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 1.135 by root, Mon Sep 10 07:03:12 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.
		202
		203	Note that this module will prefer an existing persistent connection,
		204	even if that connection was made using another proxy. If you need to
		205	ensure that a new conneciton is made in this case, you can either force
		206	C<persistent> to false or e.g. use the proxy address in your C<sessionid>.
190		207
191	=item body => $string	208	=item body => $string
192		209
193	The request body, usually empty. Will be sent as-is (future versions of	210	The request body, usually empty. Will be sent as-is (future versions of
194	this module might offer more options).	211	this module might offer more options).
…		…
196	=item cookie_jar => $hash_ref	213	=item cookie_jar => $hash_ref
197		214
198	Passing this parameter enables (simplified) cookie-processing, loosely	215	Passing this parameter enables (simplified) cookie-processing, loosely
199	based on the original netscape specification.	216	based on the original netscape specification.
200		217
201	The C<$hash_ref> must be an (initially empty) hash reference which will	218	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	219	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	220	to persistent storage with something like JSON or Storable - see the
204	recommended, as session-only cookies might survive longer than expected.	221	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		222	expired or session-only cookies, and also for documentation on the format
		223	of the cookie jar.
205		224
206	Note that this cookie implementation is not meant to be complete. If	225	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	226	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	227	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	228	working. Cookies are a privacy disaster, do not use them unless required
210	to.	229	to.
211		230
212	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>	231	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	232	headers will be set and handled by this module, otherwise they will be
…		…
223	verification) TLS context.	242	verification) TLS context.
224		243
225	The default for this option is C<low>, which could be interpreted as "give	244	The default for this option is C<low>, which could be interpreted as "give
226	me the page, no matter what".	245	me the page, no matter what".
227		246
		247	See also the C<sessionid> parameter.
		248
		249	=item sessionid => $string
		250
		251	The module might reuse connections to the same host internally (regardless
		252	of other settings, such as C<tcp_connect> or C<proxy>). Sometimes (e.g.
		253	when using TLS or a specfic proxy), you do not want to reuse connections
		254	from other sessions. This can be achieved by setting this parameter to
		255	some unique ID (such as the address of an object storing your state data
		256	or the TLS context, or the proxy IP) - only connections using the same
		257	unique ID will be reused.
		258
228	=item on_prepare => $callback->($fh)	259	=item on_prepare => $callback->($fh)
229		260
230	In rare cases you need to "tune" the socket before it is used to	261	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	262	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>	263	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	264	and behaves exactly the same way (e.g. it has to provide a
234	timeout). See the description for the C<$prepare_cb> argument of	265	timeout). See the description for the C<$prepare_cb> argument of
235	C<AnyEvent::Socket::tcp_connect> for details.	266	C<AnyEvent::Socket::tcp_connect> for details.
236		267
…		…
239	In even rarer cases you want total control over how AnyEvent::HTTP	270	In even rarer cases you want total control over how AnyEvent::HTTP
240	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>	271	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
241	to do this, but you can provide your own C<tcp_connect> function -	272	to do this, but you can provide your own C<tcp_connect> function -
242	obviously, it has to follow the same calling conventions, except that it	273	obviously, it has to follow the same calling conventions, except that it
243	may always return a connection guard object.	274	may always return a connection guard object.
		275
		276	The connections made by this hook will be treated as equivalent to
		277	connecitons made the built-in way, specifically, they will be put into
		278	and taken from the persistent conneciton cache. If your C<$tcp_connect>
		279	function is incompatible with this kind of re-use, consider switching off
		280	C<persistent> connections and/or providing a C<sessionid> identifier.
244		281
245	There are probably lots of weird uses for this function, starting from	282	There are probably lots of weird uses for this function, starting from
246	tracing the hosts C<http_request> actually tries to connect, to (inexact	283	tracing the hosts C<http_request> actually tries to connect, to (inexact
247	but fast) host => IP address caching or even socks protocol support.	284	but fast) host => IP address caching or even socks protocol support.
248		285
…		…
301	called. Instead of the C<$body> argument containing the body data, the	338	called. Instead of the C<$body> argument containing the body data, the
302	callback will receive the L<AnyEvent::Handle> object associated with the	339	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	340	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.	341	(e.g. status C<304>), the empty string will be passed.
305		342
306	The handle object might or might not be in TLS mode, might be connected to	343	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	344	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	345	etc., and configured in unspecified ways. The user is responsible for this
309	module anymore).	346	handle (it will not be used by this module anymore).
310		347
311	This is useful with some push-type services, where, after the initial	348	This is useful with some push-type services, where, after the initial
312	headers, an interactive protocol is used (typical example would be the	349	headers, an interactive protocol is used (typical example would be the
313	push-style twitter API which starts a JSON/XML stream).	350	push-style twitter API which starts a JSON/XML stream).
314		351
315	If you think you need this, first have a look at C<on_body>, to see if	352	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.	353	that doesn't solve your problem in a better way.
		354
		355	=item persistent => $boolean
		356
		357	Try to create/reuse a persistent connection. When this flag is set
		358	(default: true for idempotent requests, false for all others), then
		359	C<http_request> tries to re-use an existing (previously-created)
		360	persistent connection to same host (i.e. identical URL scheme, hostname,
		361	port and sessionid) and, failing that, tries to create a new one.
