PostgreSQL Source Code  git master
trgm.h
Go to the documentation of this file.
1 /*
2  * contrib/pg_trgm/trgm.h
3  */
4 #ifndef __TRGM_H__
5 #define __TRGM_H__
6 
7 #include "access/gist.h"
8 #include "access/itup.h"
9 #include "access/stratnum.h"
10 #include "storage/bufpage.h"
11 
12 /*
13  * Options ... but note that trgm_regexp.c effectively assumes these values
14  * of LPADDING and RPADDING.
15  */
16 #define LPADDING 2
17 #define RPADDING 1
18 #define KEEPONLYALNUM
19 /*
20  * Caution: IGNORECASE macro means that trigrams are case-insensitive.
21  * If this macro is disabled, the ~* and ~~* operators must be removed from
22  * the operator classes, because we can't handle case-insensitive wildcard
23  * search with case-sensitive trigrams. Failure to do this will result in
24  * "cannot handle ~*(~~*) with case-sensitive trigrams" errors.
25  */
26 #define IGNORECASE
27 #define DIVUNION
28 
29 /* operator strategy numbers */
30 #define SimilarityStrategyNumber 1
31 #define DistanceStrategyNumber 2
32 #define LikeStrategyNumber 3
33 #define ILikeStrategyNumber 4
34 #define RegExpStrategyNumber 5
35 #define RegExpICaseStrategyNumber 6
36 #define WordSimilarityStrategyNumber 7
37 #define WordDistanceStrategyNumber 8
38 #define StrictWordSimilarityStrategyNumber 9
39 #define StrictWordDistanceStrategyNumber 10
40 
41 typedef char trgm[3];
42 
43 #define CMPCHAR(a,b) ( ((a)==(b)) ? 0 : ( ((a)<(b)) ? -1 : 1 ) )
44 #define CMPPCHAR(a,b,i) CMPCHAR( *(((const char*)(a))+i), *(((const char*)(b))+i) )
45 #define CMPTRGM(a,b) ( CMPPCHAR(a,b,0) ? CMPPCHAR(a,b,0) : ( CMPPCHAR(a,b,1) ? CMPPCHAR(a,b,1) : CMPPCHAR(a,b,2) ) )
46 
47 #define CPTRGM(a,b) do { \
48  *(((char*)(a))+0) = *(((char*)(b))+0); \
49  *(((char*)(a))+1) = *(((char*)(b))+1); \
50  *(((char*)(a))+2) = *(((char*)(b))+2); \
51 } while(0);
52 
53 #ifdef KEEPONLYALNUM
54 #define ISWORDCHR(c) (t_isalpha(c) || t_isdigit(c))
55 #define ISPRINTABLECHAR(a) ( isascii( *(unsigned char*)(a) ) && (isalnum( *(unsigned char*)(a) ) || *(unsigned char*)(a)==' ') )
56 #else
57 #define ISWORDCHR(c) (!t_isspace(c))
58 #define ISPRINTABLECHAR(a) ( isascii( *(unsigned char*)(a) ) && isprint( *(unsigned char*)(a) ) )
59 #endif
60 #define ISPRINTABLETRGM(t) ( ISPRINTABLECHAR( ((char*)(t)) ) && ISPRINTABLECHAR( ((char*)(t))+1 ) && ISPRINTABLECHAR( ((char*)(t))+2 ) )
61 
62 #define ISESCAPECHAR(x) (*(x) == '\\') /* Wildcard escape character */
63 #define ISWILDCARDCHAR(x) (*(x) == '_' || *(x) == '%') /* Wildcard
64  * meta-character */
65 
66 typedef struct
67 {
68  int32 vl_len_; /* varlena header (do not touch directly!) */
70  char data[FLEXIBLE_ARRAY_MEMBER];
71 } TRGM;
72 
73 #define TRGMHDRSIZE (VARHDRSZ + sizeof(uint8))
74 
75 /* gist */
76 #define BITBYTE 8
77 #define SIGLENINT 3 /* >122 => key will toast, so very slow!!! */
78 #define SIGLEN ( sizeof(int)*SIGLENINT )
79 
80 #define SIGLENBIT (SIGLEN*BITBYTE - 1) /* see makesign */
81 
82 typedef char BITVEC[SIGLEN];
83 typedef char *BITVECP;
84 
85 #define LOOPBYTE \
86  for(i=0;i<SIGLEN;i++)
87 
88 #define GETBYTE(x,i) ( *( (BITVECP)(x) + (int)( (i) / BITBYTE ) ) )
89 #define GETBITBYTE(x,i) ( (((char)(x)) >> (i)) & 0x01 )
90 #define CLRBIT(x,i) GETBYTE(x,i) &= ~( 0x01 << ( (i) % BITBYTE ) )
91 #define SETBIT(x,i) GETBYTE(x,i) |= ( 0x01 << ( (i) % BITBYTE ) )
92 #define GETBIT(x,i) ( (GETBYTE(x,i) >> ( (i) % BITBYTE )) & 0x01 )
93 
94 #define HASHVAL(val) (((unsigned int)(val)) % SIGLENBIT)
