http.cc 33 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328
  1. // -*- mode: cpp; mode: fold -*-
  2. // Description /*{{{*/
  3. // $Id: http.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
  4. /* ######################################################################
  5. HTTP Acquire Method - This is the HTTP aquire method for APT.
  6. It uses HTTP/1.1 and many of the fancy options there-in, such as
  7. pipelining, range, if-range and so on.
  8. It is based on a doubly buffered select loop. A groupe of requests are
  9. fed into a single output buffer that is constantly fed out the
  10. socket. This provides ideal pipelining as in many cases all of the
  11. requests will fit into a single packet. The input socket is buffered
  12. the same way and fed into the fd for the file (may be a pipe in future).
  13. This double buffering provides fairly substantial transfer rates,
  14. compared to wget the http method is about 4% faster. Most importantly,
  15. when HTTP is compared with FTP as a protocol the speed difference is
  16. huge. In tests over the internet from two sites to llug (via ATM) this
  17. program got 230k/s sustained http transfer rates. FTP on the other
  18. hand topped out at 170k/s. That combined with the time to setup the
  19. FTP connection makes HTTP a vastly superior protocol.
  20. ##################################################################### */
  21. /*}}}*/
  22. // Include Files /*{{{*/
  23. #include <apt-pkg/fileutl.h>
  24. #include <apt-pkg/acquire-method.h>
  25. #include <apt-pkg/error.h>
  26. #include <apt-pkg/hashes.h>
  27. #include <sys/stat.h>
  28. #include <sys/time.h>
  29. #include <utime.h>
  30. #include <unistd.h>
  31. #include <signal.h>
  32. #include <stdio.h>
  33. #include <errno.h>
  34. #include <string.h>
  35. #include <iostream>
  36. #include <map>
  37. #include <apti18n.h>
  38. // Internet stuff
  39. #include <netdb.h>
  40. #include "config.h"
  41. #include "connect.h"
  42. #include "rfc2553emu.h"
  43. #include "http.h"
  44. /*}}}*/
  45. using namespace std;
  46. string HttpMethod::FailFile;
  47. int HttpMethod::FailFd = -1;
  48. time_t HttpMethod::FailTime = 0;
  49. unsigned long PipelineDepth = 10;
  50. unsigned long TimeOut = 120;
  51. bool AllowRedirect = false;
  52. bool Debug = false;
  53. URI Proxy;
  54. unsigned long CircleBuf::BwReadLimit=0;
  55. unsigned long CircleBuf::BwTickReadData=0;
  56. struct timeval CircleBuf::BwReadTick={0,0};
  57. const unsigned int CircleBuf::BW_HZ=10;
  58. // CircleBuf::CircleBuf - Circular input buffer /*{{{*/
  59. // ---------------------------------------------------------------------
  60. /* */
  61. CircleBuf::CircleBuf(unsigned long Size) : Size(Size), Hash(0)
  62. {
  63. Buf = new unsigned char[Size];
  64. Reset();
  65. CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  66. }
  67. /*}}}*/
  68. // CircleBuf::Reset - Reset to the default state /*{{{*/
  69. // ---------------------------------------------------------------------
  70. /* */
  71. void CircleBuf::Reset()
  72. {
  73. InP = 0;
  74. OutP = 0;
  75. StrPos = 0;
  76. MaxGet = (unsigned int)-1;
  77. OutQueue = string();
  78. if (Hash != 0)
  79. {
  80. delete Hash;
  81. Hash = new Hashes;
  82. }
  83. };
  84. /*}}}*/
  85. // CircleBuf::Read - Read from a FD into the circular buffer /*{{{*/
  86. // ---------------------------------------------------------------------
  87. /* This fills up the buffer with as much data as is in the FD, assuming it
  88. is non-blocking.. */
  89. bool CircleBuf::Read(int Fd)
  90. {
  91. unsigned long BwReadMax;
  92. while (1)
  93. {
  94. // Woops, buffer is full
  95. if (InP - OutP == Size)
  96. return true;
  97. // what's left to read in this tick
  98. BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
  99. if(CircleBuf::BwReadLimit) {
  100. struct timeval now;
  101. gettimeofday(&now,0);
  102. unsigned long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
  103. now.tv_usec-CircleBuf::BwReadTick.tv_usec;
  104. if(d > 1000000/BW_HZ) {
  105. CircleBuf::BwReadTick = now;
  106. CircleBuf::BwTickReadData = 0;
  107. }
  108. if(CircleBuf::BwTickReadData >= BwReadMax) {
  109. usleep(1000000/BW_HZ);
  110. return true;
  111. }
  112. }
  113. // Write the buffer segment
  114. int Res;
  115. if(CircleBuf::BwReadLimit) {
  116. Res = read(Fd,Buf + (InP%Size),