		362
		363	Requests failing in certain ways will be automatically retried once, which
		364	is dangerous for non-idempotent requests, which is why it defaults to off
		365	for them. The reason for this is because the bozos who designed HTTP/1.1
		366	made it impossible to distinguish between a fatal error and a normal
		367	connection timeout, so you never know whether there was a problem with
		368	your request or not.
		369
		370	When reusing an existent connection, many parameters (such as TLS context)
		371	will be ignored. See the C<sessionid> parameter for a workaround.
		372
		373	=item keepalive => $boolean
		374
		375	Only used when C<persistent> is also true. This parameter decides whether
		376	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		377	(as opposed to only a HTTP/1.1 persistent connection).
		378
		379	The default is true, except when using a proxy, in which case it defaults
		380	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		381
		382	=item handle_params => { key => value ... }
		383
		384	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		385	constructor that is called - not all requests will create a handle, and
		386	sometimes more than one is created, so this parameter is only good for
		387	setting hints.
		388
		389	Example: set the maximum read size to 4096, to potentially conserve memory
		390	at the cost of speed.
		391
		392	handle_params => {
		393	max_read_size => 4096,
		394	},
317		395
318	=back	396	=back
319		397
320	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	398	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
321	the response body.	399	the response body.
…		…
327		405
328	Example: do a HTTP HEAD request on https://www.google.com/, use a	406	Example: do a HTTP HEAD request on https://www.google.com/, use a
329	timeout of 30 seconds.	407	timeout of 30 seconds.
330		408
331	http_request	409	http_request
332	GET => "https://www.google.com",	410	HEAD => "https://www.google.com",
		411	headers => { "user-agent" => "MySearchClient 1.0" },
333	timeout => 30,	412	timeout => 30,
334	sub {	413	sub {
335	my ($body, $hdr) = @_;	414	my ($body, $hdr) = @_;
336	use Data::Dumper;	415	use Data::Dumper;
337	print Dumper $hdr;	416	print Dumper $hdr;
…		…
347	};	426	};
348		427
349	undef $request;	428	undef $request;
350		429
351	=cut	430	=cut
		431
		432	#############################################################################
		433	# wait queue/slots
352		434
353	sub _slot_schedule;	435	sub _slot_schedule;
354	sub _slot_schedule($) {	436	sub _slot_schedule($) {
355	my $host = shift;	437	my $host = shift;
356		438
…		…
378	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];	460	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
379		461
380	_slot_schedule $_[0];	462	_slot_schedule $_[0];
381	}	463	}
382		464
		465	#############################################################################
		466	# cookie handling
		467
		468	# expire cookies
		469	sub cookie_jar_expire($;$) {
		470	my ($jar, $session_end) = @_;
		471
		472	%$jar = () if $jar->{version} != 2;
		473
		474	my $anow = AE::now;
		475
		476	while (my ($chost, $paths) = each %$jar) {
		477	next unless ref $paths;
		478
		479	while (my ($cpath, $cookies) = each %$paths) {
		480	while (my ($cookie, $kv) = each %$cookies) {
		481	if (exists $kv->{_expires}) {
		482	delete $cookies->{$cookie}
		483	if $anow > $kv->{_expires};
		484	} elsif ($session_end) {
		485	delete $cookies->{$cookie};
		486	}
		487	}
		488
		489	delete $paths->{$cpath}
		490	unless %$cookies;
		491	}
		492
		493	delete $jar->{$chost}
		494	unless %$paths;
		495	}
		496	}
		497
383	# extract cookies from jar	498	# extract cookies from jar
384	sub cookie_jar_extract($$$$) {	499	sub cookie_jar_extract($$$$) {
385	my ($jar, $uscheme, $uhost, $upath) = @_;	500	my ($jar, $scheme, $host, $path) = @_;
386		501
387	%$jar = () if $jar->{version} != 1;	502	%$jar = () if $jar->{version} != 2;
		503
		504	$host = AnyEvent::Util::idn_to_ascii $host
		505	if $host =~ /[^\x00-\x7f]/;
388		506
389	my @cookies;	507	my @cookies;
390		508
391	while (my ($chost, $paths) = each %$jar) {	509	while (my ($chost, $paths) = each %$jar) {
392	next unless ref $paths;	510	next unless ref $paths;
393		511
394	if ($chost =~ /^\./) {	512	# exact match or suffix including . match
395	next unless $chost eq substr $uhost, -length $chost;	513	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
396	} elsif ($chost =~ /\./) {
397	next unless $chost eq $uhost;
398	} else {
399	next;	514	or next;
400	}
401		515
402	while (my ($cpath, $cookies) = each %$paths) {	516	while (my ($cpath, $cookies) = each %$paths) {
403	next unless $cpath eq substr $upath, 0, length $cpath;	517	next unless $cpath eq substr $path, 0, length $cpath;
404		518
405	while (my ($cookie, $kv) = each %$cookies) {	519	while (my ($cookie, $kv) = each %$cookies) {
406	next if $uscheme ne "https" && exists $kv->{secure};	520	next if $scheme ne "https" && exists $kv->{secure};
407		521
408	if (exists $kv->{expires}) {	522	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
409	if (AE::now > parse_date ($kv->{expires})) {
410	delete $cookies->{$cookie};	523	delete $cookies->{$cookie};
411	next;	524	next;
412	}
413	}	525	}
414		526
415	my $value = $kv->{value};	527	my $value = $kv->{value};
416		528
417	if ($value =~ /[=;,[:space:]]/) {	529	if ($value =~ /[=;,[:space:]]/) {
