ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent-HTTP/HTTP.pm
(Generate patch)

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.59 by root, Wed Dec 29 23:59:36 2010 UTC vs.
Revision 1.74 by root, Fri Dec 31 22:40:54 2010 UTC

47use AnyEvent::Util (); 47use AnyEvent::Util ();
48use AnyEvent::Handle (); 48use AnyEvent::Handle ();
49 49
50use base Exporter::; 50use base Exporter::;
51 51
52our $VERSION = '1.46'; 52our $VERSION = '1.5';
53 53
54our @EXPORT = qw(http_get http_post http_head http_request); 54our @EXPORT = qw(http_get http_post http_head http_request);
55 55
56our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)"; 56our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
57our $MAX_RECURSE = 10; 57our $MAX_RECURSE = 10;
94C<http_request> returns a "cancellation guard" - you have to keep the 94C<http_request> returns a "cancellation guard" - you have to keep the
95object at least alive until the callback get called. If the object gets 95object at least alive until the callback get called. If the object gets
96destroyed before the callback is called, the request will be cancelled. 96destroyed before the callback is called, the request will be cancelled.
97 97
98The callback will be called with the response body data as first argument 98The 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 as 99(or C<undef> if an error occured), and a hash-ref with response headers
100second argument. 100(and trailers) as second argument.
101 101
102All the headers in that hash are lowercased. In addition to the response 102All the headers in that hash are lowercased. In addition to the response
103headers, the "pseudo-headers" (uppercase to avoid clashing with possible 103headers, the "pseudo-headers" (uppercase to avoid clashing with possible
104response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the 104response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
105three parts of the HTTP Status-Line of the same name. 105three parts of the HTTP Status-Line of the same name. If an error occurs
106during the body phase of a request, then the original C<Status> and
107C<Reason> values from the header are available as C<OrigStatus> and
108C<OrigReason>.
106 109
107The pseudo-header C<URL> contains the actual URL (which can differ from 110The pseudo-header C<URL> contains the actual URL (which can differ from
108the requested URL when following redirects - for example, you might get 111the requested URL when following redirects - for example, you might get
109an error that your URL scheme is not supported even though your URL is a 112an error that your URL scheme is not supported even though your URL is a
110valid http URL because it redirected to an ftp URL, in which case you can 113valid http URL because it redirected to an ftp URL, in which case you can
147Whether to recurse requests or not, e.g. on redirects, authentication 150Whether to recurse requests or not, e.g. on redirects, authentication
148retries and so on, and how often to do so. 151retries and so on, and how often to do so.
149 152
150=item headers => hashref 153=item headers => hashref
151 154
152The request headers to use. Currently, C<http_request> may provide its 155The request headers to use. Currently, C<http_request> may provide its own
153own C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers 156C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
154and will provide defaults for C<User-Agent:> and C<Referer:> (this can be 157will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
155suppressed by using C<undef> for these headers in which case they won't be 158(this can be suppressed by using C<undef> for these headers in which case
156sent at all). 159they won't be sent at all).
157 160
158=item timeout => $seconds 161=item timeout => $seconds
159 162
160The time-out to use for various stages - each connect attempt will reset 163The time-out to use for various stages - each connect attempt will reset
161the timeout, as will read or write activity, i.e. this is not an overall 164the timeout, as will read or write activity, i.e. this is not an overall
171C<$scheme> must be either missing, C<http> for HTTP or C<https> for 174C<$scheme> must be either missing, C<http> for HTTP or C<https> for
172HTTPS. 175HTTPS.
173 176
174=item body => $string 177=item body => $string
175 178
176The request body, usually empty. Will be-sent as-is (future versions of 179The request body, usually empty. Will be sent as-is (future versions of
177this module might offer more options). 180this module might offer more options).
178 181
179=item cookie_jar => $hash_ref 182=item cookie_jar => $hash_ref
180 183
181Passing this parameter enables (simplified) cookie-processing, loosely 184Passing this parameter enables (simplified) cookie-processing, loosely
182based on the original netscape specification. 185based on the original netscape specification.
183 186
184The C<$hash_ref> must be an (initially empty) hash reference which will 187The C<$hash_ref> must be an (initially empty) hash reference which will
185get updated automatically. It is possible to save the cookie_jar to 188get updated automatically. It is possible to save the cookie jar to
186persistent storage with something like JSON or Storable, but this is not 189persistent storage with something like JSON or Storable, but this is not
187recommended, as expiry times are currently being ignored. 190recommended, as session-only cookies might survive longer than expected.
188 191
189Note that this cookie implementation is not of very high quality, nor 192Note that this cookie implementation is not meant to be complete. If
190meant to be complete. If you want complete cookie management you have to 193you want complete cookie management you have to do that on your
191do that on your own. C<cookie_jar> is meant as a quick fix to get some 194own. C<cookie_jar> is meant as a quick fix to get some cookie-using sites
192cookie-using sites working. Cookies are a privacy disaster, do not use 195working. Cookies are a privacy disaster, do not use them unless required
193them unless required to. 196to.
197
198When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
199headers will be set and handled by this module, otherwise they will be
200left untouched.
