apt-pkg/tagfile.cc

   1 // -*- mode: cpp; mode: fold -*-
   2 // Description                                                          /*{{{*/
   3 // $Id: tagfile.cc,v 1.17 1998/12/07 07:26:22 jgg Exp $
   4 /* ######################################################################
   5
   6    Fast scanner for RFC-822 type header information
   7
   8    This uses a rotating buffer to load the package information into.
   9    The scanner runs over it and isolates and indexes a single section.
  10
  11    ##################################################################### */
  12                                                                         /*}}}*/
  13 // Include Files                                                        /*{{{*/
  14 #ifdef __GNUG__
  15 #pragma implementation "apt-pkg/tagfile.h"
  16 #endif
  17
  18 #include <apt-pkg/tagfile.h>
  19 #include <apt-pkg/error.h>
  20
  21 #include <string>
  22 #include <stdio.h>
  23                                                                         /*}}}*/
  24
  25 // TagFile::pkgTagFile - Constructor                                    /*{{{*/
  26 // ---------------------------------------------------------------------
  27 /* */
  28 pkgTagFile::pkgTagFile(FileFd &Fd,unsigned long Size) : Fd(Fd), Size(Size)
  29 {
  30    Buffer = new char[Size];
  31    Start = End = Buffer;
  32    Left = Fd.Size();
  33    iOffset = 0;
  34    Fill();
  35 }
  36                                                                         /*}}}*/
  37 // TagFile::Step - Advance to the next section                          /*{{{*/
  38 // ---------------------------------------------------------------------
  39 /* If the Section Scanner fails we refill the buffer and try again. */
  40 bool pkgTagFile::Step(pkgTagSection &Tag)
  41 {
  42    if (Tag.Scan(Start,End - Start) == false)
  43    {
  44       if (Fill() == false)
  45          return false;
  46
  47       if (Tag.Scan(Start,End - Start) == false)
  48          return _error->Error("Unable to parse package file");
  49    }
  50    Start += Tag.size();
  51    iOffset += Tag.size();
  52
  53    return true;
  54 }
  55                                                                         /*}}}*/
  56 // TagFile::Fill - Top up the buffer                                    /*{{{*/
  57 // ---------------------------------------------------------------------
  58 /* This takes the bit at the end of the buffer and puts it at the start
  59    then fills the rest from the file */
  60 bool pkgTagFile::Fill()
  61 {
  62    unsigned long EndSize = End - Start;
  63
  64    if (Left == 0)
  65    {
  66       if (EndSize <= 1)
  67          return false;
  68       return true;
  69    }
  70
  71    memmove(Buffer,Start,EndSize);
  72    Start = Buffer;
  73    End = Buffer + EndSize;
  74
  75    // See if only a bit of the file is left
  76    if (Left < Size - (End - Buffer))
  77    {
  78       if (Fd.Read(End,Left) == false)
  79          return false;
  80
  81       End += Left;
  82       Left = 0;
  83    }
  84    else
  85    {
  86       if (Fd.Read(End,Size - (End - Buffer)) == false)
  87          return false;
  88
  89       Left -= Size - (End - Buffer);
  90       End = Buffer + Size;
  91    }
  92    return true;
  93 }
  94                                                                         /*}}}*/
  95 // TagFile::Jump - Jump to a pre-recorded location in the file          /*{{{*/
  96 // ---------------------------------------------------------------------
  97 /* This jumps to a pre-recorded file location and reads the record
  98    that is there */
  99 bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset)
 100 {
 101    iOffset = Offset;
 102    Left = Fd.Size() - Offset;
 103    if (Fd.Seek(Offset) == false)
 104       return false;
 105    End = Start = Buffer;
 106
 107    if (Fill() == false)
 108       return false;
 109
 110    if (Tag.Scan(Start,End - Start) == false)
 111       return _error->Error("Unable to parse package file");
 112    return true;
 113 }
 114                                                                         /*}}}*/
