]>
Commit | Line | Data |
---|---|---|
be4401bf AL |
1 | // -*- mode: cpp; mode: fold -*- |
2 | // Description /*{{{*/ | |
43252d15 | 3 | // $Id: http.cc,v 1.24 1999/01/30 08:23:49 jgg Exp $ |
be4401bf AL |
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. It accepts on the command line | |
10 | a list of url destination pairs and writes to stdout the status of the | |
11 | operation as defined in the APT method spec. | |
12 | ||
13 | It is based on a doubly buffered select loop. All the requests are | |
14 | fed into a single output buffer that is constantly fed out the | |
15 | socket. This provides ideal pipelining as in many cases all of the | |
16 | requests will fit into a single packet. The input socket is buffered | |
17 | the same way and fed into the fd for the file. | |
18 | ||
19 | This double buffering provides fairly substantial transfer rates, | |
20 | compared to wget the http method is about 4% faster. Most importantly, | |
21 | when HTTP is compared with FTP as a protocol the speed difference is | |
22 | huge. In tests over the internet from two sites to llug (via ATM) this | |
23 | program got 230k/s sustained http transfer rates. FTP on the other | |
24 | hand topped out at 170k/s. That combined with the time to setup the | |
25 | FTP connection makes HTTP a vastly superior protocol. | |
26 | ||
27 | ##################################################################### */ | |
28 | /*}}}*/ | |
29 | // Include Files /*{{{*/ | |
30 | #include <apt-pkg/fileutl.h> | |
31 | #include <apt-pkg/acquire-method.h> | |
32 | #include <apt-pkg/error.h> | |
33 | #include <apt-pkg/md5.h> | |
34 | ||
35 | #include <sys/stat.h> | |
36 | #include <sys/time.h> | |
37 | #include <utime.h> | |
38 | #include <unistd.h> | |
492f957a | 39 | #include <signal.h> |
be4401bf AL |
40 | #include <stdio.h> |
41 | ||
42 | // Internet stuff | |
43 | #include <netinet/in.h> | |
44 | #include <sys/socket.h> | |
45 | #include <arpa/inet.h> | |
46 | #include <netdb.h> | |
47 | ||
48 | #include "http.h" | |
49 | /*}}}*/ | |
50 | ||
492f957a AL |
51 | string HttpMethod::FailFile; |
52 | int HttpMethod::FailFd = -1; | |
53 | time_t HttpMethod::FailTime = 0; | |
3000ccea AL |
54 | unsigned long PipelineDepth = 10; |
55 | unsigned long TimeOut = 120; | |
492f957a | 56 | |
be4401bf AL |
57 | // CircleBuf::CircleBuf - Circular input buffer /*{{{*/ |
58 | // --------------------------------------------------------------------- | |
59 | /* */ | |
60 | CircleBuf::CircleBuf(unsigned long Size) : Size(Size), MD5(0) | |
61 | { | |
62 | Buf = new unsigned char[Size]; | |
63 | Reset(); | |
64 | } | |
65 | /*}}}*/ | |
66 | // CircleBuf::Reset - Reset to the default state /*{{{*/ | |
67 | // --------------------------------------------------------------------- | |
68 | /* */ | |
69 | void CircleBuf::Reset() | |
70 | { | |
71 | InP = 0; | |
72 | OutP = 0; | |
73 | StrPos = 0; | |
74 | MaxGet = (unsigned int)-1; | |
75 | OutQueue = string(); | |
76 | if (MD5 != 0) | |
77 | { | |
78 | delete MD5; | |
79 | MD5 = new MD5Summation; | |
80 | } | |
81 | }; | |
82 | /*}}}*/ | |
83 | // CircleBuf::Read - Read from a FD into the circular buffer /*{{{*/ | |
84 | // --------------------------------------------------------------------- | |
85 | /* This fills up the buffer with as much data as is in the FD, assuming it | |
86 | is non-blocking.. */ | |
87 | bool CircleBuf::Read(int Fd) | |
88 | { | |
89 | while (1) | |
90 | { | |
91 | // Woops, buffer is full | |
92 | if (InP - OutP == Size) | |
93 | return true; | |
94 | ||
95 | // Write the buffer segment | |
96 | int Res; | |
97 | Res = read(Fd,Buf + (InP%Size),LeftRead()); | |
98 | ||
99 | if (Res == 0) | |
100 | return false; | |
101 | if (Res < 0) | |
102 | { | |
103 | if (errno == EAGAIN) | |
104 | return true; | |
105 | return false; | |
106 | } | |
107 | ||
108 | if (InP == 0) | |
109 | gettimeofday(&Start,0); | |
110 | InP += Res; | |
111 | } | |
112 | } | |
113 | /*}}}*/ | |
114 | // CircleBuf::Read - Put the string into the buffer /*{{{*/ | |
115 | // --------------------------------------------------------------------- | |
116 | /* This will hold the string in and fill the buffer with it as it empties */ | |
117 | bool CircleBuf::Read(string Data) | |
118 | { | |
119 | OutQueue += Data; | |
120 | FillOut(); | |
121 | return true; | |
122 | } | |
123 | /*}}}*/ | |
124 | // CircleBuf::FillOut - Fill the buffer from the output queue /*{{{*/ | |
125 | // --------------------------------------------------------------------- | |
126 | /* */ | |
127 | void CircleBuf::FillOut() | |
128 | { | |
129 | if (OutQueue.empty() == true) | |
130 | return; | |
131 | while (1) | |
132 | { | |