194 201
195=item tls_ctx => $scheme | $tls_ctx 202=item tls_ctx => $scheme | $tls_ctx
196 203
197Specifies the AnyEvent::TLS context to be used for https connections. This 204Specifies the AnyEvent::TLS context to be used for https connections. This
198parameter follows the same rules as the C<tls_ctx> parameter to 205parameter follows the same rules as the C<tls_ctx> parameter to
216=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb) 223=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
217 224
218In even rarer cases you want total control over how AnyEvent::HTTP 225In even rarer cases you want total control over how AnyEvent::HTTP
219establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect> 226establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
220to do this, but you can provide your own C<tcp_connect> function - 227to do this, but you can provide your own C<tcp_connect> function -
221obviously, it has to follow the same calling conventions. 228obviously, it has to follow the same calling conventions, except that it
229may always return a connection guard object.
222 230
223There are probably lots of weird uses for this function, starting from 231There are probably lots of weird uses for this function, starting from
224tracing the hosts C<http_request> actually tries to connect, to (inexact 232tracing the hosts C<http_request> actually tries to connect, to (inexact
225but fast) host => IP address caching or even socks protocol support. 233but fast) host => IP address caching or even socks protocol support.
226 234
236 244
237This callback is useful, among other things, to quickly reject unwanted 245This callback is useful, among other things, to quickly reject unwanted
238content, which, if it is supposed to be rare, can be faster than first 246content, which, if it is supposed to be rare, can be faster than first
239doing a C<HEAD> request. 247doing a C<HEAD> request.
240 248
249The downside is that cancelling the request makes it impossible to re-use
250the connection. Also, the C<on_header> callback will not receive any
251trailer (headers sent after the response body).
252
241Example: cancel the request unless the content-type is "text/html". 253Example: cancel the request unless the content-type is "text/html".
242 254
243 on_header => sub { 255 on_header => sub {
244 $_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/ 256 $_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
245 }, 257 },
251string instead of the body data. 263string instead of the body data.
252 264
253It has to return either true (in which case AnyEvent::HTTP will continue), 265It has to return either true (in which case AnyEvent::HTTP will continue),
254or false, in which case AnyEvent::HTTP will cancel the download (and call 266or false, in which case AnyEvent::HTTP will cancel the download (and call
255the completion callback with an error code of C<598>). 267the completion callback with an error code of C<598>).
268
269The downside to cancelling the request is that it makes it impossible to
270re-use the connection.
256 271
257This callback is useful when the data is too large to be held in memory 272This callback is useful when the data is too large to be held in memory
258(so the callback writes it to a file) or when only some information should 273(so the callback writes it to a file) or when only some information should
259be extracted, or when the body should be processed incrementally. 274be extracted, or when the body should be processed incrementally.
260 275
286If you think you need this, first have a look at C<on_body>, to see if 301If you think you need this, first have a look at C<on_body>, to see if
287that doesn't solve your problem in a better way. 302that doesn't solve your problem in a better way.
288 303
289=back 304=back
290 305
291Example: make a simple HTTP GET request for http://www.nethype.de/ 306Example: do a simple HTTP GET request for http://www.nethype.de/ and print
307the response body.
292 308
293 http_request GET => "http://www.nethype.de/", sub { 309 http_request GET => "http://www.nethype.de/", sub {
294 my ($body, $hdr) = @_; 310 my ($body, $hdr) = @_;
295 print "$body\n"; 311 print "$body\n";
296 }; 312 };
297 313
298Example: make a HTTP HEAD request on https://www.google.com/, use a 314Example: do a HTTP HEAD request on https://www.google.com/, use a
299timeout of 30 seconds. 315timeout of 30 seconds.
300 316
301 http_request 317 http_request
302 GET => "https://www.google.com", 318 GET => "https://www.google.com",
303 timeout => 30, 319 timeout => 30,
306 use Data::Dumper; 322 use Data::Dumper;
307 print Dumper $hdr; 323 print Dumper $hdr;
308 } 324 }
309 ; 325 ;
310 326
311Example: make another simple HTTP GET request, but immediately try to 327Example: do another simple HTTP GET request, but immediately try to
312cancel it. 328cancel it.
