X-Git-Url: https://git.saurik.com/apt.git/blobdiff_plain/f3bcc38396ab3a7ad5c2e299a5f4e036b2cd06b3..0aae6d14390193e25ab6d0fd49295bd7b131954f:/apt-pkg/tagfile.cc diff --git a/apt-pkg/tagfile.cc b/apt-pkg/tagfile.cc index 9a581c3e8..aa0f1eee8 100644 --- a/apt-pkg/tagfile.cc +++ b/apt-pkg/tagfile.cc @@ -1,6 +1,6 @@ // -*- mode: cpp; mode: fold -*- // Description /*{{{*/ -// $Id: tagfile.cc,v 1.22 1999/01/31 21:52:27 jgg 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 @@ -11,46 +11,137 @@ ##################################################################### */ /*}}}*/ // Include Files /*{{{*/ -#ifdef __GNUG__ -#pragma implementation "apt-pkg/tagfile.h" -#endif +#include #include #include #include +#include #include #include +#include + +#include /*}}}*/ +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 &Fd,unsigned long Size) : Fd(Fd), Size(Size) +pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long long Size) { - Buffer = new char[Size]; - Start = End = Buffer; - Left = Fd.Size(); - iOffset = 0; + /* 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) + { + d->Start = d->End = d->Buffer = 0; + d->Done = true; + d->iOffset = 0; + return; + } + + d->Buffer = new char[Size]; + d->Start = d->End = d->Buffer; + d->Done = false; + d->iOffset = 0; Fill(); } /*}}}*/ +// TagFile::~pkgTagFile - Destructor /*{{{*/ +// --------------------------------------------------------------------- +/* */ +pkgTagFile::~pkgTagFile() +{ + 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) { - if (Tag.Scan(Start,End - Start) == false) + while (Tag.Scan(d->Start,d->End - d->Start) == false) { if (Fill() == false) return false; - if (Tag.Scan(Start,End - Start) == false) - return _error->Error("Unable to parse package file %s",Fd.Name().c_str()); - } - Start += Tag.size(); - iOffset += Tag.size(); - + 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()); + } + d->Start += Tag.size(); + d->iOffset += Tag.size(); + + Tag.Trim(); return true; } /*}}}*/ @@ -60,47 +151,47 @@ bool pkgTagFile::Step(pkgTagSection &Tag) then fills the rest from the file */ bool pkgTagFile::Fill() { - unsigned long EndSize = End - Start; + 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 (d->Done == false) + { + // See if only a bit of the file is left + 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 != dataSize || d->Fd.Eof() == true) + d->Done = true; + d->End += Actual; + } - if (Left == 0) + if (d->Done == true) { - if (EndSize <= 3) + 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; } - // See if only a bit of the file is left - if (Left < Size - (End - Buffer)) - { - if (Fd.Read(End,Left) == false) - return false; - - End += Left; - Left = 0; - } - else - { - if (Fd.Read(End,Size - (End - Buffer)) == false) - return false; - - Left -= Size - (End - Buffer); - End = Buffer + Size; - } return true; } /*}}}*/ @@ -108,27 +199,44 @@ bool pkgTagFile::Fill() // --------------------------------------------------------------------- /* 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) { - iOffset = Offset; - Left = Fd.Size() - Offset; - if (Fd.Seek(Offset) == false) + // We are within a buffer space of the next hit.. + if (Offset >= d->iOffset && d->iOffset + (d->End - d->Start) > Offset) + { + unsigned long long Dist = Offset - d->iOffset; + d->Start += Dist; + d->iOffset += Dist; + return Step(Tag); + } + + // Reposition and reload.. + 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; + + 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()); - if (Tag.Scan(Start,End - Start) == false) - return _error->Error("Unable to parse package file"); 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 */ +/* 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; @@ -137,37 +245,94 @@ bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength) if (Stop == 0) return false; - + TagCount = 0; - while (TagCount < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End) + 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) { Indexes[TagCount++] = Stop - Section; - unsigned char A = tolower(Stop[0]) - 'a'; - unsigned char B = tolower(Stop[1]) - 'a'; - unsigned char C = tolower(Stop[3]) - 'a'; - AlphaIndexes[((A + C/3)%26) + 26*((B + C/2)%26)] = TagCount; + AlphaIndexes[AlphaHash(Stop,End)] = TagCount; } Stop = (const char *)memchr(Stop,'\n',End - Stop); if (Stop == 0) return false; - for (; Stop[1] == '\r' && Stop+1 < End; Stop++); + + 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[0] == '\n' || Stop[0] == '\r') && Stop < End; Stop++); + TrimRecord(false,End); return true; } Stop++; } + + 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 /*{{{*/ +// --------------------------------------------------------------------- +/* There should be exactly 1 newline at the end of the buffer, no more. */ +void pkgTagSection::Trim() +{ + 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 int &Pos) const +{ + unsigned int Length = strlen(Tag); + unsigned int I = AlphaIndexes[AlphaHash(Tag)]; + if (I == 0) + return false; + I--; + for (unsigned int Counter = 0; Counter != TagCount; Counter++, + I = (I+1)%TagCount) + { + const char *St; + St = Section + Indexes[I]; + if (strncasecmp(Tag,St,Length) != 0) + continue; + + // Make sure the colon is in the right place + const char *C = St + Length; + for (; isspace(*C) != 0; C++); + if (*C != ':') + continue; + Pos = I; + return true; + } + + Pos = 0; return false; } /*}}}*/ @@ -175,13 +340,10 @@ bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength) // --------------------------------------------------------------------- /* This searches the section for a tag that matches the given string. */ bool pkgTagSection::Find(const char *Tag,const char *&Start, - const char *&End) + const char *&End) const { unsigned int Length = strlen(Tag); - unsigned char A = tolower(Tag[0]) - 'a'; - unsigned char B = tolower(Tag[1]) - 'a'; - unsigned char C = tolower(Tag[3]) - 'a'; - unsigned int I = AlphaIndexes[((A + C/3)%26) + 26*((B + C/2)%26)]; + unsigned int I = AlphaIndexes[AlphaHash(Tag)]; if (I == 0) return false; I--; @@ -193,7 +355,7 @@ bool pkgTagSection::Find(const char *Tag,const char *&Start, St = Section + Indexes[I]; if (strncasecmp(Tag,St,Length) != 0) continue; - + // Make sure the colon is in the right place const char *C = St + Length; for (; isspace(*C) != 0; C++); @@ -203,9 +365,12 @@ bool pkgTagSection::Find(const char *Tag,const char *&Start, // Strip off the gunk from the start end Start = C; End = Section + Indexes[I+1]; + if (Start >= End) + return _error->Error("Internal parsing error"); + for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++); for (; isspace(End[-1]) != 0 && End > Start; End--); - + return true; } @@ -216,7 +381,7 @@ bool pkgTagSection::Find(const char *Tag,const char *&Start, // TagSection::FindS - Find a string /*{{{*/ // --------------------------------------------------------------------- /* */ -string pkgTagSection::FindS(const char *Tag) +string pkgTagSection::FindS(const char *Tag) const { const char *Start; const char *End; @@ -228,7 +393,7 @@ string pkgTagSection::FindS(const char *Tag) // TagSection::FindI - Find an integer /*{{{*/ // --------------------------------------------------------------------- /* */ -signed int pkgTagSection::FindI(const char *Tag,signed long Default) +signed int pkgTagSection::FindI(const char *Tag,signed long Default) const { const char *Start; const char *Stop; @@ -249,19 +414,47 @@ signed int pkgTagSection::FindI(const char *Tag,signed long Default) 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 */ bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags, - unsigned long Flag) + unsigned long Flag) const { const char *Start; 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; @@ -271,11 +464,196 @@ bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags, return true; default: - _error->Warning("Unknown flag value"); + _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str()); return true; } return true; } /*}}}*/ +// TFRewrite - Rewrite a control record /*{{{*/ +// --------------------------------------------------------------------- +/* This writes the control record to stdout rewriting it as necessary. The + override map item specificies the rewriting rules to follow. This also + takes the time to sort the feild list. */ + +/* The order of this list is taken from dpkg source lib/parse.c the fieldinfos + array. */ +static