133 | // Woops, buffer is full | |
134 | if (InP - OutP == Size) | |
135 | return; | |
136 | ||
137 | // Write the buffer segment | |
138 | unsigned long Sz = LeftRead(); | |
139 | if (OutQueue.length() - StrPos < Sz) | |
140 | Sz = OutQueue.length() - StrPos; | |
141 | memcpy(Buf + (InP%Size),OutQueue.begin() + StrPos,Sz); | |
142 | ||
143 | // Advance | |
144 | StrPos += Sz; | |
145 | InP += Sz; | |
146 | if (OutQueue.length() == StrPos) | |
147 | { | |
148 | StrPos = 0; | |
149 | OutQueue = ""; | |
150 | return; | |
151 | } | |
152 | } | |
153 | } | |
154 | /*}}}*/ | |
155 | // CircleBuf::Write - Write from the buffer into a FD /*{{{*/ | |
156 | // --------------------------------------------------------------------- | |
157 | /* This empties the buffer into the FD. */ | |
158 | bool CircleBuf::Write(int Fd) | |
159 | { | |
160 | while (1) | |
161 | { | |
162 | FillOut(); | |
163 | ||
164 | // Woops, buffer is empty | |
165 | if (OutP == InP) | |
166 | return true; | |
167 | ||
168 | if (OutP == MaxGet) | |
169 | return true; | |
170 | ||
171 | // Write the buffer segment | |
172 | int Res; | |
173 | Res = write(Fd,Buf + (OutP%Size),LeftWrite()); | |
174 | ||
175 | if (Res == 0) | |
176 | return false; | |
177 | if (Res < 0) | |
178 | { | |
179 | if (errno == EAGAIN) | |
180 | return true; | |
181 | ||
182 | return false; | |
183 | } | |
184 | ||
185 | if (MD5 != 0) | |
186 | MD5->Add(Buf + (OutP%Size),Res); | |
187 | ||
188 | OutP += Res; | |
189 | } | |
190 | } | |
191 | /*}}}*/ | |
192 | // CircleBuf::WriteTillEl - Write from the buffer to a string /*{{{*/ | |
193 | // --------------------------------------------------------------------- | |
194 | /* This copies till the first empty line */ | |
195 | bool CircleBuf::WriteTillEl(string &Data,bool Single) | |
196 | { | |
197 | // We cheat and assume it is unneeded to have more than one buffer load | |
198 | for (unsigned long I = OutP; I < InP; I++) | |
199 | { | |
200 | if (Buf[I%Size] != '\n') | |
201 | continue; | |
202 | for (I++; I < InP && Buf[I%Size] == '\r'; I++); | |
203 | ||
204 | if (Single == false) | |
205 | { | |
206 | if (Buf[I%Size] != '\n') | |
207 | continue; | |
208 | for (I++; I < InP && Buf[I%Size] == '\r'; I++); | |
209 | } | |
210 | ||
211 | if (I > InP) | |
212 | I = InP; | |
213 | ||
214 | Data = ""; | |
215 | while (OutP < I) | |
216 | { | |
217 | unsigned long Sz = LeftWrite(); | |
218 | if (Sz == 0) | |
219 | return false; | |
220 | if (I - OutP < LeftWrite()) | |
221 | Sz = I - OutP; | |
222 | Data += string((char *)(Buf + (OutP%Size)),Sz); | |
223 | OutP += Sz; | |
224 | } | |
225 | return true; | |
226 | } | |
227 | return false; | |
228 | } | |
229 | /*}}}*/ | |
230 | // CircleBuf::Stats - Print out stats information /*{{{*/ | |
231 | // --------------------------------------------------------------------- | |
232 | /* */ | |
233 | void CircleBuf::Stats() | |
234 | { | |
235 | if (InP == 0) | |
236 | return; | |
237 | ||
238 | struct timeval Stop; | |
239 | gettimeofday(&Stop,0); | |
240 | /* float Diff = Stop.tv_sec - Start.tv_sec + | |
241 | (float)(Stop.tv_usec - Start.tv_usec)/1000000; | |
242 | clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/ | |
243 | } | |
244 | /*}}}*/ | |
245 | ||
246 | // ServerState::ServerState - Constructor /*{{{*/ | |
247 | // --------------------------------------------------------------------- | |
248 | /* */ | |
249 | ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner), | |
3000ccea | 250 | In(64*1024), Out(4*1024), |
be4401bf AL |
251 | ServerName(Srv) |
252 | { | |
253 | Reset(); | |
254 | } | |
255 | /*}}}*/ | |
256 | // ServerState::Open - Open a connection to the server /*{{{*/ | |
257 | // --------------------------------------------------------------------- | |
258 | /* This opens a connection to the server. */ | |
259 | string LastHost; | |
260 | in_addr LastHostA; | |
261 | bool ServerState::Open() | |
262 | { | |
92e889c8 AL |
263 | // Use the already open connection if possible. |
264 | if (ServerFd != -1) | |
265 | return true; | |
266 | ||
be4401bf | 267 | Close(); |
492f957a AL |
268 | In.Reset(); |
269 | Out.Reset(); | |
270 | ||
271 | // Determine the proxy setting | |
52e7839a | 272 | if (getenv("http_proxy") == 0) |
492f957a | 273 | { |
352c2768 AL |
274 | string DefProxy = _config->Find("Acquire::http::Proxy"); |
275 | string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host); | |
276 | if (SpecificProxy.empty() == false) | |
277 | { | |
278 | if (SpecificProxy == "DIRECT") | |
279 | Proxy = ""; | |
280 | else | |
281 | Proxy = SpecificProxy; | |
282 | } | |
492f957a | 283 | else |
352c2768 AL |
284 | Proxy = DefProxy; |
285 | } | |
492f957a | 286 | else |
