1072	}
		1073	}
		1074
		1075	$finish->($body, undef, undef, 1);
		1076	});
		1077	}
		1078	};
		1079
		1080	$_[0]->push_read (line => $state{read_chunk});
		1081
		1082	} elsif (!$redirect && $arg{on_body}) {
		1083	if (defined $len) {
		1084	$_[0]->on_read (sub {
		1085	$len -= length $_[0]{rbuf};
		1086
		1087	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1088	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1089
		1090	$len > 0
		1091	or $finish->("", undef, undef, 1);
		1092	});
		1093	} else {
		1094	$_[0]->on_eof (sub {
		1095	$finish->("");
		1096	});
		1097	$_[0]->on_read (sub {
		1098	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1099	or $finish->(undef, 598 => "Request cancelled by on_body");
		1100	});
		1101	}
		1102	} else {
		1103	$_[0]->on_eof (undef);
		1104
		1105	if (defined $len) {
		1106	$_[0]->on_read (sub {
		1107	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1108	if $len <= length $_[0]{rbuf};
		1109	});
		1110	} else {
		1111	$_[0]->on_error (sub {
		1112	($! == Errno::EPIPE || !$!)
		1113	? $finish->(delete $_[0]{rbuf})
		1114	: $finish->(undef, $ae_error => $_[2]);
		1115	});
		1116	$_[0]->on_read (sub { });
		1117	}
		1118	}
		1119	};
		1120
		1121	# if keepalive is enabled, then the server closing the connection
		1122	# before a response can happen legally - we retry on idempotent methods.
		1123	if ($was_persistent && $idempotent) {
		1124	my $old_eof = $hdl->{on_eof};
		1125	$hdl->{on_eof} = sub {
		1126	_destroy_state %state;
		1127
		1128	%state = ();
		1129	$state{recurse} =
		1130	http_request (
		1131	$method => $url,
		1132	%arg,
		1133	recurse => $recurse - 1,
		1134	persistent => 0,
		1135	sub {
		1136	%state = ();
		1137	&$cb
		1138	}
		1139	);
		1140	};
		1141	$hdl->on_read (sub {
		1142	return unless %state;
		1143
		1144	# as soon as we receive something, a connection close
		1145	# once more becomes a hard error
		1146	$hdl->{on_eof} = $old_eof;
		1147	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1148	});
		1149	} else {
		1150	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1151	}
		1152	};
		1153
		1154	my $prepare_handle = sub {
		1155	my ($hdl) = $state{handle};
		1156
		1157	$hdl->on_error (sub {
		1158	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1159	});
		1160	$hdl->on_eof (sub {
		1161	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1162	});
		1163	$hdl->timeout_reset;
		1164	$hdl->timeout ($timeout);
		1165	};
		1166
		1167	# connected to proxy (or origin server)
		1168	my $connect_cb = sub {
		1169	my $fh = shift
		1170	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1171
		1172	return unless delete $state{connect_guard};
		1173
		1174	# get handle
		1175	$state{handle} = new AnyEvent::Handle
		1176	%{ $arg{handle_params} },
		1177	fh => $fh,
		1178	peername => $uhost,
		1179	tls_ctx => $arg{tls_ctx},
		1180	;
		1181
		1182	$prepare_handle->();
		1183
		1184	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1185
		1186	# now handle proxy-CONNECT method
		1187	if ($proxy && $uscheme eq "https") {
		1188	# oh dear, we have to wrap it into a connect request
		1189
		1190	my $auth = exists $hdr{"proxy-authorization"}
		1191	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1192	: "";
		1193
		1194	# maybe re-use $uauthority with patched port?