95 #define HASH(sign, val) SETBIT((sign), HASHVAL(val))
96 
97 #define ARRKEY 0x01
98 #define SIGNKEY 0x02
99 #define ALLISTRUE 0x04
100 
101 #define ISARRKEY(x) ( ((TRGM*)x)->flag & ARRKEY )
102 #define ISSIGNKEY(x) ( ((TRGM*)x)->flag & SIGNKEY )
103 #define ISALLTRUE(x) ( ((TRGM*)x)->flag & ALLISTRUE )
104 
105 #define CALCGTSIZE(flag, len) ( TRGMHDRSIZE + ( ( (flag) & ARRKEY ) ? ((len)*sizeof(trgm)) : (((flag) & ALLISTRUE) ? 0 : SIGLEN) ) )
106 #define GETSIGN(x) ( (BITVECP)( (char*)x+TRGMHDRSIZE ) )
107 #define GETARR(x) ( (trgm*)( (char*)x+TRGMHDRSIZE ) )
108 #define ARRNELEM(x) ( ( VARSIZE(x) - TRGMHDRSIZE )/sizeof(trgm) )
109 
110 /*
111  * If DIVUNION is defined then similarity formula is:
112  * count / (len1 + len2 - count)
113  * else if DIVUNION is not defined then similarity formula is:
114  * count / max(len1, len2)
115  */
116 #ifdef DIVUNION
117 #define CALCSML(count, len1, len2) ((float4) (count)) / ((float4) ((len1) + (len2) - (count)))
118 #else
119 #define CALCSML(count, len1, len2) ((float4) (count)) / ((float4) (((len1) > (len2)) ? (len1) : (len2)))
120 #endif
121 
123 
124 extern double similarity_threshold;
125 extern double word_similarity_threshold;
127 
128 extern double index_strategy_get_limit(StrategyNumber strategy);
129 extern uint32 trgm2int(trgm *ptr);
130 extern void compact_trigram(trgm *tptr, char *str, int bytelen);
131 extern TRGM *generate_trgm(char *str, int slen);
132 extern TRGM *generate_wildcard_trgm(const char *str, int slen);
133 extern float4 cnt_sml(TRGM *trg1, TRGM *trg2, bool inexact);
134 extern bool trgm_contained_by(TRGM *trg1, TRGM *trg2);
135 extern bool *trgm_presence_map(TRGM *query, TRGM *key);
136 extern TRGM *createTrgmNFA(text *text_re, Oid collation,
137  TrgmPackedGraph **graph, MemoryContext rcontext);
138 extern bool trigramsMatchGraph(TrgmPackedGraph *graph, bool *check);
139 
140 #endif /* __TRGM_H__ */
bool trigramsMatchGraph(TrgmPackedGraph *graph, bool *check)
Definition: trgm_regexp.c:641
#define SIGLEN
Definition: trgm.h:78
char * BITVECP
Definition: trgm.h:83
TRGM * generate_wildcard_trgm(const char *str, int slen)
Definition: trgm_op.c:886
TRGM * createTrgmNFA(text *text_re, Oid collation, TrgmPackedGraph **graph, MemoryContext rcontext)
Definition: trgm_regexp.c:523
double similarity_threshold
Definition: trgm_op.c:19
unsigned char uint8
Definition: c.h:323
uint16 StrategyNumber
Definition: stratnum.h:22
float4 cnt_sml(TRGM *trg1, TRGM *trg2, bool inexact)
Definition: trgm_op.c:1018
char BITVEC[SIGLEN]
Definition: trgm.h:82
double word_similarity_threshold
Definition: trgm_op.c:20
bool * trgm_presence_map(TRGM *query, TRGM *key)
Definition: trgm_op.c:1105
int32 vl_len_
Definition: trgm.h:68
unsigned int Oid
Definition: postgres_ext.h:31
signed int int32
Definition: c.h:313
Definition: trgm.h:66
unsigned int uint32
Definition: c.h:325
double index_strategy_get_limit(StrategyNumber strategy)
Definition: trgm_op.c:132
bool trgm_contained_by(TRGM *trg1, TRGM *trg2)
Definition: trgm_op.c:1066
uint32 trgm2int(trgm *ptr)
Definition: trgm_op.c:955
float float4
Definition: c.h:457
uint8 flag
Definition: trgm.h:69
double strict_word_similarity_threshold
Definition: trgm_op.c:21
void compact_trigram(trgm *tptr, char *str, int bytelen)
Definition: trgm_op.c:218
char trgm[3]
Definition: trgm.h:41
TRGM * generate_trgm(char *str, int slen)
Definition: trgm_op.c:376
Definition: c.h:516