ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.85 by root, Sun Jan 2 05:31:56 2011 UTC vs.
Revision 1.138 by root, Fri Aug 5 20:45:09 2022 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
…		…
46	use AnyEvent::Util ();	46	use AnyEvent::Util ();
47	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
48		48
49	use base Exporter::;	49	use base Exporter::;
50		50
51	our $VERSION = '1.5';	51	our $VERSION = 2.25;
52		52
53	our @EXPORT = qw(http_get http_post http_head http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
54		54
55	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)";
56	our $MAX_RECURSE = 10;	56	our $MAX_RECURSE = 10;
57	our $MAX_PERSISTENT = 8;
58	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
59	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
60		59	our $MAX_PER_HOST = 4; # changing this is evil
61	# changing these is evil
62	our $MAX_PERSISTENT_PER_HOST = 0;
63	our $MAX_PER_HOST = 4;
64		60
65	our $PROXY;	61	our $PROXY;
66	our $ACTIVE = 0;	62	our $ACTIVE = 0;
67		63
68	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
69	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
70		66
71	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
72		68
73	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
…		…
93	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
94	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
95	destroyed before the callback is called, the request will be cancelled.	91	destroyed before the callback is called, the request will be cancelled.
96		92
97	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
98	(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
99	(and trailers) as second argument.	95	(and trailers) as second argument.
100		96
101	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
102	headers, the "pseudo-headers" (uppercase to avoid clashing with possible	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
103	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the	99	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
…		…
127	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
128	message. Currently the following status codes are used:	124	message. Currently the following status codes are used:
129		125
130	=over 4	126	=over 4
131		127
132	=item 595 - errors during connection etsbalishment, proxy handshake.	128	=item 595 - errors during connection establishment, proxy handshake.
133		129
134	=item 596 - errors during TLS negotiation, request sending and header processing.	130	=item 596 - errors during TLS negotiation, request sending and header processing.
135		131
136	=item 597 - errors during body receiving or processing.	132	=item 597 - errors during body receiving or processing.
137		133
…		…
158		154
159	=over 4	155	=over 4
160		156
161	=item recurse => $count (default: $MAX_RECURSE)	157	=item recurse => $count (default: $MAX_RECURSE)
162		158
163	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
164	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.
165		166
166	=item headers => hashref	167	=item headers => hashref
167		168
168	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
169	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
170	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:>
171	(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
172	they won't be sent at all).	173	they won't be sent at all).
173		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
174	=item timeout => $seconds	182	=item timeout => $seconds
175		183
176	The time-out to use for various stages - each connect attempt will reset	184	The time-out to use for various stages - each connect attempt will reset
177	the timeout, as will read or write activity, i.e. this is not an overall	185	the timeout, as will read or write activity, i.e. this is not an overall
178	timeout.	186	timeout.
179		187
180	Default timeout is 5 minutes.	188	Default timeout is 5 minutes.
181		189
182	=item proxy => [$host, $port[, $scheme]] or undef	190	=item proxy => [$host, $port[, $scheme]] or undef
183		191
184	Use the given http proxy for all requests. If not specified, then the	192	Use the given http proxy for all requests, or no proxy if C<undef> is
185	default proxy (as specified by C<$ENV{http_proxy}>) is used.	193	used.
186		194
187	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.
188	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 connection 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>.
189		207
190	=item body => $string	208	=item body => $string
191		209
192	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
193	this module might offer more options).	211	this module might offer more options).
…		…
224	verification) TLS context.	242	verification) TLS context.
225		243
226	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
227	me the page, no matter what".	245	me the page, no matter what".
228		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
229	=item on_prepare => $callback->($fh)	259	=item on_prepare => $callback->($fh)
230		260
231	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
232	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
233	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>	263	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
234	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
235	timeout). See the description for the C<$prepare_cb> argument of	265	timeout). See the description for the C<$prepare_cb> argument of
236	C<AnyEvent::Socket::tcp_connect> for details.	266	C<AnyEvent::Socket::tcp_connect> for details.
237		267
…		…
240	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
241	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>	271	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
242	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 -
243	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
244	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	connections made the built-in way, specifically, they will be put into
		278	and taken from the persistent connection cache. If your C<$tcp_connect>
		279	function is incompatible with this kind of re-use, consider switching off
		280	C<persistent> connections and/or providing a C<sessionid> identifier.
245		281
246	There are probably lots of weird uses for this function, starting from	282	There are probably lots of weird uses for this function, starting from
247	tracing the hosts C<http_request> actually tries to connect, to (inexact	283	tracing the hosts C<http_request> actually tries to connect, to (inexact
248	but fast) host => IP address caching or even socks protocol support.	284	but fast) host => IP address caching or even socks protocol support.
249		285
…		…
302	called. Instead of the C<$body> argument containing the body data, the	338	called. Instead of the C<$body> argument containing the body data, the
303	callback will receive the L<AnyEvent::Handle> object associated with the	339	callback will receive the L<AnyEvent::Handle> object associated with the
304	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
305	(e.g. status C<304>), the empty string will be passed.	341	(e.g. status C<304>), the empty string will be passed.
306		342
307	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
308	a proxy, be a persistent connection etc., and configured in unspecified	344	to a proxy, be a persistent connection, use chunked transfer encoding
309	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
310	module anymore).	346	handle (it will not be used by this module anymore).
311		347
312	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
313	headers, an interactive protocol is used (typical example would be the	349	headers, an interactive protocol is used (typical example would be the
314	push-style twitter API which starts a JSON/XML stream).	350	push-style twitter API which starts a JSON/XML stream).
315		351
316	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
317	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 => $boolean
		374
		375	Only used when C<persistent> is also true. This parameter decides whether
		376	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		377	(as opposed to only a HTTP/1.1 persistent connection).
		378
		379	The default is true, except when using a proxy, in which case it defaults
		380	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		381
		382	=item handle_params => { key => value ... }
		383
		384	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		385	constructor that is called - not all requests will create a handle, and
		386	sometimes more than one is created, so this parameter is only good for
		387	setting hints.
		388
		389	Example: set the maximum read size to 4096, to potentially conserve memory
		390	at the cost of speed.
