* methods/http.cc:
[ntk/apt.git] / methods / http.cc
CommitLineData
be4401bf
AL
1// -*- mode: cpp; mode: fold -*-
2// Description /*{{{*/
2cbcabd8 3// $Id: http.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
be4401bf
AL
4/* ######################################################################
5
ae58a985 6 HTTP Acquire Method - This is the HTTP aquire method for APT.
be4401bf
AL
7
8 It uses HTTP/1.1 and many of the fancy options there-in, such as
e836f356
AL
9 pipelining, range, if-range and so on.
10
11 It is based on a doubly buffered select loop. A groupe of requests are
be4401bf
AL
12 fed into a single output buffer that is constantly fed out the
13 socket. This provides ideal pipelining as in many cases all of the
14 requests will fit into a single packet. The input socket is buffered
e836f356 15 the same way and fed into the fd for the file (may be a pipe in future).
be4401bf
AL
16
17 This double buffering provides fairly substantial transfer rates,
18 compared to wget the http method is about 4% faster. Most importantly,
19 when HTTP is compared with FTP as a protocol the speed difference is
20 huge. In tests over the internet from two sites to llug (via ATM) this
21 program got 230k/s sustained http transfer rates. FTP on the other
22 hand topped out at 170k/s. That combined with the time to setup the
23 FTP connection makes HTTP a vastly superior protocol.
24
25 ##################################################################### */
26 /*}}}*/
27// Include Files /*{{{*/
28#include <apt-pkg/fileutl.h>
29#include <apt-pkg/acquire-method.h>
30#include <apt-pkg/error.h>
63b1700f 31#include <apt-pkg/hashes.h>
592b7800 32#include <apt-pkg/netrc.h>
be4401bf
AL
33
34#include <sys/stat.h>
35#include <sys/time.h>
36#include <utime.h>
37#include <unistd.h>
492f957a 38#include <signal.h>
be4401bf 39#include <stdio.h>
65a1e968 40#include <errno.h>
42195eb2
AL
41#include <string.h>
42#include <iostream>
15d7e515 43#include <map>
d77559ac 44#include <apti18n.h>
be4401bf 45
592b7800 46
be4401bf 47// Internet stuff
0837bd25 48#include <netdb.h>
be4401bf 49
59b46c41 50#include "config.h"
0837bd25 51#include "connect.h"
934b6582 52#include "rfc2553emu.h"
be4401bf
AL
53#include "http.h"
54 /*}}}*/
42195eb2 55using namespace std;
be4401bf 56
492f957a
AL
57string HttpMethod::FailFile;
58int HttpMethod::FailFd = -1;
59time_t HttpMethod::FailTime = 0;
c37030c2 60unsigned long PipelineDepth = 10;
3000ccea 61unsigned long TimeOut = 120;
15d7e515 62bool AllowRedirect = false;
c98b1307 63bool Debug = false;
c37030c2 64URI Proxy;
492f957a 65
7c6e2dc7
MV
66unsigned long CircleBuf::BwReadLimit=0;
67unsigned long CircleBuf::BwTickReadData=0;
68struct timeval CircleBuf::BwReadTick={0,0};
69const unsigned int CircleBuf::BW_HZ=10;
70
be4401bf
AL
71// CircleBuf::CircleBuf - Circular input buffer /*{{{*/
72// ---------------------------------------------------------------------
73/* */
63b1700f 74CircleBuf::CircleBuf(unsigned long Size) : Size(Size), Hash(0)
be4401bf
AL
75{
76 Buf = new unsigned char[Size];
77 Reset();
7c6e2dc7
MV
78
79 CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
be4401bf
AL
80}
81 /*}}}*/
82// CircleBuf::Reset - Reset to the default state /*{{{*/
83// ---------------------------------------------------------------------
84/* */
85void CircleBuf::Reset()
86{
87 InP = 0;
88 OutP = 0;
89 StrPos = 0;
90 MaxGet = (unsigned int)-1;
91 OutQueue = string();
63b1700f 92 if (Hash != 0)
be4401bf 93 {
63b1700f
AL
94 delete Hash;
95 Hash = new Hashes;
be4401bf
AL
96 }
97};
98 /*}}}*/
99// CircleBuf::Read - Read from a FD into the circular buffer /*{{{*/
100// ---------------------------------------------------------------------
101/* This fills up the buffer with as much data as is in the FD, assuming it
102 is non-blocking.. */
103bool CircleBuf::Read(int Fd)
104{
7c6e2dc7
MV
105 unsigned long BwReadMax;
106
be4401bf
AL
107 while (1)
108 {
109 // Woops, buffer is full
110 if (InP - OutP == Size)
111 return true;
7c6e2dc7
MV
112
113 // what's left to read in this tick
114 BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
115
116 if(CircleBuf::BwReadLimit) {
117 struct timeval now;
118 gettimeofday(&now,0);
119
120 unsigned long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
121 now.tv_usec-CircleBuf::BwReadTick.tv_usec;
122 if(d > 1000000/BW_HZ) {
123 CircleBuf::BwReadTick = now;
124 CircleBuf::BwTickReadData = 0;
125 }
126
127 if(CircleBuf::BwTickReadData >= BwReadMax) {