  117. BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
  118. } else
  119. Res = read(Fd,Buf + (InP%Size),LeftRead());
  120. if(Res > 0 && BwReadLimit > 0)
  121. CircleBuf::BwTickReadData += Res;
  122. if (Res == 0)
  123. return false;
  124. if (Res < 0)
  125. {
  126. if (errno == EAGAIN)
  127. return true;
  128. return false;
  129. }
  130. if (InP == 0)
  131. gettimeofday(&Start,0);
  132. InP += Res;
  133. }
  134. }
  135. /*}}}*/
  136. // CircleBuf::Read - Put the string into the buffer /*{{{*/
  137. // ---------------------------------------------------------------------
  138. /* This will hold the string in and fill the buffer with it as it empties */
  139. bool CircleBuf::Read(string Data)
  140. {
  141. OutQueue += Data;
  142. FillOut();
  143. return true;
  144. }
  145. /*}}}*/
  146. // CircleBuf::FillOut - Fill the buffer from the output queue /*{{{*/
  147. // ---------------------------------------------------------------------
  148. /* */
  149. void CircleBuf::FillOut()
  150. {
  151. if (OutQueue.empty() == true)
  152. return;
  153. while (1)
  154. {
  155. // Woops, buffer is full
  156. if (InP - OutP == Size)
  157. return;
  158. // Write the buffer segment
  159. unsigned long Sz = LeftRead();
  160. if (OutQueue.length() - StrPos < Sz)
  161. Sz = OutQueue.length() - StrPos;
  162. memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
  163. // Advance
  164. StrPos += Sz;
  165. InP += Sz;
  166. if (OutQueue.length() == StrPos)
  167. {
  168. StrPos = 0;
  169. OutQueue = "";
  170. return;
  171. }
  172. }
  173. }
  174. /*}}}*/
  175. // CircleBuf::Write - Write from the buffer into a FD /*{{{*/
  176. // ---------------------------------------------------------------------
  177. /* This empties the buffer into the FD. */
  178. bool CircleBuf::Write(int Fd)
  179. {
  180. while (1)
  181. {
  182. FillOut();
  183. // Woops, buffer is empty
  184. if (OutP == InP)
  185. return true;
  186. if (OutP == MaxGet)
  187. return true;
  188. // Write the buffer segment
  189. int Res;
  190. Res = write(Fd,Buf + (OutP%Size),LeftWrite());
  191. if (Res == 0)
  192. return false;
  193. if (Res < 0)
  194. {
  195. if (errno == EAGAIN)
  196. return true;
  197. return false;
  198. }
  199. if (Hash != 0)
  200. Hash->Add(Buf + (OutP%Size),Res);
  201. OutP += Res;
  202. }
  203. }
  204. /*}}}*/
  205. // CircleBuf::WriteTillEl - Write from the buffer to a string /*{{{*/
  206. // ---------------------------------------------------------------------
  207. /* This copies till the first empty line */
  208. bool CircleBuf::WriteTillEl(string &Data,bool Single)
  209. {
  210. // We cheat and assume it is unneeded to have more than one buffer load
  211. for (unsigned long I = OutP; I < InP; I++)
  212. {
  213. if (Buf[I%Size] != '\n')
  214. continue;
  215. ++I;
  216. if (Single == false)
  217. {
  218. if (I < InP && Buf[I%Size] == '\r')
  219. ++I;
  220. if (I >= InP || Buf[I%Size] != '\n')
  221. continue;
  222. ++I;
  223. }
  224. Data = "";
  225. while (OutP < I)
  226. {
  227. unsigned long Sz = LeftWrite();
  228. if (Sz == 0)
  229. return false;
  230. if (I - OutP < Sz)
  231. Sz = I - OutP;
  232. Data += string((char *)(Buf + (OutP%Size)),Sz);
  233. OutP += Sz;
  234. }
  235. return true;
  236. }
  237. return false;
  238. }
  239. /*}}}*/
  240. // CircleBuf::Stats - Print out stats information /*{{{*/
  241. // ---------------------------------------------------------------------
  242. /* */
  243. void CircleBuf::Stats()
  244. {
  245. if (InP == 0)
  246. return;
  247. struct timeval Stop;
  248. gettimeofday(&Stop,0);
  249. /* float Diff = Stop.tv_sec - Start.tv_sec +
  250. (float)(Stop.tv_usec - Start.tv_usec)/1000000;
  251. clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
  252. }
  253. /*}}}*/
  254. // ServerState::ServerState - Constructor /*{{{*/
  255. // ---------------------------------------------------------------------
  256. /* */
  257. ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
  258. In(64*1024), Out(4*1024),
  259. ServerName(Srv)
  260. {
  261. Reset();
  262. }
  263. /*}}}*/
  264. // ServerState::Open - Open a connection to the server /*{{{*/
  265. // ---------------------------------------------------------------------
  266. /* This opens a connection to the server. */
  267. bool ServerState::Open()
  268. {
  269. // Use the already open connection if possible.