		391
		392	handle_params => {
		393	max_read_size => 4096,
		394	},
318		395
319	=back	396	=back
320		397
321	Example: do a simple HTTP GET request for http://www.nethype.de/ and print	398	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
322	the response body.	399	the response body.
…		…
328		405
329	Example: do a HTTP HEAD request on https://www.google.com/, use a	406	Example: do a HTTP HEAD request on https://www.google.com/, use a
330	timeout of 30 seconds.	407	timeout of 30 seconds.
331		408
332	http_request	409	http_request
333	GET => "https://www.google.com",	410	HEAD => "https://www.google.com",
		411	headers => { "user-agent" => "MySearchClient 1.0" },
334	timeout => 30,	412	timeout => 30,
335	sub {	413	sub {
336	my ($body, $hdr) = @_;	414	my ($body, $hdr) = @_;
337	use Data::Dumper;	415	use Data::Dumper;
338	print Dumper $hdr;	416	print Dumper $hdr;
…		…
348	};	426	};
349		427
350	undef $request;	428	undef $request;
351		429
352	=cut	430	=cut
		431
		432	#############################################################################
		433	# wait queue/slots
353		434
354	sub _slot_schedule;	435	sub _slot_schedule;
355	sub _slot_schedule($) {	436	sub _slot_schedule($) {
356	my $host = shift;	437	my $host = shift;
357		438
…		…
380		461
381	_slot_schedule $_[0];	462	_slot_schedule $_[0];
382	}	463	}
383		464
384	#############################################################################	465	#############################################################################
		466	# cookie handling
385		467
386	# expire cookies	468	# expire cookies
387	sub cookie_jar_expire($;$) {	469	sub cookie_jar_expire($;$) {
388	my ($jar, $session_end) = @_;	470	my ($jar, $session_end) = @_;
389		471
390	%$jar = () if $jar->{version} != 1;	472	%$jar = () if $jar->{version} != 2;
391		473
392	my $anow = AE::now;	474	my $anow = AE::now;
393		475
394	while (my ($chost, $paths) = each %$jar) {	476	while (my ($chost, $paths) = each %$jar) {
395	next unless ref $paths;	477	next unless ref $paths;
…		…
413	}	495	}
414	}	496	}
415		497
416	# extract cookies from jar	498	# extract cookies from jar
417	sub cookie_jar_extract($$$$) {	499	sub cookie_jar_extract($$$$) {
418	my ($jar, $uscheme, $uhost, $upath) = @_;	500	my ($jar, $scheme, $host, $path) = @_;
419		501
420	$uhost = lc $uhost;
421
422	%$jar = () if $jar->{version} != 1;	502	%$jar = () if $jar->{version} != 2;
		503
		504	$host = AnyEvent::Util::idn_to_ascii $host
		505	if $host =~ /[^\x00-\x7f]/;
423		506
424	my @cookies;	507	my @cookies;
425		508
426	while (my ($chost, $paths) = each %$jar) {	509	while (my ($chost, $paths) = each %$jar) {
427	next unless ref $paths;	510	next unless ref $paths;
428		511
429	if ($chost =~ /^\./) {	512	# exact match or suffix including . match
430	next unless $chost eq substr $uhost, -length $chost;	513	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
431	} elsif ($chost =~ /\./) {
432	next unless $chost eq $uhost;
433	} else {
434	next;	514	or next;
435	}
436		515
437	while (my ($cpath, $cookies) = each %$paths) {	516	while (my ($cpath, $cookies) = each %$paths) {
438	next unless $cpath eq substr $upath, 0, length $cpath;	517	next unless $cpath eq substr $path, 0, length $cpath;
439		518
440	while (my ($cookie, $kv) = each %$cookies) {	519	while (my ($cookie, $kv) = each %$cookies) {
441	next if $uscheme ne "https" && exists $kv->{secure};	520	next if $scheme ne "https" && exists $kv->{secure};
442		521
443	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {	522	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
444	delete $cookies->{$cookie};	523	delete $cookies->{$cookie};
445	next;	524	next;
446	}	525	}
…		…
460	\@cookies	539	\@cookies
461	}	540	}
462		541
463	# parse set_cookie header into jar	542	# parse set_cookie header into jar
464	sub cookie_jar_set_cookie($$$$) {	543	sub cookie_jar_set_cookie($$$$) {
465	my ($jar, $set_cookie, $uhost, $date) = @_;	544	my ($jar, $set_cookie, $host, $date) = @_;
		545
		546	%$jar = () if $jar->{version} != 2;
466		547
467	my $anow = int AE::now;	548	my $anow = int AE::now;
468	my $snow; # server-now	549	my $snow; # server-now
469		550
470	for ($set_cookie) {	551	for ($set_cookie) {
…		…
476	while (	557	while (
477	m{	558	m{
478	\G\s*	559	\G\s*
479	(?:	560	(?:
480	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)	561	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
481	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) ) )?	562	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
482	)	563	)
483	}gcxsi	564	}gcxsi
484	) {	565	) {
485	my $name = $2;	566	my $name = $2;
486	my $value = $4;	567	my $value = $4;
…		…
493	# quoted	574	# quoted
494	$value = $3;	575	$value = $3;
495	$value =~ s/\\(.)/$1/gs;	576	$value =~ s/\\(.)/$1/gs;
496	}	577	}
497		578
498	push @kv, lc $name, $value;	579	push @kv, @kv ? lc $name : $name, $value;
499		580
500	last unless /\G\s*;/gc;	581	last unless /\G\s*;/gc;
501	}	582	}
502		583
503	last unless @kv;	584	last unless @kv;
…		…
516		597
517	my $cdom;	598	my $cdom;
518	my $cpath = (delete $kv{path}) || "/";	599	my $cpath = (delete $kv{path}) || "/";
519		600
520	if (exists $kv{domain}) {	601	if (exists $kv{domain}) {
521	$cdom = delete $kv{domain};	602	$cdom = $kv{domain};
522		603
523	$cdom =~ s/^\.?/./; # make sure it starts with a "."	604	$cdom =~ s/^\.?/./; # make sure it starts with a "."
524		605
525	next if $cdom =~ /\.$/;	606	next if $cdom =~ /\.$/;
526		607
527	# this is not rfc-like and not netscape-like. go figure.	608	# this is not rfc-like and not netscape-like. go figure.