313 329
314 my $request = http_request GET => "http://www.nethype.de/", sub { 330 my $request = http_request GET => "http://www.nethype.de/", sub {
315 my ($body, $hdr) = @_; 331 my ($body, $hdr) = @_;
316 print "$body\n"; 332 print "$body\n";
348 push @{ $CO_SLOT{$_[0]}[1] }, $_[1]; 364 push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
349 365
350 _slot_schedule $_[0]; 366 _slot_schedule $_[0];
351} 367}
352 368
369# extract cookies from jar
370sub cookie_jar_extract($$$$) {
371 my ($jar, $uscheme, $uhost, $upath) = @_;
372
373 %$jar = () if $jar->{version} != 1;
374
375 my @cookies;
376
377 while (my ($chost, $paths) = each %$jar) {
378 next unless ref $paths;
379
380 if ($chost =~ /^\./) {
381 next unless $chost eq substr $uhost, -length $chost;
382 } elsif ($chost =~ /\./) {
383 next unless $chost eq $uhost;
384 } else {
385 next;
386 }
387
388 while (my ($cpath, $cookies) = each %$paths) {
389 next unless $cpath eq substr $upath, 0, length $cpath;
390
391 while (my ($cookie, $kv) = each %$cookies) {
392 next if $uscheme ne "https" && exists $kv->{secure};
393
394 if (exists $kv->{expires}) {
395 if (AE::now > parse_date ($kv->{expires})) {
396 delete $cookies->{$cookie};
397 next;
398 }
399 }
400
401 my $value = $kv->{value};
402
403 if ($value =~ /[=;,[:space:]]/) {
404 $value =~ s/([\\"])/\\$1/g;
405 $value = "\"$value\"";
406 }
407
408 push @cookies, "$cookie=$value";
409 }
410 }
411 }
412
413 \@cookies
414}
415
416# parse set_cookie header into jar
417sub cookie_jar_set_cookie($$$) {
418 my ($jar, $set_cookie, $uhost) = @_;
419
420 for ($set_cookie) {
421 # parse NAME=VALUE
422 my @kv;
423
424 while (
425 m{
426 \G\s*
427 (?:
428 expires \s*=\s* ([A-Z][a-z][a-z],\ [^,;]+)
429 | ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )
430 )
431 }gcxsi
432 ) {
433 my $name = $2;
434 my $value = $4;
435
436 unless (defined $name) {
437 # expires
438 $name = "expires";
439 $value = $1;
440 } elsif (!defined $value) {
441 # quoted
442 $value = $3;
443 $value =~ s/\\(.)/$1/gs;
444 }
445
446 push @kv, lc $name, $value;
447
448 last unless /\G\s*;/gc;
449 }
450
451 last unless @kv;
452
453 my $name = shift @kv;
454 my %kv = (value => shift @kv, @kv);
455
456 $kv{expires} ||= format_date (AE::now + $kv{"max-age"})
457 if exists $kv{"max-age"};
458
459 my $cdom;
460 my $cpath = (delete $kv{path}) || "/";
461
462 if (exists $kv{domain}) {
463 $cdom = delete $kv{domain};
464
465 $cdom =~ s/^\.?/./; # make sure it starts with a "."
466
467 next if $cdom =~ /\.$/;
468
469 # this is not rfc-like and not netscape-like. go figure.
470 my $ndots = $cdom =~ y/.//;
471 next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
472 } else {
473 $cdom = $uhost;
474 }
475
476 # store it
477 $jar->{version} = 1;
478 $jar->{$cdom}{$cpath}{$name} = \%kv;
479
480 redo if /\G\s*,/gc;
481 }
482}
483
484# continue to parse $_ for headers and place them into the arg
485sub parse_hdr() {
486 my %hdr;
487
488 # things seen, not parsed:
489 # p3pP="NON CUR OTPi OUR NOR UNI"
490
491 $hdr{lc $1} .= ",$2"
492 while /\G
493 ([^:\000-\037]*):
494 [\011\040]*
495 ((?: [^\012]+ | \012[\011\040] )*)
496 \012
497 /gxc;
498
499 /\G$/
500 or return;
501
502 # remove the "," prefix we added to all headers above
503 substr $_, 0, 1, ""
504 for values %hdr;
505
506 \%hdr
507}
508
353our $qr_nlnl = qr{(?<![^\012])\015?\012}; 509our $qr_nlnl = qr{(?<![^\012])\015?\012};
354 510
355our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 }; 511our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
356our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" }; 512our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
357 513
376 my @pseudo = (URL => $url); 532 my @pseudo = (URL => $url);
377 push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect}; 533 push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
378 534
379 my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE; 535 my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
380 536
381 return $cb->(undef, { Status => 599, Reason => "Too many redirections", @pseudo }) 537 return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
382 if $recurse < 0; 538 if $recurse < 0;
383 539
384 my $proxy = $arg{proxy} || $PROXY; 540 my $proxy = $arg{proxy} || $PROXY;
385 my $timeout = $arg{timeout} || $TIMEOUT; 541 my $timeout = $arg{timeout} || $TIMEOUT;
386 542
389 545
390 $uscheme = lc $uscheme; 546 $uscheme = lc $uscheme;
391 547
392 my $uport = $uscheme eq "http" ? 80 548 my $uport = $uscheme eq "http" ? 80
393 : $uscheme eq "https" ? 443 549 : $uscheme eq "https" ? 443
394 : return $cb->(undef, { Status => 599, Reason => "Only http and https URL schemes supported", @pseudo }); 550 : return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
395 551
396 $uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x 552 $uauthority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
397 or return $cb->(undef, { Status => 599, Reason => "Unparsable URL", @pseudo }); 553 or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
398 554
399 my $uhost = $1; 555 my $uhost = $1;
400 $uport = $2 if defined $2; 556 $uport = $2 if defined $2;
401 557
402 $hdr{host} = defined $2 ? "$uhost:$2" : "$uhost" 558 $hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
407 563
408 $upath =~ s%^/?%/%; 564 $upath =~ s%^/?%/%;
409 565
410 # cookie processing 566 # cookie processing
411 if (my $jar = $arg{cookie_jar}) { 567 if (my $jar = $arg{cookie_jar}) {
412 %$jar = () if $jar->{version} != 1; 568 my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
413 569
414 my @cookie;
415
416 while (my ($chost, $v) = each %$jar) {
417 if ($chost =~ /^\./) {
418 next unless $chost eq substr $uhost, -length $chost;
