]>
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 aquire 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>
54 #include "rfc2553emu.h"
61 unsigned long long CircleBuf::BwReadLimit
=0;
62 unsigned long long CircleBuf::BwTickReadData
=0;
63 struct timeval
CircleBuf::BwReadTick
={0,0};
64 const unsigned int CircleBuf::BW_HZ
=10;
66 // CircleBuf::CircleBuf - Circular input buffer /*{{{*/
67 // ---------------------------------------------------------------------
69 CircleBuf::CircleBuf(unsigned long long Size
) : Size(Size
), Hash(0)
71 Buf
= new unsigned char[Size
];
74 CircleBuf::BwReadLimit
= _config
->FindI("Acquire::http::Dl-Limit",0)*1024;
77 // CircleBuf::Reset - Reset to the default state /*{{{*/
78 // ---------------------------------------------------------------------
80 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
)
100 unsigned long long BwReadMax
;
104 // Woops, buffer is full
105 if (InP
- OutP
== Size
)
108 // what's left to read in this tick
109 BwReadMax
= CircleBuf::BwReadLimit
/BW_HZ
;
111 if(CircleBuf::BwReadLimit
) {
113 gettimeofday(&now
,0);
115 unsigned long long d
= (now
.tv_sec
-CircleBuf::BwReadTick
.tv_sec
)*1000000 +
116 now
.tv_usec
-CircleBuf::BwReadTick
.tv_usec
;
117 if(d
> 1000000/BW_HZ
) {
118 CircleBuf::BwReadTick
= now
;
119 CircleBuf::BwTickReadData
= 0;
122 if(CircleBuf::BwTickReadData
>= BwReadMax
) {
123 usleep(1000000/BW_HZ
);
128 // Write the buffer segment
130 if(CircleBuf::BwReadLimit
) {
131 Res
= read(Fd
,Buf
+ (InP%Size
),
132 BwReadMax
> LeftRead() ? LeftRead() : BwReadMax
);
134 Res
= read(Fd
,Buf
+ (InP%Size
),LeftRead());
136 if(Res
> 0 && BwReadLimit
> 0)
137 CircleBuf::BwTickReadData
+= Res
;
149 gettimeofday(&Start
,0);
154 // CircleBuf::Read - Put the string into the buffer /*{{{*/
155 // ---------------------------------------------------------------------
156 /* This will hold the string in and fill the buffer with it as it empties */
157 bool CircleBuf::Read(string Data
)
164 // CircleBuf::FillOut - Fill the buffer from the output queue /*{{{*/
165 // ---------------------------------------------------------------------
167 void CircleBuf::FillOut()
169 if (OutQueue
.empty() == true)
173 // Woops, buffer is full
174 if (InP
- OutP
== Size
)
177 // Write the buffer segment
178 unsigned long long Sz
= LeftRead();
179 if (OutQueue
.length() - StrPos
< Sz
)
180 Sz
= OutQueue
.length() - StrPos
;
181 memcpy(Buf
+ (InP%Size
),OutQueue
.c_str() + StrPos
,Sz
);
186 if (OutQueue
.length() == StrPos
)
195 // CircleBuf::Write - Write from the buffer into a FD /*{{{*/
196 // ---------------------------------------------------------------------
197 /* This empties the buffer into the FD. */
198 bool CircleBuf::Write(int Fd
)
204 // Woops, buffer is empty
211 // Write the buffer segment
213 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 string SpecificProxy
= _config
->Find("Acquire::http::Proxy::" + ServerName
.Host
);
313 if (!SpecificProxy
.empty())
315 if (SpecificProxy
== "DIRECT")
318 Proxy
= SpecificProxy
;
322 string DefProxy
= _config
->Find("Acquire::http::Proxy");
323 if (!DefProxy
.empty())
329 char* result
= getenv("http_proxy");
330 Proxy
= result
? result
: "";
334 // Parse no_proxy, a , separated list of domains
335 if (getenv("no_proxy") != 0)
337 if (CheckDomainList(ServerName
.Host
,getenv("no_proxy")) == true)
341 // Determine what host and port to use based on the proxy settings
344 if (Proxy
.empty() == true || Proxy
.Host
.empty() == true)
346 if (ServerName
.Port
!= 0)
347 Port
= ServerName
.Port
;
348 Host
= ServerName
.Host
;
357 // Connect to the remote server
358 if (Connect(Host
,Port
,"http",80,ServerFd
,TimeOut
,Owner
) == false)
364 // HttpServerState::Close - Close a connection to the server /*{{{*/
365 // ---------------------------------------------------------------------
367 bool HttpServerState::Close()
374 // HttpServerState::RunData - Transfer the data from the socket /*{{{*/
375 bool HttpServerState::RunData(FileFd
* const File
)
379 // Chunked transfer encoding is fun..