528	my $ndots = $cdom =~ y/.//;	609	my $ndots = $cdom =~ y/.//;
529	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);	610	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		611
		612	$cdom = substr $cdom, 1; # remove initial .
530	} else {	613	} else {
531	$cdom = $uhost;	614	$cdom = $host;
532	}	615	}
533		616
534	# store it	617	# store it
535	$jar->{version} = 1;	618	$jar->{version} = 2;
536	$jar->{lc $cdom}{$cpath}{$name} = \%kv;	619	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
537		620
538	redo if /\G\s*,/gc;	621	redo if /\G\s*,/gc;
539	}	622	}
540	}	623	}
541		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
542	# continue to parse $_ for headers and place them into the arg	667	# continue to parse $_ for headers and place them into the arg
543	sub parse_hdr() {	668	sub _parse_hdr() {
544	my %hdr;	669	my %hdr;
545		670
546	# things seen, not parsed:	671	# things seen, not parsed:
547	# p3pP="NON CUR OTPi OUR NOR UNI"	672	# p3pP="NON CUR OTPi OUR NOR UNI"
548		673
…		…
562	for values %hdr;	687	for values %hdr;
563		688
564	\%hdr	689	\%hdr
565	}	690	}
566		691
		692	#############################################################################
		693	# http_get
		694
567	our $qr_nlnl = qr{(?<![^\012])\015?\012};	695	our $qr_nlnl = qr{(?<![^\012])\015?\012};
568		696
569	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };	697	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
570	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(\%$$) {
		710	my ($state, $cb, $hdr) = @_;
		711
		712	&_destroy_state ($state);
		713
		714	$cb->(undef, $hdr);
		715	()
		716	}
		717
		718	our %IDEMPOTENT = (
		719	DELETE => 1,
		720	GET => 1,
		721	QUERY => 1,
		722	HEAD => 1,
		723	OPTIONS => 1,
		724	PUT => 1,
		725	TRACE => 1,
		726
		727	ACL => 1,
		728	"BASELINE-CONTROL" => 1,
		729	BIND => 1,
		730	CHECKIN => 1,
		731	CHECKOUT => 1,
		732	COPY => 1,
		733	LABEL => 1,
		734	LINK => 1,
		735	MERGE => 1,
		736	MKACTIVITY => 1,
		737	MKCALENDAR => 1,
		738	MKCOL => 1,
		739	MKREDIRECTREF => 1,
		740	MKWORKSPACE => 1,
		741	MOVE => 1,
		742	ORDERPATCH => 1,
		743	PROPFIND => 1,
		744	PROPPATCH => 1,
		745	REBIND => 1,
		746	REPORT => 1,
		747	SEARCH => 1,
		748	UNBIND => 1,
		749	UNCHECKOUT => 1,
		750	UNLINK => 1,
		751	UNLOCK => 1,
		752	UPDATE => 1,
		753	UPDATEREDIRECTREF => 1,
		754	"VERSION-CONTROL" => 1,
		755	);
571		756
572	sub http_request($$@) {	757	sub http_request($$@) {
573	my $cb = pop;	758	my $cb = pop;
574	my ($method, $url, %arg) = @_;	759	my ($method, $url, %arg) = @_;
575		760
…		…
593	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;	778	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
594		779
595	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })	780	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
596	if $recurse < 0;	781	if $recurse < 0;
597		782
598	my $proxy = $arg{proxy} || $PROXY;	783	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
599	my $timeout = $arg{timeout} || $TIMEOUT;	784	my $timeout = $arg{timeout} || $TIMEOUT;
600		785
601	my ($uscheme, $uauthority, $upath, $query, $fragment) =	786	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
602	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?|;	787	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
603		788
604	$uscheme = lc $uscheme;	789	$uscheme = lc $uscheme;
605		790
606	my $uport = $uscheme eq "http" ? 80	791	my $uport = $uscheme eq "http" ? 80
607	: $uscheme eq "https" ? 443	792	: $uscheme eq "https" ? 443
608	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });	793	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
609		794
610	$uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x	795	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
611	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });	796	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
612		797
613	my $uhost = $1;	798	my $uhost = lc $1;
614	$uport = $2 if defined $2;	799	$uport = $2 if defined $2;
615		800
616	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"	801	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
617	unless exists $hdr{host};	802	unless exists $hdr{host};
618		803
…		…
637	$rscheme = "http" unless defined $rscheme;	822	$rscheme = "http" unless defined $rscheme;
638		823
639	# don't support https requests over https-proxy transport,	824	# don't support https requests over https-proxy transport,
640	# can't be done with tls as spec'ed, unless you double-encrypt.	825	# can't be done with tls as spec'ed, unless you double-encrypt.
641	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";	826	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		827
		828	$rhost = lc $rhost;
		829	$rscheme = lc $rscheme;
642	} else {	830	} else {
643	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);	831	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
644	}	832	}
645		833
646	# leave out fragment and query string, just a heuristic	834	# leave out fragment and query string, just a heuristic
…		…
648	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};	836	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
649		837
650	$hdr{"content-length"} = length $arg{body}	838	$hdr{"content-length"} = length $arg{body}
651	if length $arg{body} || $method ne "GET";	839	if length $arg{body} || $method ne "GET";
652		840
653	$hdr{connection} = "close TE"; #1.1	841	my $idempotent = $IDEMPOTENT{$method};
		842
		843	# default value for keepalive is true iff the request is for an idempotent method
		844	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		845	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		846	my $was_persistent; # true if this is actually a recycled connection
		847
		848	# the key to use in the keepalive cache
		849	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		850
		851	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
654	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1	852	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
655		853
656	my %state = (connect_guard => 1);	854	my %state = (connect_guard => 1);
657		855
658	_get_slot $uhost, sub {
659	$state{slot_guard} = shift;
660
661	return unless $state{connect_guard};
662
663	my $ae_error = 595; # connecting	856	my $ae_error = 595; # connecting
664		857
665	# handle actual, non-tunneled, request	858	# handle actual, non-tunneled, request
666	my $handle_actual_request = sub {	859	my $handle_actual_request = sub {
667	$ae_error = 596; # request phase	860	$ae_error = 596; # request phase
668		861
		862	my $hdl = $state{handle};
		863
669	$state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};	864	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
670		865
671	# send request	866	# send request
672	$state{handle}->push_write (	867	$hdl->push_write (
673	"$method $rpath HTTP/1.1\015\012"	868	"$method $rpath HTTP/1.1\015\012"
674	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)	869	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
675	. "\015\012"	870	. "\015\012"
676	. (delete $arg{body})	871	. $arg{body}
677	);	872	);
678		873
679	# return if error occured during push_write()	874	# return if error occurred during push_write()
		875	return unless %state;
		876
		877	# reduce memory usage, save a kitten, also re-use it for the response headers.
