PostgreSQL Source Code  git master
Data Structures | Macros | Typedefs | Functions | Variables
trgm.h File Reference
#include "access/gist.h"
#include "access/itup.h"
#include "access/stratnum.h"
#include "storage/bufpage.h"
Include dependency graph for trgm.h:
This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Data Structures

struct  TRGM
 

Macros

#define LPADDING   2
 
#define RPADDING   1
 
#define KEEPONLYALNUM
 
#define IGNORECASE
 
#define DIVUNION
 
#define SimilarityStrategyNumber   1
 
#define DistanceStrategyNumber   2
 
#define LikeStrategyNumber   3
 
#define ILikeStrategyNumber   4
 
#define RegExpStrategyNumber   5
 
#define RegExpICaseStrategyNumber   6
 
#define WordSimilarityStrategyNumber   7
 
#define WordDistanceStrategyNumber   8
 
#define StrictWordSimilarityStrategyNumber   9
 
#define StrictWordDistanceStrategyNumber   10
 
#define CMPCHAR(a, b)   ( ((a)==(b)) ? 0 : ( ((a)<(b)) ? -1 : 1 ) )
 
#define CMPPCHAR(a, b, i)   CMPCHAR( *(((const char*)(a))+i), *(((const char*)(b))+i) )
 
#define CMPTRGM(a, b)   ( CMPPCHAR(a,b,0) ? CMPPCHAR(a,b,0) : ( CMPPCHAR(a,b,1) ? CMPPCHAR(a,b,1) : CMPPCHAR(a,b,2) ) )
 
#define CPTRGM(a, b)
 
#define ISWORDCHR(c)   (t_isalpha(c) || t_isdigit(c))
 
#define ISPRINTABLECHAR(a)   ( isascii( *(unsigned char*)(a) ) && (isalnum( *(unsigned char*)(a) ) || *(unsigned char*)(a)==' ') )
 
#define ISPRINTABLETRGM(t)   ( ISPRINTABLECHAR( ((char*)(t)) ) && ISPRINTABLECHAR( ((char*)(t))+1 ) && ISPRINTABLECHAR( ((char*)(t))+2 ) )
 
#define ISESCAPECHAR(x)   (*(x) == '\\') /* Wildcard escape character */
 
#define ISWILDCARDCHAR(x)
 
#define TRGMHDRSIZE   (VARHDRSZ + sizeof(uint8))
 
#define BITBYTE   8
 
#define SIGLENINT   3 /* >122 => key will toast, so very slow!!! */
 
#define SIGLEN   ( sizeof(int)*SIGLENINT )
 
#define SIGLENBIT   (SIGLEN*BITBYTE - 1) /* see makesign */
 
#define LOOPBYTE   for(i=0;i<SIGLEN;i++)
 
#define GETBYTE(x, i)   ( *( (BITVECP)(x) + (int)( (i) / BITBYTE ) ) )
 
#define GETBITBYTE(x, i)   ( (((char)(x)) >> (i)) & 0x01 )
 
#define CLRBIT(x, i)   GETBYTE(x,i) &= ~( 0x01 << ( (i) % BITBYTE ) )
 
#define SETBIT(x, i)   GETBYTE(x,i) |= ( 0x01 << ( (i) % BITBYTE ) )
 
#define GETBIT(x, i)   ( (GETBYTE(x,i) >> ( (i) % BITBYTE )) & 0x01 )
 
#define HASHVAL(val)   (((unsigned int)(val)) % SIGLENBIT)
 
#define HASH(sign, val)   SETBIT((sign), HASHVAL(val))
 
#define ARRKEY   0x01
 
#define SIGNKEY   0x02
 
#define ALLISTRUE   0x04
 
#define ISARRKEY(x)   ( ((TRGM*)x)->flag & ARRKEY )
 
#define ISSIGNKEY(x)   ( ((TRGM*)x)->flag & SIGNKEY )
 
#define ISALLTRUE(x)   ( ((TRGM*)x)->flag & ALLISTRUE )
 
#define CALCGTSIZE(flag, len)   ( TRGMHDRSIZE + ( ( (flag) & ARRKEY ) ? ((len)*sizeof(trgm)) : (((flag) & ALLISTRUE) ? 0 : SIGLEN) ) )
 
#define GETSIGN(x)   ( (BITVECP)( (char*)x+TRGMHDRSIZE ) )
 
#define GETARR(x)   ( (trgm*)( (char*)x+TRGMHDRSIZE ) )
 
#define ARRNELEM(x)   ( ( VARSIZE(x) - TRGMHDRSIZE )/sizeof(trgm) )
 
#define CALCSML(count, len1, len2)   ((float4) (count)) / ((float4) ((len1) + (len2) - (count)))
 

Typedefs

typedef char trgm[3]
 
typedef char BITVEC[SIGLEN]
 
typedef char * BITVECP
 
typedef struct TrgmPackedGraph TrgmPackedGraph
 

Functions

double index_strategy_get_limit (StrategyNumber strategy)
 
uint32 trgm2int (trgm *ptr)
 
void compact_trigram (trgm *tptr, char *str, int bytelen)
 