…		…
426		538
427	\@cookies	539	\@cookies
428	}	540	}
429		541
430	# parse set_cookie header into jar	542	# parse set_cookie header into jar
431	sub cookie_jar_set_cookie($$$) {	543	sub cookie_jar_set_cookie($$$$) {
432	my ($jar, $set_cookie, $uhost) = @_;	544	my ($jar, $set_cookie, $host, $date) = @_;
		545
		546	%$jar = () if $jar->{version} != 2;
		547
		548	my $anow = int AE::now;
		549	my $snow; # server-now
433		550
434	for ($set_cookie) {	551	for ($set_cookie) {
435	# parse NAME=VALUE	552	# parse NAME=VALUE
436	my @kv;	553	my @kv;
437		554
…		…
440	while (	557	while (
441	m{	558	m{
442	\G\s*	559	\G\s*
443	(?:	560	(?:
444	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)	561	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
445	| ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )	562	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
446	)	563	)
447	}gcxsi	564	}gcxsi
448	) {	565	) {
449	my $name = $2;	566	my $name = $2;
450	my $value = $4;	567	my $value = $4;
451		568
452	unless (defined $name) {	569	if (defined $1) {
453	# expires	570	# expires
454	$name = "expires";	571	$name = "expires";
455	$value = $1;	572	$value = $1;
456	} elsif (!defined $value) {	573	} elsif (defined $3) {
457	# quoted	574	# quoted
458	$value = $3;	575	$value = $3;
459	$value =~ s/\\(.)/$1/gs;	576	$value =~ s/\\(.)/$1/gs;
460	}	577	}
461		578
462	push @kv, lc $name, $value;	579	push @kv, @kv ? lc $name : $name, $value;
463		580
464	last unless /\G\s*;/gc;	581	last unless /\G\s*;/gc;
465	}	582	}
466		583
467	last unless @kv;	584	last unless @kv;
468		585
469	my $name = shift @kv;	586	my $name = shift @kv;
470	my %kv = (value => shift @kv, @kv);	587	my %kv = (value => shift @kv, @kv);
471		588
472	$kv{expires} ||= format_date (AE::now + $kv{"max-age"})
473	if exists $kv{"max-age"};	589	if (exists $kv{"max-age"}) {
		590	$kv{_expires} = $anow + delete $kv{"max-age"};
		591	} elsif (exists $kv{expires}) {
		592	$snow ||= parse_date ($date) || $anow;
		593	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		594	} else {
		595	delete $kv{_expires};
		596	}
474		597
475	my $cdom;	598	my $cdom;
476	my $cpath = (delete $kv{path}) || "/";	599	my $cpath = (delete $kv{path}) || "/";
477		600
478	if (exists $kv{domain}) {	601	if (exists $kv{domain}) {
479	$cdom = delete $kv{domain};	602	$cdom = $kv{domain};
480		603
481	$cdom =~ s/^\.?/./; # make sure it starts with a "."	604	$cdom =~ s/^\.?/./; # make sure it starts with a "."
482		605
483	next if $cdom =~ /\.$/;	606	next if $cdom =~ /\.$/;
484		607
485	# this is not rfc-like and not netscape-like. go figure.	608	# this is not rfc-like and not netscape-like. go figure.
486	my $ndots = $cdom =~ y/.//;	609	my $ndots = $cdom =~ y/.//;
487	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);	610	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		611
		612	$cdom = substr $cdom, 1; # remove initial .
488	} else {	613	} else {
489	$cdom = $uhost;	614	$cdom = $host;
490	}	615	}
491		616
492	# store it	617	# store it
493	$jar->{version} = 1;	618	$jar->{version} = 2;
494	$jar->{$cdom}{$cpath}{$name} = \%kv;	619	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
495		620
496	redo if /\G\s*,/gc;	621	redo if /\G\s*,/gc;
497	}	622	}
498	}	623	}
499		624
		625	#############################################################################
		626	# keepalive/persistent connection cache
		627
		628	# fetch a connection from the keepalive cache
		629	sub ka_fetch($) {
		630	my $ka_key = shift;
		631
		632	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		633	delete $KA_CACHE{$ka_key}
		634	unless @{ $KA_CACHE{$ka_key} };
		635
		636	$hdl
		637	}
		638
		639	sub ka_store($$) {
		640	my ($ka_key, $hdl) = @_;
		641
		642	my $kaa = $KA_CACHE{$ka_key} ||= [];
		643
		644	my $destroy = sub {
		645	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		646
		647	$hdl->destroy;
		648
		649	@ka
		650	? $KA_CACHE{$ka_key} = \@ka
		651	: delete $KA_CACHE{$ka_key};
		652	};
		653
		654	# on error etc., destroy
		655	$hdl->on_error ($destroy);
		656	$hdl->on_eof ($destroy);
		657	$hdl->on_read ($destroy);
		658	$hdl->timeout ($PERSISTENT_TIMEOUT);
		659
		660	push @$kaa, $hdl;
		661	shift @$kaa while @$kaa > $MAX_PER_HOST;
		662	}
		663
		664	#############################################################################
		665	# utilities
		666
500	# continue to parse $_ for headers and place them into the arg	667	# continue to parse $_ for headers and place them into the arg
501	sub parse_hdr() {	668	sub _parse_hdr() {
502	my %hdr;	669	my %hdr;
503		670
504	# things seen, not parsed:	671	# things seen, not parsed:
505	# p3pP="NON CUR OTPi OUR NOR UNI"	672	# p3pP="NON CUR OTPi OUR NOR UNI"
506		673
…		…
520	for values %hdr;	687	for values %hdr;
521		688
522	\%hdr	689	\%hdr
523	}	690	}
524		691
		692	#############################################################################
		693	# http_get
		694
525	our $qr_nlnl = qr{(?<![^\012])\015?\012};	695	our $qr_nlnl = qr{(?<![^\012])\015?\012};
526		696
527	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	697	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
528	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };	698	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		699
		700	# maybe it should just become a normal object :/
		701
		702	sub _destroy_state(\%) {
