// -*- mode: cpp; mode: fold -*-
// Description /*{{{*/
-// $Id: tagfile.cc,v 1.36 2003/04/27 05:59:14 doogie Exp $
+// $Id: tagfile.cc,v 1.37.2.2 2003/12/31 16:02:30 mdz Exp $
/* ######################################################################
Fast scanner for RFC-822 type header information
##################################################################### */
/*}}}*/
// Include Files /*{{{*/
-#ifdef __GNUG__
-#pragma implementation "apt-pkg/tagfile.h"
-#endif
+#include<config.h>
#include <apt-pkg/tagfile.h>
#include <apt-pkg/error.h>
#include <apt-pkg/strutl.h>
+#include <apt-pkg/fileutl.h>
-#include <apti18n.h>
-
#include <string>
#include <stdio.h>
#include <ctype.h>
+
+#include <apti18n.h>
/*}}}*/
using std::string;
+class pkgTagFilePrivate
+{
+public:
+ pkgTagFilePrivate(FileFd *pFd, unsigned long long Size) : Fd(*pFd), Buffer(NULL),
+ Start(NULL), End(NULL),
+ Done(false), iOffset(0),
+ Size(Size)
+ {
+ }
+ FileFd &Fd;
+ char *Buffer;
+ char *Start;
+ char *End;
+ bool Done;
+ unsigned long long iOffset;
+ unsigned long long Size;
+};
+
// TagFile::pkgTagFile - Constructor /*{{{*/
// ---------------------------------------------------------------------
/* */
-pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long Size) : Fd(*pFd), Size(Size)
+pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long long Size)
{
- if (Fd.IsOpen() == false)
+ /* The size is increased by 4 because if we start with the Size of the
+ filename we need to try to read 1 char more to see an EOF faster, 1
+ char the end-pointer can be on and maybe 2 newlines need to be added
+ to the end of the file -> 4 extra chars */
+ Size += 4;
+ d = new pkgTagFilePrivate(pFd, Size);
+
+ if (d->Fd.IsOpen() == false)
{
- Buffer = 0;
- Start = End = Buffer = 0;
- Done = true;
- iOffset = 0;
+ d->Start = d->End = d->Buffer = 0;
+ d->Done = true;
+ d->iOffset = 0;
return;
}
- CurSize = Size;
- Buffer = new char[Size];
- Start = End = Buffer;
- Done = false;
- iOffset = 0;
+ d->Buffer = new char[Size];
+ d->Start = d->End = d->Buffer;
+ d->Done = false;
+ d->iOffset = 0;
Fill();
}
/*}}}*/
/* */
pkgTagFile::~pkgTagFile()
{
- delete [] Buffer;
+ delete [] d->Buffer;
+ delete d;
+}
+ /*}}}*/
+// TagFile::Offset - Return the current offset in the buffer /*{{{*/
+unsigned long pkgTagFile::Offset()
+{
+ return d->iOffset;
+}
+ /*}}}*/
+// TagFile::Resize - Resize the internal buffer /*{{{*/
+// ---------------------------------------------------------------------
+/* Resize the internal buffer (double it in size). Fail if a maximum size
+ * size is reached.
+ */
+bool pkgTagFile::Resize()
+{
+ // fail is the buffer grows too big
+ if(d->Size > 1024*1024+1)
+ return false;
+
+ return Resize(d->Size * 2);
+}
+bool pkgTagFile::Resize(unsigned long long const newSize)
+{
+ unsigned long long const EndSize = d->End - d->Start;
+ char *tmp;
+
+ // get new buffer and use it
+ tmp = new char[newSize];
+ memcpy(tmp, d->Buffer, d->Size);
+ d->Size = newSize;
+ delete [] d->Buffer;
+ d->Buffer = tmp;
+
+ // update the start/end pointers to the new buffer
+ d->Start = d->Buffer;
+ d->End = d->Start + EndSize;
+ return true;
}
/*}}}*/
// TagFile::Step - Advance to the next section /*{{{*/
// ---------------------------------------------------------------------
-/* If the Section Scanner fails we refill the buffer and try again. */
+/* If the Section Scanner fails we refill the buffer and try again.
+ * If that fails too, double the buffer size and try again until a
+ * maximum buffer is reached.