  270. if (ServerFd != -1)
  271. return true;
  272. Close();
  273. In.Reset();
  274. Out.Reset();
  275. Persistent = true;
  276. // Determine the proxy setting
  277. if (getenv("http_proxy") == 0)
  278. {
  279. string DefProxy = _config->Find("Acquire::http::Proxy");
  280. string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
  281. if (SpecificProxy.empty() == false)
  282. {
  283. if (SpecificProxy == "DIRECT")
  284. Proxy = "";
  285. else
  286. Proxy = SpecificProxy;
  287. }
  288. else
  289. Proxy = DefProxy;
  290. }
  291. else
  292. Proxy = getenv("http_proxy");
  293. // Parse no_proxy, a , separated list of domains
  294. if (getenv("no_proxy") != 0)
  295. {
  296. if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
  297. Proxy = "";
  298. }
  299. // Determine what host and port to use based on the proxy settings
  300. int Port = 0;
  301. string Host;
  302. if (Proxy.empty() == true || Proxy.Host.empty() == true)
  303. {
  304. if (ServerName.Port != 0)
  305. Port = ServerName.Port;
  306. Host = ServerName.Host;
  307. }
  308. else
  309. {
  310. if (Proxy.Port != 0)
  311. Port = Proxy.Port;
  312. Host = Proxy.Host;
  313. }
  314. // Connect to the remote server
  315. if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
  316. return false;
  317. return true;
  318. }
  319. /*}}}*/
  320. // ServerState::Close - Close a connection to the server /*{{{*/
  321. // ---------------------------------------------------------------------
  322. /* */
  323. bool ServerState::Close()
  324. {
  325. close(ServerFd);
  326. ServerFd = -1;
  327. return true;
  328. }
  329. /*}}}*/
  330. // ServerState::RunHeaders - Get the headers before the data /*{{{*/
  331. // ---------------------------------------------------------------------
  332. /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
  333. parse error occurred */
  334. int ServerState::RunHeaders()
  335. {
  336. State = Header;
  337. Owner->Status(_("Waiting for headers"));
  338. Major = 0;
  339. Minor = 0;
  340. Result = 0;
  341. Size = 0;
  342. StartPos = 0;
  343. Encoding = Closes;
  344. HaveContent = false;
  345. time(&Date);
  346. do
  347. {
  348. string Data;
  349. if (In.WriteTillEl(Data) == false)
  350. continue;
  351. if (Debug == true)
  352. clog << Data;
  353. for (string::const_iterator I = Data.begin(); I < Data.end(); I++)
  354. {
  355. string::const_iterator J = I;
  356. for (; J != Data.end() && *J != '\n' && *J != '\r';J++);
  357. if (HeaderLine(string(I,J)) == false)
  358. return 2;
  359. I = J;
  360. }
  361. // 100 Continue is a Nop...
  362. if (Result == 100)
  363. continue;
  364. // Tidy up the connection persistance state.
  365. if (Encoding == Closes && HaveContent == true)
  366. Persistent = false;
  367. return 0;
  368. }
  369. while (Owner->Go(false,this) == true);
  370. return 1;
  371. }
  372. /*}}}*/
  373. // ServerState::RunData - Transfer the data from the socket /*{{{*/
  374. // ---------------------------------------------------------------------
  375. /* */
  376. bool ServerState::RunData()
  377. {
  378. State = Data;
  379. // Chunked transfer encoding is fun..
  380. if (Encoding == Chunked)
  381. {
  382. while (1)
  383. {
  384. // Grab the block size
  385. bool Last = true;
  386. string Data;
  387. In.Limit(-1);
  388. do
  389. {
  390. if (In.WriteTillEl(Data,true) == true)
  391. break;
  392. }
  393. while ((Last = Owner->Go(false,this)) == true);
  394. if (Last == false)
  395. return false;
  396. // See if we are done
  397. unsigned long Len = strtol(Data.c_str(),0,16);
  398. if (Len == 0)
  399. {
  400. In.Limit(-1);
  401. // We have to remove the entity trailer
  402. Last = true;
  403. do
  404. {
  405. if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
  406. break;
  407. }
  408. while ((Last = Owner->Go(false,this)) == true);
  409. if (Last == false)
  410. return false;
  411. return !_error->PendingError();
  412. }
  413. // Transfer the block
  414. In.Limit(Len);
  415. while (Owner->Go(true,this) == true)
  416. if (In.IsLimit() == true)
  417. break;
  418. // Error
  419. if (In.IsLimit() == false)
  420. return false;
  421. // The server sends an extra new line before the next block specifier..
  422. In.Limit(-1);
  423. Last = true;
  424. do
  425. {
  426. if (In.WriteTillEl(Data,true) == true)
  427. break;
  428. }
  429. while ((Last = Owner->Go(false,this)) == true);
  430. if (Last == false)
  431. return false;
  432. }
  433. }
  434. else
  435. {
  436. /* Closes encoding is used when the server did not specify a size, the
  437. loss of the connection means we are done */
  438. if (Encoding == Closes)
  439. In.Limit(-1);
  440. else
  441. In.Limit(Size - StartPos);
  442. // Just transfer the whole block.