352c2768 AL |
287 | Proxy = getenv("http_proxy"); |
288 | ||
492f957a | 289 | // Determine what host and port to use based on the proxy settings |
92e889c8 | 290 | int Port = 80; |
492f957a | 291 | string Host; |
92e889c8 | 292 | if (Proxy.empty() == true) |
be4401bf | 293 | { |
92e889c8 AL |
294 | if (ServerName.Port != 0) |
295 | Port = ServerName.Port; | |
be4401bf AL |
296 | Host = ServerName.Host; |
297 | } | |
298 | else | |
299 | { | |
92e889c8 AL |
300 | if (Proxy.Port != 0) |
301 | Port = Proxy.Port; | |
be4401bf AL |
302 | Host = Proxy.Host; |
303 | } | |
304 | ||
492f957a AL |
305 | /* We used a cached address record.. Yes this is against the spec but |
306 | the way we have setup our rotating dns suggests that this is more | |
307 | sensible */ | |
be4401bf AL |
308 | if (LastHost != Host) |
309 | { | |
310 | Owner->Status("Connecting to %s",Host.c_str()); | |
311 | ||
312 | // Lookup the host | |
313 | hostent *Addr = gethostbyname(Host.c_str()); | |
3000ccea | 314 | if (Addr == 0 || Addr->h_addr_list[0] == 0) |
92e889c8 | 315 | return _error->Error("Could not resolve '%s'",Host.c_str()); |
be4401bf AL |
316 | LastHost = Host; |
317 | LastHostA = *(in_addr *)(Addr->h_addr_list[0]); | |
318 | } | |
319 | ||
320 | Owner->Status("Connecting to %s (%s)",Host.c_str(),inet_ntoa(LastHostA)); | |
321 | ||
322 | // Get a socket | |
323 | if ((ServerFd = socket(AF_INET,SOCK_STREAM,0)) < 0) | |
324 | return _error->Errno("socket","Could not create a socket"); | |
325 | ||
326 | // Connect to the server | |
327 | struct sockaddr_in server; | |
328 | server.sin_family = AF_INET; | |
329 | server.sin_port = htons(Port); | |
330 | server.sin_addr = LastHostA; | |
3000ccea AL |
331 | SetNonBlock(ServerFd,true); |
332 | if (connect(ServerFd,(sockaddr *)&server,sizeof(server)) < 0 && | |
333 | errno != EINPROGRESS) | |
be4401bf AL |
334 | return _error->Errno("socket","Could not create a socket"); |
335 | ||
3000ccea AL |
336 | /* This implements a timeout for connect by opening the connection |
337 | nonblocking */ | |
338 | fd_set wfds; | |
339 | FD_ZERO(&wfds); | |
340 | FD_SET(ServerFd,&wfds); | |
341 | struct timeval tv; | |
342 | tv.tv_sec = TimeOut; | |
343 | tv.tv_usec = 0; | |
344 | int Res = 0; | |
345 | if ((Res = select(ServerFd+1,0,&wfds,0,&tv)) < 0) | |
346 | return _error->Errno("select","Select failed"); | |
347 | if (Res == 0) | |
348 | return _error->Error("Could not connect, connection timed out"); | |
7f9b781e | 349 | unsigned int Err,Len=sizeof(Err); |
3000ccea AL |
350 | if (getsockopt(ServerFd,SOL_SOCKET,SO_ERROR,&Err,&Len) != 0) |
351 | return _error->Errno("getsockopt","Failed"); | |
352 | if (Err != 0) | |
353 | return _error->Error("Could not connect."); | |
354 | ||
be4401bf AL |
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; | |
be4401bf AL |
365 | return true; |
366 | } | |
367 | /*}}}*/ | |
368 | // ServerState::RunHeaders - Get the headers before the data /*{{{*/ | |
369 | // --------------------------------------------------------------------- | |
92e889c8 AL |
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() | |
be4401bf AL |
373 | { |
374 | State = Header; | |
375 | ||
376 | Owner->Status("Waiting for file"); | |
377 | ||
378 | Major = 0; | |
379 | Minor = 0; | |
380 | Result = 0; | |
381 | Size = 0; | |
382 | StartPos = 0; | |
92e889c8 AL |
383 | Encoding = Closes; |
384 | HaveContent = false; | |
be4401bf AL |
385 | time(&Date); |
386 | ||
387 | do | |
388 | { | |
389 | string Data; | |
390 | if (In.WriteTillEl(Data) == false) | |
391 | continue; | |
392 | ||
393 | for (string::const_iterator I = Data.begin(); I < Data.end(); I++) | |
394 | { | |
395 | string::const_iterator J = I; | |
396 | for (; J != Data.end() && *J != '\n' && *J != '\r';J++); | |
397 | if (HeaderLine(string(I,J-I)) == false) | |
92e889c8 | 398 | return 2; |
be4401bf AL |
399 | I = J; |
400 | } | |
92e889c8 | 401 | return 0; |
be4401bf AL |
402 | } |
403 | while (Owner->Go(false,this) == true); | |
92e889c8 AL |
404 | |
405 | return 1; | |
be4401bf AL |
406 | } |
407 | /*}}}*/ | |
408 | // ServerState::RunData - Transfer the data from the socket /*{{{*/ | |
409 | // --------------------------------------------------------------------- | |
410 | /* */ | |
411 | bool ServerState::RunData() | |
412 | { | |
413 | State = Data; | |
414 | ||
415 | // Chunked transfer encoding is fun.. | |
416 | if (Encoding == Chunked) | |
417 | { | |
418 | while (1) | |
419 | { | |
420 | // Grab the block size | |
421 | bool Last = true; | |
422 | string Data; | |
423 | In.Limit(-1); | |
424 | do | |
425 | { | |
426 | if (In.WriteTillEl(Data,true) == true) | |
427 | break; | |
428 | } | |