const char *iTFRewritePackageOrder[] = { + "Package", + "Essential", + "Status", + "Priority", + "Section", + "Installed-Size", + "Maintainer", + "Original-Maintainer", + "Architecture", + "Source", + "Version", + "Revision", // Obsolete + "Config-Version", // Obsolete + "Replaces", + "Provides", + "Depends", + "Pre-Depends", + "Recommends", + "Suggests", + "Conflicts", + "Breaks", + "Conffiles", + "Filename", + "Size", + "MD5Sum", + "SHA1", + "SHA256", + "SHA512", + "MSDOS-Filename", // Obsolete + "Description", + 0}; +static const char *iTFRewriteSourceOrder[] = {"Package", + "Source", + "Binary", + "Version", + "Priority", + "Section", + "Maintainer", + "Original-Maintainer", + "Build-Depends", + "Build-Depends-Indep", + "Build-Conflicts", + "Build-Conflicts-Indep", + "Architecture", + "Standards-Version", + "Format", + "Directory", + "Files", + 0}; - +/* Two levels of initialization are used because gcc will set the symbol + size of an array to the length of the array, causing dynamic relinking + errors. Doing this makes the symbol size constant */ +const char **TFRewritePackageOrder = iTFRewritePackageOrder; +const char **TFRewriteSourceOrder = iTFRewriteSourceOrder; + +bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[], + TFRewriteData *Rewrite) +{ + unsigned char Visited[256]; // Bit 1 is Order, Bit 2 is Rewrite + for (unsigned I = 0; I != 256; I++) + Visited[I] = 0; + + // Set new tag up as necessary. + for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) + { + if (Rewrite[J].NewTag == 0) + Rewrite[J].NewTag = Rewrite[J].Tag; + } + + // Write all all of the tags, in order. + for (unsigned int I = 0; Order[I] != 0; I++) + { + bool Rewritten = false; + + // See if this is a field that needs to be rewritten + for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) + { + if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0) + { + Visited[J] |= 2; + if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) + { + if (isspace(Rewrite[J].Rewrite[0])) + fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); + else + fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); + } + + Rewritten = true; + break; + } + } + + // See if it is in the fragment + unsigned Pos; + if (Tags.Find(Order[I],Pos) == false) + continue; + Visited[Pos] |= 1; + + if (Rewritten == true) + continue; + + /* Write out this element, taking a moment to rewrite the tag + in case of changes of case. */ + const char *Start; + const char *Stop; + Tags.Get(Start,Stop,Pos); + + if (fputs(Order[I],Output) < 0) + return _error->Errno("fputs","IO Error to output"); + Start += strlen(Order[I]); + if (fwrite(Start,Stop - Start,1,Output) != 1) + return _error->Errno("fwrite","IO Error to output"); + if (Stop[-1] != '\n') + fprintf(Output,"\n"); + } + + // Now write all the old tags that were missed. + for (unsigned int I = 0; I != Tags.Count(); I++) + { + if ((Visited[I] & 1) == 1) + continue; + + const char *Start; + const char *Stop; + Tags.Get(Start,Stop,I); + const char *End = Start; + for (; End < Stop && *End != ':'; End++); + + // See if this is a field that needs to be rewritten + bool Rewritten = false; + for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) + { + if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0) + { + Visited[J] |= 2; + if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) + { + if (isspace(Rewrite[J].Rewrite[0])) + fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); + else + fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); + } + + Rewritten = true; + break; + } + } + + if (Rewritten == true) + continue; + + // Write out this element + if (fwrite(Start,Stop - Start,1,Output) != 1) + return _error->Errno("fwrite","IO Error to output"); + if (Stop[-1] != '\n') + fprintf(Output,"\n"); + } + + // Now write all the rewrites that were missed + for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) + { + if ((Visited[J] & 2) == 2) + continue; + + if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) + { + if (isspace(Rewrite[J].Rewrite[0])) + fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); + else + fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); + } + } + + return true; +} + /*}}}*/