419 } elsif ($chost =~ /\./) {
420 next unless $chost eq $uhost;
421 } else {
422 next;
423 }
424
425 while (my ($cpath, $v) = each %$v) {
426 next unless $cpath eq substr $upath, 0, length $cpath;
427
428 while (my ($k, $v) = each %$v) {
429 next if $uscheme ne "https" && exists $v->{secure};
430 my $value = $v->{value};
431 $value =~ s/([\\"])/\\$1/g;
432 push @cookie, "$k=\"$value\"";
433 }
434 }
435 }
436
437 $hdr{cookie} = join "; ", @cookie 570 $hdr{cookie} = join "; ", @$cookies
438 if @cookie; 571 if @$cookies;
439 } 572 }
440 573
441 my ($rhost, $rport, $rscheme, $rpath); # request host, port, path 574 my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
442 575
443 if ($proxy) { 576 if ($proxy) {
451 } else { 584 } else {
452 ($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath); 585 ($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
453 } 586 }
454 587
455 # leave out fragment and query string, just a heuristic 588 # leave out fragment and query string, just a heuristic
456 $hdr{referer} ||= "$uscheme://$uauthority$upath" unless exists $hdr{referer}; 589 $hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
457 $hdr{"user-agent"} ||= $USERAGENT unless exists $hdr{"user-agent"}; 590 $hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
458 591
459 $hdr{"content-length"} = length $arg{body} 592 $hdr{"content-length"} = length $arg{body}
460 if length $arg{body} || $method ne "GET"; 593 if length $arg{body} || $method ne "GET";
461 594
595 $hdr{connection} = "close TE"; #1.1
596 $hdr{te} = "trailers" unless exists $hdr{te}; #1.1
597
462 my %state = (connect_guard => 1); 598 my %state = (connect_guard => 1);
463 599
464 _get_slot $uhost, sub { 600 _get_slot $uhost, sub {
465 $state{slot_guard} = shift; 601 $state{slot_guard} = shift;
466 602
467 return unless $state{connect_guard}; 603 return unless $state{connect_guard};
468 604
469 my $tcp_connect = $arg{tcp_connect} 605 my $connect_cb = sub {
470 || do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
471
472 $state{connect_guard} = $tcp_connect->(
473 $rhost,
474 $rport,
475 sub {
476 $state{fh} = shift 606 $state{fh} = shift
477 or do { 607 or do {
478 my $err = "$!"; 608 my $err = "$!";
479 %state = (); 609 %state = ();
480 return $cb->(undef, { Status => 599, Reason => $err, @pseudo }); 610 return $cb->(undef, { @pseudo, Status => 599, Reason => $err });
481 }; 611 };
482 612
483 pop; # free memory, save a tree 613 pop; # free memory, save a tree
484 614
485 return unless delete $state{connect_guard}; 615 return unless delete $state{connect_guard};
486 616
487 # get handle 617 # get handle
488 $state{handle} = new AnyEvent::Handle 618 $state{handle} = new AnyEvent::Handle
489 fh => $state{fh}, 619 fh => $state{fh},
490 peername => $rhost, 620 peername => $rhost,
491 tls_ctx => $arg{tls_ctx}, 621 tls_ctx => $arg{tls_ctx},
492 # these need to be reconfigured on keepalive handles 622 # these need to be reconfigured on keepalive handles
493 timeout => $timeout, 623 timeout => $timeout,
494 on_error => sub { 624 on_error => sub {
495 %state = (); 625 %state = ();
496 $cb->(undef, { Status => 599, Reason => $_[2], @pseudo }); 626 $cb->(undef, { @pseudo, Status => 599, Reason => $_[2] });
497 }, 627 },
498 on_eof => sub { 628 on_eof => sub {
499 %state = (); 629 %state = ();
500 $cb->(undef, { Status => 599, Reason => "Unexpected end-of-file", @pseudo }); 630 $cb->(undef, { @pseudo, Status => 599, Reason => "Unexpected end-of-file" });
501 }, 631 },
502 ; 632 ;
503 633
504 # limit the number of persistent connections 634 # limit the number of persistent connections
505 # keepalive not yet supported 635 # keepalive not yet supported
506# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) { 636# if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
507# ++$KA_COUNT{$_[1]}; 637# ++$KA_COUNT{$_[1]};
508# $state{handle}{ka_count_guard} = AnyEvent::Util::guard { 638# $state{handle}{ka_count_guard} = AnyEvent::Util::guard {
509# --$KA_COUNT{$_[1]} 639# --$KA_COUNT{$_[1]}
510# }; 640# };
511# $hdr{connection} = "keep-alive"; 641# $hdr{connection} = "keep-alive";
512# } else {
513 delete $hdr{connection};
514# } 642# }
515 643
516 $state{handle}->starttls ("connect") if $rscheme eq "https"; 644 $state{handle}->starttls ("connect") if $rscheme eq "https";
517 645
518 # handle actual, non-tunneled, request 646 # handle actual, non-tunneled, request
519 my $handle_actual_request = sub { 647 my $handle_actual_request = sub {
520 $state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls}; 648 $state{handle}->starttls ("connect") if $uscheme eq "https" && !exists $state{handle}{tls};
521 649
522 # send request 650 # send request
523 $state{handle}->push_write ( 651 $state{handle}->push_write (
524 "$method $rpath HTTP/1.0\015\012" 652 "$method $rpath HTTP/1.1\015\012"
525 . (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr) 653 . (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
526 . "\015\012" 654 . "\015\012"
527 . (delete $arg{body}) 655 . (delete $arg{body})
528 ); 656 );
529 657
530 # return if error occured during push_write() 658 # return if error occured during push_write()
531 return unless %state; 659 return unless %state;
532 660
533 %hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use 661 %hdr = (); # reduce memory usage, save a kitten, also make it possible to re-use
534 662
535 # status line and headers 663 # status line and headers
536 $state{handle}->push_read (line => $qr_nlnl, sub { 664 $state{read_response} = sub {
537 for ("$_[1]") { 665 for ("$_[1]") {
538 y/\015//d; # weed out any \015, as they show up in the weirdest of places. 666 y/\015//d; # weed out any \015, as they show up in the weirdest of places.