  443. do
  444. {
  445. if (In.IsLimit() == false)
  446. continue;
  447. In.Limit(-1);
  448. return !_error->PendingError();
  449. }
  450. while (Owner->Go(true,this) == true);
  451. }
  452. return Owner->Flush(this) && !_error->PendingError();
  453. }
  454. /*}}}*/
  455. // ServerState::HeaderLine - Process a header line /*{{{*/
  456. // ---------------------------------------------------------------------
  457. /* */
  458. bool ServerState::HeaderLine(string Line)
  459. {
  460. if (Line.empty() == true)
  461. return true;
  462. // The http server might be trying to do something evil.
  463. if (Line.length() >= MAXLEN)
  464. return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
  465. string::size_type Pos = Line.find(' ');
  466. if (Pos == string::npos || Pos+1 > Line.length())
  467. {
  468. // Blah, some servers use "connection:closes", evil.
  469. Pos = Line.find(':');
  470. if (Pos == string::npos || Pos + 2 > Line.length())
  471. return _error->Error(_("Bad header line"));
  472. Pos++;
  473. }
  474. // Parse off any trailing spaces between the : and the next word.
  475. string::size_type Pos2 = Pos;
  476. while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
  477. Pos2++;
  478. string Tag = string(Line,0,Pos);
  479. string Val = string(Line,Pos2);
  480. if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
  481. {
  482. // Evil servers return no version
  483. if (Line[4] == '/')
  484. {
  485. if (sscanf(Line.c_str(),"HTTP/%u.%u %u%[^\n]",&Major,&Minor,
  486. &Result,Code) != 4)
  487. return _error->Error(_("The HTTP server sent an invalid reply header"));
  488. }
  489. else
  490. {
  491. Major = 0;
  492. Minor = 9;
  493. if (sscanf(Line.c_str(),"HTTP %u%[^\n]",&Result,Code) != 2)
  494. return _error->Error(_("The HTTP server sent an invalid reply header"));
  495. }
  496. /* Check the HTTP response header to get the default persistance
  497. state. */
  498. if (Major < 1)
  499. Persistent = false;
  500. else
  501. {
  502. if (Major == 1 && Minor <= 0)
  503. Persistent = false;
  504. else
  505. Persistent = true;
  506. }
  507. return true;
  508. }
  509. if (stringcasecmp(Tag,"Content-Length:") == 0)
  510. {
  511. if (Encoding == Closes)
  512. Encoding = Stream;
  513. HaveContent = true;
  514. // The length is already set from the Content-Range header
  515. if (StartPos != 0)
  516. return true;
  517. if (sscanf(Val.c_str(),"%lu",&Size) != 1)
  518. return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
  519. return true;
  520. }
  521. if (stringcasecmp(Tag,"Content-Type:") == 0)
  522. {
  523. HaveContent = true;
  524. return true;
  525. }
  526. if (stringcasecmp(Tag,"Content-Range:") == 0)
  527. {
  528. HaveContent = true;
  529. if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2)
  530. return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
  531. if ((unsigned)StartPos > Size)
  532. return _error->Error(_("This HTTP server has broken range support"));
  533. return true;
  534. }
  535. if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
  536. {
  537. HaveContent = true;
  538. if (stringcasecmp(Val,"chunked") == 0)
  539. Encoding = Chunked;
  540. return true;
  541. }
  542. if (stringcasecmp(Tag,"Connection:") == 0)
  543. {
  544. if (stringcasecmp(Val,"close") == 0)
  545. Persistent = false;
  546. if (stringcasecmp(Val,"keep-alive") == 0)
  547. Persistent = true;
  548. return true;
  549. }
  550. if (stringcasecmp(Tag,"Last-Modified:") == 0)
  551. {
  552. if (StrToTime(Val,Date) == false)
  553. return _error->Error(_("Unknown date format"));
  554. return true;
  555. }
  556. if (stringcasecmp(Tag,"Location:") == 0)
  557. {
  558. Location = Val;
  559. return true;
  560. }
  561. return true;
  562. }
  563. /*}}}*/
  564. // HttpMethod::SendReq - Send the HTTP request /*{{{*/
  565. // ---------------------------------------------------------------------
  566. /* This places the http request in the outbound buffer */
  567. void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
  568. {
  569. URI Uri = Itm->Uri;
  570. // The HTTP server expects a hostname with a trailing :port
  571. char Buf[1000];
  572. string ProperHost = Uri.Host;
  573. if (Uri.Port != 0)
  574. {
  575. sprintf(Buf,":%u",Uri.Port);
  576. ProperHost += Buf;
  577. }
  578. // Just in case.