		878	%hdr = ();
		879
		880	# status line and headers
		881	$state{read_response} = sub {
680	return unless %state;	882	return unless %state;
681		883
682	%hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
683
684	# status line and headers
685	$state{read_response} = sub {
686	for ("$_[1]") {	884	for ("$_[1]") {
687	y/\015//d; # weed out any \015, as they show up in the weirdest of places.	885	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
688		886
689	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci	887	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
690	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid server response" }));	888	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
691		889
692	# 100 Continue handling	890	# 100 Continue handling
693	# should not happen as we don't send expect: 100-continue,	891	# should not happen as we don't send expect: 100-continue,
694	# but we handle it just in case.	892	# but we handle it just in case.
695	# since we send the request body regardless, if we get an error	893	# since we send the request body regardless, if we get an error
696	# we are out of-sync, which we currently do NOT handle correctly.	894	# we are out of-sync, which we currently do NOT handle correctly.
697	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})	895	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
698	if $2 eq 100;	896	if $2 eq 100;
699		897
700	push @pseudo,	898	push @pseudo,
701	HTTPVersion => $1,	899	HTTPVersion => $1,
702	Status => $2,	900	Status => $2,
703	Reason => $3,	901	Reason => $3,
704	;	902	;
705		903
706	my $hdr = parse_hdr	904	my $hdr = _parse_hdr
707	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Garbled response headers" }));	905	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
708		906
709	%hdr = (%$hdr, @pseudo);	907	%hdr = (%$hdr, @pseudo);
710	}	908	}
711		909
712	# redirect handling	910	# redirect handling
713	# microsoft and other shitheads don't give a shit for following standards,	911	# relative uri handling forced by microsoft and other shitheads.
714	# try to support some common forms of broken Location headers.	912	# we give our best and fall back to URI if available.
		913	if (exists $hdr{location}) {
		914	my $loc = $hdr{location};
		915
		916	if ($loc =~ m%^//%) { # //
		917	$loc = "$uscheme:$loc";
		918
		919	} elsif ($loc eq "") {
		920	$loc = $url;
		921
715	if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {	922	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
716	$hdr{location} =~ s/^\.\/+//;	923	$loc =~ s/^\.\/+//;
717		924
718	my $url = "$rscheme://$uhost:$uport";	925	if ($loc !~ m%^[.?#]%) {
		926	my $prefix = "$uscheme://$uauthority";
719		927
720	unless ($hdr{location} =~ s/^\///) {	928	unless ($loc =~ s/^\///) {
721	$url .= $upath;	929	$prefix .= $upath;
722	$url =~ s/\/[^\/]*$//;	930	$prefix =~ s/\/[^\/]*$//;
723	}	931	}
724		932
725	$hdr{location} = "$url/$hdr{location}";	933	$loc = "$prefix/$loc";
726	}
727		934
728	my $redirect;	935	} elsif (eval { require URI }) { # uri
		936	$loc = URI->new_abs ($loc, $url)->as_string;
729		937
730	if ($recurse) {	938	} else {
731	my $status = $hdr{Status};	939	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
732		940	#$hdr{Status} = 599;
733	# industry standard is to redirect POST as GET for	941	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
734	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
735	# also, the UA should ask the user for 301 and 307 and POST,
736	# industry standard seems to be to simply follow.
737	# we go with the industry standard.
738	if ($status == 301 or $status == 302 or $status == 303) {
739	# HTTP/1.1 is unclear on how to mutate the method
740	$method = "GET" unless $method eq "HEAD";
741	$redirect = 1;	942	#$recurse = 0;
742	} elsif ($status == 307) {
743	$redirect = 1;
744	}	943	}
745	}	944	}
746		945
747	my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])	946	$hdr{location} = $loc;
748	my $may_keep_alive = $_[3];	947	}
749		948
750	$state{handle}->destroy if $state{handle};	949	my $redirect;
751	%state = ();
752		950
753	if (defined $_[1]) {	951	if ($recurse) {
754	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];	952	my $status = $hdr{Status};
755	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];	953
		954	# industry standard is to redirect POST as GET for
		955	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		956	# also, the UA should ask the user for 301 and 307 and POST,
		957	# industry standard seems to be to simply follow.