539 667
540 /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )? \015?\012/igxc 668 /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/igxc
541 or return (%state = (), $cb->(undef, { Status => 599, Reason => "Invalid server response", @pseudo })); 669 or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid server response" }));
542 670
671 # 100 Continue handling
672 # should not happen as we don't send expect: 100-continue,
673 # but we handle it just in case.
674 # since we send the request body regardless, if we get an error
675 # we are out of-sync, which we currently do NOT handle correctly.
676 return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
677 if $2 eq 100;
678
543 push @pseudo, 679 push @pseudo,
544 HTTPVersion => $1, 680 HTTPVersion => $1,
545 Status => $2, 681 Status => $2,
546 Reason => $3, 682 Reason => $3,
547 ; 683 ;
548 684
549 # things seen, not parsed: 685 my $hdr = parse_hdr
550 # p3pP="NON CUR OTPi OUR NOR UNI"
551
552 $hdr{lc $1} .= ",$2"
553 while /\G
554 ([^:\000-\037]*):
555 [\011\040]*
556 ((?: [^\012]+ | \012[\011\040] )*)
557 \012
558 /gxc;
559
560 /\G$/
561 or return (%state = (), $cb->(undef, { Status => 599, Reason => "Garbled response headers", @pseudo })); 686 or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Garbled response headers" }));
687
688 %hdr = (%$hdr, @pseudo);
689 }
690
691 # redirect handling
692 # microsoft and other shitheads don't give a shit for following standards,
693 # try to support some common forms of broken Location headers.
694 if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) {
695 $hdr{location} =~ s/^\.\/+//;
696
697 my $url = "$rscheme://$uhost:$uport";
698
699 unless ($hdr{location} =~ s/^\///) {
700 $url .= $upath;
701 $url =~ s/\/[^\/]*$//;
562 } 702 }
563 703
564 # remove the "," prefix we added to all headers above 704 $hdr{location} = "$url/$hdr{location}";
565 substr $_, 0, 1, "" 705 }
566 for values %hdr;
567 706
568 # patch in all pseudo headers 707 my $redirect;
569 %hdr = (%hdr, @pseudo);
570 708
709 if ($recurse) {
710 my $status = $hdr{Status};
711
712 # industry standard is to redirect POST as GET for
713 # 301, 302 and 303, in contrast to http/1.0 and 1.1.
714 # also, the UA should ask the user for 301 and 307 and POST,
715 # industry standard seems to be to simply follow.
716 # we go with the industry standard.
717 if ($status == 301 or $status == 302 or $status == 303) {
718 # HTTP/1.1 is unclear on how to mutate the method
719 $method = "GET" unless $method eq "HEAD";
571 # redirect handling 720 $redirect = 1;
572 # microsoft and other shitheads don't give a shit for following standards, 721 } elsif ($status == 307) {
573 # try to support some common forms of broken Location headers. 722 $redirect = 1;
574 if ($hdr{location} !~ /^(?: $ | [^:\/?\#]+ : )/x) { 723 }
575 $hdr{location} =~ s/^\.\/+//; 724 }
576 725
577 my $url = "$rscheme://$uhost:$uport"; 726 my $finish = sub { # ($data, $err_status, $err_reason[, $keepalive])
727 my $may_keep_alive = $_[3];
578 728
579 unless ($hdr{location} =~ s/^\///) { 729 $state{handle}->destroy if $state{handle};
730 %state = ();
731
732 if (defined $_[1]) {
733 $hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
734 $hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
735 }
736
737 # set-cookie processing
738 if ($arg{cookie_jar}) {
739 cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost;
740 }
741
742 if ($redirect && exists $hdr{location}) {
743 # we ignore any errors, as it is very common to receive
744 # Content-Length != 0 but no actual body
745 # we also access %hdr, as $_[1] might be an erro
746 http_request (
747 $method => $hdr{location},
748 %arg,
749 recurse => $recurse - 1,
750 Redirect => [$_[0], \%hdr],
751 $cb);
752 } else {
753 $cb->($_[0], \%hdr);
754 }
755 };
756
757 my $len = $hdr{"content-length"};
758
759 if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
760 $finish->(undef, 598 => "Request cancelled by on_header");
761 } elsif (
762 $hdr{Status} =~ /^(?:1..|204|205|304)$/
763 or $method eq "HEAD"
764 or (defined $len && !$len)
765 ) {
766 # no body
767 $finish->("", undef, undef, 1);
768 } else {
769 # body handling, many different code paths
