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