		958	# we go with the industry standard. 308 is defined
		959	# by rfc7538
		960	if ($status == 301 or $status == 302 or $status == 303) {
		961	$redirect = 1;
		962	# HTTP/1.1 is unclear on how to mutate the method
		963	unless ($method eq "HEAD") {
		964	$method = "GET";
		965	delete $arg{body};
756	}	966	}
		967	} elsif ($status == 307 or $status == 308) {
		968	$redirect = 1;
		969	}
		970	}
757		971
758	# set-cookie processing	972	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
759	if ($arg{cookie_jar}) {	973	if ($state{handle}) {
760	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};	974	# handle keepalive
		975	if (
		976	$persistent
		977	&& $_[3]
		978	&& ($hdr{HTTPVersion} < 1.1
		979	? $hdr{connection} =~ /\bkeep-?alive\b/i
		980	: $hdr{connection} !~ /\bclose\b/i)
		981	) {
		982	ka_store $ka_key, delete $state{handle};
		983	} else {
		984	# no keepalive, destroy the handle
		985	$state{handle}->destroy;
761	}	986	}
		987	}
762		988
		989	%state = ();
		990
		991	if (defined $_[1]) {
		992	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		993	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		994	}
		995
		996	# set-cookie processing
		997	if ($arg{cookie_jar}) {
		998	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		999	}
		1000
763	if ($redirect && exists $hdr{location}) {	1001	if ($redirect && exists $hdr{location}) {
764	# we ignore any errors, as it is very common to receive	1002	# we ignore any errors, as it is very common to receive
765	# Content-Length != 0 but no actual body	1003	# Content-Length != 0 but no actual body
766	# we also access %hdr, as $_[1] might be an erro	1004	# we also access %hdr, as $_[1] might be an erro
		1005	$state{recurse} =
767	http_request (	1006	http_request (
768	$method => $hdr{location},	1007	$method => $hdr{location},
769	%arg,	1008	%arg,
770	recurse => $recurse - 1,	1009	recurse => $recurse - 1,
771	Redirect => [$_[0], \%hdr],	1010	Redirect => [$_[0], \%hdr],
		1011	sub {
		1012	%state = ();
		1013	&$cb
		1014	},
		1015	);
		1016	} else {
		1017	$cb->($_[0], \%hdr);
		1018	}
		1019	};
		1020
		1021	$ae_error = 597; # body phase
		1022
		1023	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1024
		1025	my $len = $chunked ? undef : $hdr{"content-length"};
		1026
		1027	# body handling, many different code paths
		1028	# - no body expected
		1029	# - want_body_handle
		1030	# - te chunked
		1031	# - 2x length known (with or without on_body)
		1032	# - 2x length not known (with or without on_body)
		1033	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1034	$finish->(undef, 598 => "Request cancelled by on_header");
		1035	} elsif (
		1036	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1037	or $method eq "HEAD"
		1038	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1039	) {
		1040	# no body
		1041	$finish->("", undef, undef, 1);
		1042
		1043	} elsif (!$redirect && $arg{want_body_handle}) {
		1044	$_[0]->on_eof (undef);
		1045	$_[0]->on_error (undef);
		1046	$_[0]->on_read (undef);
		1047
		1048	$finish->(delete $state{handle});
		1049
		1050	} elsif ($chunked) {
		1051	my $cl = 0;
		1052	my $body = "";
		1053	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1054
		1055	$state{read_chunk} = sub {
		1056	$_[1] =~ /^([0-9a-fA-F]+)/
		1057	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1058
		1059	my $len = hex $1;
		1060
		1061	if ($len) {
		1062	$cl += $len;
		1063
		1064	$_[0]->push_read (chunk => $len, sub {
		1065	$on_body->($_[1], \%hdr)
		1066	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1067
		1068	$_[0]->push_read (line => sub {
		1069	length $_[1]
		1070	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1071	$_[0]->push_read (line => $state{read_chunk});
772	$cb);	1072	});
		1073	});
773	} else {	1074	} else {
774	$cb->($_[0], \%hdr);	1075	$hdr{"content-length"} ||= $cl;
		1076
		1077	$_[0]->push_read (line => $qr_nlnl, sub {
		1078	if (length $_[1]) {
		1079	for ("$_[1]") {
		1080	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1081
		1082	my $hdr = _parse_hdr
		1083	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1084
		1085	%hdr = (%hdr, %$hdr);
		1086	}
		1087	}
		1088
		1089	$finish->($body, undef, undef, 1);
		1090	});
775	}	1091	}
776	};	1092	};
777		1093
778	$ae_error = 597; # body phase
779
780	my $len = $hdr{"content-length"};
781
782	# body handling, many different code paths
783	# - no body expected
784	# - want_body_handle
785	# - te chunked
786	# - 2x length known (with or without on_body)
787	# - 2x length not known (with or without on_body)
788	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
789	$finish->(undef, 598 => "Request cancelled by on_header");
790	} elsif (
791	$hdr{Status} =~ /^(?:1..|204|205|304)$/
792	or $method eq "HEAD"
793	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
794	) {
795	# no body
796	$finish->("", undef, undef, 1);
797
798	} elsif (!$redirect && $arg{want_body_handle}) {
799	$_[0]->on_eof (undef);
800	$_[0]->on_error (undef);
801	$_[0]->on_read (undef);
802
803	$finish->(delete $state{handle});
804
805	} elsif ($hdr{"transfer-encoding"} =~ /\bchunked\b/i) {
806	my $cl = 0;
807	my $body = undef;
808	my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
809
810	$state{read_chunk} = sub {
811	$_[1] =~ /^([0-9a-fA-F]+)/
812	or $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
813
814	my $len = hex $1;
815
816	if ($len) {
817	$cl += $len;
818
819	$_[0]->push_read (chunk => $len, sub {
820	$on_body->($_[1], \%hdr)
821	or return $finish->(undef, 598 => "Request cancelled by on_body");
822
823	$_[0]->push_read (line => sub {
824	length $_[1]
825	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
826	$_[0]->push_read (line => $state{read_chunk});
827	});
828	});
829	} else {
830	$hdr{"content-length"} ||= $cl;
831
832	$_[0]->push_read (line => $qr_nlnl, sub {
833	if (length $_[1]) {
834	for ("$_[1]") {
835	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
836
837	my $hdr = parse_hdr
838	or return $finish->(undef, $ae_error => "Garbled response trailers");
839
840	%hdr = (%hdr, %$hdr);
841	}
842	}
843
844	$finish->($body, undef, undef, 1);
845	});
846	}
847	};
848
849	$_[0]->push_read (line => $state{read_chunk});	1094	$_[0]->push_read (line => $state{read_chunk});
850		1095
851	} elsif ($arg{on_body}) {	1096	} elsif (!$redirect && $arg{on_body}) {
852	if (defined $len) {	1097	if (defined $len) {
853	$_[0]->on_read (sub {	1098	$_[0]->on_read (sub {
854	$len -= length $_[0]{rbuf};	1099	$len -= length $_[0]{rbuf};
855		1100
856	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)	1101	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
857	or return $finish->(undef, 598 => "Request cancelled by on_body");	1102	or return $finish->(undef, 598 => "Request cancelled by on_body");
858		1103
859	$len > 0	1104	$len > 0
860	or $finish->("", undef, undef, 1);	1105	or $finish->("", undef, undef, 1);
861	});
862	} else {
863	$_[0]->on_eof (sub {
864	$finish->("");
865	});
866	$_[0]->on_read (sub {
867	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
868	or $finish->(undef, 598 => "Request cancelled by on_body");
869	});
870	}	1106	});
871	} else {	1107	} else {
872	$_[0]->on_eof (undef);	1108	$_[0]->on_eof (sub {
873		1109	$finish->("");
874	if (defined $len) {	1110	});
875	$_[0]->on_read (sub {	1111	$_[0]->on_read (sub {
876	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)	1112	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
877	if $len <= length $_[0]{rbuf};	1113	or $finish->(undef, 598 => "Request cancelled by on_body");
878	});
879	} else {
880	$_[0]->on_error (sub {
881	($! == Errno::EPIPE || !$!)