770 # - no body expected
771 # - want_body_handle
772 # - te chunked
773 # - 2x length known (with or without on_body)
774 # - 2x length not known (with or without on_body)
775 if (!$redirect && $arg{want_body_handle}) {
776 $_[0]->on_eof (undef);
777 $_[0]->on_error (undef);
778 $_[0]->on_read (undef);
779
780 $finish->(delete $state{handle});
781
782 } elsif ($hdr{"transfer-encoding"} =~ /\bchunked\b/i) {
783 my $cl = 0;
784 my $body = undef;
785 my $on_body = $arg{on_body} || sub { $body .= shift; 1 };
786
787 $_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
788
789 my $read_chunk; $read_chunk = sub {
790 $_[1] =~ /^([0-9a-fA-F]+)/
791 or $finish->(undef, 599 => "Garbled chunked transfer encoding");
792
793 my $len = hex $1;
794
795 if ($len) {
580 $url .= $upath; 796 $cl += $len;
581 $url =~ s/\/[^\/]*$//; 797
798 $_[0]->push_read (chunk => $len, sub {
799 $on_body->($_[1], \%hdr)
800 or return $finish->(undef, 598 => "Request cancelled by on_body");
801
802 $_[0]->push_read (line => sub {
803 length $_[1]
804 and return $finish->(undef, 599 => "Garbled chunked transfer encoding");
805 $_[0]->push_read (line => $read_chunk);
806 });
807 });
808 } else {
809 $hdr{"content-length"} ||= $cl;
810
811 $_[0]->push_read (line => $qr_nlnl, sub {
812 if (length $_[1]) {
813 for ("$_[1]") {
814 y/\015//d; # weed out any \015, as they show up in the weirdest of places.
815
816 my $hdr = parse_hdr
817 or return $finish->(undef, 599 => "Garbled response trailers");
818
819 %hdr = (%hdr, %$hdr);
820 }
821 }
822
823 $finish->($body, undef, undef, 1);
824 });
825 }
826 };
827
828 $_[0]->push_read (line => $read_chunk);
829
830 } elsif ($arg{on_body}) {
831 $_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
832
833 if ($len) {
834 $_[0]->on_read (sub {
835 $len -= length $_[0]{rbuf};
836
837 $arg{on_body}(delete $_[0]{rbuf}, \%hdr)
838 or return $finish->(undef, 598 => "Request cancelled by on_body");
839
840 $len > 0
841 or $finish->("", undef, undef, 1);
842 });
843 } else {
844 $_[0]->on_eof (sub {
845 $finish->("");
846 });
847 $_[0]->on_read (sub {
848 $arg{on_body}(delete $_[0]{rbuf}, \%hdr)
849 or $finish->(undef, 598 => "Request cancelled by on_body");
850 });
582 } 851 }
583
584 $hdr{location} = "$url/$hdr{location}";
585 } 852 } else {
853 $_[0]->on_eof (undef);
586 854
587 my $redirect;
588
589 if ($recurse) { 855 if ($len) {
590 my $status = $hdr{Status}; 856 $_[0]->on_error (sub { $finish->(undef, 599 => $_[2]) });
591 857 $_[0]->on_read (sub {
592 # industry standard is to redirect POST as GET for 858 $finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
593 # 301, 302 and 303, in contrast to http/1.0 and 1.1. 859 if $len <= length $_[0]{rbuf};
594 # also, the UA should ask the user for 301 and 307 and POST,
595 # industry standard seems to be to simply follow.
596 # we go with the industry standard.
597 if ($status == 301 or $status == 302 or $status == 303) {
598 # HTTP/1.1 is unclear on how to mutate the method
599 $method = "GET" unless $method eq "HEAD";
600 $redirect = 1; 860 });
601 } elsif ($status == 307) { 861 } else {
862 $_[0]->on_error (sub {
863 ($! == Errno::EPIPE || !$!)
864 ? $finish->(delete $_[0]{rbuf})
865 : $finish->(undef, 599 => $_[2]);
602 $redirect = 1; 866 });
867 $_[0]->on_read (sub { });
603 } 868 }
604 } 869 }
605
606 my $finish = sub {
607 $state{handle}->destroy if $state{handle};
608 %state = ();
609
610 # set-cookie processing
611 if ($arg{cookie_jar}) {
612 for ($_[1]{"set-cookie"}) {
613 # parse NAME=VALUE
614 my @kv;
615
616 while (/\G\s* ([^=;,[:space:]]+) \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^=;,[:space:]]*) )/gcxs) {
617 my $name = $1;
618 my $value = $3;
619
620 unless ($value) {
621 $value = $2;
622 $value =~ s/\\(.)/$1/gs;
623 }
624
625 push @kv, $name => $value;
626
627 last unless /\G\s*;/gc;
628 }
629
630 last unless @kv;
631
632 my $name = shift @kv;
633 my %kv = (value => shift @kv, @kv);
634
635 my $cdom;
636 my $cpath = (delete $kv{path}) || "/";
637
638 if (exists $kv{domain}) {
639 $cdom = delete $kv{domain};
640
641 $cdom =~ s/^\.?/./; # make sure it starts with a "."
642
643 next if $cdom =~ /\.$/;
644
645 # this is not rfc-like and not netscape-like. go figure.