429 | while ((Last = Owner->Go(false,this)) == true); | |
430 | ||
431 | if (Last == false) | |
432 | return false; | |
433 | ||
434 | // See if we are done | |
435 | unsigned long Len = strtol(Data.c_str(),0,16); | |
436 | if (Len == 0) | |
437 | { | |
438 | In.Limit(-1); | |
439 | ||
440 | // We have to remove the entity trailer | |
441 | Last = true; | |
442 | do | |
443 | { | |
444 | if (In.WriteTillEl(Data,true) == true && Data.length() <= 2) | |
445 | break; | |
446 | } | |
447 | while ((Last = Owner->Go(false,this)) == true); | |
448 | if (Last == false) | |
449 | return false; | |
450 | return true; | |
451 | } | |
452 | ||
453 | // Transfer the block | |
454 | In.Limit(Len); | |
455 | while (Owner->Go(true,this) == true) | |
456 | if (In.IsLimit() == true) | |
457 | break; | |
458 | ||
459 | // Error | |
460 | if (In.IsLimit() == false) | |
461 | return false; | |
462 | ||
463 | // The server sends an extra new line before the next block specifier.. | |
464 | In.Limit(-1); | |
465 | Last = true; | |
466 | do | |
467 | { | |
468 | if (In.WriteTillEl(Data,true) == true) | |
469 | break; | |
470 | } | |
471 | while ((Last = Owner->Go(false,this)) == true); | |
472 | if (Last == false) | |
473 | return false; | |
92e889c8 | 474 | } |
be4401bf AL |
475 | } |
476 | else | |
477 | { | |
478 | /* Closes encoding is used when the server did not specify a size, the | |
479 | loss of the connection means we are done */ | |
480 | if (Encoding == Closes) | |
481 | In.Limit(-1); | |
482 | else | |
483 | In.Limit(Size - StartPos); | |
484 | ||
485 | // Just transfer the whole block. | |
486 | do | |
487 | { | |
488 | if (In.IsLimit() == false) | |
489 | continue; | |
490 | ||
491 | In.Limit(-1); | |
492 | return true; | |
493 | } | |
494 | while (Owner->Go(true,this) == true); | |
495 | } | |
496 | ||
497 | return Owner->Flush(this); | |
498 | } | |
499 | /*}}}*/ | |
500 | // ServerState::HeaderLine - Process a header line /*{{{*/ | |
501 | // --------------------------------------------------------------------- | |
502 | /* */ | |
503 | bool ServerState::HeaderLine(string Line) | |
504 | { | |
505 | if (Line.empty() == true) | |
506 | return true; | |
30456e14 | 507 | |
be4401bf AL |
508 | // The http server might be trying to do something evil. |
509 | if (Line.length() >= MAXLEN) | |
510 | return _error->Error("Got a single header line over %u chars",MAXLEN); | |
511 | ||
512 | string::size_type Pos = Line.find(' '); | |
513 | if (Pos == string::npos || Pos+1 > Line.length()) | |
514 | return _error->Error("Bad header line"); | |
515 | ||
516 | string Tag = string(Line,0,Pos); | |
517 | string Val = string(Line,Pos+1); | |
518 | ||
92e889c8 | 519 | if (stringcasecmp(Tag.begin(),Tag.begin()+4,"HTTP") == 0) |
be4401bf AL |
520 | { |
521 | // Evil servers return no version | |
522 | if (Line[4] == '/') | |
523 | { | |
524 | if (sscanf(Line.c_str(),"HTTP/%u.%u %u %[^\n]",&Major,&Minor, | |
525 | &Result,Code) != 4) | |
526 | return _error->Error("The http server sent an invalid reply header"); | |
527 | } | |
528 | else | |
529 | { | |
530 | Major = 0; | |
531 | Minor = 9; | |
532 | if (sscanf(Line.c_str(),"HTTP %u %[^\n]",&Result,Code) != 2) | |
533 | return _error->Error("The http server sent an invalid reply header"); | |
534 | } | |
535 | ||
536 | return true; | |
537 | } | |
538 | ||
92e889c8 | 539 | if (stringcasecmp(Tag,"Content-Length:") == 0) |
be4401bf AL |
540 | { |
541 | if (Encoding == Closes) | |
542 | Encoding = Stream; | |
92e889c8 | 543 | HaveContent = true; |
be4401bf AL |
544 | |
545 | // The length is already set from the Content-Range header | |
546 | if (StartPos != 0) | |
547 | return true; | |
548 | ||
549 | if (sscanf(Val.c_str(),"%lu",&Size) != 1) | |
550 | return _error->Error("The http server sent an invalid Content-Length header"); | |
551 | return true; | |
552 | } | |
553 | ||
92e889c8 AL |
554 | if (stringcasecmp(Tag,"Content-Type:") == 0) |
555 | { | |
556 | HaveContent = true; | |
557 | return true; | |
558 | } | |
559 | ||
560 | if (stringcasecmp(Tag,"Content-Range:") == 0) | |
be4401bf | 561 | { |
92e889c8 AL |
562 | HaveContent = true; |
563 | ||
be4401bf AL |
564 | if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2) |
565 | return _error->Error("The http server sent an invalid Content-Range header"); | |
566 | if ((unsigned)StartPos > Size) | |
567 | return _error->Error("This http server has broken range support"); | |
568 | return true; | |
569 | } | |
570 | ||
92e889c8 | 571 | if (stringcasecmp(Tag,"Transfer-Encoding:") == 0) |
be4401bf | 572 | { |
92e889c8 AL |
573 | HaveContent = true; |
574 | if (stringcasecmp(Val,"chunked") == 0) | |
be4401bf | 575 | Encoding = Chunked; |
92e889c8 | 576 | |