882	? $finish->(delete $_[0]{rbuf})
883	: $finish->(undef, $ae_error => $_[2]);
884	});
885	$_[0]->on_read (sub { });
886	}	1114	});
887	}	1115	}
888	};
889
890	$state{handle}->push_read (line => $qr_nlnl, $state{read_response});
891	};
892
893	my $connect_cb = sub {
894	$state{fh} = shift
895	or do {
896	my $err = "$!";
897	%state = ();
898	return $cb->(undef, { @pseudo, Status => $ae_error, Reason => $err });
899	};
900
901	return unless delete $state{connect_guard};
902
903	# get handle
904	$state{handle} = new AnyEvent::Handle
905	fh => $state{fh},
906	peername => $rhost,
907	tls_ctx => $arg{tls_ctx},
908	# these need to be reconfigured on keepalive handles
909	timeout => $timeout,
910	on_error => sub {
911	%state = ();
912	$cb->(undef, { @pseudo, Status => $ae_error, Reason => $_[2] });
913	},
914	on_eof => sub {
915	%state = ();
916	$cb->(undef, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" });
917	},
918	;
919
920	# limit the number of persistent connections
921	# keepalive not yet supported
922	# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
923	# ++$KA_COUNT{$_[1]};
924	# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
925	# --$KA_COUNT{$_[1]}
926	# };
927	# $hdr{connection} = "keep-alive";
928	# }
929
930	$state{handle}->starttls ("connect") if $rscheme eq "https";
931
932	# now handle proxy-CONNECT method
933	if ($proxy && $uscheme eq "https") {
934	# oh dear, we have to wrap it into a connect request
935
936	# maybe re-use $uauthority with patched port?
937	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
938	$state{handle}->push_read (line => $qr_nlnl, sub {
939	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
940	or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" }));
941
942	if ($2 == 200) {
943	$rpath = $upath;
944	$handle_actual_request->();
945	} else {
946	%state = ();
947	$cb->(undef, { @pseudo, Status => $2, Reason => $3 });
948	}
949	});
950	} else {	1116	} else {
951	$handle_actual_request->();	1117	$_[0]->on_eof (undef);
		1118
		1119	if (defined $len) {
		1120	$_[0]->on_read (sub {
		1121	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1122	if $len <= length $_[0]{rbuf};
		1123	});
		1124	} else {
		1125	$_[0]->on_error (sub {
		1126	($! == Errno::EPIPE || !$!)
		1127	? $finish->(delete $_[0]{rbuf})
		1128	: $finish->(undef, $ae_error => $_[2]);
		1129	});
		1130	$_[0]->on_read (sub { });
		1131	}
952	}	1132	}
953	};	1133	};
954		1134
955	my $tcp_connect = $arg{tcp_connect}	1135	# if keepalive is enabled, then the server closing the connection
956	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };	1136	# before a response can happen legally - we retry on idempotent methods.
		1137	if ($was_persistent && $idempotent) {
		1138	my $old_eof = $hdl->{on_eof};
		1139	$hdl->{on_eof} = sub {
		1140	_destroy_state %state;
957		1141
958	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });	1142	%state = ();
		1143	$state{recurse} =
		1144	http_request (
		1145	$method => $url,
		1146	%arg,
		1147	recurse => $recurse - 1,
		1148	persistent => 0,
		1149	sub {
		1150	%state = ();
		1151	&$cb
		1152	}
		1153	);
		1154	};
		1155	$hdl->on_read (sub {
		1156	return unless %state;
		1157
		1158	# as soon as we receive something, a connection close
		1159	# once more becomes a hard error
		1160	$hdl->{on_eof} = $old_eof;
		1161	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1162	});
		1163	} else {
		1164	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1165	}
959	};	1166	};
960		1167
		1168	my $prepare_handle = sub {
		1169	my ($hdl) = $state{handle};
		1170
		1171	$hdl->on_error (sub {
		1172	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1173	});
		1174	$hdl->on_eof (sub {
		1175	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1176	});
		1177	$hdl->timeout_reset;
		1178	$hdl->timeout ($timeout);
		1179	};
		1180
		1181	# connected to proxy (or origin server)
		1182	my $connect_cb = sub {
		1183	my $fh = shift
		1184	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
		1185
		1186	return unless delete $state{connect_guard};
		1187
		1188	# get handle
		1189	$state{handle} = new AnyEvent::Handle
		1190	%{ $arg{handle_params} },
		1191	fh => $fh,
		1192	peername => $uhost,
		1193	tls_ctx => $arg{tls_ctx},
		1194	;
		1195
		1196	$prepare_handle->();
		1197
		1198	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1199
		1200	# now handle proxy-CONNECT method
		1201	if ($proxy && $uscheme eq "https") {
		1202	# oh dear, we have to wrap it into a connect request
		1203
		1204	my $auth = exists $hdr{"proxy-authorization"}
		1205	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1206	: "";
		1207
		1208	# maybe re-use $uauthority with patched port?
		1209	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
		1210	$state{handle}->push_read (line => $qr_nlnl, sub {
		1211	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
		1212	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
		1213
		1214	if ($2 == 200) {
		1215	$rpath = $upath;
		1216	$handle_actual_request->();
		1217	} else {
		1218	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
		1219	}
		1220	});
		1221	} else {
		1222	delete $hdr{"proxy-authorization"} unless $proxy;
		1223
		1224	$handle_actual_request->();
		1225	}
		1226	};
		1227
		1228	_get_slot $uhost, sub {
		1229	$state{slot_guard} = shift;
		1230
		1231	return unless $state{connect_guard};
		1232
		1233	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1234	# on a keepalive request (in theory, this should be a separate config option).