		703	my ($state) = @_;
		704
		705	$state->{handle}->destroy if $state->{handle};
		706	%$state = ();
		707	}
		708
		709	sub _error(\%$$) {
		710	my ($state, $cb, $hdr) = @_;
		711
		712	&_destroy_state ($state);
		713
		714	$cb->(undef, $hdr);
		715	()
		716	}
		717
		718	our %IDEMPOTENT = (
		719	DELETE => 1,
		720	GET => 1,
		721	HEAD => 1,
		722	OPTIONS => 1,
		723	PUT => 1,
		724	TRACE => 1,
		725
		726	ACL => 1,
		727	"BASELINE-CONTROL" => 1,
		728	BIND => 1,
		729	CHECKIN => 1,
		730	CHECKOUT => 1,
		731	COPY => 1,
		732	LABEL => 1,
		733	LINK => 1,
		734	MERGE => 1,
		735	MKACTIVITY => 1,
		736	MKCALENDAR => 1,
		737	MKCOL => 1,
		738	MKREDIRECTREF => 1,
		739	MKWORKSPACE => 1,
		740	MOVE => 1,
		741	ORDERPATCH => 1,
		742	PROPFIND => 1,
		743	PROPPATCH => 1,
		744	REBIND => 1,
		745	REPORT => 1,
		746	SEARCH => 1,
		747	UNBIND => 1,
		748	UNCHECKOUT => 1,
		749	UNLINK => 1,
		750	UNLOCK => 1,
		751	UPDATE => 1,
		752	UPDATEREDIRECTREF => 1,
		753	"VERSION-CONTROL" => 1,
		754	);
529		755
530	sub http_request($$@) {	756	sub http_request($$@) {
531	my $cb = pop;	757	my $cb = pop;
532	my ($method, $url, %arg) = @_;	758	my ($method, $url, %arg) = @_;
533		759
…		…
551	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	777	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
552		778
553	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	779	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
554	if $recurse < 0;	780	if $recurse < 0;
555		781
556	my $proxy = $arg{proxy} || $PROXY;	782	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
557	my $timeout = $arg{timeout} || $TIMEOUT;	783	my $timeout = $arg{timeout} || $TIMEOUT;
558		784
559	my ($uscheme, $uauthority, $upath, $query, $fragment) =	785	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
560	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	786	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
561		787
562	$uscheme = lc $uscheme;	788	$uscheme = lc $uscheme;
563		789
564	my $uport = $uscheme eq "http" ? 80	790	my $uport = $uscheme eq "http" ? 80
565	: $uscheme eq "https" ? 443	791	: $uscheme eq "https" ? 443
566	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });	792	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
567		793
568	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	794	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
569	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	795	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
570		796
571	my $uhost = $1;	797	my $uhost = lc $1;
572	$uport = $2 if defined $2;	798	$uport = $2 if defined $2;
573		799
574	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	800	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
575	unless exists $hdr{host};	801	unless exists $hdr{host};
576		802
…		…
595	$rscheme = "http" unless defined $rscheme;	821	$rscheme = "http" unless defined $rscheme;
596		822
597	# don't support https requests over https-proxy transport,	823	# don't support https requests over https-proxy transport,
598	# can't be done with tls as spec'ed, unless you double-encrypt.	824	# can't be done with tls as spec'ed, unless you double-encrypt.
599	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";	825	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		826
		827	$rhost = lc $rhost;
		828	$rscheme = lc $rscheme;
600	} else {	829	} else {
601	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	830	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
602	}	831	}
603		832
604	# leave out fragment and query string, just a heuristic	833	# leave out fragment and query string, just a heuristic
…		…
606	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	835	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
607		836
608	$hdr{"content-length"} = length $arg{body}	837	$hdr{"content-length"} = length $arg{body}
609	if length $arg{body} || $method ne "GET";	838	if length $arg{body} || $method ne "GET";
610		839
611	$hdr{connection} = "close TE"; #1.1	840	my $idempotent = $IDEMPOTENT{$method};
		841
		842	# default value for keepalive is true iff the request is for an idempotent method
		843	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		844	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		845	my $was_persistent; # true if this is actually a recycled connection
		846
		847	# the key to use in the keepalive cache
		848	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		849
		850	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
612	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	851	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
613		852
614	my %state = (connect_guard => 1);	853	my %state = (connect_guard => 1);
		854
		855	my $ae_error = 595; # connecting
		856
		857	# handle actual, non-tunneled, request
		858	my $handle_actual_request = sub {
		859	$ae_error = 596; # request phase
		860
		861	my $hdl = $state{handle};
		862
		863	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		864
		865	# send request
		866	$hdl->push_write (
		867	"$method $rpath HTTP/1.1\015\012"
		868	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		869	. "\015\012"
		870	. $arg{body}
		871	);
		872
		873	# return if error occurred during push_write()
		874	return unless %state;
		875
		876	# reduce memory usage, save a kitten, also re-use it for the response headers.
