X-Git-Url: https://git.saurik.com/apt.git/blobdiff_plain/6db194289ece36e62cb8dab0aa178209b46c59f2..2737f28a1cb2d03c66d2a7edd04215566903dbf1:/apt-pkg/tagfile.cc diff --git a/apt-pkg/tagfile.cc b/apt-pkg/tagfile.cc index a8f04b23a..7085e7d69 100644 --- a/apt-pkg/tagfile.cc +++ b/apt-pkg/tagfile.cc @@ -11,15 +11,20 @@ ##################################################################### */ /*}}}*/ // Include Files /*{{{*/ +#include + #include #include #include +#include -#include - #include #include #include +#include +#include + +#include /*}}}*/ using std::string; @@ -27,7 +32,10 @@ using std::string; class pkgTagFilePrivate { public: - pkgTagFilePrivate(FileFd *pFd, unsigned long Size) : Fd(*pFd), Size(Size) + pkgTagFilePrivate(FileFd *pFd, unsigned long long Size) : Fd(*pFd), Buffer(NULL), + Start(NULL), End(NULL), + Done(false), iOffset(0), + Size(Size) { } FileFd &Fd; @@ -35,30 +43,63 @@ public: char *Start; char *End; bool Done; - unsigned long iOffset; - unsigned long Size; + unsigned long long iOffset; + unsigned long long Size; }; +static unsigned long AlphaHash(const char *Text, size_t Length) /*{{{*/ +{ + /* This very simple hash function for the last 8 letters gives + very good performance on the debian package files */ + if (Length > 8) + { + Text += (Length - 8); + Length = 8; + } + unsigned long Res = 0; + for (size_t i = 0; i < Length; ++i) + Res = ((unsigned long)(Text[i]) & 0xDF) ^ (Res << 1); + return Res & 0xFF; +} + /*}}}*/ + // TagFile::pkgTagFile - Constructor /*{{{*/ // --------------------------------------------------------------------- /* */ -pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long Size) +pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long long Size) + : d(NULL) +{ + Init(pFd, Size); +} + +void pkgTagFile::Init(FileFd *pFd,unsigned long long Size) { + /* 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; + if(d != NULL) + { + free(d->Buffer); + delete d; + } d = new pkgTagFilePrivate(pFd, Size); if (d->Fd.IsOpen() == false) - { d->Start = d->End = d->Buffer = 0; + else + d->Buffer = (char*)malloc(sizeof(char) * Size); + + if (d->Buffer == NULL) d->Done = true; - d->iOffset = 0; - return; - } - - d->Buffer = new char[Size]; + else + d->Done = false; + d->Start = d->End = d->Buffer; - d->Done = false; d->iOffset = 0; - Fill(); + if (d->Done == false) + Fill(); } /*}}}*/ // TagFile::~pkgTagFile - Destructor /*{{{*/ @@ -66,12 +107,12 @@ pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long Size) /* */ pkgTagFile::~pkgTagFile() { - delete [] d->Buffer; + free(d->Buffer); delete d; } /*}}}*/ -// TagFile::Offset - Return the current offset in the buffer /*{{{*/ -unsigned long pkgTagFile::Offset() +// TagFile::Offset - Return the current offset in the buffer /*{{{*/ +APT_PURE unsigned long pkgTagFile::Offset() { return d->iOffset; } @@ -83,19 +124,22 @@ unsigned long pkgTagFile::Offset() */ bool pkgTagFile::Resize() { - char *tmp; - unsigned long EndSize = d->End - d->Start; - // 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; + // get new buffer and use it - tmp = new char[2*d->Size]; - memcpy(tmp, d->Buffer, d->Size); - d->Size = d->Size*2; - delete [] d->Buffer; - d->Buffer = tmp; + char* newBuffer = (char*)realloc(d->Buffer, sizeof(char) * newSize); + if (newBuffer == NULL) + return false; + d->Buffer = newBuffer; + d->Size = newSize; // update the start/end pointers to the new buffer d->Start = d->Buffer; @@ -111,18 +155,23 @@ bool pkgTagFile::Resize() */ bool pkgTagFile::Step(pkgTagSection &Tag) { - while (Tag.Scan(d->Start,d->End - d->Start) == false) + if(Tag.Scan(d->Start,d->End - d->Start) == false) { - if (Fill() == false) - return false; - - if(Tag.Scan(d->Start,d->End - d->Start)) - break; + do + { + if (Fill() == false) + return false; + + if(Tag.Scan(d->Start,d->End - d->Start, false)) + break; + + if (Resize() == false) + return _error->Error(_("Unable to parse package file %s (1)"), + d->Fd.Name().c_str()); - if (Resize() == false) - return _error->Error(_("Unable to parse package file %s (1)"), - d->Fd.Name().c_str()); + } while (Tag.Scan(d->Start,d->End - d->Start, false) == false); } + d->Start += Tag.size(); d->iOffset += Tag.size(); @@ -136,8 +185,8 @@ bool pkgTagFile::Step(pkgTagSection &Tag) then fills the rest from the file */ bool pkgTagFile::Fill() { - unsigned long EndSize = d->End - d->Start; - unsigned long Actual = 0; + unsigned long long EndSize = d->End - d->Start; + unsigned long long Actual = 0; memmove(d->Buffer,d->Start,EndSize); d->Start = d->Buffer; @@ -146,9 +195,10 @@ bool pkgTagFile::Fill() if (d->Done == false) { // See if only a bit of the file is left - if (d->Fd.Read(d->End, d->Size - (d->End - d->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 != d->Size - (d->End - d->Buffer)) + if (Actual != dataSize) d->Done = true; d->End += Actual; } @@ -165,8 +215,13 @@ bool pkgTagFile::Fill() for (const char *E = d->End - 1; E - d->End < 6 && (*E == '\n' || *E == '\r'); E--) if (*E == '\n') LineCount++; - for (; LineCount < 2; LineCount++) - *d->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; } @@ -178,15 +233,19 @@ 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) { // We are within a buffer space of the next hit.. if (Offset >= d->iOffset && d->iOffset + (d->End - d->Start) > Offset) { - unsigned long Dist = Offset - d->iOffset; + unsigned long long Dist = Offset - d->iOffset; d->Start += Dist; d->iOffset += Dist; - return Step(Tag); + // if we have seen the end, don't ask for more + if (d->Done == true) + return Tag.Scan(d->Start, d->End - d->Start); + else + return Step(Tag); } // Reposition and reload.. @@ -206,48 +265,120 @@ bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset) if (Fill() == false) return false; - if (Tag.Scan(d->Start, d->End - d->Start) == false) + if (Tag.Scan(d->Start, d->End - d->Start, false) == 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 /*{{{*/ +// pkgTagSection::pkgTagSection - Constructor /*{{{*/ // --------------------------------------------------------------------- -/* 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) +/* */ +pkgTagSection::pkgTagSection() + : Section(0), d(NULL), Stop(0) +{ + memset(&LookupTable, 0, sizeof(LookupTable)); +} + /*}}}*/ +// TagSection::Scan - Scan for the end of the header information /*{{{*/ +bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength, bool const Restart) { + Section = Start; const char *End = Start + MaxLength; - Stop = Section = Start; - memset(AlphaIndexes,0,sizeof(AlphaIndexes)); + + if (Restart == false && Tags.empty() == false) + { + Stop = Section + Tags.back().StartTag; + if (End <= Stop) + return false; + Stop = (const char *)memchr(Stop,'\n',End - Stop); + if (Stop == NULL) + return false; + ++Stop; + } + else + { + Stop = Section; + if (Tags.empty() == false) + { + memset(&LookupTable, 0, sizeof(LookupTable)); + Tags.clear(); + } + Tags.reserve(0x100); + } + size_t TagCount = Tags.size(); if (Stop == 0) return false; - TagCount = 0; - while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End) + TagData lastTagData(0); + lastTagData.EndTag = 0; + unsigned long lastTagHash = 0; + while (Stop < End) { - TrimRecord(true,End); + TrimRecord(true,End); + + // this can happen when TrimRecord trims away the entire Record + // (e.g. because it just contains comments) + if(Stop == End) + return true; // Start a new index and add it to the hash if (isspace(Stop[0]) == 0) { - Indexes[TagCount++] = Stop - Section; - AlphaIndexes[AlphaHash(Stop,End)] = TagCount; + // store the last found tag + if (lastTagData.EndTag != 0) + { + if (LookupTable[lastTagHash] != 0) + lastTagData.NextInBucket = LookupTable[lastTagHash]; + LookupTable[lastTagHash] = TagCount; + Tags.push_back(lastTagData); + } + + ++TagCount; + lastTagData = TagData(Stop - Section); + // find the colon separating tag and value + char const * Colon = (char const *) memchr(Stop, ':', End - Stop); + if (Colon == NULL) + return false; + // find the end of the tag (which might or might not be the colon) + char const * EndTag = Colon; + --EndTag; + for (; EndTag > Stop && isspace(*EndTag) != 0; --EndTag) + ; + ++EndTag; + lastTagData.EndTag = EndTag - Section; + lastTagHash = AlphaHash(Stop, EndTag - Stop); + // find the beginning of the value + Stop = Colon + 1; + for (; isspace(*Stop) != 0; ++Stop); + if (Stop >= End) + return false; + lastTagData.StartValue = Stop - Section; } Stop = (const char *)memchr(Stop,'\n',End - Stop); - + if (Stop == 0) return false; - for (; Stop+1 < End && Stop[1] == '\r'; Stop++); + for (; Stop+1 < End && Stop[1] == '\r'; Stop++) + /* nothing */ + ; // Double newline marks the end of the record if (Stop+1 < End && Stop[1] == '\n') { - Indexes[TagCount] = Stop - Section; + if (lastTagData.EndTag != 0) + { + if (LookupTable[lastTagHash] != 0) + lastTagData.NextInBucket = LookupTable[lastTagHash]; + LookupTable[lastTagHash] = TagCount; + Tags.push_back(lastTagData); + } + + TagData const td(Stop - Section); + Tags.push_back(td); TrimRecord(false,End); return true; } @@ -276,78 +407,55 @@ 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) const +{ + 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)]; - if (I == 0) + size_t const Length = strlen(Tag); + unsigned int Bucket = LookupTable[AlphaHash(Tag, Length)]; + if (Bucket == 0) return false; - I--; - - for (unsigned int Counter = 0; Counter != TagCount; Counter++, - I = (I+1)%TagCount) + + for (; Bucket != 0; Bucket = Tags[Bucket - 1].NextInBucket) { - const char *St; - St = Section + Indexes[I]; - if (strncasecmp(Tag,St,Length) != 0) + if ((Tags[Bucket - 1].EndTag - Tags[Bucket - 1].StartTag) != Length) continue; - // Make sure the colon is in the right place - const char *C = St + Length; - for (; isspace(*C) != 0; C++); - if (*C != ':') + char const * const St = Section + Tags[Bucket - 1].StartTag; + if (strncasecmp(Tag,St,Length) != 0) continue; - Pos = I; + + Pos = Bucket - 1; return true; } Pos = 0; return false; } - /*}}}*/ -// TagSection::Find - Locate a tag /*{{{*/ -// --------------------------------------------------------------------- -/* 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 { - unsigned int Length = strlen(Tag); - unsigned int I = AlphaIndexes[AlphaHash(Tag)]; - if (I == 0) + unsigned int Pos; + if (Find(Tag, Pos) == false) 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; - // 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; - } - - Start = End = 0; - return false; + Start = Section + Tags[Pos].StartValue; + // Strip off the gunk from the end + End = Section + Tags[Pos + 1].StartTag; + if (unlikely(Start > End)) + return _error->Error("Internal parsing error"); + + for (; isspace(End[-1]) != 0 && End > Start; --End); + + return true; } /*}}}*/ // TagSection::FindS - Find a string /*{{{*/ @@ -410,6 +518,17 @@ unsigned long long pkgTagSection::FindULL(const char *Tag, unsigned long long co return Result; } /*}}}*/ +// TagSection::FindB - Find boolean value /*{{{*/ +// --------------------------------------------------------------------- +/* */ +bool pkgTagSection::FindB(const char *Tag, bool const &Default) const +{ + const char *Start, *Stop; + if (Find(Tag, Start, Stop) == false) + return Default; + return StringToBool(string(Start, Stop)); +} + /*}}}*/ // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/ // --------------------------------------------------------------------- /* The bits marked in Flag are masked on/off in Flags */ @@ -422,7 +541,7 @@ bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags, return true; return FindFlag(Flags, Flag, Start, Stop); } -bool const pkgTagSection::FindFlag(unsigned long &Flags, unsigned long Flag, +bool pkgTagSection::FindFlag(unsigned long &Flags, unsigned long Flag, char const* Start, char const* Stop) { switch (StringToBool(string(Start, Stop))) @@ -442,6 +561,13 @@ bool const pkgTagSection::FindFlag(unsigned long &Flags, unsigned long Flag, return true; } /*}}}*/ +APT_PURE unsigned int pkgTagSection::Count() const { /*{{{*/ + if (Tags.empty() == true) + return 0; + // the last element is just marking the end and isn't a real one + return Tags.size() - 1; +} + /*}}}*/ // TFRewrite - Rewrite a control record /*{{{*/ // --------------------------------------------------------------------- /* This writes the control record to stdout rewriting it as necessary. The @@ -522,52 +648,54 @@ bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[], } // Write all all of the tags, in order. - for (unsigned int I = 0; Order[I] != 0; I++) + if (Order != NULL) { - 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++) + for (unsigned int I = 0; Order[I] != 0; I++) { - 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); - } + 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; + } + } - 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; + // 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); + /* 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"); - } + 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++)