380 if (Encoding
== Chunked
)
384 // Grab the block size
390 if (In
.WriteTillEl(Data
,true) == true)
393 while ((Last
= Go(false, File
)) == true);
398 // See if we are done
399 unsigned long long Len
= strtoull(Data
.c_str(),0,16);
404 // We have to remove the entity trailer
408 if (In
.WriteTillEl(Data
,true) == true && Data
.length() <= 2)
411 while ((Last
= Go(false, File
)) == true);
414 return !_error
->PendingError();
417 // Transfer the block
419 while (Go(true, File
) == true)
420 if (In
.IsLimit() == true)
424 if (In
.IsLimit() == false)
427 // The server sends an extra new line before the next block specifier..
432 if (In
.WriteTillEl(Data
,true) == true)
435 while ((Last
= Go(false, File
)) == true);
442 /* Closes encoding is used when the server did not specify a size, the
443 loss of the connection means we are done */
444 if (Encoding
== Closes
)
447 In
.Limit(Size
- StartPos
);
449 // Just transfer the whole block.
452 if (In
.IsLimit() == false)
456 return !_error
->PendingError();
458 while (Go(true, File
) == true);
461 return Owner
->Flush() && !_error
->PendingError();
464 bool HttpServerState::ReadHeaderLines(std::string
&Data
) /*{{{*/
466 return In
.WriteTillEl(Data
);
469 bool HttpServerState::LoadNextResponse(bool const ToFile
, FileFd
* const File
)/*{{{*/
471 return Go(ToFile
, File
);
474 bool HttpServerState::WriteResponse(const std::string
&Data
) /*{{{*/
476 return Out
.Read(Data
);
479 bool HttpServerState::IsOpen() /*{{{*/
481 return (ServerFd
!= -1);
484 bool HttpServerState::InitHashes(FileFd
&File
) /*{{{*/
487 In
.Hash
= new Hashes
;
489 // Set the expected size and read file for the hashes
492 File
.Truncate(StartPos
);
494 return In
.Hash
->AddFD(File
, StartPos
);
499 Hashes
* HttpServerState::GetHashes() /*{{{*/
504 // HttpServerState::Die - The server has closed the connection. /*{{{*/
505 bool HttpServerState::Die(FileFd
&File
)
507 unsigned int LErrno
= errno
;
509 // Dump the buffer to the file
510 if (State
== ServerState::Data
)
512 // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
514 if (File
.Name() != "/dev/null")
515 SetNonBlock(File
.Fd(),false);
516 while (In
.WriteSpace() == true)
518 if (In
.Write(File
.Fd()) == false)
519 return _error
->Errno("write",_("Error writing to the file"));
522 if (In
.IsLimit() == true)
527 // See if this is because the server finished the data stream
528 if (In
.IsLimit() == false && State
!= HttpServerState::Header
&&
529 Encoding
!= HttpServerState::Closes
)
533 return _error
->Error(_("Error reading from server. Remote end closed connection"));
535 return _error
->Errno("read",_("Error reading from server"));
541 // Nothing left in the buffer
542 if (In
.WriteSpace() == false)
545 // We may have got multiple responses back in one packet..