		877	%hdr = ();
		878
		879	# status line and headers
		880	$state{read_response} = sub {
		881	return unless %state;
		882
		883	for ("$_[1]") {
		884	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		885
		886	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		887	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		888
		889	# 100 Continue handling
		890	# should not happen as we don't send expect: 100-continue,
		891	# but we handle it just in case.
		892	# since we send the request body regardless, if we get an error
		893	# we are out of-sync, which we currently do NOT handle correctly.
		894	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		895	if $2 eq 100;
		896
		897	push @pseudo,
		898	HTTPVersion => $1,
		899	Status => $2,
		900	Reason => $3,
		901	;
		902
		903	my $hdr = _parse_hdr
		904	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		905
		906	%hdr = (%$hdr, @pseudo);
		907	}
		908
		909	# redirect handling
		910	# relative uri handling forced by microsoft and other shitheads.
		911	# we give our best and fall back to URI if available.
		912	if (exists $hdr{location}) {
		913	my $loc = $hdr{location};
		914
		915	if ($loc =~ m%^//%) { # //
		916	$loc = "$uscheme:$loc";
		917
		918	} elsif ($loc eq "") {
		919	$loc = $url;
		920
		921	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		922	$loc =~ s/^\.\/+//;
		923
		924	if ($loc !~ m%^[.?#]%) {
		925	my $prefix = "$uscheme://$uauthority";
		926
		927	unless ($loc =~ s/^\///) {
		928	$prefix .= $upath;
		929	$prefix =~ s/\/[^\/]*$//;
		930	}
		931
		932	$loc = "$prefix/$loc";
		933
		934	} elsif (eval { require URI }) { # uri
		935	$loc = URI->new_abs ($loc, $url)->as_string;
		936
		937	} else {
		938	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		939	#$hdr{Status} = 599;
		940	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		941	#$recurse = 0;
		942	}
		943	}
		944
		945	$hdr{location} = $loc;
		946	}
		947
		948	my $redirect;
		949
		950	if ($recurse) {
		951	my $status = $hdr{Status};
		952
		953	# industry standard is to redirect POST as GET for
		954	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		955	# also, the UA should ask the user for 301 and 307 and POST,
		956	# industry standard seems to be to simply follow.
		957	# we go with the industry standard. 308 is defined
		958	# by rfc7538
		959	if ($status == 301 or $status == 302 or $status == 303) {
		960	$redirect = 1;
		961	# HTTP/1.1 is unclear on how to mutate the method
		962	unless ($method eq "HEAD") {
		963	$method = "GET";
		964	delete $arg{body};
		965	}
		966	} elsif ($status == 307 or $status == 308) {
		967	$redirect = 1;
		968	}
		969	}
		970
		971	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		972	if ($state{handle}) {
		973	# handle keepalive
		974	if (
		975	$persistent
		976	&& $_[3]
		977	&& ($hdr{HTTPVersion} < 1.1
		978	? $hdr{connection} =~ /\bkeep-?alive\b/i
		979	: $hdr{connection} !~ /\bclose\b/i)
		980	) {
		981	ka_store $ka_key, delete $state{handle};
		982	} else {
		983	# no keepalive, destroy the handle
		984	$state{handle}->destroy;
		985	}
		986	}
		987
		988	%state = ();
		989
		990	if (defined $_[1]) {
		991	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		992	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		993	}
		994
		995	# set-cookie processing
		996	if ($arg{cookie_jar}) {
		997	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		998	}
		999
		1000	if ($redirect && exists $hdr{location}) {
		1001	# we ignore any errors, as it is very common to receive
		1002	# Content-Length != 0 but no actual body
		1003	# we also access %hdr, as $_[1] might be an erro
		1004	$state{recurse} =
		1005	http_request (
		1006	$method => $hdr{location},
		1007	%arg,
		1008	recurse => $recurse - 1,
		1009	Redirect => [$_[0], \%hdr],
		1010	sub {
		1011	%state = ();
		1012	&$cb
		1013	},
		1014	);
		1015	} else {
		1016	$cb->($_[0], \%hdr);
		1017	}
		1018	};
		1019
		1020	$ae_error = 597; # body phase
		1021
		1022	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1023
		1024	my $len = $chunked ? undef : $hdr{"content-length"};
		1025
		1026	# body handling, many different code paths
		1027	# - no body expected
		1028	# - want_body_handle
		1029	# - te chunked
		1030	# - 2x length known (with or without on_body)
		1031	# - 2x length not known (with or without on_body)
		1032	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1033	$finish->(undef, 598 => "Request cancelled by on_header");
		1034	} elsif (
		1035	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1036	or $method eq "HEAD"
		1037	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1038	) {
		1039	# no body
		1040	$finish->("", undef, undef, 1);
		1041
		1042	} elsif (!$redirect && $arg{want_body_handle}) {
		1043	$_[0]->on_eof (undef);
		1044	$_[0]->on_error (undef);
		1045	$_[0]->on_read (undef);
		1046
		1047	$finish->(delete $state{handle});
		1048
		1049	} elsif ($chunked) {
		1050	my $cl = 0;
		1051	my $body = "";
		1052	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1053
		1054	$state{read_chunk} = sub {
		1055	$_[1] =~ /^([0-9a-fA-F]+)/
