]>
git.saurik.com Git - apt.git/blob - methods/http.cc
1 // -*- mode: cpp; mode: fold -*-
3 // $Id: http.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
4 /* ######################################################################
6 HTTP Acquire Method - This is the HTTP acquire method for APT.
8 It uses HTTP/1.1 and many of the fancy options there-in, such as
9 pipelining, range, if-range and so on.
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).
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.
25 ##################################################################### */
27 // Include Files /*{{{*/
30 #include <apt-pkg/fileutl.h>
31 #include <apt-pkg/acquire-method.h>
32 #include <apt-pkg/configuration.h>
33 #include <apt-pkg/error.h>
34 #include <apt-pkg/hashes.h>
35 #include <apt-pkg/netrc.h>
36 #include <apt-pkg/strutl.h>
37 #include <apt-pkg/proxy.h>
41 #include <sys/select.h>
59 unsigned long long CircleBuf::BwReadLimit
=0;
60 unsigned long long CircleBuf::BwTickReadData
=0;
61 struct timeval
CircleBuf::BwReadTick
={0,0};
62 const unsigned int CircleBuf::BW_HZ
=10;
64 // CircleBuf::CircleBuf - Circular input buffer /*{{{*/
65 // ---------------------------------------------------------------------
67 CircleBuf::CircleBuf(unsigned long long Size
)
68 : Size(Size
), Hash(0), TotalWriten(0)
70 Buf
= new unsigned char[Size
];
73 CircleBuf::BwReadLimit
= _config
->FindI("Acquire::http::Dl-Limit",0)*1024;
76 // CircleBuf::Reset - Reset to the default state /*{{{*/
77 // ---------------------------------------------------------------------
79 void CircleBuf::Reset()
85 MaxGet
= (unsigned long long)-1;
94 // CircleBuf::Read - Read from a FD into the circular buffer /*{{{*/
95 // ---------------------------------------------------------------------
96 /* This fills up the buffer with as much data as is in the FD, assuming it
98 bool CircleBuf::Read(int Fd
)
102 // Woops, buffer is full
103 if (InP
- OutP
== Size
)
106 // what's left to read in this tick
107 unsigned long long const BwReadMax
= CircleBuf::BwReadLimit
/BW_HZ
;
109 if(CircleBuf::BwReadLimit
) {
111 gettimeofday(&now
,0);
113 unsigned long long d
= (now
.tv_sec
-CircleBuf::BwReadTick
.tv_sec
)*1000000 +
114 now
.tv_usec
-CircleBuf::BwReadTick
.tv_usec
;
115 if(d
> 1000000/BW_HZ
) {
116 CircleBuf::BwReadTick
= now
;
117 CircleBuf::BwTickReadData
= 0;
120 if(CircleBuf::BwTickReadData
>= BwReadMax
) {
121 usleep(1000000/BW_HZ
);
126 // Write the buffer segment
128 if(CircleBuf::BwReadLimit
) {
129 Res
= read(Fd
,Buf
+ (InP%Size
),
130 BwReadMax
> LeftRead() ? LeftRead() : BwReadMax
);
132 Res
= read(Fd
,Buf
+ (InP%Size
),LeftRead());
134 if(Res
> 0 && BwReadLimit
> 0)
135 CircleBuf::BwTickReadData
+= Res
;
147 gettimeofday(&Start
,0);
152 // CircleBuf::Read - Put the string into the buffer /*{{{*/
153 // ---------------------------------------------------------------------
154 /* This will hold the string in and fill the buffer with it as it empties */
155 bool CircleBuf::Read(string Data
)
162 // CircleBuf::FillOut - Fill the buffer from the output queue /*{{{*/
163 // ---------------------------------------------------------------------
165 void CircleBuf::FillOut()
167 if (OutQueue
.empty() == true)
171 // Woops, buffer is full
172 if (InP
- OutP
== Size
)
175 // Write the buffer segment
176 unsigned long long Sz
= LeftRead();
177 if (OutQueue
.length() - StrPos
< Sz
)
178 Sz
= OutQueue
.length() - StrPos
;
179 memcpy(Buf
+ (InP%Size
),OutQueue
.c_str() + StrPos
,Sz
);
184 if (OutQueue
.length() == StrPos
)
193 // CircleBuf::Write - Write from the buffer into a FD /*{{{*/
194 // ---------------------------------------------------------------------
195 /* This empties the buffer into the FD. */
196 bool CircleBuf::Write(int Fd
)
202 // Woops, buffer is empty
209 // Write the buffer segment