		1195	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
		1196	$state{handle}->push_read (line => $qr_nlnl, sub {
		1197	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1198	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1199
		1200	if ($2 == 200) {
		1201	$rpath = $upath;
		1202	$handle_actual_request->();
		1203	} else {
		1204	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1205	}
		1206	});
		1207	} else {
		1208	delete $hdr{"proxy-authorization"} unless $proxy;
		1209
		1210	$handle_actual_request->();
		1211	}
		1212	};
615		1213
616	_get_slot $uhost, sub {	1214	_get_slot $uhost, sub {
617	$state{slot_guard} = shift;	1215	$state{slot_guard} = shift;
618		1216
619	return unless $state{connect_guard};	1217	return unless $state{connect_guard};
620		1218
621	my $ae_error = 595; # connecting	1219	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1220	# on a keepalive request (in theory, this should be a separate config option).
		1221	if ($persistent && $KA_CACHE{$ka_key}) {
		1222	$was_persistent = 1;
622		1223
623	my $connect_cb = sub {	1224	$state{handle} = ka_fetch $ka_key;
624	$state{fh} = shift	1225	# $state{handle}->destroyed
625	or do {	1226	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
626	my $err = "$!";	1227	$prepare_handle->();
627	%state = ();	1228	# $state{handle}->destroyed
628	return $cb->(undef, { @pseudo, Status => $ae_error, Reason => $err });	1229	# 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;
903	&$handle_actual_request;	1230	$handle_actual_request->();
904	} else {	1231
905	%state = ();
906	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });
907	}
908	});
909	} else {	1232	} else {
910	&$handle_actual_request;
911	}
912	};
913
914	my $tcp_connect = $arg{tcp_connect}	1233	my $tcp_connect = $arg{tcp_connect}
915	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	1234	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
916		1235
917	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	1236	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
918		1237	}
919	};	1238	};
920		1239
921	defined wantarray && AnyEvent::Util::guard { %state = () }	1240	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
922	}	1241	}
923		1242
924	sub http_get($@) {	1243	sub http_get($@) {
925	unshift @_, "GET";	1244	unshift @_, "GET";
926	&http_request	1245	&http_request
…		…
944	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1263	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
945	the actual connection, which in turn uses AnyEvent::DNS to resolve	1264	the actual connection, which in turn uses AnyEvent::DNS to resolve
946	hostnames. The latter is a simple stub resolver and does no caching	1265	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	1266	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	1267	your own default resolver (by storing a suitable resolver object in
949	C<$AnyEvent::DNS::RESOLVER>).	1268	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
950		1269
951	=head2 GLOBAL FUNCTIONS AND VARIABLES	1270	=head2 GLOBAL FUNCTIONS AND VARIABLES
952		1271
953	=over 4	1272	=over 4
954		1273
955	=item AnyEvent::HTTP::set_proxy "proxy-url"	1274	=item AnyEvent::HTTP::set_proxy "proxy-url"
956		1275
957	Sets the default proxy server to use. The proxy-url must begin with a	1276	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	1277	string of the form C<http://host:port>, croaks otherwise.
959	otherwise.
960		1278
961	To clear an already-set proxy, use C<undef>.	1279	To clear an already-set proxy, use C<undef>.
		1280
		1281	When AnyEvent::HTTP is loaded for the first time it will query the
		1282	default proxy from the operating system, currently by looking at
		1283	C<$ENV{http_proxy>}.
		1284
		1285	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1286
		1287	Remove all cookies from the cookie jar that have been expired. If
		1288	C<$session_end> is given and true, then additionally remove all session
		1289	cookies.
		1290
		1291	You should call this function (with a true C<$session_end>) before you
		1292	save cookies to disk, and you should call this function after loading them
		1293	again. If you have a long-running program you can additionally call this
		1294	function from time to time.