		1056	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1057
		1058	my $len = hex $1;
		1059
		1060	if ($len) {
		1061	$cl += $len;
		1062
		1063	$_[0]->push_read (chunk => $len, sub {
		1064	$on_body->($_[1], \%hdr)
		1065	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1066
		1067	$_[0]->push_read (line => sub {
		1068	length $_[1]
		1069	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1070	$_[0]->push_read (line => $state{read_chunk});
		1071	});
		1072	});
		1073	} else {
		1074	$hdr{"content-length"} ||= $cl;
		1075
		1076	$_[0]->push_read (line => $qr_nlnl, sub {
		1077	if (length $_[1]) {
		1078	for ("$_[1]") {
		1079	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1080
		1081	my $hdr = _parse_hdr
		1082	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1083
		1084	%hdr = (%hdr, %$hdr);
		1085	}
		1086	}
		1087
		1088	$finish->($body, undef, undef, 1);
		1089	});
		1090	}
		1091	};
		1092
		1093	$_[0]->push_read (line => $state{read_chunk});
		1094
		1095	} elsif (!$redirect && $arg{on_body}) {
		1096	if (defined $len) {
		1097	$_[0]->on_read (sub {
		1098	$len -= length $_[0]{rbuf};
		1099
		1100	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1101	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1102
		1103	$len > 0
		1104	or $finish->("", undef, undef, 1);
		1105	});
		1106	} else {
		1107	$_[0]->on_eof (sub {
		1108	$finish->("");
		1109	});
		1110	$_[0]->on_read (sub {
		1111	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1112	or $finish->(undef, 598 => "Request cancelled by on_body");
		1113	});
		1114	}
		1115	} else {
		1116	$_[0]->on_eof (undef);
		1117
		1118	if (defined $len) {
		1119	$_[0]->on_read (sub {
		1120	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1121	if $len <= length $_[0]{rbuf};
		1122	});
		1123	} else {
		1124	$_[0]->on_error (sub {
		1125	($! == Errno::EPIPE || !$!)
		1126	? $finish->(delete $_[0]{rbuf})
		1127	: $finish->(undef, $ae_error => $_[2]);
		1128	});
		1129	$_[0]->on_read (sub { });
		1130	}
		1131	}
		1132	};
		1133
		1134	# if keepalive is enabled, then the server closing the connection
		1135	# before a response can happen legally - we retry on idempotent methods.
		1136	if ($was_persistent && $idempotent) {
		1137	my $old_eof = $hdl->{on_eof};
		1138	$hdl->{on_eof} = sub {
		1139	_destroy_state %state;
		1140
		1141	%state = ();
		1142	$state{recurse} =
		1143	http_request (
		1144	$method => $url,
		1145	%arg,
		1146	recurse => $recurse - 1,
		1147	persistent => 0,
		1148	sub {
		1149	%state = ();
		1150	&$cb
		1151	}
		1152	);
		1153	};
		1154	$hdl->on_read (sub {
		1155	return unless %state;
		1156
		1157	# as soon as we receive something, a connection close
		1158	# once more becomes a hard error
		1159	$hdl->{on_eof} = $old_eof;
		1160	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1161	});
		1162	} else {
		1163	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1164	}
		1165	};
		1166
		1167	my $prepare_handle = sub {
		1168	my ($hdl) = $state{handle};
		1169
		1170	$hdl->on_error (sub {
		1171	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1172	});
		1173	$hdl->on_eof (sub {
		1174	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1175	});
		1176	$hdl->timeout_reset;
		1177	$hdl->timeout ($timeout);
		1178	};
		1179
		1180	# connected to proxy (or origin server)
		1181	my $connect_cb = sub {
		1182	my $fh = shift
		1183	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1184
		1185	return unless delete $state{connect_guard};
		1186
		1187	# get handle
		1188	$state{handle} = new AnyEvent::Handle
		1189	%{ $arg{handle_params} },
		1190	fh => $fh,
		1191	peername => $uhost,
		1192	tls_ctx => $arg{tls_ctx},
		1193	;
		1194
		1195	$prepare_handle->();
		1196
		1197	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1198
		1199	# now handle proxy-CONNECT method
		1200	if ($proxy && $uscheme eq "https") {
		1201	# oh dear, we have to wrap it into a connect request
		1202
		1203	my $auth = exists $hdr{"proxy-authorization"}
		1204	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1205	: "";
		1206
		1207	# maybe re-use $uauthority with patched port?
		1208	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
		1209	$state{handle}->push_read (line => $qr_nlnl, sub {
		1210	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1211	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1212
		1213	if ($2 == 200) {
		1214	$rpath = $upath;
		1215	$handle_actual_request->();
		1216	} else {
		1217	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1218	}
		1219	});
		1220	} else {
		1221	delete $hdr{"proxy-authorization"} unless $proxy;
		1222
		1223	$handle_actual_request->();
		1224	}
		1225	};
615		1226
616	_get_slot $uhost, sub {	1227	_get_slot $uhost, sub {
617	$state{slot_guard} = shift;	1228	$state{slot_guard} = shift;
618		1229
619	return unless $state{connect_guard};	1230	return unless $state{connect_guard};
620		1231
621	my $ae_error = 595; # connecting	1232	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1233	# on a keepalive request (in theory, this should be a separate config option).