+ */
bool pkgTagFile::Step(pkgTagSection &Tag)
{
- pkgTagSection::ScanFlags ret = Tag.Scan(Start,End - Start);
- if (ret == pkgTagSection::ScanEOF) {
- CurSize <<= 1;
+ while (Tag.Scan(d->Start,d->End - d->Start) == false)
+ {
if (Fill() == false)
return false;
- do {
- ret = Tag.Scan(Start,End - Start);
- if (ret == pkgTagSection::ScanEOF) {
- CurSize <<= 1;
- if (Fill() == false)
- break;
- }
- } while (ret == pkgTagSection::ScanEOF);
+
+ if(Tag.Scan(d->Start,d->End - d->Start))
+ break;
+
+ if (Resize() == false)
+ return _error->Error(_("Unable to parse package file %s (1)"),
+ d->Fd.Name().c_str());
}
- if (ret != pkgTagSection::ScanSuccess)
- return _error->Error(_("Unable to parse package file %s (1)"),
- Fd.Name().c_str());
- Start += Tag.size();
- iOffset += Tag.size();
+ d->Start += Tag.size();
+ d->iOffset += Tag.size();
Tag.Trim();
return true;
then fills the rest from the file */
bool pkgTagFile::Fill()
{
- unsigned long EndSize = End - Start;
- unsigned long Actual = 0;
+ unsigned long long EndSize = d->End - d->Start;
+ unsigned long long Actual = 0;
- memmove(Buffer,Start,EndSize);
- Start = Buffer;
- End = Buffer + EndSize;
+ memmove(d->Buffer,d->Start,EndSize);
+ d->Start = d->Buffer;
+ d->End = d->Buffer + EndSize;
- if (Done == false)
+ if (d->Done == false)
{
// See if only a bit of the file is left
- if (Fd.Read(End,Size - (End - Buffer),&Actual) == false)
+ unsigned long long const dataSize = d->Size - ((d->End - d->Buffer) + 1);
+ if (d->Fd.Read(d->End, dataSize, &Actual) == false)
return false;
- if (Actual != Size - (End - Buffer))
- Done = true;
- End += Actual;
+ if (Actual != dataSize || d->Fd.Eof() == true)
+ d->Done = true;
+ d->End += Actual;
}
- if (Done == true)
+ if (d->Done == true)
{
if (EndSize <= 3 && Actual == 0)
return false;
- if (Size - (End - Buffer) < 4)
+ if (d->Size - (d->End - d->Buffer) < 4)
return true;
// Append a double new line if one does not exist
unsigned int LineCount = 0;
- for (const char *E = End - 1; E - End < 6 && (*E == '\n' || *E == '\r'); E--)
+ for (const char *E = d->End - 1; E - d->End < 6 && (*E == '\n' || *E == '\r'); E--)
if (*E == '\n')
LineCount++;
- for (; LineCount < 2; LineCount++)
- *End++ = '\n';
+ if (LineCount < 2)
+ {
+ if ((unsigned)(d->End - d->Buffer) >= d->Size)
+ Resize(d->Size + 3);
+ for (; LineCount < 2; LineCount++)
+ *d->End++ = '\n';
+ }
return true;
}
// ---------------------------------------------------------------------
/* This jumps to a pre-recorded file location and reads the record
that is there */
-bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset)
+bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long long Offset)
{
// We are within a buffer space of the next hit..
- if (Offset >= iOffset && iOffset + (End - Start) > Offset)
+ if (Offset >= d->iOffset && d->iOffset + (d->End - d->Start) > Offset)
{
- unsigned long Dist = Offset - iOffset;
- Start += Dist;
- iOffset += Dist;
+ unsigned long long Dist = Offset - d->iOffset;
+ d->Start += Dist;
+ d->iOffset += Dist;
return Step(Tag);
}
// Reposition and reload..
- iOffset = Offset;
- Done = false;
- if (Fd.Seek(Offset) == false)
+ d->iOffset = Offset;
+ d->Done = false;
+ if (d->Fd.Seek(Offset) == false)
return false;
- End = Start = Buffer;
+ d->End = d->Start = d->Buffer;
if (Fill() == false)
return false;
- pkgTagSection::ScanFlags ret = Tag.Scan(Start,End - Start);
- if (ret == pkgTagSection::ScanEOF) {
- CurSize <<= 1;
- if (Fill() == false)
- return false;
- do {
- ret = Tag.Scan(Start,End - Start);
- if (ret == pkgTagSection::ScanEOF) {
- CurSize <<= 1;
- if (Fill() == false)
- break;
- }
- } while (ret == pkgTagSection::ScanEOF);
- }
- if (ret != pkgTagSection::ScanSuccess)
- return _error->Error(_("Unable to parse package file %s (2)"),Fd.Name().c_str());
+ if (Tag.Scan(d->Start, d->End - d->Start) == true)
+ return true;
+
+ // This appends a double new line (for the real eof handling)
+ if (Fill() == false)
+ return false;
+
+ if (Tag.Scan(d->Start, d->End - d->Start) == false)
+ return _error->Error(_("Unable to parse package file %s (2)"),d->Fd.Name().c_str());
return true;
}
/*}}}*/
// TagSection::Scan - Scan for the end of the header information /*{{{*/
// ---------------------------------------------------------------------
-/* This looks for the first double new line in the data stream. It also
- indexes the tags in the section. This very simple hash function for the
- first 3 letters gives very good performance on the debian package files */
-inline static unsigned long AlphaHash(const char *Text, const char *End = 0)
-{
- unsigned long Res = 0;
- for (; Text != End && *Text != ':' && *Text != 0; Text++)
- Res = (unsigned long)(*Text) ^ (Res << 2);
- return Res & 0xFF;
-}
-
-enum pkgTagSection::ScanFlags pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
+/* This looks for the first double new line in the data stream.