646 my $ndots = $cdom =~ y/.//;
647 next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
648 } else {
649 $cdom = $uhost;
650 }
651
652 # store it
653 $arg{cookie_jar}{version} = 1;
654 $arg{cookie_jar}{$cdom}{$cpath}{$name} = \%kv;
655
656 redo if /\G\s*,/gc;
657 }
658 }
659
660 if ($redirect && exists $hdr{location}) {
661 # we ignore any errors, as it is very common to receive
662 # Content-Length != 0 but no actual body
663 # we also access %hdr, as $_[1] might be an erro
664 http_request (
665 $method => $hdr{location},
666 %arg,
667 recurse => $recurse - 1,
668 Redirect => \@_,
669 $cb);
670 } else {
671 $cb->($_[0], $_[1]);
672 }
673 };
674
675 my $len = $hdr{"content-length"};
676
677 if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
678 $finish->(undef, { Status => 598, Reason => "Request cancelled by on_header", @pseudo });
679 } elsif (
680 $hdr{Status} =~ /^(?:1..|[23]04)$/
681 or $method eq "HEAD"
682 or (defined $len && !$len)
683 ) {
684 # no body
685 $finish->("", \%hdr);
686 } else {
687 # body handling, four different code paths
688 # for want_body_handle, on_body (2x), normal (2x)
689 # we might read too much here, but it does not matter yet (no pers. connections)
690 if (!$redirect && $arg{want_body_handle}) {
691 $_[0]->on_eof (undef);
692 $_[0]->on_error (undef);
693 $_[0]->on_read (undef);
694
695 $finish->(delete $state{handle}, \%hdr);
696
697 } elsif ($arg{on_body}) {
698 $_[0]->on_error (sub { $finish->(undef, { Status => 599, Reason => $_[2], @pseudo }) });
699 if ($len) {
700 $_[0]->on_eof (undef);
701 $_[0]->on_read (sub {
702 $len -= length $_[0]{rbuf};
703
704 $arg{on_body}(delete $_[0]{rbuf}, \%hdr)
705 or $finish->(undef, { Status => 598, Reason => "Request cancelled by on_body", @pseudo });
706
707 $len > 0
708 or $finish->("", \%hdr);
709 });
710 } else {
711 $_[0]->on_eof (sub {
712 $finish->("", \%hdr);
713 });
714 $_[0]->on_read (sub {
715 $arg{on_body}(delete $_[0]{rbuf}, \%hdr)
716 or $finish->(undef, { Status => 598, Reason => "Request cancelled by on_body", @pseudo });
717 });
718 }
719 } else {
720 $_[0]->on_eof (undef);
721
722 if ($len) {
723 $_[0]->on_error (sub { $finish->(undef, { Status => 599, Reason => $_[2], @pseudo }) });
724 $_[0]->on_read (sub {
725 $finish->((substr delete $_[0]{rbuf}, 0, $len, ""), \%hdr)
726 if $len <= length $_[0]{rbuf};
727 });
728 } else {
729 $_[0]->on_error (sub {
730 ($! == Errno::EPIPE || !$!)
731 ? $finish->(delete $_[0]{rbuf}, \%hdr)
732 : $finish->(undef, { Status => 599, Reason => $_[2], @pseudo });
733 });
734 $_[0]->on_read (sub { });
735 }
736 }
737 }
738 }); 870 }
739 }; 871 };
740 872
873 $state{handle}->push_read (line => $qr_nlnl, $state{read_response});
874 };
875
741 # now handle proxy-CONNECT method 876 # now handle proxy-CONNECT method
742 if ($proxy && $uscheme eq "https") { 877 if ($proxy && $uscheme eq "https") {
743 # oh dear, we have to wrap it into a connect request 878 # oh dear, we have to wrap it into a connect request
744 879
745 # maybe re-use $uauthority with patched port? 880 # maybe re-use $uauthority with patched port?
746 $state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012"); 881 $state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012Host: $uhost\015\012\015\012");
747 $state{handle}->push_read (line => $qr_nlnl, sub { 882 $state{handle}->push_read (line => $qr_nlnl, sub {
748 $_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix 883 $_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
749 or return (%state = (), $cb->(undef, { Status => 599, Reason => "Invalid proxy connect response ($_[1])", @pseudo })); 884 or return (%state = (), $cb->(undef, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" }));
750 885
751 if ($2 == 200) { 886 if ($2 == 200) {
752 $rpath = $upath; 887 $rpath = $upath;
753 &$handle_actual_request; 888 &$handle_actual_request;
754 } else { 889 } else {
755 %state = (); 890 %state = ();
756 $cb->(undef, { Status => $2, Reason => $3, @pseudo }); 891 $cb->(undef, { @pseudo, Status => $2, Reason => $3 });
757 }
758 }); 892 }
893 });
759 } else { 894 } else {
760 &$handle_actual_request; 895 &$handle_actual_request;
761 }
762
763 }, 896 }
764 $arg{on_prepare} || sub { $timeout }
765 ); 897 };
898
899 my $tcp_connect = $arg{tcp_connect}
900 || do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
901
902 $state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
903
766 }; 904 };
767 905
768 defined wantarray && AnyEvent::Util::guard { %state = () } 906 defined wantarray && AnyEvent::Util::guard { %state = () }
769} 907}
770 908
805string of the form C<http://host:port> (optionally C<https:...>), croaks 943string of the form C<http://host:port> (optionally C<https:...>), croaks
806otherwise. 944otherwise.