  579. if (Itm->Uri.length() >= sizeof(Buf))
  580. abort();
  581. /* Build the request. We include a keep-alive header only for non-proxy
  582. requests. This is to tweak old http/1.0 servers that do support keep-alive
  583. but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
  584. will glitch HTTP/1.0 proxies because they do not filter it out and
  585. pass it on, HTTP/1.1 says the connection should default to keep alive
  586. and we expect the proxy to do this */
  587. if (Proxy.empty() == true || Proxy.Host.empty())
  588. sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
  589. QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
  590. else
  591. {
  592. /* Generate a cache control header if necessary. We place a max
  593. cache age on index files, optionally set a no-cache directive
  594. and a no-store directive for archives. */
  595. sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
  596. Itm->Uri.c_str(),ProperHost.c_str());
  597. // only generate a cache control header if we actually want to
  598. // use a cache
  599. if (_config->FindB("Acquire::http::No-Cache",false) == false)
  600. {
  601. if (Itm->IndexFile == true)
  602. sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
  603. _config->FindI("Acquire::http::Max-Age",0));
  604. else
  605. {
  606. if (_config->FindB("Acquire::http::No-Store",false) == true)
  607. strcat(Buf,"Cache-Control: no-store\r\n");
  608. }
  609. }
  610. }
  611. // generate a no-cache header if needed
  612. if (_config->FindB("Acquire::http::No-Cache",false) == true)
  613. strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
  614. string Req = Buf;
  615. // Check for a partial file
  616. struct stat SBuf;
  617. if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
  618. {
  619. // In this case we send an if-range query with a range header
  620. sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",(long)SBuf.st_size - 1,
  621. TimeRFC1123(SBuf.st_mtime).c_str());
  622. Req += Buf;
  623. }
  624. else
  625. {
  626. if (Itm->LastModified != 0)
  627. {
  628. sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
  629. Req += Buf;
  630. }
  631. }
  632. if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
  633. Req += string("Proxy-Authorization: Basic ") +
  634. Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
  635. if (Uri.User.empty() == false || Uri.Password.empty() == false)
  636. Req += string("Authorization: Basic ") +
  637. Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
  638. Req += "User-Agent: Debian APT-HTTP/1.3 ("VERSION")\r\n\r\n";
  639. if (Debug == true)
  640. cerr << Req << endl;
  641. Out.Read(Req);
  642. }
  643. /*}}}*/
  644. // HttpMethod::Go - Run a single loop /*{{{*/
  645. // ---------------------------------------------------------------------
  646. /* This runs the select loop over the server FDs, Output file FDs and
  647. stdin. */
  648. bool HttpMethod::Go(bool ToFile,ServerState *Srv)
  649. {
  650. // Server has closed the connection
  651. if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
  652. ToFile == false))
  653. return false;
  654. fd_set rfds,wfds;
  655. FD_ZERO(&rfds);
  656. FD_ZERO(&wfds);
  657. /* Add the server. We only send more requests if the connection will
  658. be persisting */
  659. if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
  660. && Srv->Persistent == true)
  661. FD_SET(Srv->ServerFd,&wfds);
  662. if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
  663. FD_SET(Srv->ServerFd,&rfds);
  664. // Add the file
  665. int FileFD = -1;
  666. if (File != 0)
  667. FileFD = File->Fd();
  668. if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
  669. FD_SET(FileFD,&wfds);
  670. // Add stdin
  671. FD_SET(STDIN_FILENO,&rfds);
  672. // Figure out the max fd
  673. int MaxFd = FileFD;
  674. if (MaxFd < Srv->ServerFd)
  675. MaxFd = Srv->ServerFd;
  676. // Select
  677. struct timeval tv;
  678. tv.tv_sec = TimeOut;
  679. tv.tv_usec = 0;
  680. int Res = 0;
  681. if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
  682. {
  683. if (errno == EINTR)
  684. return true;
  685. return _error->Errno("select",_("Select failed"));
  686. }
  687. if (Res == 0)
  688. {
  689. _error->Error(_("Connection timed out"));
  690. return ServerDie(Srv);
  691. }
  692. // Handle server IO
  693. if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
  694. {
  695. errno = 0;
  696. if (Srv->In.Read(Srv->ServerFd) == false)
  697. return ServerDie(Srv);
  698. }
  699. if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
  700. {
  701. errno = 0;
  702. if (Srv->Out.Write(Srv->ServerFd) == false)
  703. return ServerDie(Srv);
  704. }
  705. // Send data to the file
  706. if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
  707. {