211 Res
= write(Fd
,Buf
+ (OutP%Size
),LeftWrite());
226 Hash
->Add(Buf
+ (OutP%Size
),Res
);
232 // CircleBuf::WriteTillEl - Write from the buffer to a string /*{{{*/
233 // ---------------------------------------------------------------------
234 /* This copies till the first empty line */
235 bool CircleBuf::WriteTillEl(string
&Data
,bool Single
)
237 // We cheat and assume it is unneeded to have more than one buffer load
238 for (unsigned long long I
= OutP
; I
< InP
; I
++)
240 if (Buf
[I%Size
] != '\n')
246 if (I
< InP
&& Buf
[I%Size
] == '\r')
248 if (I
>= InP
|| Buf
[I%Size
] != '\n')
256 unsigned long long Sz
= LeftWrite();
261 Data
+= string((char *)(Buf
+ (OutP%Size
)),Sz
);
269 // CircleBuf::Stats - Print out stats information /*{{{*/
270 // ---------------------------------------------------------------------
272 void CircleBuf::Stats()
278 gettimeofday(&Stop
,0);
279 /* float Diff = Stop.tv_sec - Start.tv_sec +
280 (float)(Stop.tv_usec - Start.tv_usec)/1000000;
281 clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
284 CircleBuf::~CircleBuf()
290 // HttpServerState::HttpServerState - Constructor /*{{{*/
291 HttpServerState::HttpServerState(URI Srv
,HttpMethod
*Owner
) : ServerState(Srv
, Owner
), In(64*1024), Out(4*1024)
293 TimeOut
= _config
->FindI("Acquire::http::Timeout",TimeOut
);
297 // HttpServerState::Open - Open a connection to the server /*{{{*/
298 // ---------------------------------------------------------------------
299 /* This opens a connection to the server. */
300 bool HttpServerState::Open()
302 // Use the already open connection if possible.
311 // Determine the proxy setting
312 AutoDetectProxy(ServerName
);
313 string SpecificProxy
= _config
->Find("Acquire::http::Proxy::" + ServerName
.Host
);
314 if (!SpecificProxy
.empty())
316 if (SpecificProxy
== "DIRECT")
319 Proxy
= SpecificProxy
;
323 string DefProxy
= _config
->Find("Acquire::http::Proxy");
324 if (!DefProxy
.empty())
330 char* result
= getenv("http_proxy");
331 Proxy
= result
? result
: "";
335 // Parse no_proxy, a , separated list of domains
336 if (getenv("no_proxy") != 0)
338 if (CheckDomainList(ServerName
.Host
,getenv("no_proxy")) == true)
342 // Determine what host and port to use based on the proxy settings
345 if (Proxy
.empty() == true || Proxy
.Host
.empty() == true)
347 if (ServerName
.Port
!= 0)
348 Port
= ServerName
.Port
;
349 Host
= ServerName
.Host
;
358 // Connect to the remote server
359 if (Connect(Host
,Port
,"http",80,ServerFd
,TimeOut
,Owner
) == false)
365 // HttpServerState::Close - Close a connection to the server /*{{{*/
366 // ---------------------------------------------------------------------
368 bool HttpServerState::Close()
375 // HttpServerState::RunData - Transfer the data from the socket /*{{{*/
376 bool HttpServerState::RunData(FileFd
* const File
)
380 // Chunked transfer encoding is fun..
381 if (Encoding
== Chunked
)
385 // Grab the block size
391 if (In
.WriteTillEl(Data
,true) == true)
394 while ((Last
= Go(false, File
)) == true);
399 // See if we are done
400 unsigned long long Len
= strtoull(Data
.c_str(),0,16);
405 // We have to remove the entity trailer
409 if (In
.WriteTillEl(Data
,true) == true && Data
.length() <= 2)
412 while ((Last
= Go(false, File
)) == true);
415 return !_error
->PendingError();
418 // Transfer the block
420 while (Go(true, File
) == true)
421 if (In
.IsLimit() == true)
425 if (In
.IsLimit() == false)
428 // The server sends an extra new line before the next block specifier..
433 if (In
.WriteTillEl(Data
,true) == true)
436 while ((Last
= Go(false, File
)) == true);
443 /* Closes encoding is used when the server did not specify a size, the
444 loss of the connection means we are done */
445 if (Encoding
== Closes
)
448 In
.Limit(Size
- StartPos
);
450 // Just transfer the whole block.