128 usleep(1000000/BW_HZ);
129 return true;
130 }
131 }
132
be4401bf
AL
133 // Write the buffer segment
134 int Res;
7c6e2dc7
MV
135 if(CircleBuf::BwReadLimit) {
136 Res = read(Fd,Buf + (InP%Size),
137 BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
138 } else
139 Res = read(Fd,Buf + (InP%Size),LeftRead());
be4401bf 140
7c6e2dc7
MV
141 if(Res > 0 && BwReadLimit > 0)
142 CircleBuf::BwTickReadData += Res;
143
be4401bf
AL
144 if (Res == 0)
145 return false;
146 if (Res < 0)
147 {
148 if (errno == EAGAIN)
149 return true;
150 return false;
151 }
152
153 if (InP == 0)
154 gettimeofday(&Start,0);
155 InP += Res;
156 }
157}
158 /*}}}*/
159// CircleBuf::Read - Put the string into the buffer /*{{{*/
160// ---------------------------------------------------------------------
161/* This will hold the string in and fill the buffer with it as it empties */
162bool CircleBuf::Read(string Data)
163{
164 OutQueue += Data;
165 FillOut();
166 return true;
167}
168 /*}}}*/
169// CircleBuf::FillOut - Fill the buffer from the output queue /*{{{*/
170// ---------------------------------------------------------------------
171/* */
172void CircleBuf::FillOut()
173{
174 if (OutQueue.empty() == true)
175 return;
176 while (1)
177 {
178 // Woops, buffer is full
179 if (InP - OutP == Size)
180 return;
181
182 // Write the buffer segment
183 unsigned long Sz = LeftRead();
184 if (OutQueue.length() - StrPos < Sz)
185 Sz = OutQueue.length() - StrPos;
42195eb2 186 memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
be4401bf
AL
187
188 // Advance
189 StrPos += Sz;
190 InP += Sz;
191 if (OutQueue.length() == StrPos)
192 {
193 StrPos = 0;
194 OutQueue = "";
195 return;
196 }
197 }
198}
199 /*}}}*/
200// CircleBuf::Write - Write from the buffer into a FD /*{{{*/
201// ---------------------------------------------------------------------
202/* This empties the buffer into the FD. */
203bool CircleBuf::Write(int Fd)
204{
205 while (1)
206 {
207 FillOut();
208
209 // Woops, buffer is empty
210 if (OutP == InP)
211 return true;
212
213 if (OutP == MaxGet)
214 return true;
215
216 // Write the buffer segment
217 int Res;
218 Res = write(Fd,Buf + (OutP%Size),LeftWrite());
219
220 if (Res == 0)
221 return false;
222 if (Res < 0)
223 {
224 if (errno == EAGAIN)
225 return true;
226
227 return false;
228 }
229
63b1700f
AL
230 if (Hash != 0)
231 Hash->Add(Buf + (OutP%Size),Res);
be4401bf
AL
232
233 OutP += Res;
234 }
235}
236 /*}}}*/
237// CircleBuf::WriteTillEl - Write from the buffer to a string /*{{{*/
238// ---------------------------------------------------------------------
239/* This copies till the first empty line */
240bool CircleBuf::WriteTillEl(string &Data,bool Single)
241{
242 // We cheat and assume it is unneeded to have more than one buffer load
243 for (unsigned long I = OutP; I < InP; I++)
244 {
245 if (Buf[I%Size] != '\n')
246 continue;
2cbcabd8 247 ++I;
be4401bf
AL
248
249 if (Single == false)
250 {
2cbcabd8
AL
251 if (I < InP && Buf[I%Size] == '\r')
252 ++I;
927c393f
MV
253 if (I >= InP || Buf[I%Size] != '\n')
254 continue;
255 ++I;
be4401bf
AL
256 }
257
be4401bf
AL
258 Data = "";
259 while (OutP < I)
260 {
261 unsigned long Sz = LeftWrite();
262 if (Sz == 0)
263 return false;
927c393f 264 if (I - OutP < Sz)
be4401bf
AL
265 Sz = I - OutP;
266 Data += string((char *)(Buf + (OutP%Size)),Sz);
267 OutP += Sz;
268 }
269 return true;
270 }
271 return false;
272}
273 /*}}}*/
274// CircleBuf::Stats - Print out stats information /*{{{*/
275// ---------------------------------------------------------------------
276/* */
277void CircleBuf::Stats()
278{
279 if (InP == 0)
280 return;
281
282 struct timeval Stop;
283 gettimeofday(&Stop,0);
284/* float Diff = Stop.tv_sec - Start.tv_sec +
285 (float)(Stop.tv_usec - Start.tv_usec)/1000000;
286 clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
287}
288 /*}}}*/
289
290// ServerState::ServerState - Constructor /*{{{*/
291// ---------------------------------------------------------------------
292/* */
293ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
3000ccea 294 In(64*1024), Out(4*1024),
be4401bf
AL
295 ServerName(Srv)
296{
297 Reset();
298}
299 /*}}}*/
300// ServerState::Open - Open a connection to the server /*{{{*/
301// ---------------------------------------------------------------------
302/* This opens a connection to the server. */
be4401bf
AL
303bool ServerState::Open()
304{
92e889c8
AL
305 // Use the already open connection if possible.