  708. if (Srv->In.Write(FileFD) == false)
  709. return _error->Errno("write",_("Error writing to output file"));
  710. }
  711. // Handle commands from APT
  712. if (FD_ISSET(STDIN_FILENO,&rfds))
  713. {
  714. if (Run(true) != -1)
  715. exit(100);
  716. }
  717. return true;
  718. }
  719. /*}}}*/
  720. // HttpMethod::Flush - Dump the buffer into the file /*{{{*/
  721. // ---------------------------------------------------------------------
  722. /* This takes the current input buffer from the Server FD and writes it
  723. into the file */
  724. bool HttpMethod::Flush(ServerState *Srv)
  725. {
  726. if (File != 0)
  727. {
  728. // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
  729. // can't be set
  730. if (File->Name() != "/dev/null")
  731. SetNonBlock(File->Fd(),false);
  732. if (Srv->In.WriteSpace() == false)
  733. return true;
  734. while (Srv->In.WriteSpace() == true)
  735. {
  736. if (Srv->In.Write(File->Fd()) == false)
  737. return _error->Errno("write",_("Error writing to file"));
  738. if (Srv->In.IsLimit() == true)
  739. return true;
  740. }
  741. if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
  742. return true;
  743. }
  744. return false;
  745. }
  746. /*}}}*/
  747. // HttpMethod::ServerDie - The server has closed the connection. /*{{{*/
  748. // ---------------------------------------------------------------------
  749. /* */
  750. bool HttpMethod::ServerDie(ServerState *Srv)
  751. {
  752. unsigned int LErrno = errno;
  753. // Dump the buffer to the file
  754. if (Srv->State == ServerState::Data)
  755. {
  756. // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
  757. // can't be set
  758. if (File->Name() != "/dev/null")
  759. SetNonBlock(File->Fd(),false);
  760. while (Srv->In.WriteSpace() == true)
  761. {
  762. if (Srv->In.Write(File->Fd()) == false)
  763. return _error->Errno("write",_("Error writing to the file"));
  764. // Done
  765. if (Srv->In.IsLimit() == true)
  766. return true;
  767. }
  768. }
  769. // See if this is because the server finished the data stream
  770. if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
  771. Srv->Encoding != ServerState::Closes)
  772. {
  773. Srv->Close();
  774. if (LErrno == 0)
  775. return _error->Error(_("Error reading from server. Remote end closed connection"));
  776. errno = LErrno;
  777. return _error->Errno("read",_("Error reading from server"));
  778. }
  779. else
  780. {
  781. Srv->In.Limit(-1);
  782. // Nothing left in the buffer
  783. if (Srv->In.WriteSpace() == false)
  784. return false;
  785. // We may have got multiple responses back in one packet..
  786. Srv->Close();
  787. return true;
  788. }
  789. return false;
  790. }
  791. /*}}}*/
  792. // HttpMethod::DealWithHeaders - Handle the retrieved header data /*{{{*/
  793. // ---------------------------------------------------------------------
  794. /* We look at the header data we got back from the server and decide what
  795. to do. Returns
  796. 0 - File is open,
  797. 1 - IMS hit
  798. 3 - Unrecoverable error
  799. 4 - Error with error content page
  800. 5 - Unrecoverable non-server error (close the connection)
  801. 6 - Try again with a new or changed URI
  802. */
  803. int HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
  804. {
  805. // Not Modified
  806. if (Srv->Result == 304)
  807. {
  808. unlink(Queue->DestFile.c_str());
  809. Res.IMSHit = true;
  810. Res.LastModified = Queue->LastModified;
  811. return 1;
  812. }
  813. /* Redirect
  814. *
  815. * Note that it is only OK for us to treat all redirection the same
  816. * because we *always* use GET, not other HTTP methods. There are
  817. * three redirection codes for which it is not appropriate that we
  818. * redirect. Pass on those codes so the error handling kicks in.
  819. */
  820. if (AllowRedirect
  821. && (Srv->Result > 300 && Srv->Result < 400)
  822. && (Srv->Result != 300 // Multiple Choices
  823. && Srv->Result != 304 // Not Modified
  824. && Srv->Result != 306)) // (Not part of HTTP/1.1, reserved)
  825. {
  826. if (!Srv->Location.empty())
  827. {
  828. NextURI = Srv->Location;
  829. return 6;
  830. }
  831. /* else pass through for error message */
  832. }
  833. /* We have a reply we dont handle. This should indicate a perm server
  834. failure */
  835. if (Srv->Result < 200 || Srv->Result >= 300)
  836. {
  837. _error->Error("%u %s",Srv->Result,Srv->Code);
  838. if (Srv->HaveContent == true)
  839. return 4;
  840. return 3;
  841. }
  842. // This is some sort of 2xx 'data follows' reply
  843. Res.LastModified = Srv->Date;
  844. Res.Size = Srv->Size;
  845. // Open the file
  846. delete File;
  847. File = new FileFd(Queue->DestFile,FileFd::WriteAny);
  848. if (_error->PendingError() == true)
  849. return 5;