		1295
		1296	A cookie jar is initially an empty hash-reference that is managed by this
		1297	module. Its format is subject to change, but currently it is as follows:
		1298
		1299	The key C<version> has to contain C<2>, otherwise the hash gets
		1300	cleared. All other keys are hostnames or IP addresses pointing to
		1301	hash-references. The key for these inner hash references is the
		1302	server path for which this cookie is meant, and the values are again
		1303	hash-references. Each key of those hash-references is a cookie name, and
		1304	the value, you guessed it, is another hash-reference, this time with the
		1305	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1306	which have been replaced by a C<_expires> key that contains the cookie
		1307	expiry timestamp. Session cookies are indicated by not having an
		1308	C<_expires> key.
		1309
		1310	Here is an example of a cookie jar with a single cookie, so you have a
		1311	chance of understanding the above paragraph:
		1312
		1313	{
		1314	version => 1,
		1315	"10.0.0.1" => {
		1316	"/" => {
		1317	"mythweb_id" => {
		1318	_expires => 1293917923,
		1319	value => "ooRung9dThee3ooyXooM1Ohm",
		1320	},
		1321	},
		1322	},
		1323	}
962		1324
963	=item $date = AnyEvent::HTTP::format_date $timestamp	1325	=item $date = AnyEvent::HTTP::format_date $timestamp
964		1326
965	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP	1327	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
966	Date (RFC 2616).	1328	Date (RFC 2616).
…		…
973		1335
974	=item $AnyEvent::HTTP::MAX_RECURSE	1336	=item $AnyEvent::HTTP::MAX_RECURSE
975		1337
976	The default value for the C<recurse> request parameter (default: C<10>).	1338	The default value for the C<recurse> request parameter (default: C<10>).
977		1339
		1340	=item $AnyEvent::HTTP::TIMEOUT
		1341
		1342	The default timeout for connection operations (default: C<300>).
		1343
978	=item $AnyEvent::HTTP::USERAGENT	1344	=item $AnyEvent::HTTP::USERAGENT
979		1345
980	The default value for the C<User-Agent> header (the default is	1346	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)>).	1347	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
982		1348
983	=item $AnyEvent::HTTP::MAX_PER_HOST	1349	=item $AnyEvent::HTTP::MAX_PER_HOST
984		1350
985	The maximum number of concurrent connections to the same host (identified	1351	The maximum number of concurrent connections to the same host (identified
986	by the hostname). If the limit is exceeded, then the additional requests	1352	by the hostname). If the limit is exceeded, then additional requests
987	are queued until previous connections are closed.	1353	are queued until previous connections are closed. Both persistent and
		1354	non-persistent connections are counted in this limit.
988		1355
989	The default value for this is C<4>, and it is highly advisable to not	1356	The default value for this is C<4>, and it is highly advisable to not
990	increase it.	1357	increase it much.
		1358
		1359	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1360	connections, older browsers used 2, newer ones (such as firefox 3)
		1361	typically use 6, and Opera uses 8 because like, they have the fastest
		1362	browser and give a shit for everybody else on the planet.
		1363
		1364	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1365
		1366	The time after which idle persistent connections get closed by
		1367	AnyEvent::HTTP (default: C<3>).
991		1368
992	=item $AnyEvent::HTTP::ACTIVE	1369	=item $AnyEvent::HTTP::ACTIVE
993		1370
994	The number of active connections. This is not the number of currently	1371	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	1372	running requests, but the number of currently open and non-idle TCP
996	connections. This number of can be useful for load-leveling.	1373	connections. This number can be useful for load-leveling.