306 if (ServerFd != -1)
307 return true;
308
be4401bf 309 Close();
492f957a
AL
310 In.Reset();
311 Out.Reset();
e836f356
AL
312 Persistent = true;
313
492f957a 314 // Determine the proxy setting
788a8f42
EL
315 string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
316 if (!SpecificProxy.empty())
492f957a 317 {
788a8f42
EL
318 if (SpecificProxy == "DIRECT")
319 Proxy = "";
320 else
321 Proxy = SpecificProxy;
352c2768 322 }
492f957a 323 else
788a8f42
EL
324 {
325 string DefProxy = _config->Find("Acquire::http::Proxy");
326 if (!DefProxy.empty())
327 {
328 Proxy = DefProxy;
329 }
330 else
331 {
332 char* result = getenv("http_proxy");
333 Proxy = result ? result : "";
334 }
335 }
352c2768 336
f8081133 337 // Parse no_proxy, a , separated list of domains
9e2a06ff
AL
338 if (getenv("no_proxy") != 0)
339 {
f8081133
AL
340 if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
341 Proxy = "";
342 }
343
492f957a 344 // Determine what host and port to use based on the proxy settings
934b6582 345 int Port = 0;
492f957a 346 string Host;
dd1fd92b 347 if (Proxy.empty() == true || Proxy.Host.empty() == true)
be4401bf 348 {
92e889c8
AL
349 if (ServerName.Port != 0)
350 Port = ServerName.Port;
be4401bf
AL
351 Host = ServerName.Host;
352 }
353 else
354 {
92e889c8
AL
355 if (Proxy.Port != 0)
356 Port = Proxy.Port;
be4401bf
AL
357 Host = Proxy.Host;
358 }
359
0837bd25 360 // Connect to the remote server
9505213b 361 if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
0837bd25 362 return false;
3000ccea 363
be4401bf
AL
364 return true;
365}
366 /*}}}*/
367// ServerState::Close - Close a connection to the server /*{{{*/
368// ---------------------------------------------------------------------
369/* */
370bool ServerState::Close()
371{
372 close(ServerFd);
373 ServerFd = -1;
be4401bf
AL
374 return true;
375}
376 /*}}}*/
377// ServerState::RunHeaders - Get the headers before the data /*{{{*/
378// ---------------------------------------------------------------------
38965a34
MV
379/* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
380 parse error occurred */
92e889c8 381int ServerState::RunHeaders()
be4401bf
AL
382{
383 State = Header;
384
519c5591 385 Owner->Status(_("Waiting for headers"));
be4401bf
AL
386
387 Major = 0;
388 Minor = 0;
389 Result = 0;
390 Size = 0;
391 StartPos = 0;
92e889c8
AL
392 Encoding = Closes;
393 HaveContent = false;
be4401bf
AL
394 time(&Date);
395
396 do
397 {
398 string Data;
399 if (In.WriteTillEl(Data) == false)
400 continue;
9d95e726
AL
401
402 if (Debug == true)
403 clog << Data;
be4401bf
AL
404
405 for (string::const_iterator I = Data.begin(); I < Data.end(); I++)
406 {
407 string::const_iterator J = I;
408 for (; J != Data.end() && *J != '\n' && *J != '\r';J++);
42195eb2 409 if (HeaderLine(string(I,J)) == false)
92e889c8 410 return 2;
be4401bf
AL
411 I = J;
412 }
e836f356 413
b2e465d6
AL
414 // 100 Continue is a Nop...
415 if (Result == 100)
416 continue;
417
e836f356
AL
418 // Tidy up the connection persistance state.
419 if (Encoding == Closes && HaveContent == true)
420 Persistent = false;
421
92e889c8 422 return 0;
be4401bf
AL
423 }
424 while (Owner->Go(false,this) == true);
e836f356 425
92e889c8 426 return 1;
be4401bf
AL
427}
428 /*}}}*/
429// ServerState::RunData - Transfer the data from the socket /*{{{*/
430// ---------------------------------------------------------------------
431/* */
432bool ServerState::RunData()
433{
434 State = Data;
435
436 // Chunked transfer encoding is fun..
437 if (Encoding == Chunked)
438 {
439 while (1)
440 {
441 // Grab the block size
442 bool Last = true;
443 string Data;
444 In.Limit(-1);
445 do
446 {
447 if (In.WriteTillEl(Data,true) == true)
448 break;
449 }
450 while ((Last = Owner->Go(false,this)) == true);
451
452 if (Last == false)
453 return false;
454
455 // See if we are done
456 unsigned long Len = strtol(Data.c_str(),0,16);
457 if (Len == 0)
458 {
459 In.Limit(-1);
460
461 // We have to remove the entity trailer
462 Last = true;
463 do
464 {
465 if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
466 break;
467 }
468 while ((Last = Owner->Go(false,this)) == true);
469 if (Last == false)
470 return false;
e1b96638 471 return !_error->PendingError();
be4401bf
AL
472 }
473
474 // Transfer the block
475 In.Limit(Len);
476 while (Owner->Go(true,this) == true)
477 if (In.IsLimit() == true)
478 break;
479
480 // Error
481 if (In.IsLimit() == false)
482 return false;
483
484 // The server sends an extra new line before the next block specifier..
485 In.Limit(-1);
486 Last = true;
487 do
488 {
489 if (In.WriteTillEl(Data,true) == true)
490 break;
491 }
492 while ((Last = Owner->Go(false,this)) == true);
493 if (Last == false)
494 return false;
92e889c8 495 }
be4401bf
AL
496 }
497 else
498 {
499 /* Closes encoding is used when the server did not specify a size, the
500 loss of the connection means we are done */
501 if (Encoding == Closes)
502 In.Limit(-1);
503 else
504 In.Limit(Size - StartPos);
505
506 // Just transfer the whole block.
507 do
508 {
509 if (In.IsLimit() == false)
510 continue;
511
512 In.Limit(-1);
e1b96638 513 return !_error->PendingError();
be4401bf
AL
514 }
515 while (Owner->Go(true,this) == true);
516 }
517
e1b96638 518 return Owner->Flush(this) && !_error->PendingError();
be4401bf
AL
519}
520 /*}}}*/
521// ServerState::HeaderLine - Process a header line /*{{{*/
522// ---------------------------------------------------------------------
523/* */
524bool ServerState::HeaderLine(string Line)
525{
526 if (Line.empty() == true)
527 return true;
30456e14 528
be4401bf
AL
529 // The http server might be trying to do something evil.
530 if (Line.length() >= MAXLEN)
dc738e7a 531 return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
be4401bf
AL
532
533 string::size_type Pos = Line.find(' ');
534 if (Pos == string::npos || Pos+1 > Line.length())
c901051d
AL
535 {
536 // Blah, some servers use "connection:closes", evil.
537 Pos = Line.find(':');
538 if (Pos == string::npos || Pos + 2 > Line.length())
dc738e7a 539 return _error->Error(_("Bad header line"));
c901051d
AL
540 Pos++;
541 }
be4401bf 542
c901051d
AL
543 // Parse off any trailing spaces between the : and the next word.
544 string::size_type Pos2 = Pos;
545 while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
546 Pos2++;
547
548 string Tag = string(Line,0,Pos);
549 string Val = string(Line,Pos2);
550
42195eb2 551 if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
be4401bf
AL
552 {
553 // Evil servers return no version
554 if (Line[4] == '/')
555 {
dda7233c 556 if (sscanf(Line.c_str(),"HTTP/%u.%u %u%[^\n]",&Major,&Minor,
be4401bf 557 &Result,Code) != 4)
db0db9fe 558 return _error->Error(_("The HTTP server sent an invalid reply header"));
be4401bf
AL
559 }
560 else
561 {
562 Major = 0;
563 Minor = 9;
dda7233c 564 if (sscanf(Line.c_str(),"HTTP %u%[^\n]",&Result,Code) != 2)
db0db9fe 565 return _error->Error(_("The HTTP server sent an invalid reply header"));
be4401bf 566 }
e836f356
AL
567
568 /* Check the HTTP response header to get the default persistance
569 state. */
570 if (Major < 1)
571 Persistent = false;
572 else
573 {
574 if (Major == 1 && Minor <= 0)
575 Persistent = false;
576 else
577 Persistent = true;
578 }
b2e465d6 579
be4401bf
AL
580 return true;
581 }
582
92e889c8 583 if (stringcasecmp(Tag,"Content-Length:") == 0)
be4401bf
AL
584 {
585 if (Encoding == Closes)
586 Encoding = Stream;
92e889c8 587 HaveContent = true;
be4401bf
AL
588
589 // The length is already set from the Content-Range header
590 if (StartPos != 0)
591 return true;
592
593 if (sscanf(Val.c_str(),"%lu",&Size) != 1)
db0db9fe 594 return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
be4401bf
AL
595 return true;
596 }
597
92e889c8
AL
598 if (stringcasecmp(Tag,"Content-Type:") == 0)
599 {
600 HaveContent = true;
601 return true;
602 }
603
604 if (stringcasecmp(Tag,"Content-Range:") == 0)
be4401bf 605 {
92e889c8
AL
606 HaveContent = true;
607
be4401bf 608 if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2)
db0db9fe 609 return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
be4401bf 610 if ((unsigned)StartPos > Size)
db0db9fe 611 return _error->Error(_("This HTTP server has broken range support"));
be4401bf
AL
612 return true;
613 }
614
92e889c8 615 if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
be4401bf 616 {
92e889c8
AL
617 HaveContent = true;
618 if (stringcasecmp(Val,"chunked") == 0)
e836f356 619 Encoding = Chunked;
be4401bf
AL
620 return true;
621 }
622
e836f356
AL
623 if (stringcasecmp(Tag,"Connection:") == 0)
624 {
625 if (stringcasecmp(Val,"close") == 0)
626 Persistent = false;
627 if (stringcasecmp(Val,"keep-alive") == 0)
628 Persistent = true;
629 return true;
630 }
631
92e889c8 632 if (stringcasecmp(Tag,"Last-Modified:") == 0)
be4401bf
AL
633 {
634 if (StrToTime(Val,Date) == false)
dc738e7a 635 return _error->Error(_("Unknown date format"));
be4401bf
AL
636 return true;
637 }
638
15d7e515
MV
639 if (stringcasecmp(Tag,"Location:") == 0)
640 {
641 Location = Val;
642 return true;
643 }
644
be4401bf
AL
645 return true;
646}
647 /*}}}*/
648
649// HttpMethod::SendReq - Send the HTTP request /*{{{*/
650// ---------------------------------------------------------------------
651/* This places the http request in the outbound buffer */
652void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
653{
654 URI Uri = Itm->Uri;
c1a22377 655
be4401bf 656 // The HTTP server expects a hostname with a trailing :port
c1a22377 657 char Buf[1000];
be4401bf
AL
658 string ProperHost = Uri.Host;
659 if (Uri.Port != 0)
660 {
661 sprintf(Buf,":%u",Uri.Port);
662 ProperHost += Buf;
663 }
664
c1a22377
AL
665 // Just in case.
666 if (Itm->Uri.length() >= sizeof(Buf))
667 abort();
668
492f957a
AL
669 /* Build the request. We include a keep-alive header only for non-proxy
670 requests. This is to tweak old http/1.0 servers that do support keep-alive
671 but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
672 will glitch HTTP/1.0 proxies because they do not filter it out and
673 pass it on, HTTP/1.1 says the connection should default to keep alive
674 and we expect the proxy to do this */
02b7ddb1 675 if (Proxy.empty() == true || Proxy.Host.empty())
be4401bf 676 sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
a4edf53b 677 QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
be4401bf 678 else
c1a22377
AL
679 {
680 /* Generate a cache control header if necessary. We place a max
681 cache age on index files, optionally set a no-cache directive
682 and a no-store directive for archives. */
be4401bf
AL
683 sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
684 Itm->Uri.c_str(),ProperHost.c_str());
c9cd3b70
MV
685 }
686 // generate a cache control header (if needed)
687 if (_config->FindB("Acquire::http::No-Cache",false) == true)
688 {
689 strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
690 }
691 else
692 {
693 if (Itm->IndexFile == true)
c1a22377 694 {
c9cd3b70
MV
695 sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
696 _config->FindI("Acquire::http::Max-Age",0));
697 }
698 else
699 {
700 if (_config->FindB("Acquire::http::No-Store",false) == true)
701 strcat(Buf,"Cache-Control: no-store\r\n");
c1a22377
AL
702 }
703 }
106e6740 704
c1a22377 705
be4401bf 706 string Req = Buf;
492f957a 707
be4401bf
AL
708 // Check for a partial file
709 struct stat SBuf;
710 if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
711 {
712 // In this case we send an if-range query with a range header
1ae93c94 713 sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",(long)SBuf.st_size - 1,
be4401bf
AL
714 TimeRFC1123(SBuf.st_mtime).c_str());
715 Req += Buf;
716 }
717 else
718 {
719 if (Itm->LastModified != 0)
720 {
721 sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
722 Req += Buf;
723 }
724 }
725
8d64c395
AL
726 if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
727 Req += string("Proxy-Authorization: Basic ") +
728 Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
be4401bf 729
1de1f703 730 maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
b2e465d6 731 if (Uri.User.empty() == false || Uri.Password.empty() == false)
592b7800 732 {
b2e465d6
AL
733 Req += string("Authorization: Basic ") +
734 Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
592b7800 735 }
9f542bae
DK
736 Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
737 "Debian APT-HTTP/1.3 ("VERSION")") + "\r\n\r\n";
c98b1307
AL
738
739 if (Debug == true)
740 cerr << Req << endl;
c1a22377 741
be4401bf
AL
742 Out.Read(Req);
743}
744 /*}}}*/
745// HttpMethod::Go - Run a single loop /*{{{*/
746// ---------------------------------------------------------------------
747/* This runs the select loop over the server FDs, Output file FDs and
748 stdin. */
749bool HttpMethod::Go(bool ToFile,ServerState *Srv)
750{
751 // Server has closed the connection
8195ae46
AL
752 if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
753 ToFile == false))
be4401bf
AL
754 return false;
755
d955fe80 756 fd_set rfds,wfds;
be4401bf
AL
757 FD_ZERO(&rfds);
758 FD_ZERO(&wfds);
be4401bf 759
e836f356
AL
760 /* Add the server. We only send more requests if the connection will
761 be persisting */
762 if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
763 && Srv->Persistent == true)
be4401bf 764 FD_SET(Srv->ServerFd,&wfds);
e836f356 765 if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
be4401bf
AL
766 FD_SET(Srv->ServerFd,&rfds);
767
768 // Add the file
769 int FileFD = -1;
770 if (File != 0)
771 FileFD = File->Fd();
772
773 if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
774 FD_SET(FileFD,&wfds);
775
776 // Add stdin
777 FD_SET(STDIN_FILENO,&rfds);
778
be4401bf
AL
779 // Figure out the max fd
780 int MaxFd = FileFD;
781 if (MaxFd < Srv->ServerFd)
782 MaxFd = Srv->ServerFd;
8195ae46 783
be4401bf
AL
784 // Select
785 struct timeval tv;
3000ccea 786 tv.tv_sec = TimeOut;
be4401bf
AL
787 tv.tv_usec = 0;
788 int Res = 0;
d955fe80 789 if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
c37b9502
AL
790 {
791 if (errno == EINTR)
792 return true;
dc738e7a 793 return _error->Errno("select",_("Select failed"));
c37b9502 794 }
be4401bf
AL
795
796 if (Res == 0)
797 {
dc738e7a 798 _error->Error(_("Connection timed out"));
be4401bf
AL
799 return ServerDie(Srv);
800 }
801
be4401bf
AL
802 // Handle server IO
803 if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
804 {
805 errno = 0;
806 if (Srv->In.Read(Srv->ServerFd) == false)
807 return ServerDie(Srv);
808 }
809
810 if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
811 {
812 errno = 0;
813 if (Srv->Out.Write(Srv->ServerFd) == false)
814 return ServerDie(Srv);
815 }
816
817 // Send data to the file
818 if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
819 {
820 if (Srv->In.Write(FileFD) == false)
dc738e7a 821 return _error->Errno("write",_("Error writing to output file"));
be4401bf
AL
822 }
823
824 // Handle commands from APT
825 if (FD_ISSET(STDIN_FILENO,&rfds))
826 {
6920216d 827 if (Run(true) != -1)
be4401bf
AL
828 exit(100);
829 }
830
831 return true;
832}
833 /*}}}*/
834// HttpMethod::Flush - Dump the buffer into the file /*{{{*/
835// ---------------------------------------------------------------------
836/* This takes the current input buffer from the Server FD and writes it
837 into the file */
838bool HttpMethod::Flush(ServerState *Srv)
839{
840 if (File != 0)
841 {
b57c8bb4
MV
842 // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
843 // can't be set
844 if (File->Name() != "/dev/null")
845 SetNonBlock(File->Fd(),false);
be4401bf
AL
846 if (Srv->In.WriteSpace() == false)
847 return true;
848
849 while (Srv->In.WriteSpace() == true)
850 {
851 if (Srv->In.Write(File->Fd()) == false)
dc738e7a 852 return _error->Errno("write",_("Error writing to file"));
92e889c8
AL
853 if (Srv->In.IsLimit() == true)
854 return true;
be4401bf
AL
855 }
856
857 if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
858 return true;
859 }
860 return false;
861}
862 /*}}}*/
863// HttpMethod::ServerDie - The server has closed the connection. /*{{{*/
864// ---------------------------------------------------------------------
865/* */
866bool HttpMethod::ServerDie(ServerState *Srv)
867{
2b154e53
AL
868 unsigned int LErrno = errno;
869
be4401bf
AL
870 // Dump the buffer to the file
871 if (Srv->State == ServerState::Data)
872 {
b57c8bb4
MV
873 // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
874 // can't be set
875 if (File->Name() != "/dev/null")
876 SetNonBlock(File->Fd(),false);
be4401bf
AL
877 while (Srv->In.WriteSpace() == true)
878 {
879 if (Srv->In.Write(File->Fd()) == false)
dc738e7a 880 return _error->Errno("write",_("Error writing to the file"));
92e889c8
AL
881
882 // Done
883 if (Srv->In.IsLimit() == true)
884 return true;
be4401bf
AL
885 }
886 }
887
888 // See if this is because the server finished the data stream
889 if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
890 Srv->Encoding != ServerState::Closes)
891 {
3d615484 892 Srv->Close();
2b154e53 893 if (LErrno == 0)
db0db9fe 894 return _error->Error(_("Error reading from server. Remote end closed connection"));
2b154e53 895 errno = LErrno;
dc738e7a 896 return _error->Errno("read",_("Error reading from server"));
be4401bf
AL
897 }
898 else
899 {
900 Srv->In.Limit(-1);
901
902 // Nothing left in the buffer
903 if (Srv->In.WriteSpace() == false)
904 return false;
905
906 // We may have got multiple responses back in one packet..
907 Srv->Close();
908 return true;
909 }
910
911 return false;
912}
913 /*}}}*/
914// HttpMethod::DealWithHeaders - Handle the retrieved header data /*{{{*/
915// ---------------------------------------------------------------------
916/* We look at the header data we got back from the server and decide what
917 to do. Returns
918 0 - File is open,
919 1 - IMS hit
92e889c8 920 3 - Unrecoverable error
94235cfb 921 4 - Error with error content page
15d7e515
MV
922 5 - Unrecoverable non-server error (close the connection)
923 6 - Try again with a new or changed URI
924 */
be4401bf
AL
925int HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
926{
927 // Not Modified
928 if (Srv->Result == 304)
929 {
930 unlink(Queue->DestFile.c_str());
931 Res.IMSHit = true;
932 Res.LastModified = Queue->LastModified;
933 return 1;
934 }
935
15d7e515
MV
936 /* Redirect
937 *
938 * Note that it is only OK for us to treat all redirection the same
939 * because we *always* use GET, not other HTTP methods. There are
940 * three redirection codes for which it is not appropriate that we
941 * redirect. Pass on those codes so the error handling kicks in.
942 */
943 if (AllowRedirect
944 && (Srv->Result > 300 && Srv->Result < 400)
945 && (Srv->Result != 300 // Multiple Choices
946 && Srv->Result != 304 // Not Modified
947 && Srv->Result != 306)) // (Not part of HTTP/1.1, reserved)
948 {
949 if (!Srv->Location.empty())
950 {
951 NextURI = Srv->Location;
952 return 6;
953 }
954 /* else pass through for error message */
955 }
956
be4401bf
AL
957 /* We have a reply we dont handle. This should indicate a perm server
958 failure */
959 if (Srv->Result < 200 || Srv->Result >= 300)
960 {
961 _error->Error("%u %s",Srv->Result,Srv->Code);
92e889c8
AL
962 if (Srv->HaveContent == true)
963 return 4;
be4401bf
AL
964 return 3;
965 }
966
967 // This is some sort of 2xx 'data follows' reply
968 Res.LastModified = Srv->Date;
969 Res.Size = Srv->Size;
970
971 // Open the file
972 delete File;
973 File = new FileFd(Queue->DestFile,FileFd::WriteAny);
974 if (_error->PendingError() == true)
94235cfb 975 return 5;
492f957a
AL
976
977 FailFile = Queue->DestFile;
30b30ec1 978 FailFile.c_str(); // Make sure we dont do a malloc in the signal handler
492f957a
AL
979 FailFd = File->Fd();
980 FailTime = Srv->Date;
981
be4401bf
AL
982 // Set the expected size
983 if (Srv->StartPos >= 0)
984 {
985 Res.ResumePoint = Srv->StartPos;
9b5d79ec
MV
986 if (ftruncate(File->Fd(),Srv->StartPos) < 0)
987 _error->Errno("ftruncate", _("Failed to truncate file"));
be4401bf
AL
988 }
989
990 // Set the start point
991 lseek(File->Fd(),0,SEEK_END);
992
63b1700f
AL
993 delete Srv->In.Hash;
994 Srv->In.Hash = new Hashes;
be4401bf 995
63b1700f 996 // Fill the Hash if the file is non-empty (resume)
be4401bf
AL
997 if (Srv->StartPos > 0)
998 {
999 lseek(File->Fd(),0,SEEK_SET);
63b1700f 1000 if (Srv->In.Hash->AddFD(File->Fd(),Srv->StartPos) == false)
be4401bf 1001 {
dc738e7a 1002 _error->Errno("read",_("Problem hashing file"));
94235cfb 1003 return 5;
be4401bf
AL
1004 }
1005 lseek(File->Fd(),0,SEEK_END);
1006 }
1007
1008 SetNonBlock(File->Fd(),true);
1009 return 0;
1010}
1011 /*}}}*/
492f957a
AL
1012// HttpMethod::SigTerm - Handle a fatal signal /*{{{*/
1013// ---------------------------------------------------------------------
1014/* This closes and timestamps the open file. This is neccessary to get
1015 resume behavoir on user abort */
1016void HttpMethod::SigTerm(int)
1017{
1018 if (FailFd == -1)
ffe9323a 1019 _exit(100);
492f957a
AL
1020 close(FailFd);
1021
1022 // Timestamp
1023 struct utimbuf UBuf;
492f957a
AL
1024 UBuf.actime = FailTime;
1025 UBuf.modtime = FailTime;
1026 utime(FailFile.c_str(),&UBuf);
1027
ffe9323a 1028 _exit(100);
492f957a
AL
1029}
1030 /*}}}*/
5cb5d8dc
AL
1031// HttpMethod::Fetch - Fetch an item /*{{{*/
1032// ---------------------------------------------------------------------
1033/* This adds an item to the pipeline. We keep the pipeline at a fixed
1034 depth. */
1035bool HttpMethod::Fetch(FetchItem *)
1036{
1037 if (Server == 0)
1038 return true;
3000ccea 1039
5cb5d8dc
AL
1040 // Queue the requests
1041 int Depth = -1;
f93d1355
AL
1042 for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1043 I = I->Next, Depth++)
5cb5d8dc 1044 {
f93d1355
AL
1045 // If pipelining is disabled, we only queue 1 request
1046 if (Server->Pipeline == false && Depth >= 0)
1047 break;
1048
5cb5d8dc
AL
1049 // Make sure we stick with the same server
1050 if (Server->Comp(I->Uri) == false)
1051 break;
5cb5d8dc 1052 if (QueueBack == I)
5cb5d8dc 1053 {
5cb5d8dc
AL
1054 QueueBack = I->Next;
1055 SendReq(I,Server->Out);
1056 continue;
f93d1355 1057 }
5cb5d8dc
AL
1058 }
1059
1060 return true;
1061};
1062 /*}}}*/
85f72a56
AL
1063// HttpMethod::Configuration - Handle a configuration message /*{{{*/
1064// ---------------------------------------------------------------------
1065/* We stash the desired pipeline depth */
1066bool HttpMethod::Configuration(string Message)
1067{
1068 if (pkgAcqMethod::Configuration(Message) == false)
1069 return false;
1070
15d7e515 1071 AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
30456e14
AL
1072 TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1073 PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1074 PipelineDepth);
c98b1307 1075 Debug = _config->FindB("Debug::Acquire::http",false);
3000ccea 1076
85f72a56
AL
1077 return true;
1078}
1079 /*}}}*/
492f957a 1080// HttpMethod::Loop - Main loop /*{{{*/
be4401bf
AL
1081// ---------------------------------------------------------------------
1082/* */
1083int HttpMethod::Loop()
1084{
15d7e515
MV
1085 typedef vector<string> StringVector;
1086 typedef vector<string>::iterator StringVectorIterator;
1087 map<string, StringVector> Redirected;
1088
492f957a
AL
1089 signal(SIGTERM,SigTerm);
1090 signal(SIGINT,SigTerm);
1091
5cb5d8dc 1092 Server = 0;
be4401bf 1093
92e889c8 1094 int FailCounter = 0;
be4401bf 1095 while (1)
2b154e53 1096 {
be4401bf
AL
1097 // We have no commands, wait for some to arrive
1098 if (Queue == 0)
1099 {
1100 if (WaitFd(STDIN_FILENO) == false)
1101 return 0;
1102 }
1103
6920216d
AL
1104 /* Run messages, we can accept 0 (no message) if we didn't
1105 do a WaitFd above.. Otherwise the FD is closed. */
1106 int Result = Run(true);
1107 if (Result != -1 && (Result != 0 || Queue == 0))
be4401bf
AL
1108 return 100;
1109
1110 if (Queue == 0)
1111 continue;
1112
1113 // Connect to the server
1114 if (Server == 0 || Server->Comp(Queue->Uri) == false)
1115 {
1116 delete Server;
1117 Server = new ServerState(Queue->Uri,this);
1118 }
e836f356
AL
1119 /* If the server has explicitly said this is the last connection
1120 then we pre-emptively shut down the pipeline and tear down
1121 the connection. This will speed up HTTP/1.0 servers a tad
1122 since we don't have to wait for the close sequence to
1123 complete */
1124 if (Server->Persistent == false)
1125 Server->Close();
1126
a7fb252c
AL
1127 // Reset the pipeline
1128 if (Server->ServerFd == -1)
1129 QueueBack = Queue;
1130
be4401bf
AL
1131 // Connnect to the host
1132 if (Server->Open() == false)
1133 {
43252d15 1134 Fail(true);
a1459f52
AL
1135 delete Server;
1136 Server = 0;
be4401bf
AL
1137 continue;
1138 }
be4401bf 1139
5cb5d8dc
AL
1140 // Fill the pipeline.
1141 Fetch(0);
1142
92e889c8
AL
1143 // Fetch the next URL header data from the server.
1144 switch (Server->RunHeaders())
be4401bf 1145 {
92e889c8
AL
1146 case 0:
1147 break;
1148
1149 // The header data is bad
1150 case 2:
1151 {
db0db9fe 1152 _error->Error(_("Bad header data"));
43252d15 1153 Fail(true);
b2e465d6 1154 RotateDNS();
92e889c8
AL
1155 continue;
1156 }
1157
1158 // The server closed a connection during the header get..
1159 default:
1160 case 1:
1161 {
1162 FailCounter++;
3d615484 1163 _error->Discard();
92e889c8 1164 Server->Close();
f93d1355
AL
1165 Server->Pipeline = false;
1166
2b154e53
AL
1167 if (FailCounter >= 2)
1168 {
dc738e7a 1169 Fail(_("Connection failed"),true);
2b154e53
AL
1170 FailCounter = 0;
1171 }
1172
b2e465d6 1173 RotateDNS();
92e889c8
AL
1174 continue;
1175 }
1176 };
5cb5d8dc 1177
be4401bf
AL
1178 // Decide what to do.
1179 FetchResult Res;
bfd22fc0 1180 Res.Filename = Queue->DestFile;
be4401bf
AL
1181 switch (DealWithHeaders(Res,Server))
1182 {
1183 // Ok, the file is Open
1184 case 0:
1185 {
1186 URIStart(Res);
1187
1188 // Run the data
492f957a
AL
1189 bool Result = Server->RunData();
1190
b2e465d6
AL
1191 /* If the server is sending back sizeless responses then fill in
1192 the size now */
1193 if (Res.Size == 0)
1194 Res.Size = File->Size();
1195
492f957a
AL
1196 // Close the file, destroy the FD object and timestamp it
1197 FailFd = -1;
1198 delete File;
1199 File = 0;
1200
1201 // Timestamp
1202 struct utimbuf UBuf;
1203 time(&UBuf.actime);
1204 UBuf.actime = Server->Date;
1205 UBuf.modtime = Server->Date;
1206 utime(Queue->DestFile.c_str(),&UBuf);
1207
1208 // Send status to APT
1209 if (Result == true)
92e889c8 1210 {
a7c835af 1211 Res.TakeHashes(*Server->In.Hash);
92e889c8
AL
1212 URIDone(Res);
1213 }
492f957a 1214 else
82d0afc2
MV
1215 {
1216 if (Server->ServerFd == -1)
1217 {
1218 FailCounter++;
1219 _error->Discard();
1220 Server->Close();
1221
1222 if (FailCounter >= 2)
9a52beaa 1223 {
82d0afc2
MV
1224 Fail(_("Connection failed"),true);
1225 FailCounter = 0;
9a52beaa 1226 }
82d0afc2
MV
1227
1228 QueueBack = Queue;
1229 }
1230 else
1231 Fail(true);
1232 }
be4401bf
AL
1233 break;
1234 }
1235
1236 // IMS hit
1237 case 1:
1238 {
1239 URIDone(Res);
1240 break;
1241 }
1242
1243 // Hard server error, not found or something
1244 case 3:
1245 {
1246 Fail();
1247 break;
1248 }
94235cfb
AL
1249
1250 // Hard internal error, kill the connection and fail
1251 case 5:
1252 {
a305f593
AL
1253 delete File;
1254 File = 0;
1255
94235cfb 1256 Fail();
b2e465d6 1257 RotateDNS();
94235cfb
AL
1258 Server->Close();
1259 break;
1260 }
92e889c8
AL
1261
1262 // We need to flush the data, the header is like a 404 w/ error text
1263 case 4:
1264 {
1265 Fail();
1266
1267 // Send to content to dev/null
1268 File = new FileFd("/dev/null",FileFd::WriteExists);
1269 Server->RunData();
1270 delete File;
1271 File = 0;
1272 break;
1273 }
be4401bf 1274
15d7e515
MV
1275 // Try again with a new URL
1276 case 6:
1277 {
1278 // Clear rest of response if there is content
1279 if (Server->HaveContent)
1280 {
1281 File = new FileFd("/dev/null",FileFd::WriteExists);
1282 Server->RunData();
1283 delete File;
1284 File = 0;
1285 }
1286
1287 /* Detect redirect loops. No more redirects are allowed
1288 after the same URI is seen twice in a queue item. */
1289 StringVector &R = Redirected[Queue->DestFile];
1290 bool StopRedirects = false;
1291 if (R.size() == 0)
1292 R.push_back(Queue->Uri);
1293 else if (R[0] == "STOP" || R.size() > 10)
1294 StopRedirects = true;
1295 else
1296 {
1297 for (StringVectorIterator I = R.begin(); I != R.end(); I++)
1298 if (Queue->Uri == *I)
1299 {
1300 R[0] = "STOP";
1301 break;
1302 }
1303
1304 R.push_back(Queue->Uri);
1305 }
1306
1307 if (StopRedirects == false)
1308 Redirect(NextURI);
1309 else
1310 Fail();
1311
1312 break;
1313 }
1314
be4401bf 1315 default:
dc738e7a 1316 Fail(_("Internal error"));
be4401bf 1317 break;
92e889c8
AL
1318 }
1319
1320 FailCounter = 0;
be4401bf
AL
1321 }
1322
1323 return 0;
1324}
1325 /*}}}*/
1326
1327int main()
1328{
049c0171 1329 setlocale(LC_ALL, "");
ab7f4d7c
MV
1330 // ignore SIGPIPE, this can happen on write() if the socket
1331 // closes the connection (this is dealt with via ServerDie())
1332 signal(SIGPIPE, SIG_IGN);
049c0171 1333
be4401bf 1334 HttpMethod Mth;
be4401bf
AL
1335 return Mth.Loop();
1336}
a305f593
AL
1337
1338