453 if (In
.IsLimit() == false)
457 return !_error
->PendingError();
459 while (Go(true, File
) == true);
462 return Owner
->Flush() && !_error
->PendingError();
465 bool HttpServerState::ReadHeaderLines(std::string
&Data
) /*{{{*/
467 return In
.WriteTillEl(Data
);
470 bool HttpServerState::LoadNextResponse(bool const ToFile
, FileFd
* const File
)/*{{{*/
472 return Go(ToFile
, File
);
475 bool HttpServerState::WriteResponse(const std::string
&Data
) /*{{{*/
477 return Out
.Read(Data
);
480 APT_PURE
bool HttpServerState::IsOpen() /*{{{*/
482 return (ServerFd
!= -1);
485 bool HttpServerState::InitHashes(FileFd
&File
) /*{{{*/
488 In
.Hash
= new Hashes
;
490 // Set the expected size and read file for the hashes
491 File
.Truncate(StartPos
);
492 return In
.Hash
->AddFD(File
, StartPos
);
495 APT_PURE Hashes
* HttpServerState::GetHashes() /*{{{*/
500 // HttpServerState::Die - The server has closed the connection. /*{{{*/
501 bool HttpServerState::Die(FileFd
&File
)
503 unsigned int LErrno
= errno
;
505 // Dump the buffer to the file
506 if (State
== ServerState::Data
)
508 // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
510 if (File
.Name() != "/dev/null")
511 SetNonBlock(File
.Fd(),false);
512 while (In
.WriteSpace() == true)
514 if (In
.Write(File
.Fd()) == false)
515 return _error
->Errno("write",_("Error writing to the file"));
518 if (In
.IsLimit() == true)
523 // See if this is because the server finished the data stream
524 if (In
.IsLimit() == false && State
!= HttpServerState::Header
&&
525 Encoding
!= HttpServerState::Closes
)
529 return _error
->Error(_("Error reading from server. Remote end closed connection"));
531 return _error
->Errno("read",_("Error reading from server"));
537 // Nothing left in the buffer
538 if (In
.WriteSpace() == false)
541 // We may have got multiple responses back in one packet..
549 // HttpServerState::Flush - Dump the buffer into the file /*{{{*/
550 // ---------------------------------------------------------------------
551 /* This takes the current input buffer from the Server FD and writes it
553 bool HttpServerState::Flush(FileFd
* const File
)
557 // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
559 if (File
->Name() != "/dev/null")
560 SetNonBlock(File
->Fd(),false);
561 if (In
.WriteSpace() == false)
564 while (In
.WriteSpace() == true)
566 if (In
.Write(File
->Fd()) == false)
567 return _error
->Errno("write",_("Error writing to file"));
568 if (In
.IsLimit() == true)
572 if (In
.IsLimit() == true || Encoding
== ServerState::Closes
)
578 // HttpServerState::Go - Run a single loop /*{{{*/
579 // ---------------------------------------------------------------------
580 /* This runs the select loop over the server FDs, Output file FDs and
582 bool HttpServerState::Go(bool ToFile
, FileFd
* const File
)
584 // Server has closed the connection
585 if (ServerFd
== -1 && (In
.WriteSpace() == false ||
593 /* Add the server. We only send more requests if the connection will
595 if (Out
.WriteSpace() == true && ServerFd
!= -1
596 && Persistent
== true)
597 FD_SET(ServerFd
,&wfds
);
598 if (In
.ReadSpace() == true && ServerFd
!= -1)
599 FD_SET(ServerFd
,&rfds
);
606 if (In
.WriteSpace() == true && ToFile
== true && FileFD
!= -1)
607 FD_SET(FileFD
,&wfds
);
610 if (_config
->FindB("Acquire::http::DependOnSTDIN", true) == true)
611 FD_SET(STDIN_FILENO
,&rfds
);
613 // Figure out the max fd
615 if (MaxFd
< ServerFd
)
623 if ((Res
= select(MaxFd
+1,&rfds
,&wfds
,0,&tv
)) < 0)
627 return _error
->Errno("select",_("Select failed"));
632 _error
->Error(_("Connection timed out"));
637 if (ServerFd
!= -1 && FD_ISSET(ServerFd
,&rfds
))
640 if (In
.Read(ServerFd
) == false)
644 if (ServerFd
!= -1 && FD_ISSET(ServerFd
,&wfds
))
647 if (Out
.Write(ServerFd
) == false)
651 // Send data to the file
652 if (FileFD
!= -1 && FD_ISSET(FileFD
,&wfds
))
654 if (In
.Write(FileFD
) == false)
655 return _error
->Errno("write",_("Error writing to output file"));
658 if (MaximumSize
> 0 && File
&& File
->Tell() > MaximumSize
)
660 return _error
->Error("Writing more data than expected (%llu > %llu)",
661 File
->Tell(), MaximumSize
);
664 // Handle commands from APT
665 if (FD_ISSET(STDIN_FILENO
,&rfds
))
667 if (Owner
->Run(true) != -1)
675 // HttpMethod::SendReq - Send the HTTP request /*{{{*/
676 // ---------------------------------------------------------------------
677 /* This places the http request in the outbound buffer */
678 void HttpMethod::SendReq(FetchItem
*Itm
)
682 // The HTTP server expects a hostname with a trailing :port
683 std::stringstream Req
;
686 if (Uri
.Host
.find(':') != string::npos
)
687 ProperHost
= '[' + Uri
.Host
+ ']';
689 ProperHost
= Uri
.Host
;
691 /* RFC 2616 ยง5.1.2 requires absolute URIs for requests to proxies,
692 but while its a must for all servers to accept absolute URIs,
693 it is assumed clients will sent an absolute path for non-proxies */
694 std::string requesturi
;
695 if (Server
->Proxy
.empty() == true || Server
->Proxy
.Host
.empty())
696 requesturi
= Uri
.Path
;
698 requesturi
= Itm
->Uri
;
700 // The "+" is encoded as a workaround for a amazon S3 bug
701 // see LP bugs #1003633 and #1086997.