		1234	if ($persistent && $KA_CACHE{$ka_key}) {
		1235	$was_persistent = 1;
622		1236
623	my $connect_cb = sub {	1237	$state{handle} = ka_fetch $ka_key;
624	$state{fh} = shift	1238	# $state{handle}->destroyed
625	or do {	1239	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
626	my $err = "$!";	1240	$prepare_handle->();
627	%state = ();	1241	# $state{handle}->destroyed
628	return $cb->(undef, { @pseudo, Status => $ae_error, Reason => $err });	1242	# 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;	1243	$handle_actual_request->();
904	} else {	1244
905	%state = ();
906	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });
907	}
908	});
909	} else {	1245	} else {
910	&$handle_actual_request;
911	}
912	};
913
914	my $tcp_connect = $arg{tcp_connect}	1246	my $tcp_connect = $arg{tcp_connect}
915	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	1247	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
916		1248
917	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	1249	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
918		1250	}
919	};	1251	};
920		1252
921	defined wantarray && AnyEvent::Util::guard { %state = () }	1253	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
922	}	1254	}
923		1255
924	sub http_get($@) {	1256	sub http_get($@) {
925	unshift @_, "GET";	1257	unshift @_, "GET";
926	&http_request	1258	&http_request
…		…
944	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1276	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
945	the actual connection, which in turn uses AnyEvent::DNS to resolve	1277	the actual connection, which in turn uses AnyEvent::DNS to resolve
946	hostnames. The latter is a simple stub resolver and does no caching	1278	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	1279	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	1280	your own default resolver (by storing a suitable resolver object in
949	C<$AnyEvent::DNS::RESOLVER>).	1281	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
950		1282
951	=head2 GLOBAL FUNCTIONS AND VARIABLES	1283	=head2 GLOBAL FUNCTIONS AND VARIABLES
952		1284
953	=over 4	1285	=over 4
954		1286
955	=item AnyEvent::HTTP::set_proxy "proxy-url"	1287	=item AnyEvent::HTTP::set_proxy "proxy-url"
956		1288
957	Sets the default proxy server to use. The proxy-url must begin with a	1289	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	1290	string of the form C<http://host:port>, croaks otherwise.
959	otherwise.
960		1291
961	To clear an already-set proxy, use C<undef>.	1292	To clear an already-set proxy, use C<undef>.
		1293
		1294	When AnyEvent::HTTP is loaded for the first time it will query the
		1295	default proxy from the operating system, currently by looking at
		1296	C<$ENV{http_proxy>}.
		1297
		1298	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1299
		1300	Remove all cookies from the cookie jar that have been expired. If
		1301	C<$session_end> is given and true, then additionally remove all session
		1302	cookies.
		1303
		1304	You should call this function (with a true C<$session_end>) before you
		1305	save cookies to disk, and you should call this function after loading them
		1306	again. If you have a long-running program you can additionally call this
		1307	function from time to time.
		1308
		1309	A cookie jar is initially an empty hash-reference that is managed by this
		1310	module. Its format is subject to change, but currently it is as follows:
		1311
		1312	The key C<version> has to contain C<2>, otherwise the hash gets
		1313	cleared. All other keys are hostnames or IP addresses pointing to
		1314	hash-references. The key for these inner hash references is the
		1315	server path for which this cookie is meant, and the values are again
		1316	hash-references. Each key of those hash-references is a cookie name, and
		1317	the value, you guessed it, is another hash-reference, this time with the
		1318	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1319	which have been replaced by a C<_expires> key that contains the cookie
		1320	expiry timestamp. Session cookies are indicated by not having an
		1321	C<_expires> key.
		1322
		1323	Here is an example of a cookie jar with a single cookie, so you have a
		1324	chance of understanding the above paragraph:
		1325
		1326	{
		1327	version => 2,
		1328	"10.0.0.1" => {
		1329	"/" => {
		1330	"mythweb_id" => {
		1331	_expires => 1293917923,
		1332	value => "ooRung9dThee3ooyXooM1Ohm",
		1333	},
		1334	},
		1335	},
		1336	}
962		1337
963	=item $date = AnyEvent::HTTP::format_date $timestamp	1338	=item $date = AnyEvent::HTTP::format_date $timestamp
964		1339
965	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP	1340	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
966	Date (RFC 2616).	1341	Date (RFC 2616).
…		…
973		1348
974	=item $AnyEvent::HTTP::MAX_RECURSE	1349	=item $AnyEvent::HTTP::MAX_RECURSE
975		1350
976	The default value for the C<recurse> request parameter (default: C<10>).	1351	The default value for the C<recurse> request parameter (default: C<10>).
977		1352
		1353	=item $AnyEvent::HTTP::TIMEOUT
		1354
		1355	The default timeout for connection operations (default: C<300>).
		1356
978	=item $AnyEvent::HTTP::USERAGENT	1357	=item $AnyEvent::HTTP::USERAGENT
979		1358
980	The default value for the C<User-Agent> header (the default is	1359	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)>).	1360	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
982		1361
983	=item $AnyEvent::HTTP::MAX_PER_HOST	1362	=item $AnyEvent::HTTP::MAX_PER_HOST
984		1363
985	The maximum number of concurrent connections to the same host (identified	1364	The maximum number of concurrent connections to the same host (identified
986	by the hostname). If the limit is exceeded, then the additional requests	1365	by the hostname). If the limit is exceeded, then additional requests
987	are queued until previous connections are closed.	1366	are queued until previous connections are closed. Both persistent and
		1367	non-persistent connections are counted in this limit.