		1235	if ($persistent && $KA_CACHE{$ka_key}) {
		1236	$was_persistent = 1;
		1237
		1238	$state{handle} = ka_fetch $ka_key;
		1239	# $state{handle}->destroyed
		1240	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1241	$prepare_handle->();
		1242	# $state{handle}->destroyed
		1243	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1244	$rpath = $upath;
		1245	$handle_actual_request->();
		1246
		1247	} else {
		1248	my $tcp_connect = $arg{tcp_connect}
		1249	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1250
		1251	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
		1252	}
		1253	};
		1254
961	defined wantarray && AnyEvent::Util::guard { %state = () }	1255	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
962	}	1256	}
963		1257
964	sub http_get($@) {	1258	sub http_get($@) {
965	unshift @_, "GET";	1259	unshift @_, "GET";
966	&http_request	1260	&http_request
…		…
984	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for	1278	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
985	the actual connection, which in turn uses AnyEvent::DNS to resolve	1279	the actual connection, which in turn uses AnyEvent::DNS to resolve
986	hostnames. The latter is a simple stub resolver and does no caching	1280	hostnames. The latter is a simple stub resolver and does no caching
987	on its own. If you want DNS caching, you currently have to provide	1281	on its own. If you want DNS caching, you currently have to provide
988	your own default resolver (by storing a suitable resolver object in	1282	your own default resolver (by storing a suitable resolver object in
989	C<$AnyEvent::DNS::RESOLVER>).	1283	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
990		1284
991	=head2 GLOBAL FUNCTIONS AND VARIABLES	1285	=head2 GLOBAL FUNCTIONS AND VARIABLES
992		1286
993	=over 4	1287	=over 4
994		1288
995	=item AnyEvent::HTTP::set_proxy "proxy-url"	1289	=item AnyEvent::HTTP::set_proxy "proxy-url"
996		1290
997	Sets the default proxy server to use. The proxy-url must begin with a	1291	Sets the default proxy server to use. The proxy-url must begin with a
998	string of the form C<http://host:port> (optionally C<https:...>), croaks	1292	string of the form C<http://host:port>, croaks otherwise.
999	otherwise.
1000		1293
1001	To clear an already-set proxy, use C<undef>.	1294	To clear an already-set proxy, use C<undef>.
		1295
		1296	When AnyEvent::HTTP is loaded for the first time it will query the
		1297	default proxy from the operating system, currently by looking at
		1298	C<$ENV{http_proxy>}.
1002		1299
1003	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]	1300	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
1004		1301
1005	Remove all cookies from the cookie jar that have been expired. If	1302	Remove all cookies from the cookie jar that have been expired. If
1006	C<$session_end> is given and true, then additionally remove all session	1303	C<$session_end> is given and true, then additionally remove all session
1007	cookies.	1304	cookies.
1008		1305
1009	You should call this function (with a true C<$session_end>) before you	1306	You should call this function (with a true C<$session_end>) before you
1010	save cookies to disk, and you should call this function after loading them	1307	save cookies to disk, and you should call this function after loading them
1011	again. If you have a long-running program you can additonally call this	1308	again. If you have a long-running program you can additionally call this
1012	function from time to time.	1309	function from time to time.
1013		1310
1014	A cookie jar is initially an empty hash-reference that is managed by this	1311	A cookie jar is initially an empty hash-reference that is managed by this
1015	module. It's format is subject to change, but currently it is like this:	1312	module. Its format is subject to change, but currently it is as follows:
1016		1313
1017	The key C<version> has to contain C<1>, otherwise the hash gets	1314	The key C<version> has to contain C<2>, otherwise the hash gets
1018	emptied. All other keys are hostnames or IP addresses pointing to	1315	cleared. All other keys are hostnames or IP addresses pointing to
1019	hash-references. The key for these inner hash references is the	1316	hash-references. The key for these inner hash references is the
1020	server path for which this cookie is meant, and the values are again	1317	server path for which this cookie is meant, and the values are again
1021	hash-references. The keys of those hash-references is the cookie name, and	1318	hash-references. Each key of those hash-references is a cookie name, and
1022	the value, you guessed it, is another hash-reference, this time with the	1319	the value, you guessed it, is another hash-reference, this time with the
1023	key-value pairs from the cookie, except for C<expires> and C<max-age>,	1320	key-value pairs from the cookie, except for C<expires> and C<max-age>,
1024	which have been replaced by a C<_expires> key that contains the cookie	1321	which have been replaced by a C<_expires> key that contains the cookie
1025	expiry timestamp.	1322	expiry timestamp. Session cookies are indicated by not having an
		1323	C<_expires> key.
1026		1324
1027	Here is an example of a cookie jar with a single cookie, so you have a	1325	Here is an example of a cookie jar with a single cookie, so you have a
1028	chance of understanding the above paragraph:	1326	chance of understanding the above paragraph:
1029		1327
1030	{	1328	{
1031	version => 1,	1329	version => 2,
1032	"10.0.0.1" => {	1330	"10.0.0.1" => {
1033	"/" => {	1331	"/" => {
1034	"mythweb_id" => {	1332	"mythweb_id" => {
1035	_expires => 1293917923,	1333	_expires => 1293917923,
1036	value => "ooRung9dThee3ooyXooM1Ohm",	1334	value => "ooRung9dThee3ooyXooM1Ohm",
…		…
1052		1350
1053	=item $AnyEvent::HTTP::MAX_RECURSE	1351	=item $AnyEvent::HTTP::MAX_RECURSE
1054		1352
1055	The default value for the C<recurse> request parameter (default: C<10>).	1353	The default value for the C<recurse> request parameter (default: C<10>).
1056		1354
		1355	=item $AnyEvent::HTTP::TIMEOUT
		1356
		1357	The default timeout for connection operations (default: C<300>).
		1358
1057	=item $AnyEvent::HTTP::USERAGENT	1359	=item $AnyEvent::HTTP::USERAGENT
1058		1360
1059	The default value for the C<User-Agent> header (the default is	1361	The default value for the C<User-Agent> header (the default is
1060	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1362	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
1061		1363
1062	=item $AnyEvent::HTTP::MAX_PER_HOST	1364	=item $AnyEvent::HTTP::MAX_PER_HOST
1063		1365
1064	The maximum number of concurrent connections to the same host (identified	1366	The maximum number of concurrent connections to the same host (identified
1065	by the hostname). If the limit is exceeded, then the additional requests	1367	by the hostname). If the limit is exceeded, then additional requests
1066	are queued until previous connections are closed.	1368	are queued until previous connections are closed. Both persistent and
		1369	non-persistent connections are counted in this limit.