  850. FailFile = Queue->DestFile;
  851. FailFile.c_str(); // Make sure we dont do a malloc in the signal handler
  852. FailFd = File->Fd();
  853. FailTime = Srv->Date;
  854. // Set the expected size
  855. if (Srv->StartPos >= 0)
  856. {
  857. Res.ResumePoint = Srv->StartPos;
  858. if (ftruncate(File->Fd(),Srv->StartPos) < 0)
  859. _error->Errno("ftruncate", _("Failed to truncate file"));
  860. }
  861. // Set the start point
  862. lseek(File->Fd(),0,SEEK_END);
  863. delete Srv->In.Hash;
  864. Srv->In.Hash = new Hashes;
  865. // Fill the Hash if the file is non-empty (resume)
  866. if (Srv->StartPos > 0)
  867. {
  868. lseek(File->Fd(),0,SEEK_SET);
  869. if (Srv->In.Hash->AddFD(File->Fd(),Srv->StartPos) == false)
  870. {
  871. _error->Errno("read",_("Problem hashing file"));
  872. return 5;
  873. }
  874. lseek(File->Fd(),0,SEEK_END);
  875. }
  876. SetNonBlock(File->Fd(),true);
  877. return 0;
  878. }
  879. /*}}}*/
  880. // HttpMethod::SigTerm - Handle a fatal signal /*{{{*/
  881. // ---------------------------------------------------------------------
  882. /* This closes and timestamps the open file. This is neccessary to get
  883. resume behavoir on user abort */
  884. void HttpMethod::SigTerm(int)
  885. {
  886. if (FailFd == -1)
  887. _exit(100);
  888. close(FailFd);
  889. // Timestamp
  890. struct utimbuf UBuf;
  891. UBuf.actime = FailTime;
  892. UBuf.modtime = FailTime;
  893. utime(FailFile.c_str(),&UBuf);
  894. _exit(100);
  895. }
  896. /*}}}*/
  897. // HttpMethod::Fetch - Fetch an item /*{{{*/
  898. // ---------------------------------------------------------------------
  899. /* This adds an item to the pipeline. We keep the pipeline at a fixed
  900. depth. */
  901. bool HttpMethod::Fetch(FetchItem *)
  902. {
  903. if (Server == 0)
  904. return true;
  905. // Queue the requests
  906. int Depth = -1;
  907. for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
  908. I = I->Next, Depth++)
  909. {
  910. // If pipelining is disabled, we only queue 1 request
  911. if (Server->Pipeline == false && Depth >= 0)
  912. break;
  913. // Make sure we stick with the same server
  914. if (Server->Comp(I->Uri) == false)
  915. break;
  916. if (QueueBack == I)
  917. {
  918. QueueBack = I->Next;
  919. SendReq(I,Server->Out);
  920. continue;
  921. }
  922. }
  923. return true;
  924. };
  925. /*}}}*/
  926. // HttpMethod::Configuration - Handle a configuration message /*{{{*/
  927. // ---------------------------------------------------------------------
  928. /* We stash the desired pipeline depth */
  929. bool HttpMethod::Configuration(string Message)
  930. {
  931. if (pkgAcqMethod::Configuration(Message) == false)
  932. return false;
  933. AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
  934. TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
  935. PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
  936. PipelineDepth);
  937. Debug = _config->FindB("Debug::Acquire::http",false);
  938. return true;
  939. }
  940. /*}}}*/
  941. // HttpMethod::Loop - Main loop /*{{{*/
  942. // ---------------------------------------------------------------------
  943. /* */
  944. int HttpMethod::Loop()
  945. {
  946. typedef vector<string> StringVector;
  947. typedef vector<string>::iterator StringVectorIterator;
  948. map<string, StringVector> Redirected;
  949. signal(SIGTERM,SigTerm);
  950. signal(SIGINT,SigTerm);
  951. Server = 0;
  952. int FailCounter = 0;
  953. while (1)
  954. {
  955. // We have no commands, wait for some to arrive
  956. if (Queue == 0)
  957. {
  958. if (WaitFd(STDIN_FILENO) == false)
  959. return 0;
  960. }
  961. /* Run messages, we can accept 0 (no message) if we didn't
  962. do a WaitFd above.. Otherwise the FD is closed. */
  963. int Result = Run(true);
  964. if (Result != -1 && (Result != 0 || Queue == 0))
  965. return 100;
  966. if (Queue == 0)
  967. continue;
  968. // Connect to the server
  969. if (Server == 0 || Server->Comp(Queue->Uri) == false)
  970. {
  971. delete Server;
  972. Server = new ServerState(Queue->Uri,this);
  973. }
  974. /* If the server has explicitly said this is the last connection
  975. then we pre-emptively shut down the pipeline and tear down
  976. the connection. This will speed up HTTP/1.0 servers a tad
  977. since we don't have to wait for the close sequence to
  978. complete */
  979. if (Server->Persistent == false)
  980. Server->Close();
  981. // Reset the pipeline
  982. if (Server->ServerFd == -1)
  983. QueueBack = Queue;
  984. // Connnect to the host
  985. if (Server->Open() == false)
  986. {
  987. Fail(true);
  988. delete Server;
  989. Server = 0;
  990. continue;
  991. }
  992. // Fill the pipeline.
  993. Fetch(0);
  994. // Fetch the next URL header data from the server.
