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