be4401bf AL |
577 | return true; |
578 | } | |
579 | ||
92e889c8 | 580 | if (stringcasecmp(Tag,"Last-Modified:") == 0) |
be4401bf AL |
581 | { |
582 | if (StrToTime(Val,Date) == false) | |
583 | return _error->Error("Unknown date format"); | |
584 | return true; | |
585 | } | |
586 | ||
587 | return true; | |
588 | } | |
589 | /*}}}*/ | |
590 | ||
591 | // HttpMethod::SendReq - Send the HTTP request /*{{{*/ | |
592 | // --------------------------------------------------------------------- | |
593 | /* This places the http request in the outbound buffer */ | |
594 | void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out) | |
595 | { | |
596 | URI Uri = Itm->Uri; | |
c1a22377 | 597 | |
be4401bf | 598 | // The HTTP server expects a hostname with a trailing :port |
c1a22377 | 599 | char Buf[1000]; |
be4401bf AL |
600 | string ProperHost = Uri.Host; |
601 | if (Uri.Port != 0) | |
602 | { | |
603 | sprintf(Buf,":%u",Uri.Port); | |
604 | ProperHost += Buf; | |
605 | } | |
606 | ||
c1a22377 AL |
607 | // Just in case. |
608 | if (Itm->Uri.length() >= sizeof(Buf)) | |
609 | abort(); | |
610 | ||
492f957a AL |
611 | /* Build the request. We include a keep-alive header only for non-proxy |
612 | requests. This is to tweak old http/1.0 servers that do support keep-alive | |
613 | but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server | |
614 | will glitch HTTP/1.0 proxies because they do not filter it out and | |
615 | pass it on, HTTP/1.1 says the connection should default to keep alive | |
616 | and we expect the proxy to do this */ | |
be4401bf AL |
617 | if (Proxy.empty() == true) |
618 | sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n", | |
a4edf53b | 619 | QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str()); |
be4401bf | 620 | else |
c1a22377 AL |
621 | { |
622 | /* Generate a cache control header if necessary. We place a max | |
623 | cache age on index files, optionally set a no-cache directive | |
624 | and a no-store directive for archives. */ | |
be4401bf AL |
625 | sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n", |
626 | Itm->Uri.c_str(),ProperHost.c_str()); | |
c1a22377 | 627 | if (_config->FindB("Acquire::http::No-Cache",false) == true) |
b53b7926 | 628 | strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n"); |
c1a22377 AL |
629 | else |
630 | { | |
631 | if (Itm->IndexFile == true) | |
632 | sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n", | |
633 | _config->FindI("Acquire::http::Max-Age",60*60*24)); | |
634 | else | |
635 | { | |
636 | if (_config->FindB("Acquire::http::No-Store",false) == true) | |
637 | strcat(Buf,"Cache-Control: no-store\r\n"); | |
638 | } | |
639 | } | |
640 | } | |
641 | ||
be4401bf | 642 | string Req = Buf; |
492f957a | 643 | |
be4401bf AL |
644 | // Check for a partial file |
645 | struct stat SBuf; | |
646 | if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0) | |
647 | { | |
648 | // In this case we send an if-range query with a range header | |
649 | sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",SBuf.st_size - 1, | |
650 | TimeRFC1123(SBuf.st_mtime).c_str()); | |
651 | Req += Buf; | |
652 | } | |
653 | else | |
654 | { | |
655 | if (Itm->LastModified != 0) | |
656 | { | |
657 | sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str()); | |
658 | Req += Buf; | |
659 | } | |
660 | } | |
661 | ||
8d64c395 AL |
662 | if (Proxy.User.empty() == false || Proxy.Password.empty() == false) |
663 | Req += string("Proxy-Authorization: Basic ") + | |
664 | Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n"; | |
be4401bf AL |
665 | |
666 | Req += "User-Agent: Debian APT-HTTP/1.2\r\n\r\n"; | |
39536c62 | 667 | // cerr << Req << endl; |
c1a22377 | 668 | |
be4401bf AL |
669 | Out.Read(Req); |
670 | } | |
671 | /*}}}*/ | |
672 | // HttpMethod::Go - Run a single loop /*{{{*/ | |
673 | // --------------------------------------------------------------------- | |
674 | /* This runs the select loop over the server FDs, Output file FDs and | |
675 | stdin. */ | |
676 | bool HttpMethod::Go(bool ToFile,ServerState *Srv) | |
677 | { | |
678 | // Server has closed the connection | |
679 | if (Srv->ServerFd == -1 && Srv->In.WriteSpace() == false) | |
680 | return false; | |
681 | ||
682 | fd_set rfds,wfds,efds; | |
683 | FD_ZERO(&rfds); | |
684 | FD_ZERO(&wfds); | |
685 | FD_ZERO(&efds); | |
686 | ||
687 | // Add the server | |
688 | if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1) | |
689 | FD_SET(Srv->ServerFd,&wfds); | |
690 | if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1) | |
691 | FD_SET(Srv->ServerFd,&rfds); | |
692 | ||
693 | // Add the file | |
694 | int FileFD = -1; | |
695 | if (File != 0) | |