  995. switch (Server->RunHeaders())
  996. {
  997. case 0:
  998. break;
  999. // The header data is bad
  1000. case 2:
  1001. {
  1002. _error->Error(_("Bad header data"));
  1003. Fail(true);
  1004. RotateDNS();
  1005. continue;
  1006. }
  1007. // The server closed a connection during the header get..
  1008. default:
  1009. case 1:
  1010. {
  1011. FailCounter++;
  1012. _error->Discard();
  1013. Server->Close();
  1014. Server->Pipeline = false;
  1015. if (FailCounter >= 2)
  1016. {
  1017. Fail(_("Connection failed"),true);
  1018. FailCounter = 0;
  1019. }
  1020. RotateDNS();
  1021. continue;
  1022. }
  1023. };
  1024. // Decide what to do.
  1025. FetchResult Res;
  1026. Res.Filename = Queue->DestFile;
  1027. switch (DealWithHeaders(Res,Server))
  1028. {
  1029. // Ok, the file is Open
  1030. case 0:
  1031. {
  1032. URIStart(Res);
  1033. // Run the data
  1034. bool Result = Server->RunData();
  1035. /* If the server is sending back sizeless responses then fill in
  1036. the size now */
  1037. if (Res.Size == 0)
  1038. Res.Size = File->Size();
  1039. // Close the file, destroy the FD object and timestamp it
  1040. FailFd = -1;
  1041. delete File;
  1042. File = 0;
  1043. // Timestamp
  1044. struct utimbuf UBuf;
  1045. time(&UBuf.actime);
  1046. UBuf.actime = Server->Date;
  1047. UBuf.modtime = Server->Date;
  1048. utime(Queue->DestFile.c_str(),&UBuf);
  1049. // Send status to APT
  1050. if (Result == true)
  1051. {
  1052. Res.TakeHashes(*Server->In.Hash);
  1053. URIDone(Res);
  1054. }
  1055. else
  1056. {
  1057. if (Server->ServerFd == -1)
  1058. {
  1059. FailCounter++;
  1060. _error->Discard();
  1061. Server->Close();
  1062. if (FailCounter >= 2)
  1063. {
  1064. Fail(_("Connection failed"),true);
  1065. FailCounter = 0;
  1066. }
  1067. QueueBack = Queue;
  1068. }
  1069. else
  1070. Fail(true);
  1071. }
  1072. break;
  1073. }
  1074. // IMS hit
  1075. case 1:
  1076. {
  1077. URIDone(Res);
  1078. break;
  1079. }
  1080. // Hard server error, not found or something
  1081. case 3:
  1082. {
  1083. Fail();
  1084. break;
  1085. }
  1086. // Hard internal error, kill the connection and fail
  1087. case 5:
  1088. {
  1089. delete File;
  1090. File = 0;
  1091. Fail();
  1092. RotateDNS();
  1093. Server->Close();
  1094. break;
  1095. }
  1096. // We need to flush the data, the header is like a 404 w/ error text
  1097. case 4:
  1098. {
  1099. Fail();
  1100. // Send to content to dev/null
  1101. File = new FileFd("/dev/null",FileFd::WriteExists);
  1102. Server->RunData();
  1103. delete File;
  1104. File = 0;
  1105. break;
  1106. }
  1107. // Try again with a new URL
  1108. case 6:
  1109. {
  1110. // Clear rest of response if there is content
  1111. if (Server->HaveContent)
  1112. {
  1113. File = new FileFd("/dev/null",FileFd::WriteExists);
  1114. Server->RunData();
  1115. delete File;
  1116. File = 0;
  1117. }
  1118. /* Detect redirect loops. No more redirects are allowed
  1119. after the same URI is seen twice in a queue item. */
  1120. StringVector &R = Redirected[Queue->DestFile];
  1121. bool StopRedirects = false;
  1122. if (R.size() == 0)
  1123. R.push_back(Queue->Uri);
  1124. else if (R[0] == "STOP" || R.size() > 10)
  1125. StopRedirects = true;
  1126. else
  1127. {
  1128. for (StringVectorIterator I = R.begin(); I != R.end(); I++)
  1129. if (Queue->Uri == *I)
  1130. {
  1131. R[0] = "STOP";
  1132. break;
  1133. }
  1134. R.push_back(Queue->Uri);
  1135. }
  1136. if (StopRedirects == false)
  1137. Redirect(NextURI);
  1138. else
  1139. Fail();
  1140. break;
  1141. }
  1142. default:
  1143. Fail(_("Internal error"));
  1144. break;
  1145. }
  1146. FailCounter = 0;
  1147. }
  1148. return 0;
  1149. }
  1150. /*}}}*/
  1151. int main()
  1152. {
  1153. setlocale(LC_ALL, "");
  1154. // ignore SIGPIPE, this can happen on write() if the socket
  1155. // closes the connection (this is dealt with via ServerDie())
  1156. signal(SIGPIPE, SIG_IGN);
  1157. HttpMethod Mth;
  1158. return Mth.Loop();
  1159. }