bigm.h

   1 /*-------------------------------------------------------------------------
   2  *
   3  * Portions Copyright (c) 2017-2020, pg_bigm Development Group
   4  * Portions Copyright (c) 2013-2016, NTT DATA Corporation
   5  * Portions Copyright (c) 2004-2012, PostgreSQL Global Development Group
   6  *
   7  * Changelog:
   8  *       2013/01/09
   9  *       Support full text search using bigrams.
  10  *       Author: NTT DATA Corporation
  11  *
  12  *-------------------------------------------------------------------------
  13  */
  14 #ifndef __BIGM_H__
  15 #define __BIGM_H__
  16
  17 #include "access/itup.h"
  18 #include "storage/bufpage.h"
  19 #include "utils/builtins.h"
  20
  21 /* GUC variable */
  22 extern bool bigm_enable_recheck;
  23 extern int      bigm_gin_key_limit;
  24 extern double bigm_similarity_limit;
  25 extern char     *bigm_last_update;
  26
  27 /* options */
  28 #define LPADDING                1
  29 #define RPADDING                1
  30
  31 /* operator strategy numbers */
  32 #define LikeStrategyNumber                      1
  33 #define SimilarityStrategyNumber        2
  34
  35 typedef struct
  36 {
  37         bool            pmatch;                 /* partial match is required? */
  38         int8            bytelen;                /* byte length of bi-gram string */
  39
  40         /*
  41          * Bi-gram string; we assume here that the maximum bytes for a character
  42          * are four.
  43          */
  44         char            str[8];
  45 }       bigm;
  46
  47 #define BIGMSIZE        sizeof(bigm)
  48
  49 static inline int
  50 bigmstrcmp(char *arg1, int len1, char *arg2, int len2)
  51 {
  52         int                     i;
  53         int                     len = Min(len1, len2);
  54
  55         for (i = 0; i < len; i++, arg1++, arg2++)
  56         {
  57                 if (*arg1 == *arg2)
  58                         continue;
  59                 if (*arg1 < *arg2)
  60                         return -1;
  61                 else
  62                         return 1;
  63         }
  64
  65         return (len1 == len2) ? 0 : ((len1 < len2) ? -1 : 1);
  66 }
  67
  68 #define CMPBIGM(a,b) ( bigmstrcmp(((bigm *)a)->str, ((bigm *)a)->bytelen, ((bigm *)b)->str, ((bigm *)b)->bytelen) )
  69
  70 #define CPBIGM(bptr, s, len) do {               \
  71         Assert(len <= 8);                               \
  72         memcpy(bptr->str, s, len);              \
  73         bptr->bytelen = len;                    \
  74         bptr->pmatch = false;                   \
  75 } while(0);
  76
  77 #define ISESCAPECHAR(x) (*(x) == '\\')  /* Wildcard escape character */
  78 #define ISWILDCARDCHAR(x) (*(x) == '%' || *(x) == '_')  /* Wildcard
  79                                                                                                                  * meta-character */
  80 typedef struct
  81 {
  82         int32           vl_len_;                /* varlena header (do not touch directly!) */
  83         char            data[1];
  84 }       BIGM;
  85
  86 #define CALCGTSIZE(len) (VARHDRSZ + len * sizeof(bigm))
  87 #define GETARR(x)               ( (bigm *)( (char*)x + VARHDRSZ ) )
  88 #define ARRNELEM(x) ( ( VARSIZE(x) - VARHDRSZ )/sizeof(bigm) )
  89
  90 extern BIGM *generate_bigm(char *str, int slen);
  91 extern BIGM *generate_wildcard_bigm(const char *str, int slen, bool *removeDups);
  92
  93 #endif   /* __BIGM_H__ */