lib/index.h

   1 /*
   2  *      Sherlock: Data structures used in indices
   3  *
   4  *      (c) 2001 Martin Mares <mj@ucw.cz>
   5  */
   6
   7 /* Words */
   8
   9 #define MAX_WORD_LEN            64
  10
  11 /* Word and string types are defined in lib/custom.h */
  12
  13 /* Index card attributes */
  14
  15 struct card_attr {
  16   u32 card;                             /* Reference to card description (either oid or filepos) */
  17   u32 site_id;
  18   byte weight;
  19   byte flags;
  20   byte rfu[2];
  21 };
  22
  23 enum card_flag {
  24   CARD_FLAG_EMPTY = 1,                  /* Empty document (redirect, robot file etc.) [scanner] */
  25   CARD_FLAG_ACCENTED = 2,               /* Document contains accented characters [scanner] */
  26   CARD_FLAG_DUP = 4,                    /* Removed as a duplicate [merger] */
  27   CARD_FLAG_MERGED = 8,                 /* Destination of a merge [merger] */
  28 };
  29
  30 #define CARD_POS_SHIFT 5                /* Card positions are shifted this # of bytes to the right */
  31
  32 /* String fingerprints */
  33
  34 struct fingerprint {
  35   byte hash[12];
  36 };
  37
  38 void fingerprint(byte *string, struct fingerprint *fp);
  39
  40 static inline u32
  41 fp_hash(struct fingerprint *fp)
  42 {
  43   return (fp->hash[0] << 24) | (fp->hash[1] << 16) | (fp->hash[2] << 8) | fp->hash[3];
  44 }
  45
  46 /* Reading of tagged text (Unicode values, tags mapped to 0x80000000 and higher) */
  47
  48 #define GET_TAGGED_CHAR(p,u) do {                               \
  49   u = *p;                                                       \
  50   if (u >= 0xc0)                                                \
  51     GET_UTF8_CHAR(p,u);                                         \
  52   else if (u >= 0x80)                                           \
  53     {                                                           \
  54       p++;                                                      \
  55       if (u >= 0xb0)                                            \
  56         {                                                       \
  57           if (u != 0xb0)                                        \
  58             ASSERT(0);                                          \
  59           u += 0x80020000;                                      \
  60         }                                                       \
  61       else if (u >= 0xa0)                                       \
  62         {                                                       \
  63           ASSERT(*p >= 0x80 && *p <= 0xbf);                     \
  64           u = 0x80010000 + ((u & 0x0f) << 6) + (*p++ & 0x3f);   \
  65         }                                                       \
  66       else                                                      \
  67         u += 0x80000000;                                        \
  68     }                                                           \
  69   else                                                          \
  70     p++;                                                        \
  71 } while (0)