+ It also indexes the tags in the section. */
+bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
{
const char *End = Start + MaxLength;
Stop = Section = Start;
memset(AlphaIndexes,0,sizeof(AlphaIndexes));
if (Stop == 0)
- return ScanError;
-
+ return false;
+
TagCount = 0;
while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End)
{
+ TrimRecord(true,End);
+
// Start a new index and add it to the hash
if (isspace(Stop[0]) == 0)
{
Stop = (const char *)memchr(Stop,'\n',End - Stop);
if (Stop == 0)
- return ScanEOF;
-
+ return false;
+
for (; Stop+1 < End && Stop[1] == '\r'; Stop++);
// Double newline marks the end of the record
if (Stop+1 < End && Stop[1] == '\n')
{
Indexes[TagCount] = Stop - Section;
- for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
- return ScanSuccess;
+ TrimRecord(false,End);
+ return true;
}
Stop++;
}
- return ScanEOF;
+ return false;
+}
+ /*}}}*/
+// TagSection::TrimRecord - Trim off any garbage before/after a record /*{{{*/
+// ---------------------------------------------------------------------
+/* There should be exactly 2 newline at the end of the record, no more. */
+void pkgTagSection::TrimRecord(bool BeforeRecord, const char*& End)
+{
+ if (BeforeRecord == true)
+ return;
+ for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
}
/*}}}*/
// TagSection::Trim - Trim off any trailing garbage /*{{{*/
for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--);
}
/*}}}*/
+// TagSection::Exists - return True if a tag exists /*{{{*/
+bool pkgTagSection::Exists(const char* const Tag)
+{
+ unsigned int tmp;
+ return Find(Tag, tmp);
+}
+ /*}}}*/
// TagSection::Find - Locate a tag /*{{{*/
// ---------------------------------------------------------------------
/* This searches the section for a tag that matches the given string. */
-bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const
+bool pkgTagSection::Find(const char *Tag,unsigned int &Pos) const
{
unsigned int Length = strlen(Tag);
unsigned int I = AlphaIndexes[AlphaHash(Tag)];
return Result;
}
/*}}}*/
+// TagSection::FindULL - Find an unsigned long long integer /*{{{*/
+// ---------------------------------------------------------------------
+/* */
+unsigned long long pkgTagSection::FindULL(const char *Tag, unsigned long long const &Default) const
+{
+ const char *Start;
+ const char *Stop;
+ if (Find(Tag,Start,Stop) == false)
+ return Default;
+
+ // Copy it into a temp buffer so we can use strtoull
+ char S[100];
+ if ((unsigned)(Stop - Start) >= sizeof(S))
+ return Default;
+ strncpy(S,Start,Stop-Start);
+ S[Stop - Start] = 0;
+
+ char *End;
+ unsigned long long Result = strtoull(S,&End,10);
+ if (S == End)
+ return Default;
+ return Result;
+}
+ /*}}}*/
// TagSection::FindFlag - Locate a yes/no type flag /*{{{*/
// ---------------------------------------------------------------------
/* The bits marked in Flag are masked on/off in Flags */
const char *Stop;
if (Find(Tag,Start,Stop) == false)
return true;
-
- switch (StringToBool(string(Start,Stop)))
- {
+ return FindFlag(Flags, Flag, Start, Stop);
+}
+bool const pkgTagSection::FindFlag(unsigned long &Flags, unsigned long Flag,
+ char const* Start, char const* Stop)
+{
+ switch (StringToBool(string(Start, Stop)))
+ {
case 0:
Flags &= ~Flag;
return true;
return true;
}
/*}}}*/
-
// TFRewrite - Rewrite a control record /*{{{*/
// ---------------------------------------------------------------------
/* This writes the control record to stdout rewriting it as necessary. The
"Section",
"Installed-Size",
"Maintainer",
+ "Original-Maintainer",
"Architecture",
"Source",
"Version",
"Recommends",
"Suggests",
"Conflicts",
+ "Breaks",
"Conffiles",
"Filename",
"Size",
"MD5Sum",
- "SHA1Sum",
+ "SHA1",
+ "SHA256",
+ "SHA512",
"MSDOS-Filename", // Obsolete
"Description",
0};
"Priority",
"Section",
"Maintainer",
+ "Original-Maintainer",
"Build-Depends",
"Build-Depends-Indep",
"Build-Conflicts",