553 // HttpServerState::Flush - Dump the buffer into the file /*{{{*/
554 // ---------------------------------------------------------------------
555 /* This takes the current input buffer from the Server FD and writes it
557 bool HttpServerState::Flush(FileFd
* const File
)
561 // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
563 if (File
->Name() != "/dev/null")
564 SetNonBlock(File
->Fd(),false);
565 if (In
.WriteSpace() == false)
568 while (In
.WriteSpace() == true)
570 if (In
.Write(File
->Fd()) == false)
571 return _error
->Errno("write",_("Error writing to file"));
572 if (In
.IsLimit() == true)
576 if (In
.IsLimit() == true || Encoding
== ServerState::Closes
)
582 // HttpServerState::Go - Run a single loop /*{{{*/
583 // ---------------------------------------------------------------------
584 /* This runs the select loop over the server FDs, Output file FDs and
586 bool HttpServerState::Go(bool ToFile
, FileFd
* const File
)
588 // Server has closed the connection
589 if (ServerFd
== -1 && (In
.WriteSpace() == false ||
597 /* Add the server. We only send more requests if the connection will
599 if (Out
.WriteSpace() == true && ServerFd
!= -1
600 && Persistent
== true)
601 FD_SET(ServerFd
,&wfds
);
602 if (In
.ReadSpace() == true && ServerFd
!= -1)
603 FD_SET(ServerFd
,&rfds
);
610 if (In
.WriteSpace() == true && ToFile
== true && FileFD
!= -1)
611 FD_SET(FileFD
,&wfds
);
614 if (_config
->FindB("Acquire::http::DependOnSTDIN", true) == true)
615 FD_SET(STDIN_FILENO
,&rfds
);
617 // Figure out the max fd
619 if (MaxFd
< ServerFd
)
627 if ((Res
= select(MaxFd
+1,&rfds
,&wfds
,0,&tv
)) < 0)
631 return _error
->Errno("select",_("Select failed"));
636 _error
->Error(_("Connection timed out"));
641 if (ServerFd
!= -1 && FD_ISSET(ServerFd
,&rfds
))
644 if (In
.Read(ServerFd
) == false)
648 if (ServerFd
!= -1 && FD_ISSET(ServerFd
,&wfds
))
651 if (Out
.Write(ServerFd
) == false)
655 // Send data to the file
656 if (FileFD
!= -1 && FD_ISSET(FileFD
,&wfds
))
658 if (In
.Write(FileFD
) == false)
659 return _error
->Errno("write",_("Error writing to output file"));
662 // Handle commands from APT
663 if (FD_ISSET(STDIN_FILENO
,&rfds
))
665 if (Owner
->Run(true) != -1)
673 // HttpMethod::SendReq - Send the HTTP request /*{{{*/
674 // ---------------------------------------------------------------------
675 /* This places the http request in the outbound buffer */
676 void HttpMethod::SendReq(FetchItem
*Itm
)
680 // The HTTP server expects a hostname with a trailing :port
684 if (Uri
.Host
.find(':') != string::npos
)
685 ProperHost
= '[' + Uri
.Host
+ ']';
687 ProperHost
= Uri
.Host
;
690 sprintf(Buf
,":%u",Uri
.Port
);
695 if (Itm
->Uri
.length() >= sizeof(Buf
))
698 /* RFC 2616 ยง5.1.2 requires absolute URIs for requests to proxies,
699 but while its a must for all servers to accept absolute URIs,
700 it is assumed clients will sent an absolute path for non-proxies */
701 std::string requesturi
;
702 if (Server
->Proxy
.empty() == true || Server
->Proxy
.Host
.empty())
703 requesturi
= Uri
.Path
;
705 requesturi
= Itm
->Uri
;
707 // The "+" is encoded as a workaround for a amazon S3 bug
708 // see LP bugs #1003633 and #1086997.
709 requesturi
= QuoteString(requesturi
, "+~ ");
711 /* Build the request. No keep-alive is included as it is the default
712 in 1.1, can cause problems with proxies, and we are an HTTP/1.1
714 C.f. https://tools.ietf.org/wg/httpbis/trac/ticket/158 */
715 sprintf(Buf
,"GET %s HTTP/1.1\r\nHost: %s\r\n",
716 requesturi
.c_str(),ProperHost
.c_str());
718 // generate a cache control header (if needed)
719 if (_config
->FindB("Acquire::http::No-Cache",false) == true)
721 strcat(Buf
,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
725 if (Itm
->IndexFile
== true)
727 sprintf(Buf
+strlen(Buf
),"Cache-Control: max-age=%u\r\n",
728 _config
->FindI("Acquire::http::Max-Age",0));
732 if (_config
->FindB("Acquire::http::No-Store",false) == true)
733 strcat(Buf
,"Cache-Control: no-store\r\n");
737 // If we ask for uncompressed files servers might respond with content-
738 // negotation which lets us end up with compressed files we do not support,
739 // see 657029, 657560 and co, so if we have no extension on the request
740 // ask for text only. As a sidenote: If there is nothing to negotate servers
741 // seem to be nice and ignore it.