 115 // TagSection::Scan - Scan for the end of the header information        /*{{{*/
 116 // ---------------------------------------------------------------------
 117 /* This looks for the first double new line in the data stream. It also
 118    indexes the tags in the section. This very simple hash function for the
 119    first 3 letters gives very good performance on the debian package files */
 120 bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
 121 {
 122    const char *End = Start + MaxLength;
 123    Stop = Section = Start;
 124    memset(AlphaIndexes,0,sizeof(AlphaIndexes));
 125
 126    TagCount = 0;
 127    while (TagCount < sizeof(Indexes)/sizeof(Indexes[0]))
 128    {
 129       if (isspace(Stop[0]) == 0)
 130       {
 131          Indexes[TagCount++] = Stop - Section;
 132          unsigned char A = tolower(Stop[0]) - 'a';
 133          unsigned char B = tolower(Stop[1]) - 'a';
 134          unsigned char C = tolower(Stop[3]) - 'a';
 135          AlphaIndexes[((A + C/3)%26) + 26*((B + C/2)%26)] = TagCount;
 136       }
 137
 138       Stop = (const char *)memchr(Stop,'\n',End - Stop);
 139
 140       if (Stop == 0)
 141          return false;
 142       for (; Stop[1] == '\r' && Stop < End; Stop++);
 143
 144       if (Stop[1] == '\n')
 145       {
 146          Indexes[TagCount] = Stop - Section;
 147          for (; (Stop[0] == '\n' || Stop[0] == '\r') && Stop < End; Stop++);
 148          return true;
 149       }
 150
 151       Stop++;
 152    }
 153
 154    return false;
 155 }
 156                                                                         /*}}}*/
 157 // TagSection::Find - Locate a tag                                      /*{{{*/
 158 // ---------------------------------------------------------------------
 159 /* This searches the section for a tag that matches the given string. */
 160 bool pkgTagSection::Find(const char *Tag,const char *&Start,
 161                          const char *&End)
 162 {
 163    unsigned int Length = strlen(Tag);
 164    unsigned char A = tolower(Tag[0]) - 'a';
 165    unsigned char B = tolower(Tag[1]) - 'a';
 166    unsigned char C = tolower(Tag[3]) - 'a';
 167    unsigned int I = AlphaIndexes[((A + C/3)%26) + 26*((B + C/2)%26)];
 168    if (I == 0)
 169       return false;
 170    I--;
 171
 172    for (unsigned int Counter = 0; Counter != TagCount; Counter++,
 173         I = (I+1)%TagCount)
 174    {
 175       const char *St;
 176       St = Section + Indexes[I];
 177       if (strncasecmp(Tag,St,Length) != 0)
 178          continue;
 179
 180       // Make sure the colon is in the right place
 181       const char *C = St + Length;
 182       for (; isspace(*C) != 0; C++);
 183       if (*C != ':')
 184          continue;
 185
 186       // Strip off the gunk from the start end
 187       Start = C;
 188       End = Section + Indexes[I+1];
 189       for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++);
 190       for (; isspace(End[-1]) != 0 && End > Start; End--);
 191
 192       return true;
 193    }
 194
 195    Start = End = 0;
 196    return false;
 197 }
 198                                                                         /*}}}*/
 199 // TagSection::FindS - Find a string                                    /*{{{*/
 200 // ---------------------------------------------------------------------
 201 /* */
 202 string pkgTagSection::FindS(const char *Tag)
 203 {
 204    const char *Start;
 205    const char *End;
 206    if (Find(Tag,Start,End) == false)
 207       return string();
 208    return string(Start,End);
 209 }
 210                                                                         /*}}}*/
 211 // TagSection::FindI - Find an integer                                  /*{{{*/
 212 // ---------------------------------------------------------------------
 213 /* */
 214 unsigned int pkgTagSection::FindI(const char *Tag)
 215 {
 216    const char *Start;
 217    const char *End;
 218    if (Find(Tag,Start,End) == false)
 219       return 0;
 220
 221    return atoi(string(Start,End).c_str());
 222 }
 223                                                                         /*}}}*/
 224