1067		1370
1068	The default value for this is C<4>, and it is highly advisable to not	1371	The default value for this is C<4>, and it is highly advisable to not
1069	increase it.	1372	increase it much.
		1373
		1374	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1375	connections, older browsers used 2, newer ones (such as firefox 3)
		1376	typically use 6, and Opera uses 8 because like, they have the fastest
		1377	browser and give a shit for everybody else on the planet.
		1378
		1379	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
		1380
		1381	The time after which idle persistent connections get closed by
		1382	AnyEvent::HTTP (default: C<3>).
1070		1383
1071	=item $AnyEvent::HTTP::ACTIVE	1384	=item $AnyEvent::HTTP::ACTIVE
1072		1385
1073	The number of active connections. This is not the number of currently	1386	The number of active connections. This is not the number of currently
1074	running requests, but the number of currently open and non-idle TCP	1387	running requests, but the number of currently open and non-idle TCP
1075	connections. This number of can be useful for load-leveling.	1388	connections. This number can be useful for load-leveling.
1076		1389
1077	=back	1390	=back
1078		1391
1079	=cut	1392	=cut
1080		1393
…		…
1114	# other formats fail in the loop below	1427	# other formats fail in the loop below
1115		1428
1116	for (0..11) {	1429	for (0..11) {
1117	if ($m eq $month[$_]) {	1430	if ($m eq $month[$_]) {
1118	require Time::Local;	1431	require Time::Local;
1119	return Time::Local::timegm ($S, $M, $H, $d, $_, $y);	1432	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
1120	}	1433	}
1121	}	1434	}
1122		1435
1123	undef	1436	undef
1124	}	1437	}
1125		1438
1126	sub set_proxy($) {	1439	sub set_proxy($) {
1127	if (length $_[0]) {	1440	if (length $_[0]) {
1128	$_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix	1441	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
1129	or Carp::croak "$_[0]: invalid proxy URL";	1442	or Carp::croak "$_[0]: invalid proxy URL";
1130	$PROXY = [$2, $3 || 3128, $1]	1443	$PROXY = [$2, $3 || 3128, $1]
1131	} else {	1444	} else {
1132	undef $PROXY;	1445	undef $PROXY;
1133	}	1446	}
…		…
1136	# initialise proxy from environment	1449	# initialise proxy from environment
1137	eval {	1450	eval {
1138	set_proxy $ENV{http_proxy};	1451	set_proxy $ENV{http_proxy};
1139	};	1452	};
1140		1453
		1454	=head2 SHOWCASE
		1455
		1456	This section contains some more elaborate "real-world" examples or code
		1457	snippets.
		1458
		1459	=head2 HTTP/1.1 FILE DOWNLOAD
		1460
		1461	Downloading files with HTTP can be quite tricky, especially when something
		1462	goes wrong and you want to resume.
		1463
		1464	Here is a function that initiates and resumes a download. It uses the
		1465	last modified time to check for file content changes, and works with many
		1466	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1467	on older servers.
		1468
		1469	It calls the completion callback with either C<undef>, which means a
		1470	nonretryable error occurred, C<0> when the download was partial and should
		1471	be retried, and C<1> if it was successful.
		1472
		1473	use AnyEvent::HTTP;
		1474
		1475	sub download($$$) {
		1476	my ($url, $file, $cb) = @_;
		1477
		1478	open my $fh, "+<", $file
		1479	or die "$file: $!";
		1480
		1481	my %hdr;
		1482	my $ofs = 0;
		1483
		1484	if (stat $fh and -s _) {
		1485	$ofs = -s _;
		1486	warn "-s is ", $ofs;
		1487	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1488	$hdr{"range"} = "bytes=$ofs-";
		1489	}
		1490
		1491	http_get $url,
		1492	headers => \%hdr,
		1493	on_header => sub {
		1494	my ($hdr) = @_;
		1495
		1496	if ($hdr->{Status} == 200 && $ofs) {
		1497	# resume failed
		1498	truncate $fh, $ofs = 0;
		1499	}
		1500
		1501	sysseek $fh, $ofs, 0;
		1502
		1503	1
		1504	},
		1505	on_body => sub {
		1506	my ($data, $hdr) = @_;
		1507
		1508	if ($hdr->{Status} =~ /^2/) {
		1509	length $data == syswrite $fh, $data
		1510	or return; # abort on write errors
		1511	}
		1512
		1513	1
		1514	},
		1515	sub {
		1516	my (undef, $hdr) = @_;
		1517
		1518	my $status = $hdr->{Status};
		1519
		1520	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1521	utime $time, $time, $fh;
		1522	}
		1523
		1524	if ($status == 200 || $status == 206 || $status == 416) {
		1525	# download ok || resume ok || file already fully downloaded
		1526	$cb->(1, $hdr);
		1527
		1528	} elsif ($status == 412) {
		1529	# file has changed while resuming, delete and retry
		1530	unlink $file;
		1531	$cb->(0, $hdr);
		1532
		1533	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1534	# retry later
		1535	$cb->(0, $hdr);
		1536
		1537	} else {
		1538	$cb->(undef, $hdr);
		1539	}
		1540	}
		1541	;
		1542	}
		1543
		1544	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1545	if ($_[0]) {
		1546	print "OK!\n";
		1547	} elsif (defined $_[0]) {
		1548	print "please retry later\n";
		1549	} else {
		1550	print "ERROR\n";
		1551	}
		1552	};
		1553
1141	=head2 SOCKS PROXIES	1554	=head3 SOCKS PROXIES
1142		1555
1143	Socks proxies are not directly supported by AnyEvent::HTTP. You can	1556	Socks proxies are not directly supported by AnyEvent::HTTP. You can
1144	compile your perl to support socks, or use an external program such as	1557	compile your perl to support socks, or use an external program such as
1145	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy	1558	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1146	transparently.	1559	transparently.

Diff Legend

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