702 requesturi
= QuoteString(requesturi
, "+~ ");
704 /* Build the request. No keep-alive is included as it is the default
705 in 1.1, can cause problems with proxies, and we are an HTTP/1.1
707 C.f. https://tools.ietf.org/wg/httpbis/trac/ticket/158 */
708 Req
<< "GET " << requesturi
<< " HTTP/1.1\r\n";
710 Req
<< "Host: " << ProperHost
<< ":" << Uri
.Port
<< "\r\n";
712 Req
<< "Host: " << ProperHost
<< "\r\n";
714 // generate a cache control header (if needed)
715 if (_config
->FindB("Acquire::http::No-Cache",false) == true)
716 Req
<< "Cache-Control: no-cache\r\n"
717 << "Pragma: no-cache\r\n";
718 else if (Itm
->IndexFile
== true)
719 Req
<< "Cache-Control: max-age=" << _config
->FindI("Acquire::http::Max-Age",0) << "\r\n";
720 else if (_config
->FindB("Acquire::http::No-Store",false) == true)
721 Req
<< "Cache-Control: no-store\r\n";
723 // If we ask for uncompressed files servers might respond with content-
724 // negotiation which lets us end up with compressed files we do not support,
725 // see 657029, 657560 and co, so if we have no extension on the request
726 // ask for text only. As a sidenote: If there is nothing to negotate servers
727 // seem to be nice and ignore it.
728 if (_config
->FindB("Acquire::http::SendAccept", true) == true)
730 size_t const filepos
= Itm
->Uri
.find_last_of('/');
731 string
const file
= Itm
->Uri
.substr(filepos
+ 1);
732 if (flExtension(file
) == file
)
733 Req
<< "Accept: text/*\r\n";
736 // Check for a partial file and send if-queries accordingly
738 if (stat(Itm
->DestFile
.c_str(),&SBuf
) >= 0 && SBuf
.st_size
> 0)
739 Req
<< "Range: bytes=" << SBuf
.st_size
<< "-\r\n"
740 << "If-Range: " << TimeRFC1123(SBuf
.st_mtime
) << "\r\n";
741 else if (Itm
->LastModified
!= 0)
742 Req
<< "If-Modified-Since: " << TimeRFC1123(Itm
->LastModified
).c_str() << "\r\n";
744 if (Server
->Proxy
.User
.empty() == false || Server
->Proxy
.Password
.empty() == false)
745 Req
<< "Proxy-Authorization: Basic "
746 << Base64Encode(Server
->Proxy
.User
+ ":" + Server
->Proxy
.Password
) << "\r\n";
748 maybe_add_auth (Uri
, _config
->FindFile("Dir::Etc::netrc"));
749 if (Uri
.User
.empty() == false || Uri
.Password
.empty() == false)
750 Req
<< "Authorization: Basic "
751 << Base64Encode(Uri
.User
+ ":" + Uri
.Password
) << "\r\n";
753 Req
<< "User-Agent: " << _config
->Find("Acquire::http::User-Agent",
754 "Debian APT-HTTP/1.3 (" PACKAGE_VERSION
")") << "\r\n";
759 cerr
<< Req
.str() << endl
;
761 Server
->WriteResponse(Req
.str());
764 // HttpMethod::Configuration - Handle a configuration message /*{{{*/
765 // ---------------------------------------------------------------------
766 /* We stash the desired pipeline depth */
767 bool HttpMethod::Configuration(string Message
)
769 if (ServerMethod::Configuration(Message
) == false)
772 AllowRedirect
= _config
->FindB("Acquire::http::AllowRedirect",true);
773 PipelineDepth
= _config
->FindI("Acquire::http::Pipeline-Depth",
775 Debug
= _config
->FindB("Debug::Acquire::http",false);
780 ServerState
* HttpMethod::CreateServerState(URI uri
) /*{{{*/
782 return new HttpServerState(uri
, this);
785 void HttpMethod::RotateDNS() /*{{{*/