807 945
808To clear an already-set proxy, use C<undef>. 946To clear an already-set proxy, use C<undef>.
809 947
948=item $date = AnyEvent::HTTP::format_date $timestamp
949
950Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
951Date (RFC 2616).
952
953=item $timestamp = AnyEvent::HTTP::parse_date $date
954
955Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) and
956returns the corresponding POSIX timestamp, or C<undef> if the date cannot
957be parsed.
958
810=item $AnyEvent::HTTP::MAX_RECURSE 959=item $AnyEvent::HTTP::MAX_RECURSE
811 960
812The default value for the C<recurse> request parameter (default: C<10>). 961The default value for the C<recurse> request parameter (default: C<10>).
813 962
814=item $AnyEvent::HTTP::USERAGENT 963=item $AnyEvent::HTTP::USERAGENT
832connections. This number of can be useful for load-leveling. 981connections. This number of can be useful for load-leveling.
833 982
834=back 983=back
835 984
836=cut 985=cut
986
987our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
988our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
989
990sub format_date($) {
991 my ($time) = @_;
992
993 # RFC 822/1123 format
994 my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
995
996 sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
997 $weekday[$wday], $mday, $month[$mon], $year + 1900,
998 $H, $M, $S;
999}
1000
1001sub parse_date($) {
1002 my ($date) = @_;
1003
1004 my ($d, $m, $y, $H, $M, $S);
1005
1006 if ($date =~ /^[A-Z][a-z][a-z], ([0-9][0-9])[\- ]([A-Z][a-z][a-z])[\- ]([0-9][0-9][0-9][0-9]) ([0-9][0-9]):([0-9][0-9]):([0-9][0-9]) GMT$/) {
1007 # RFC 822/1123, required by RFC 2616 (with " ")
1008 # cookie dates (with "-")
1009
1010 ($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
1011
1012 } elsif ($date =~ /^[A-Z][a-z]+, ([0-9][0-9])-([A-Z][a-z][a-z])-([0-9][0-9]) ([0-9][0-9]):([0-9][0-9]):([0-9][0-9]) GMT$/) {
1013 # RFC 850
1014 ($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
1015
1016 } elsif ($date =~ /^[A-Z][a-z][a-z] ([A-Z][a-z][a-z]) ([0-9 ][0-9]) ([0-9][0-9]):([0-9][0-9]):([0-9][0-9]) ([0-9][0-9][0-9][0-9])$/) {
1017 # ISO C's asctime
1018 ($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
1019 }
1020 # other formats fail in the loop below
1021
1022 for (0..11) {
1023 if ($m eq $month[$_]) {
1024 require Time::Local;
1025 return Time::Local::timegm ($S, $M, $H, $d, $_, $y);
1026 }
1027 }
1028
1029 undef
1030}
837 1031
838sub set_proxy($) { 1032sub set_proxy($) {
839 if (length $_[0]) { 1033 if (length $_[0]) {
840 $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix 1034 $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix
841 or Carp::croak "$_[0]: invalid proxy URL"; 1035 or Carp::croak "$_[0]: invalid proxy URL";
848# initialise proxy from environment 1042# initialise proxy from environment
849eval { 1043eval {
850 set_proxy $ENV{http_proxy}; 1044 set_proxy $ENV{http_proxy};
851}; 1045};
852 1046
1047=head2 SOCKS PROXIES
1048
1049Socks proxies are not directly supported by AnyEvent::HTTP. You can
1050compile your perl to support socks, or use an external program such as
1051F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
1052transparently.
1053
1054Alternatively, for AnyEvent::HTTP only, you can use your own
1055C<tcp_connect> function that does the proxy handshake - here is an example
1056that works with socks4a proxies:
1057
1058 use Errno;
1059 use AnyEvent::Util;
1060 use AnyEvent::Socket;
1061 use AnyEvent::Handle;
1062
1063 # host, port and username of/for your socks4a proxy
1064 my $socks_host = "10.0.0.23";
1065 my $socks_port = 9050;
1066 my $socks_user = "";
1067
1068 sub socks4a_connect {
1069 my ($host, $port, $connect_cb, $prepare_cb) = @_;
1070
1071 my $hdl = new AnyEvent::Handle
1072 connect => [$socks_host, $socks_port],
1073 on_prepare => sub { $prepare_cb->($_[0]{fh}) },
1074 on_error => sub { $connect_cb->() },
1075 ;
1076
1077 $hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
1078
1079 $hdl->push_read (chunk => 8, sub {
1080 my ($hdl, $chunk) = @_;
1081 my ($status, $port, $ipn) = unpack "xCna4", $chunk;
1082
1083 if ($status == 0x5a) {
1084 $connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
1085 } else {
1086 $! = Errno::ENXIO; $connect_cb->();
1087 }
1088 });
1089
1090 $hdl
1091 }
1092
1093Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
1094possibly after switching off other proxy types:
1095
1096 AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
1097
1098 http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
1099 my ($data, $headers) = @_;
1100 ...
1101 };
1102
853=head1 SEE ALSO 1103=head1 SEE ALSO
854 1104
855L<AnyEvent>. 1105L<AnyEvent>.
856 1106
857=head1 AUTHOR 1107=head1 AUTHOR

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines