dispose http(s) 416 error page as non-content
[ntk/apt.git] / methods / server.cc
CommitLineData
7330f4df
DK
1// -*- mode: cpp; mode: fold -*-
2// Description /*{{{*/
3/* ######################################################################
4
5 HTTP and HTTPS share a lot of common code and these classes are
6 exactly the dumping ground for this common code
7
8 ##################################################################### */
9 /*}}}*/
10// Include Files /*{{{*/
11#include <config.h>
12
7330f4df
DK
13#include <apt-pkg/acquire-method.h>
14#include <apt-pkg/configuration.h>
15#include <apt-pkg/error.h>
453b82a3
DK
16#include <apt-pkg/fileutl.h>
17#include <apt-pkg/strutl.h>
7330f4df 18
453b82a3
DK
19#include <ctype.h>
20#include <signal.h>
21#include <stdio.h>
22#include <stdlib.h>
7330f4df
DK
23#include <sys/stat.h>
24#include <sys/time.h>
453b82a3 25#include <time.h>
7330f4df 26#include <unistd.h>
7330f4df 27#include <iostream>
453b82a3 28#include <limits>
7330f4df 29#include <map>
453b82a3
DK
30#include <string>
31#include <vector>
7330f4df 32
453b82a3 33#include "server.h"
7330f4df
DK
34
35#include <apti18n.h>
36 /*}}}*/
37using namespace std;
38
39string ServerMethod::FailFile;
40int ServerMethod::FailFd = -1;
41time_t ServerMethod::FailTime = 0;
42
43// ServerState::RunHeaders - Get the headers before the data /*{{{*/
44// ---------------------------------------------------------------------
45/* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
46 parse error occurred */
9622b211
MV
47ServerState::RunHeadersResult ServerState::RunHeaders(FileFd * const File,
48 const std::string &Uri)
7330f4df
DK
49{
50 State = Header;
51
52 Owner->Status(_("Waiting for headers"));
53
54 Major = 0;
55 Minor = 0;
56 Result = 0;
57 Size = 0;
92e8c1ff 58 JunkSize = 0;
7330f4df
DK
59 StartPos = 0;
60 Encoding = Closes;
61 HaveContent = false;
62 time(&Date);
63
64 do
65 {
66 string Data;
67 if (ReadHeaderLines(Data) == false)
68 continue;
69
70 if (Owner->Debug == true)
9622b211 71 clog << "Answer for: " << Uri << endl << Data;
7330f4df
DK
72
73 for (string::const_iterator I = Data.begin(); I < Data.end(); ++I)
74 {
75 string::const_iterator J = I;
76 for (; J != Data.end() && *J != '\n' && *J != '\r'; ++J);
77 if (HeaderLine(string(I,J)) == false)
78 return RUN_HEADERS_PARSE_ERROR;
79 I = J;
80 }
81
82 // 100 Continue is a Nop...
83 if (Result == 100)
84 continue;
85
1e3f4083 86 // Tidy up the connection persistence state.
7330f4df
DK
87 if (Encoding == Closes && HaveContent == true)
88 Persistent = false;
89
90 return RUN_HEADERS_OK;
91 }
92 while (LoadNextResponse(false, File) == true);
93
94 return RUN_HEADERS_IO_ERROR;
95}
96 /*}}}*/
97// ServerState::HeaderLine - Process a header line /*{{{*/
98// ---------------------------------------------------------------------
99/* */
100bool ServerState::HeaderLine(string Line)
101{
102 if (Line.empty() == true)
103 return true;
104
105 string::size_type Pos = Line.find(' ');
106 if (Pos == string::npos || Pos+1 > Line.length())
107 {
108 // Blah, some servers use "connection:closes", evil.
109 Pos = Line.find(':');
110 if (Pos == string::npos || Pos + 2 > Line.length())
111 return _error->Error(_("Bad header line"));
112 Pos++;
113 }
114
115 // Parse off any trailing spaces between the : and the next word.
116 string::size_type Pos2 = Pos;
117 while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
118 Pos2++;
d3e8fbb3 119
7330f4df
DK
120 string Tag = string(Line,0,Pos);
121 string Val = string(Line,Pos2);
d3e8fbb3 122
7330f4df
DK
123 if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
124 {
125 // Evil servers return no version
126 if (Line[4] == '/')
127 {
128 int const elements = sscanf(Line.c_str(),"HTTP/%3u.%3u %3u%359[^\n]",&Major,&Minor,&Result,Code);
129 if (elements == 3)
130 {
131 Code[0] = '\0';
132 if (Owner->Debug == true)
133 clog << "HTTP server doesn't give Reason-Phrase for " << Result << std::endl;
134 }
135 else if (elements != 4)
136 return _error->Error(_("The HTTP server sent an invalid reply header"));
137 }
138 else
139 {
140 Major = 0;
141 Minor = 9;
142 if (sscanf(Line.c_str(),"HTTP %3u%359[^\n]",&Result,Code) != 2)
143 return _error->Error(_("The HTTP server sent an invalid reply header"));
144 }
145
1e3f4083 146 /* Check the HTTP response header to get the default persistence
7330f4df
DK
147 state. */
148 if (Major < 1)
149 Persistent = false;
150 else
151 {
152 if (Major == 1 && Minor == 0)
153 Persistent = false;
154 else
155 Persistent = true;
156 }
157
158 return true;
d3e8fbb3
DK
159 }
160
7330f4df
DK
161 if (stringcasecmp(Tag,"Content-Length:") == 0)
162 {
163 if (Encoding == Closes)
164 Encoding = Stream;
165 HaveContent = true;
d3e8fbb3 166
92e8c1ff
DK
167 unsigned long long * SizePtr = &Size;
168 if (Result == 416)
169 SizePtr = &JunkSize;
7330f4df 170
92e8c1ff
DK
171 *SizePtr = strtoull(Val.c_str(), NULL, 10);
172 if (*SizePtr >= std::numeric_limits<unsigned long long>::max())
7330f4df 173 return _error->Errno("HeaderLine", _("The HTTP server sent an invalid Content-Length header"));
92e8c1ff 174 else if (*SizePtr == 0)
7330f4df
DK
175 HaveContent = false;
176 return true;
177 }
178
179 if (stringcasecmp(Tag,"Content-Type:") == 0)
180 {
181 HaveContent = true;
182 return true;
183 }
d3e8fbb3 184
7330f4df
DK
185 if (stringcasecmp(Tag,"Content-Range:") == 0)
186 {
187 HaveContent = true;
188
189 // ยง14.16 says 'byte-range-resp-spec' should be a '*' in case of 416
190 if (Result == 416 && sscanf(Val.c_str(), "bytes */%llu",&Size) == 1)
92e8c1ff 191 ; // we got the expected filesize which is all we wanted
7330f4df
DK
192 else if (sscanf(Val.c_str(),"bytes %llu-%*u/%llu",&StartPos,&Size) != 2)
193 return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
194 if ((unsigned long long)StartPos > Size)
195 return _error->Error(_("This HTTP server has broken range support"));
196 return true;
197 }
d3e8fbb3 198
7330f4df
DK
199 if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
200 {
201 HaveContent = true;
202 if (stringcasecmp(Val,"chunked") == 0)
d3e8fbb3 203 Encoding = Chunked;
7330f4df
DK
204 return true;
205 }
206
207 if (stringcasecmp(Tag,"Connection:") == 0)
208 {
209 if (stringcasecmp(Val,"close") == 0)
210 Persistent = false;
211 if (stringcasecmp(Val,"keep-alive") == 0)
212 Persistent = true;
213 return true;
214 }
d3e8fbb3 215
7330f4df
DK
216 if (stringcasecmp(Tag,"Last-Modified:") == 0)
217 {
218 if (RFC1123StrToTime(Val.c_str(), Date) == false)
219 return _error->Error(_("Unknown date format"));
220 return true;
221 }
222
223 if (stringcasecmp(Tag,"Location:") == 0)
224 {
225 Location = Val;
226 return true;
227 }
228
229 return true;
230}
231 /*}}}*/
232// ServerState::ServerState - Constructor /*{{{*/
233ServerState::ServerState(URI Srv, ServerMethod *Owner) : ServerName(Srv), TimeOut(120), Owner(Owner)
234{
235 Reset();
236}
237 /*}}}*/
238
239bool ServerMethod::Configuration(string Message) /*{{{*/
240{
241 return pkgAcqMethod::Configuration(Message);
242}
243 /*}}}*/
244
245// ServerMethod::DealWithHeaders - Handle the retrieved header data /*{{{*/
246// ---------------------------------------------------------------------
247/* We look at the header data we got back from the server and decide what
248 to do. Returns DealWithHeadersResult (see http.h for details).
249 */
250ServerMethod::DealWithHeadersResult
251ServerMethod::DealWithHeaders(FetchResult &Res)
252{
253 // Not Modified
254 if (Server->Result == 304)
255 {
256 unlink(Queue->DestFile.c_str());
257 Res.IMSHit = true;
258 Res.LastModified = Queue->LastModified;
259 return IMS_HIT;
260 }
261
262 /* Redirect
263 *
264 * Note that it is only OK for us to treat all redirection the same
265 * because we *always* use GET, not other HTTP methods. There are
266 * three redirection codes for which it is not appropriate that we
267 * redirect. Pass on those codes so the error handling kicks in.
268 */
269 if (AllowRedirect
270 && (Server->Result > 300 && Server->Result < 400)
271 && (Server->Result != 300 // Multiple Choices
272 && Server->Result != 304 // Not Modified
273 && Server->Result != 306)) // (Not part of HTTP/1.1, reserved)
274 {
275 if (Server->Location.empty() == true);
276 else if (Server->Location[0] == '/' && Queue->Uri.empty() == false)
277 {
278 URI Uri = Queue->Uri;
279 if (Uri.Host.empty() == false)
280 NextURI = URI::SiteOnly(Uri);
281 else
282 NextURI.clear();
283 NextURI.append(DeQuoteString(Server->Location));
284 return TRY_AGAIN_OR_REDIRECT;
285 }
286 else
287 {
9082a1fc
DK
288 NextURI = DeQuoteString(Server->Location);
289 URI tmpURI = NextURI;
290 URI Uri = Queue->Uri;
291 // same protocol redirects are okay
292 if (tmpURI.Access == Uri.Access)
293 return TRY_AGAIN_OR_REDIRECT;
294 // as well as http to https
295 else if (Uri.Access == "http" && tmpURI.Access == "https")
296 return TRY_AGAIN_OR_REDIRECT;
7330f4df
DK
297 }
298 /* else pass through for error message */
299 }
300 // retry after an invalid range response without partial data
301 else if (Server->Result == 416)
302 {
303 struct stat SBuf;
304 if (stat(Queue->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
305 {
306 if ((unsigned long long)SBuf.st_size == Server->Size)
307 {
308 // the file is completely downloaded, but was not moved
92e8c1ff
DK
309 if (Server->HaveContent == true)
310 {
311 // Send to error page to dev/null
312 FileFd DevNull("/dev/null",FileFd::WriteExists);
313 Server->RunData(&DevNull);
314 }
315 Server->HaveContent = false;
7330f4df
DK
316 Server->StartPos = Server->Size;
317 Server->Result = 200;
7330f4df
DK
318 }
319 else if (unlink(Queue->DestFile.c_str()) == 0)
320 {
321 NextURI = Queue->Uri;
322 return TRY_AGAIN_OR_REDIRECT;
323 }
324 }
325 }
326
327 /* We have a reply we dont handle. This should indicate a perm server
328 failure */
329 if (Server->Result < 200 || Server->Result >= 300)
330 {
331 char err[255];
332 snprintf(err,sizeof(err)-1,"HttpError%i",Server->Result);
333 SetFailReason(err);
334 _error->Error("%u %s",Server->Result,Server->Code);
335 if (Server->HaveContent == true)
336 return ERROR_WITH_CONTENT_PAGE;
337 return ERROR_UNRECOVERABLE;
338 }
339
340 // This is some sort of 2xx 'data follows' reply
341 Res.LastModified = Server->Date;
342 Res.Size = Server->Size;
343
344 // Open the file
345 delete File;
346 File = new FileFd(Queue->DestFile,FileFd::WriteAny);
347 if (_error->PendingError() == true)
348 return ERROR_NOT_FROM_SERVER;
349
350 FailFile = Queue->DestFile;
351 FailFile.c_str(); // Make sure we dont do a malloc in the signal handler
352 FailFd = File->Fd();
353 FailTime = Server->Date;
354
355 if (Server->InitHashes(*File) == false)
356 {
357 _error->Errno("read",_("Problem hashing file"));
358 return ERROR_NOT_FROM_SERVER;
359 }
360 if (Server->StartPos > 0)
361 Res.ResumePoint = Server->StartPos;
362
363 SetNonBlock(File->Fd(),true);
364 return FILE_IS_OPEN;
365}
366 /*}}}*/
367// ServerMethod::SigTerm - Handle a fatal signal /*{{{*/
368// ---------------------------------------------------------------------
1e3f4083 369/* This closes and timestamps the open file. This is necessary to get
7330f4df
DK
370 resume behavoir on user abort */
371void ServerMethod::SigTerm(int)
372{
373 if (FailFd == -1)
374 _exit(100);
9ce3cfc9 375
246bbb61 376 struct timeval times[2];
9ce3cfc9
DK
377 times[0].tv_sec = FailTime;
378 times[1].tv_sec = FailTime;
246bbb61
DK
379 times[0].tv_usec = times[1].tv_usec = 0;
380 utimes(FailFile.c_str(), times);
7330f4df 381 close(FailFd);
9ce3cfc9 382
7330f4df
DK
383 _exit(100);
384}
385 /*}}}*/
386// ServerMethod::Fetch - Fetch an item /*{{{*/
387// ---------------------------------------------------------------------
388/* This adds an item to the pipeline. We keep the pipeline at a fixed
389 depth. */
390bool ServerMethod::Fetch(FetchItem *)
391{
392 if (Server == 0)
393 return true;
394
395 // Queue the requests
396 int Depth = -1;
397 for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
398 I = I->Next, Depth++)
399 {
400 // If pipelining is disabled, we only queue 1 request
401 if (Server->Pipeline == false && Depth >= 0)
402 break;
403
404 // Make sure we stick with the same server
405 if (Server->Comp(I->Uri) == false)
406 break;
407 if (QueueBack == I)
408 {
409 QueueBack = I->Next;
410 SendReq(I);
411 continue;
412 }
413 }
414
415 return true;
d3e8fbb3 416}
7330f4df
DK
417 /*}}}*/
418// ServerMethod::Loop - Main loop /*{{{*/
419int ServerMethod::Loop()
420{
421 typedef vector<string> StringVector;
422 typedef vector<string>::iterator StringVectorIterator;
423 map<string, StringVector> Redirected;
424
425 signal(SIGTERM,SigTerm);
426 signal(SIGINT,SigTerm);
427
428 Server = 0;
429
430 int FailCounter = 0;
431 while (1)
432 {
433 // We have no commands, wait for some to arrive
434 if (Queue == 0)
435 {
436 if (WaitFd(STDIN_FILENO) == false)
437 return 0;
438 }
439
440 /* Run messages, we can accept 0 (no message) if we didn't
441 do a WaitFd above.. Otherwise the FD is closed. */
442 int Result = Run(true);
443 if (Result != -1 && (Result != 0 || Queue == 0))
444 {
445 if(FailReason.empty() == false ||
446 _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
447 return 100;
448 else
449 return 0;
450 }
451
452 if (Queue == 0)
453 continue;
454
455 // Connect to the server
456 if (Server == 0 || Server->Comp(Queue->Uri) == false)
457 {
458 delete Server;
459 Server = CreateServerState(Queue->Uri);
460 }
461 /* If the server has explicitly said this is the last connection
462 then we pre-emptively shut down the pipeline and tear down
463 the connection. This will speed up HTTP/1.0 servers a tad
464 since we don't have to wait for the close sequence to
465 complete */
466 if (Server->Persistent == false)
467 Server->Close();
468
469 // Reset the pipeline
470 if (Server->IsOpen() == false)
471 QueueBack = Queue;
472
473 // Connnect to the host
474 if (Server->Open() == false)
475 {
476 Fail(true);
477 delete Server;
478 Server = 0;
479 continue;
480 }
481
482 // Fill the pipeline.
483 Fetch(0);
484
485 // Fetch the next URL header data from the server.
9622b211 486 switch (Server->RunHeaders(File, Queue->Uri))
7330f4df
DK
487 {
488 case ServerState::RUN_HEADERS_OK:
489 break;
490
491 // The header data is bad
492 case ServerState::RUN_HEADERS_PARSE_ERROR:
493 {
494 _error->Error(_("Bad header data"));
495 Fail(true);
496 RotateDNS();
497 continue;
498 }
499
500 // The server closed a connection during the header get..
501 default:
502 case ServerState::RUN_HEADERS_IO_ERROR:
503 {
504 FailCounter++;
505 _error->Discard();
506 Server->Close();
507 Server->Pipeline = false;
508
509 if (FailCounter >= 2)
510 {
511 Fail(_("Connection failed"),true);
512 FailCounter = 0;
513 }
514
515 RotateDNS();
516 continue;
517 }
518 };
519
520 // Decide what to do.
521 FetchResult Res;
522 Res.Filename = Queue->DestFile;
523 switch (DealWithHeaders(Res))
524 {
525 // Ok, the file is Open
526 case FILE_IS_OPEN:
527 {
528 URIStart(Res);
529
530 // Run the data
531 bool Result = true;
532 if (Server->HaveContent)
533 Result = Server->RunData(File);
534
535 /* If the server is sending back sizeless responses then fill in
536 the size now */
537 if (Res.Size == 0)
538 Res.Size = File->Size();
539
540 // Close the file, destroy the FD object and timestamp it
541 FailFd = -1;
542 delete File;
543 File = 0;
544
545 // Timestamp
246bbb61 546 struct timeval times[2];
9ce3cfc9 547 times[0].tv_sec = times[1].tv_sec = Server->Date;
246bbb61
DK
548 times[0].tv_usec = times[1].tv_usec = 0;
549 utimes(Queue->DestFile.c_str(), times);
7330f4df
DK
550
551 // Send status to APT
552 if (Result == true)
553 {
554 Res.TakeHashes(*Server->GetHashes());
555 URIDone(Res);
556 }
557 else
558 {
559 if (Server->IsOpen() == false)
560 {
561 FailCounter++;
562 _error->Discard();
563 Server->Close();
564
565 if (FailCounter >= 2)
566 {
567 Fail(_("Connection failed"),true);
568 FailCounter = 0;
569 }
570
571 QueueBack = Queue;
572 }
573 else
574 Fail(true);
575 }
576 break;
577 }
578
579 // IMS hit
580 case IMS_HIT:
581 {
582 URIDone(Res);
583 break;
584 }
585
586 // Hard server error, not found or something
587 case ERROR_UNRECOVERABLE:
588 {
589 Fail();
590 break;
591 }
592
593 // Hard internal error, kill the connection and fail
594 case ERROR_NOT_FROM_SERVER:
595 {
596 delete File;
597 File = 0;
598
599 Fail();
600 RotateDNS();
601 Server->Close();
602 break;
603 }
604
605 // We need to flush the data, the header is like a 404 w/ error text
606 case ERROR_WITH_CONTENT_PAGE:
607 {
608 Fail();
609
610 // Send to content to dev/null
611 File = new FileFd("/dev/null",FileFd::WriteExists);
612 Server->RunData(File);
613 delete File;
614 File = 0;
615 break;
616 }
617
618 // Try again with a new URL
619 case TRY_AGAIN_OR_REDIRECT:
620 {
621 // Clear rest of response if there is content
622 if (Server->HaveContent)
623 {
624 File = new FileFd("/dev/null",FileFd::WriteExists);
625 Server->RunData(File);
626 delete File;
627 File = 0;
628 }
629
630 /* Detect redirect loops. No more redirects are allowed
631 after the same URI is seen twice in a queue item. */
632 StringVector &R = Redirected[Queue->DestFile];
633 bool StopRedirects = false;
634 if (R.empty() == true)
635 R.push_back(Queue->Uri);
636 else if (R[0] == "STOP" || R.size() > 10)
637 StopRedirects = true;
638 else
639 {
640 for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
641 if (Queue->Uri == *I)
642 {
643 R[0] = "STOP";
644 break;
645 }
646
647 R.push_back(Queue->Uri);
648 }
649
650 if (StopRedirects == false)
651 Redirect(NextURI);
652 else
653 Fail();
654
655 break;
656 }
657
658 default:
659 Fail(_("Internal error"));
660 break;
661 }
662
663 FailCounter = 0;
664 }
665
666 return 0;
667}
668 /*}}}*/