742 if (_config
->FindB("Acquire::http::SendAccept", true) == true)
744 size_t const filepos
= Itm
->Uri
.find_last_of('/');
745 string
const file
= Itm
->Uri
.substr(filepos
+ 1);
746 if (flExtension(file
) == file
)
747 strcat(Buf
,"Accept: text/*\r\n");
752 // Check for a partial file
754 if (stat(Itm
->DestFile
.c_str(),&SBuf
) >= 0 && SBuf
.st_size
> 0)
756 // In this case we send an if-range query with a range header
757 sprintf(Buf
,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf
.st_size
,
758 TimeRFC1123(SBuf
.st_mtime
).c_str());
763 if (Itm
->LastModified
!= 0)
765 sprintf(Buf
,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm
->LastModified
).c_str());
770 if (Server
->Proxy
.User
.empty() == false || Server
->Proxy
.Password
.empty() == false)
771 Req
+= string("Proxy-Authorization: Basic ") +
772 Base64Encode(Server
->Proxy
.User
+ ":" + Server
->Proxy
.Password
) + "\r\n";
774 maybe_add_auth (Uri
, _config
->FindFile("Dir::Etc::netrc"));
775 if (Uri
.User
.empty() == false || Uri
.Password
.empty() == false)
777 Req
+= string("Authorization: Basic ") +
778 Base64Encode(Uri
.User
+ ":" + Uri
.Password
) + "\r\n";
780 Req
+= "User-Agent: " + _config
->Find("Acquire::http::User-Agent",
781 "Debian APT-HTTP/1.3 (" PACKAGE_VERSION
")") + "\r\n\r\n";
786 Server
->WriteResponse(Req
);
789 // HttpMethod::Configuration - Handle a configuration message /*{{{*/
790 // ---------------------------------------------------------------------
791 /* We stash the desired pipeline depth */
792 bool HttpMethod::Configuration(string Message
)
794 if (ServerMethod::Configuration(Message
) == false)
797 AllowRedirect
= _config
->FindB("Acquire::http::AllowRedirect",true);
798 PipelineDepth
= _config
->FindI("Acquire::http::Pipeline-Depth",
800 Debug
= _config
->FindB("Debug::Acquire::http",false);
802 // Get the proxy to use
808 // HttpMethod::AutoDetectProxy - auto detect proxy /*{{{*/
809 // ---------------------------------------------------------------------
811 bool HttpMethod::AutoDetectProxy()
813 // option is "Acquire::http::Proxy-Auto-Detect" but we allow the old
814 // name without the dash ("-")
815 AutoDetectProxyCmd
= _config
->Find("Acquire::http::Proxy-Auto-Detect",
816 _config
->Find("Acquire::http::ProxyAutoDetect"));
818 if (AutoDetectProxyCmd
.empty())
822 clog
<< "Using auto proxy detect command: " << AutoDetectProxyCmd
<< endl
;
824 int Pipes
[2] = {-1,-1};
825 if (pipe(Pipes
) != 0)
826 return _error
->Errno("pipe", "Failed to create Pipe");
828 pid_t Process
= ExecFork();
832 dup2(Pipes
[1],STDOUT_FILENO
);
833 SetCloseExec(STDOUT_FILENO
,false);
836 Args
[0] = AutoDetectProxyCmd
.c_str();
838 execv(Args
[0],(char **)Args
);
839 cerr
<< "Failed to exec method " << Args
[0] << endl
;
845 int res
= read(InFd
, buf
, sizeof(buf
)-1);
846 ExecWait(Process
, "ProxyAutoDetect", true);
849 return _error
->Errno("read", "Failed to read");
851 return _error
->Warning("ProxyAutoDetect returned no data");
857 clog
<< "auto detect command returned: '" << buf
<< "'" << endl
;
859 if (strstr(buf
, "http://") == buf
)
860 _config
->Set("Acquire::http::proxy", _strstrip(buf
));
865 ServerState
* HttpMethod::CreateServerState(URI uri
) /*{{{*/
867 return new HttpServerState(uri
, this);
870 void HttpMethod::RotateDNS() /*{{{*/