696 | FileFD = File->Fd(); | |
697 | ||
698 | if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1) | |
699 | FD_SET(FileFD,&wfds); | |
700 | ||
701 | // Add stdin | |
702 | FD_SET(STDIN_FILENO,&rfds); | |
703 | ||
704 | // Error Set | |
705 | if (FileFD != -1) | |
706 | FD_SET(FileFD,&efds); | |
707 | if (Srv->ServerFd != -1) | |
708 | FD_SET(Srv->ServerFd,&efds); | |
709 | ||
710 | // Figure out the max fd | |
711 | int MaxFd = FileFD; | |
712 | if (MaxFd < Srv->ServerFd) | |
713 | MaxFd = Srv->ServerFd; | |
714 | ||
715 | // Select | |
716 | struct timeval tv; | |
3000ccea | 717 | tv.tv_sec = TimeOut; |
be4401bf AL |
718 | tv.tv_usec = 0; |
719 | int Res = 0; | |
720 | if ((Res = select(MaxFd+1,&rfds,&wfds,&efds,&tv)) < 0) | |
721 | return _error->Errno("select","Select failed"); | |
722 | ||
723 | if (Res == 0) | |
724 | { | |
725 | _error->Error("Connection timed out"); | |
726 | return ServerDie(Srv); | |
727 | } | |
728 | ||
729 | // Some kind of exception (error) on the sockets, die | |
730 | if ((FileFD != -1 && FD_ISSET(FileFD,&efds)) || | |
731 | (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&efds))) | |
732 | return _error->Error("Socket Exception"); | |
733 | ||
734 | // Handle server IO | |
735 | if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds)) | |
736 | { | |
737 | errno = 0; | |
738 | if (Srv->In.Read(Srv->ServerFd) == false) | |
739 | return ServerDie(Srv); | |
740 | } | |
741 | ||
742 | if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds)) | |
743 | { | |
744 | errno = 0; | |
745 | if (Srv->Out.Write(Srv->ServerFd) == false) | |
746 | return ServerDie(Srv); | |
747 | } | |
748 | ||
749 | // Send data to the file | |
750 | if (FileFD != -1 && FD_ISSET(FileFD,&wfds)) | |
751 | { | |
752 | if (Srv->In.Write(FileFD) == false) | |
753 | return _error->Errno("write","Error writing to output file"); | |
754 | } | |
755 | ||
756 | // Handle commands from APT | |
757 | if (FD_ISSET(STDIN_FILENO,&rfds)) | |
758 | { | |
759 | if (Run(true) != 0) | |
760 | exit(100); | |
761 | } | |
762 | ||
763 | return true; | |
764 | } | |
765 | /*}}}*/ | |
766 | // HttpMethod::Flush - Dump the buffer into the file /*{{{*/ | |
767 | // --------------------------------------------------------------------- | |
768 | /* This takes the current input buffer from the Server FD and writes it | |
769 | into the file */ | |
770 | bool HttpMethod::Flush(ServerState *Srv) | |
771 | { | |
772 | if (File != 0) | |
773 | { | |
774 | SetNonBlock(File->Fd(),false); | |
775 | if (Srv->In.WriteSpace() == false) | |
776 | return true; | |
777 | ||
778 | while (Srv->In.WriteSpace() == true) | |
779 | { | |
780 | if (Srv->In.Write(File->Fd()) == false) | |
781 | return _error->Errno("write","Error writing to file"); | |
92e889c8 AL |
782 | if (Srv->In.IsLimit() == true) |
783 | return true; | |
be4401bf AL |
784 | } |
785 | ||
786 | if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes) | |
787 | return true; | |
788 | } | |
789 | return false; | |
790 | } | |
791 | /*}}}*/ | |
792 | // HttpMethod::ServerDie - The server has closed the connection. /*{{{*/ | |
793 | // --------------------------------------------------------------------- | |
794 | /* */ | |
795 | bool HttpMethod::ServerDie(ServerState *Srv) | |
796 | { | |
797 | // Dump the buffer to the file | |
798 | if (Srv->State == ServerState::Data) | |
799 | { | |
800 | SetNonBlock(File->Fd(),false); | |
801 | while (Srv->In.WriteSpace() == true) | |
802 | { | |
803 | if (Srv->In.Write(File->Fd()) == false) | |
804 | return _error->Errno("write","Error writing to the file"); | |
92e889c8 AL |
805 | |
806 | // Done | |
807 | if (Srv->In.IsLimit() == true) | |
808 | return true; | |
be4401bf AL |
809 | } |
810 | } | |
811 | ||
812 | // See if this is because the server finished the data stream | |
813 | if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header && | |
814 | Srv->Encoding != ServerState::Closes) | |
815 | { | |
3d615484 | 816 | Srv->Close(); |
be4401bf AL |
817 | if (errno == 0) |
818 | return _error->Error("Error reading from server Remote end closed connection"); | |
819 | return _error->Errno("read","Error reading from server"); | |
820 | } | |
821 | else | |
822 | { | |
823 | Srv->In.Limit(-1); | |
824 | ||
825 | // Nothing left in the buffer | |
826 | if (Srv->In.WriteSpace() == false) | |
827 | return false; | |
828 | ||
829 | // We may have got multiple responses back in one packet.. | |
830 | Srv->Close(); | |
831 | return true; | |
832 | } | |
833 | ||
834 | return false; | |
835 | } | |
836 | /*}}}*/ | |
837 | // HttpMethod::DealWithHeaders - Handle the retrieved header data /*{{{*/ | |