TRGM * generate_trgm (char *str, int slen)
 
TRGM * generate_wildcard_trgm (const char *str, int slen)
 
float4 cnt_sml (TRGM *trg1, TRGM *trg2, bool inexact)
 
bool trgm_contained_by (TRGM *trg1, TRGM *trg2)
 
bool * trgm_presence_map (TRGM *query, TRGM *key)
 
TRGM * createTrgmNFA (text *text_re, Oid collation, TrgmPackedGraph **graph, MemoryContext rcontext)
 
bool trigramsMatchGraph (TrgmPackedGraph *graph, bool *check)
 

Variables

double similarity_threshold
 
double word_similarity_threshold
 
double strict_word_similarity_threshold
 

Macro Definition Documentation

◆ ALLISTRUE

#define ALLISTRUE   0x04

Definition at line 99 of file trgm.h.

◆ ARRKEY

#define ARRKEY   0x01

Definition at line 97 of file trgm.h.

Referenced by expandColorTrigrams(), generate_trgm(), and generate_wildcard_trgm().

◆ ARRNELEM

#define ARRNELEM (   x)    ( ( VARSIZE(x) - TRGMHDRSIZE )/sizeof(trgm) )

Definition at line 108 of file trgm.h.

Referenced by cnt_sml(), cnt_sml_sign_common(), gin_extract_query_trgm(), gin_extract_value_trgm(), gtrgm_consistent(), gtrgm_distance(), gtrgm_same(), makesign(), show_trgm(), trgm_contained_by(), trgm_presence_map(), and unionkey().

◆ BITBYTE

#define BITBYTE   8

Definition at line 76 of file trgm.h.

◆ CALCGTSIZE

#define CALCGTSIZE (   flag,
  len 
)    ( TRGMHDRSIZE + ( ( (flag) & ARRKEY ) ? ((len)*sizeof(trgm)) : (((flag) & ALLISTRUE) ? 0 : SIGLEN) ) )

Definition at line 105 of file trgm.h.

◆ CALCSML

#define CALCSML (   count,
  len1,
  len2 
)    ((float4) (count)) / ((float4) ((len1) + (len2) - (count)))

Definition at line 117 of file trgm.h.

Referenced by cnt_sml(), and iterate_word_similarity().

◆ CLRBIT

#define CLRBIT (   x,
  i 
)    GETBYTE(x,i) &= ~( 0x01 << ( (i) % BITBYTE ) )

Definition at line 90 of file trgm.h.

◆ CMPCHAR

#define CMPCHAR (   a,
  b 
)    ( ((a)==(b)) ? 0 : ( ((a)<(b)) ? -1 : 1 ) )

Definition at line 43 of file trgm.h.

◆ CMPPCHAR

#define CMPPCHAR (   a,
  b,
  i 
)    CMPCHAR( *(((const char*)(a))+i), *(((const char*)(b))+i) )

Definition at line 44 of file trgm.h.

◆ CMPTRGM

#define CMPTRGM (   a,
  b 
)    ( CMPPCHAR(a,b,0) ? CMPPCHAR(a,b,0) : ( CMPPCHAR(a,b,1) ? CMPPCHAR(a,b,1) : CMPPCHAR(a,b,2) ) )

Definition at line 45 of file trgm.h.

Referenced by calc_word_similarity(), cnt_sml(), comp_ptrgm(), comp_trgm(), gtrgm_same(), trgm_contained_by(), trgm_presence_map(), and unique_array().

◆ CPTRGM

#define CPTRGM (   a,
  b 
)
Value:
do { \
*(((char*)(a))+0) = *(((char*)(b))+0); \
*(((char*)(a))+1) = *(((char*)(b))+1); \
*(((char*)(a))+2) = *(((char*)(b))+2); \
} while(0);

Definition at line 47 of file trgm.h.

Referenced by cnt_sml_sign_common(), compact_trigram(), gtrgm_consistent(), make_trigrams(), makesign(), show_trgm(), unionkey(), and unique_array().

◆ DistanceStrategyNumber

#define DistanceStrategyNumber   2

Definition at line 31 of file trgm.h.

Referenced by gtrgm_distance().

◆ DIVUNION

#define DIVUNION

Definition at line 27 of file trgm.h.

◆ GETARR

#define GETARR (   x)    ( (trgm*)( (char*)x+TRGMHDRSIZE ) )

Definition at line 107 of file trgm.h.

Referenced by cnt_sml(), cnt_sml_sign_common(), expandColorTrigrams(), generate_trgm(), generate_wildcard_trgm(), gin_extract_query_trgm(), gin_extract_value_trgm(), gtrgm_consistent(), gtrgm_same(), makesign(), packArcInfoCmp(), show_trgm(), trgm_contained_by(), trgm_presence_map(), and unionkey().

◆ GETBIT

#define GETBIT (   x,
  i 
)    ( (GETBYTE(x,i) >> ( (i) % BITBYTE )) & 0x01 )

Definition at line 92 of file trgm.h.