988		1368
989	The default value for this is C<4>, and it is highly advisable to not	1369	The default value for this is C<4>, and it is highly advisable to not
990	increase it.	1370	increase it much.
		1371
		1372	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1373	connections, older browsers used 2, newer ones (such as firefox 3)
		1374	typically use 6, and Opera uses 8 because like, they have the fastest
		1375	browser and give a shit for everybody else on the planet.
		1376
		1377	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1378
		1379	The time after which idle persistent connections get closed by
		1380	AnyEvent::HTTP (default: C<3>).
991		1381
992	=item $AnyEvent::HTTP::ACTIVE	1382	=item $AnyEvent::HTTP::ACTIVE
993		1383
994	The number of active connections. This is not the number of currently	1384	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	1385	running requests, but the number of currently open and non-idle TCP
996	connections. This number of can be useful for load-leveling.	1386	connections. This number can be useful for load-leveling.
997		1387
998	=back	1388	=back
999		1389
1000	=cut	1390	=cut
1001		1391
…		…
1035	# other formats fail in the loop below	1425	# other formats fail in the loop below
1036		1426
1037	for (0..11) {	1427	for (0..11) {
1038	if ($m eq $month[$_]) {	1428	if ($m eq $month[$_]) {
1039	require Time::Local;	1429	require Time::Local;
1040	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);	1430	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
1041	}	1431	}
1042	}	1432	}
1043		1433
1044	undef	1434	undef
1045	}	1435	}
1046		1436
1047	sub set_proxy($) {	1437	sub set_proxy($) {
1048	if (length $_[0]) {	1438	if (length $_[0]) {
1049	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1439	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1050	or Carp::croak "$_[0]: invalid proxy URL";	1440	or Carp::croak "$_[0]: invalid proxy URL";
1051	$PROXY = [$2, $3 || 3128, $1]	1441	$PROXY = [$2, $3 || 3128, $1]
1052	} else {	1442	} else {
1053	undef $PROXY;	1443	undef $PROXY;
1054	}	1444	}
…		…
1057	# initialise proxy from environment	1447	# initialise proxy from environment
1058	eval {	1448	eval {
1059	set_proxy $ENV{http_proxy};	1449	set_proxy $ENV{http_proxy};
1060	};	1450	};
1061		1451
		1452	=head2 SHOWCASE
		1453
		1454	This section contains some more elaborate "real-world" examples or code
		1455	snippets.
		1456
		1457	=head2 HTTP/1.1 FILE DOWNLOAD
		1458
		1459	Downloading files with HTTP can be quite tricky, especially when something
		1460	goes wrong and you want to resume.
		1461
		1462	Here is a function that initiates and resumes a download. It uses the
		1463	last modified time to check for file content changes, and works with many
		1464	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1465	on older servers.
		1466
		1467	It calls the completion callback with either C<undef>, which means a
		1468	nonretryable error occurred, C<0> when the download was partial and should
		1469	be retried, and C<1> if it was successful.
		1470
		1471	use AnyEvent::HTTP;
		1472
		1473	sub download($$$) {
		1474	my ($url, $file, $cb) = @_;
		1475
		1476	open my $fh, "+<", $file
		1477	or die "$file: $!";
		1478
		1479	my %hdr;
		1480	my $ofs = 0;
		1481
		1482	if (stat $fh and -s _) {
		1483	$ofs = -s _;
		1484	warn "-s is ", $ofs;
		1485	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1486	$hdr{"range"} = "bytes=$ofs-";
		1487	}
		1488
		1489	http_get $url,
		1490	headers => \%hdr,
		1491	on_header => sub {
		1492	my ($hdr) = @_;
		1493
		1494	if ($hdr->{Status} == 200 && $ofs) {
		1495	# resume failed
		1496	truncate $fh, $ofs = 0;
		1497	}
		1498
		1499	sysseek $fh, $ofs, 0;
		1500
		1501	1
		1502	},
		1503	on_body => sub {
		1504	my ($data, $hdr) = @_;
		1505
		1506	if ($hdr->{Status} =~ /^2/) {
		1507	length $data == syswrite $fh, $data
		1508	or return; # abort on write errors
		1509	}
		1510
		1511	1
		1512	},
		1513	sub {
		1514	my (undef, $hdr) = @_;
		1515
		1516	my $status = $hdr->{Status};
		1517
		1518	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1519	utime $time, $time, $fh;
		1520	}
		1521
		1522	if ($status == 200 || $status == 206 || $status == 416) {
		1523	# download ok || resume ok || file already fully downloaded
		1524	$cb->(1, $hdr);
		1525
		1526	} elsif ($status == 412) {
		1527	# file has changed while resuming, delete and retry
		1528	unlink $file;
		1529	$cb->(0, $hdr);
		1530
		1531	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1532	# retry later
		1533	$cb->(0, $hdr);
		1534
		1535	} else {
		1536	$cb->(undef, $hdr);
		1537	}
		1538	}
		1539	;
		1540	}
		1541
		1542	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1543	if ($_[0]) {
		1544	print "OK!\n";
		1545	} elsif (defined $_[0]) {
		1546	print "please retry later\n";
		1547	} else {
		1548	print "ERROR\n";
		1549	}
		1550	};
		1551
1062	=head2 SOCKS PROXIES	1552	=head3 SOCKS PROXIES
1063		1553
1064	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1554	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	1555	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	1556	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1067	transparently.	1557	transparently.

Diff Legend

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