838 | // --------------------------------------------------------------------- | |
839 | /* We look at the header data we got back from the server and decide what | |
840 | to do. Returns | |
841 | 0 - File is open, | |
842 | 1 - IMS hit | |
92e889c8 | 843 | 3 - Unrecoverable error |
94235cfb AL |
844 | 4 - Error with error content page |
845 | 5 - Unrecoverable non-server error (close the connection) */ | |
be4401bf AL |
846 | int HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv) |
847 | { | |
848 | // Not Modified | |
849 | if (Srv->Result == 304) | |
850 | { | |
851 | unlink(Queue->DestFile.c_str()); | |
852 | Res.IMSHit = true; | |
853 | Res.LastModified = Queue->LastModified; | |
854 | return 1; | |
855 | } | |
856 | ||
857 | /* We have a reply we dont handle. This should indicate a perm server | |
858 | failure */ | |
859 | if (Srv->Result < 200 || Srv->Result >= 300) | |
860 | { | |
861 | _error->Error("%u %s",Srv->Result,Srv->Code); | |
92e889c8 AL |
862 | if (Srv->HaveContent == true) |
863 | return 4; | |
be4401bf AL |
864 | return 3; |
865 | } | |
866 | ||
867 | // This is some sort of 2xx 'data follows' reply | |
868 | Res.LastModified = Srv->Date; | |
869 | Res.Size = Srv->Size; | |
870 | ||
871 | // Open the file | |
872 | delete File; | |
873 | File = new FileFd(Queue->DestFile,FileFd::WriteAny); | |
874 | if (_error->PendingError() == true) | |
94235cfb | 875 | return 5; |
492f957a AL |
876 | |
877 | FailFile = Queue->DestFile; | |
91cb4c6b | 878 | FailFile.c_str(); // Make sure we don't do a malloc in the signal handler |
492f957a AL |
879 | FailFd = File->Fd(); |
880 | FailTime = Srv->Date; | |
881 | ||
be4401bf AL |
882 | // Set the expected size |
883 | if (Srv->StartPos >= 0) | |
884 | { | |
885 | Res.ResumePoint = Srv->StartPos; | |
886 | ftruncate(File->Fd(),Srv->StartPos); | |
887 | } | |
888 | ||
889 | // Set the start point | |
890 | lseek(File->Fd(),0,SEEK_END); | |
891 | ||
892 | delete Srv->In.MD5; | |
893 | Srv->In.MD5 = new MD5Summation; | |
894 | ||
895 | // Fill the MD5 Hash if the file is non-empty (resume) | |
896 | if (Srv->StartPos > 0) | |
897 | { | |
898 | lseek(File->Fd(),0,SEEK_SET); | |
899 | if (Srv->In.MD5->AddFD(File->Fd(),Srv->StartPos) == false) | |
900 | { | |
901 | _error->Errno("read","Problem hashing file"); | |
94235cfb | 902 | return 5; |
be4401bf AL |
903 | } |
904 | lseek(File->Fd(),0,SEEK_END); | |
905 | } | |
906 | ||
907 | SetNonBlock(File->Fd(),true); | |
908 | return 0; | |
909 | } | |
910 | /*}}}*/ | |
492f957a AL |
911 | // HttpMethod::SigTerm - Handle a fatal signal /*{{{*/ |
912 | // --------------------------------------------------------------------- | |
913 | /* This closes and timestamps the open file. This is neccessary to get | |
914 | resume behavoir on user abort */ | |
915 | void HttpMethod::SigTerm(int) | |
916 | { | |
917 | if (FailFd == -1) | |
918 | exit(100); | |
919 | close(FailFd); | |
920 | ||
921 | // Timestamp | |
922 | struct utimbuf UBuf; | |
923 | time(&UBuf.actime); | |
924 | UBuf.actime = FailTime; | |
925 | UBuf.modtime = FailTime; | |
926 | utime(FailFile.c_str(),&UBuf); | |
927 | ||
928 | exit(100); | |
929 | } | |
930 | /*}}}*/ | |
5cb5d8dc AL |
931 | // HttpMethod::Fetch - Fetch an item /*{{{*/ |
932 | // --------------------------------------------------------------------- | |
933 | /* This adds an item to the pipeline. We keep the pipeline at a fixed | |
934 | depth. */ | |
935 | bool HttpMethod::Fetch(FetchItem *) | |
936 | { | |
937 | if (Server == 0) | |
938 | return true; | |
3000ccea | 939 | |
5cb5d8dc AL |
940 | // Queue the requests |
941 | int Depth = -1; | |
942 | bool Tail = false; | |
c1a22377 | 943 | for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth; I = I->Next, Depth++) |
5cb5d8dc AL |
944 | { |
945 | // Make sure we stick with the same server | |
946 | if (Server->Comp(I->Uri) == false) | |
947 | break; | |
5cb5d8dc AL |
948 | if (QueueBack == I) |
949 | Tail = true; | |
950 | if (Tail == true) | |
951 | { | |
5cb5d8dc AL |
952 | QueueBack = I->Next; |
953 | SendReq(I,Server->Out); | |
954 | continue; | |
955 | } | |
956 | } | |
957 | ||
958 | return true; | |
959 | }; | |
960 | /*}}}*/ | |
85f72a56 AL |
961 | // HttpMethod::Configuration - Handle a configuration message /*{{{*/ |
962 | // --------------------------------------------------------------------- | |
963 | /* We stash the desired pipeline depth */ | |
964 | bool HttpMethod::Configuration(string Message) | |
965 | { | |
966 | if (pkgAcqMethod::Configuration(Message) == false) | |
967 | return false; | |
968 | ||
30456e14 AL |
969 | TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut); |
970 | PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth", | |