997		1374
998	=back	1375	=back
999		1376
1000	=cut	1377	=cut
1001		1378
…		…
1035	# other formats fail in the loop below	1412	# other formats fail in the loop below
1036		1413
1037	for (0..11) {	1414	for (0..11) {
1038	if ($m eq $month[$_]) {	1415	if ($m eq $month[$_]) {
1039	require Time::Local;	1416	require Time::Local;
1040	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);	1417	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
1041	}	1418	}
1042	}	1419	}
1043		1420
1044	undef	1421	undef
1045	}	1422	}
1046		1423
1047	sub set_proxy($) {	1424	sub set_proxy($) {
1048	if (length $_[0]) {	1425	if (length $_[0]) {
1049	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1426	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1050	or Carp::croak "$_[0]: invalid proxy URL";	1427	or Carp::croak "$_[0]: invalid proxy URL";
1051	$PROXY = [$2, $3 || 3128, $1]	1428	$PROXY = [$2, $3 || 3128, $1]
1052	} else {	1429	} else {
1053	undef $PROXY;	1430	undef $PROXY;
1054	}	1431	}
…		…
1057	# initialise proxy from environment	1434	# initialise proxy from environment
1058	eval {	1435	eval {
1059	set_proxy $ENV{http_proxy};	1436	set_proxy $ENV{http_proxy};
1060	};	1437	};
1061		1438
		1439	=head2 SHOWCASE
		1440
		1441	This section contains some more elaborate "real-world" examples or code
		1442	snippets.
		1443
		1444	=head2 HTTP/1.1 FILE DOWNLOAD
		1445
		1446	Downloading files with HTTP can be quite tricky, especially when something
		1447	goes wrong and you want to resume.
		1448
		1449	Here is a function that initiates and resumes a download. It uses the
		1450	last modified time to check for file content changes, and works with many
		1451	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1452	on older servers.
		1453
		1454	It calls the completion callback with either C<undef>, which means a
		1455	nonretryable error occurred, C<0> when the download was partial and should
		1456	be retried, and C<1> if it was successful.
		1457
		1458	use AnyEvent::HTTP;
		1459
		1460	sub download($$$) {
		1461	my ($url, $file, $cb) = @_;
		1462
		1463	open my $fh, "+<", $file
		1464	or die "$file: $!";
		1465
		1466	my %hdr;
		1467	my $ofs = 0;
		1468
		1469	if (stat $fh and -s _) {
		1470	$ofs = -s _;
		1471	warn "-s is ", $ofs;
		1472	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1473	$hdr{"range"} = "bytes=$ofs-";
		1474	}
		1475
		1476	http_get $url,
		1477	headers => \%hdr,
		1478	on_header => sub {
		1479	my ($hdr) = @_;
		1480
		1481	if ($hdr->{Status} == 200 && $ofs) {
		1482	# resume failed
		1483	truncate $fh, $ofs = 0;
		1484	}
		1485
		1486	sysseek $fh, $ofs, 0;
		1487
		1488	1
		1489	},
		1490	on_body => sub {
		1491	my ($data, $hdr) = @_;
		1492
		1493	if ($hdr->{Status} =~ /^2/) {
		1494	length $data == syswrite $fh, $data
		1495	or return; # abort on write errors
		1496	}
		1497
		1498	1
		1499	},
		1500	sub {
		1501	my (undef, $hdr) = @_;
		1502
		1503	my $status = $hdr->{Status};
		1504
		1505	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1506	utime $time, $time, $fh;
		1507	}
		1508
		1509	if ($status == 200 || $status == 206 || $status == 416) {
		1510	# download ok || resume ok || file already fully downloaded
		1511	$cb->(1, $hdr);
		1512
		1513	} elsif ($status == 412) {
		1514	# file has changed while resuming, delete and retry
		1515	unlink $file;
		1516	$cb->(0, $hdr);
		1517
		1518	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1519	# retry later
		1520	$cb->(0, $hdr);
		1521
		1522	} else {
		1523	$cb->(undef, $hdr);
		1524	}
		1525	}
		1526	;
		1527	}
		1528
		1529	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1530	if ($_[0]) {
		1531	print "OK!\n";
		1532	} elsif (defined $_[0]) {
		1533	print "please retry later\n";
		1534	} else {
		1535	print "ERROR\n";
		1536	}
		1537	};
		1538
1062	=head2 SOCKS PROXIES	1539	=head3 SOCKS PROXIES
1063		1540
1064	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1541	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	1542	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	1543	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1067	transparently.	1544	transparently.

Diff Legend

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