971 | PipelineDepth); | |
3000ccea | 972 | |
85f72a56 AL |
973 | return true; |
974 | } | |
975 | /*}}}*/ | |
492f957a | 976 | // HttpMethod::Loop - Main loop /*{{{*/ |
be4401bf AL |
977 | // --------------------------------------------------------------------- |
978 | /* */ | |
979 | int HttpMethod::Loop() | |
980 | { | |
492f957a AL |
981 | signal(SIGTERM,SigTerm); |
982 | signal(SIGINT,SigTerm); | |
983 | ||
5cb5d8dc | 984 | Server = 0; |
be4401bf | 985 | |
92e889c8 | 986 | int FailCounter = 0; |
be4401bf AL |
987 | while (1) |
988 | { | |
92e889c8 AL |
989 | if (FailCounter >= 2) |
990 | { | |
43252d15 | 991 | Fail("Massive Server Brain Damage",true); |
92e889c8 AL |
992 | FailCounter = 0; |
993 | } | |
994 | ||
be4401bf AL |
995 | // We have no commands, wait for some to arrive |
996 | if (Queue == 0) | |
997 | { | |
998 | if (WaitFd(STDIN_FILENO) == false) | |
999 | return 0; | |
1000 | } | |
1001 | ||
1002 | // Run messages | |
1003 | if (Run(true) != 0) | |
1004 | return 100; | |
1005 | ||
1006 | if (Queue == 0) | |
1007 | continue; | |
1008 | ||
1009 | // Connect to the server | |
1010 | if (Server == 0 || Server->Comp(Queue->Uri) == false) | |
1011 | { | |
1012 | delete Server; | |
1013 | Server = new ServerState(Queue->Uri,this); | |
1014 | } | |
1015 | ||
a7fb252c AL |
1016 | // Reset the pipeline |
1017 | if (Server->ServerFd == -1) | |
1018 | QueueBack = Queue; | |
1019 | ||
be4401bf AL |
1020 | // Connnect to the host |
1021 | if (Server->Open() == false) | |
1022 | { | |
43252d15 | 1023 | Fail(true); |
a1459f52 AL |
1024 | delete Server; |
1025 | Server = 0; | |
be4401bf AL |
1026 | continue; |
1027 | } | |
be4401bf | 1028 | |
5cb5d8dc AL |
1029 | // Fill the pipeline. |
1030 | Fetch(0); | |
1031 | ||
92e889c8 AL |
1032 | // Fetch the next URL header data from the server. |
1033 | switch (Server->RunHeaders()) | |
be4401bf | 1034 | { |
92e889c8 AL |
1035 | case 0: |
1036 | break; | |
1037 | ||
1038 | // The header data is bad | |
1039 | case 2: | |
1040 | { | |
1041 | _error->Error("Bad header Data"); | |
43252d15 | 1042 | Fail(true); |
92e889c8 AL |
1043 | continue; |
1044 | } | |
1045 | ||
1046 | // The server closed a connection during the header get.. | |
1047 | default: | |
1048 | case 1: | |
1049 | { | |
1050 | FailCounter++; | |
3d615484 | 1051 | _error->Discard(); |
92e889c8 AL |
1052 | Server->Close(); |
1053 | continue; | |
1054 | } | |
1055 | }; | |
5cb5d8dc | 1056 | |
be4401bf AL |
1057 | // Decide what to do. |
1058 | FetchResult Res; | |
bfd22fc0 | 1059 | Res.Filename = Queue->DestFile; |
be4401bf AL |
1060 | switch (DealWithHeaders(Res,Server)) |
1061 | { | |
1062 | // Ok, the file is Open | |
1063 | case 0: | |
1064 | { | |
1065 | URIStart(Res); | |
1066 | ||
1067 | // Run the data | |
492f957a AL |
1068 | bool Result = Server->RunData(); |
1069 | ||
1070 | // Close the file, destroy the FD object and timestamp it | |
1071 | FailFd = -1; | |
1072 | delete File; | |
1073 | File = 0; | |
1074 | ||
1075 | // Timestamp | |
1076 | struct utimbuf UBuf; | |
1077 | time(&UBuf.actime); | |
1078 | UBuf.actime = Server->Date; | |
1079 | UBuf.modtime = Server->Date; | |
1080 | utime(Queue->DestFile.c_str(),&UBuf); | |
1081 | ||
1082 | // Send status to APT | |
1083 | if (Result == true) | |
92e889c8 AL |
1084 | { |
1085 | Res.MD5Sum = Server->In.MD5->Result(); | |
1086 | URIDone(Res); | |
1087 | } | |
492f957a AL |
1088 | else |
1089 | Fail(); | |
1090 | ||
be4401bf AL |
1091 | break; |
1092 | } | |
1093 | ||
1094 | // IMS hit | |
1095 | case 1: | |
1096 | { | |
1097 | URIDone(Res); | |
1098 | break; | |
1099 | } | |
1100 | ||
1101 | // Hard server error, not found or something | |
1102 | case 3: | |
1103 | { | |
1104 | Fail(); | |
1105 | break; | |
1106 | } | |
94235cfb AL |
1107 | |
1108 | // Hard internal error, kill the connection and fail | |
1109 | case 5: | |
1110 | { | |
1111 | Fail(); | |
1112 | Server->Close(); | |
1113 | break; | |
1114 | } | |
92e889c8 AL |
1115 | |
1116 | // We need to flush the data, the header is like a 404 w/ error text | |
1117 | case 4: | |
1118 | { | |
1119 | Fail(); | |
1120 | ||
1121 | // Send to content to dev/null | |
1122 | File = new FileFd("/dev/null",FileFd::WriteExists); | |
1123 | Server->RunData(); | |
1124 | delete File; | |
1125 | File = 0; | |
1126 | break; | |
1127 | } | |
be4401bf AL |
1128 | |
1129 | default: | |
1130 | Fail("Internal error"); | |
1131 | break; | |
92e889c8 AL |
1132 | } |
1133 | ||
1134 | FailCounter = 0; | |
be4401bf AL |
1135 | } |
1136 | ||
1137 | return 0; | |
1138 | } | |
1139 | /*}}}*/ | |
1140 | ||
1141 | int main() | |
1142 | { | |
1143 | HttpMethod Mth; | |
1144 | ||
1145 | return Mth.Loop(); | |
1146 | } |