◆ GETBITBYTE

#define GETBITBYTE (   x,
  i 
)    ( (((char)(x)) >> (i)) & 0x01 )

Definition at line 89 of file trgm.h.

◆ GETBYTE

#define GETBYTE (   x,
  i 
)    ( *( (BITVECP)(x) + (int)( (i) / BITBYTE ) ) )

Definition at line 88 of file trgm.h.

◆ GETSIGN

#define GETSIGN (   x)    ( (BITVECP)( (char*)x+TRGMHDRSIZE ) )

Definition at line 106 of file trgm.h.

◆ HASH

#define HASH (   sign,
  val 
)    SETBIT((sign), HASHVAL(val))

Definition at line 95 of file trgm.h.

◆ HASHVAL

#define HASHVAL (   val)    (((unsigned int)(val)) % SIGLENBIT)

Definition at line 94 of file trgm.h.

◆ IGNORECASE

#define IGNORECASE

Definition at line 26 of file trgm.h.

◆ ILikeStrategyNumber

#define ILikeStrategyNumber   4

Definition at line 33 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), and gtrgm_consistent().

◆ ISALLTRUE

#define ISALLTRUE (   x)    ( ((TRGM*)x)->flag & ALLISTRUE )

Definition at line 103 of file trgm.h.

◆ ISARRKEY

#define ISARRKEY (   x)    ( ((TRGM*)x)->flag & ARRKEY )

Definition at line 101 of file trgm.h.

Referenced by fillcache(), and gtrgm_penalty().

◆ ISESCAPECHAR

#define ISESCAPECHAR (   x)    (*(x) == '\\') /* Wildcard escape character */

Definition at line 62 of file trgm.h.

Referenced by get_wildcard_part().

◆ ISPRINTABLECHAR

#define ISPRINTABLECHAR (   a)    ( isascii( *(unsigned char*)(a) ) && (isalnum( *(unsigned char*)(a) ) || *(unsigned char*)(a)==' ') )

Definition at line 55 of file trgm.h.

◆ ISPRINTABLETRGM

#define ISPRINTABLETRGM (   t)    ( ISPRINTABLECHAR( ((char*)(t)) ) && ISPRINTABLECHAR( ((char*)(t))+1 ) && ISPRINTABLECHAR( ((char*)(t))+2 ) )

Definition at line 60 of file trgm.h.

Referenced by show_trgm().

◆ ISSIGNKEY

#define ISSIGNKEY (   x)    ( ((TRGM*)x)->flag & SIGNKEY )

Definition at line 102 of file trgm.h.

Referenced by gtrgm_compress(), gtrgm_same(), and unionkey().

◆ ISWILDCARDCHAR

#define ISWILDCARDCHAR (   x)
Value:
(*(x) == '_' || *(x) == '%') /* Wildcard
* meta-character */

Definition at line 63 of file trgm.h.

Referenced by get_wildcard_part().

◆ ISWORDCHR

#define ISWORDCHR (   c)    (t_isalpha(c) || t_isdigit(c))

Definition at line 54 of file trgm.h.

Referenced by find_word(), get_wildcard_part(), and getColorInfo().

◆ KEEPONLYALNUM

#define KEEPONLYALNUM

Definition at line 18 of file trgm.h.

◆ LikeStrategyNumber

#define LikeStrategyNumber   3

Definition at line 32 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), and gtrgm_consistent().

◆ LOOPBYTE

#define LOOPBYTE   for(i=0;i<SIGLEN;i++)

Definition at line 85 of file trgm.h.

◆ LPADDING

#define LPADDING   2

Definition at line 16 of file trgm.h.

Referenced by generate_trgm_only(), generate_wildcard_trgm(), and get_wildcard_part().

◆ RegExpICaseStrategyNumber

#define RegExpICaseStrategyNumber   6

Definition at line 35 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), and gtrgm_consistent().

◆ RegExpStrategyNumber

#define RegExpStrategyNumber   5

Definition at line 34 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), and gtrgm_consistent().

◆ RPADDING

#define RPADDING   1

Definition at line 17 of file trgm.h.

Referenced by generate_trgm_only(), generate_wildcard_trgm(), and get_wildcard_part().

◆ SETBIT

#define SETBIT (   x,
  i 
)    GETBYTE(x,i) |= ( 0x01 << ( (i) % BITBYTE ) )

Definition at line 91 of file trgm.h.

◆ SIGLEN

#define SIGLEN   ( sizeof(int)*SIGLENINT )

Definition at line 78 of file trgm.h.

◆ SIGLENBIT

#define SIGLENBIT   (SIGLEN*BITBYTE - 1) /* see makesign */

Definition at line 80 of file trgm.h.

◆ SIGLENINT

#define SIGLENINT   3 /* >122 => key will toast, so very slow!!! */

Definition at line 77 of file trgm.h.

◆ SIGNKEY

#define SIGNKEY   0x02

Definition at line 98 of file trgm.h.

Referenced by gtrgm_compress(), gtrgm_picksplit(), and gtrgm_union().

◆ SimilarityStrategyNumber

#define SimilarityStrategyNumber   1

Definition at line 30 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), gtrgm_consistent(), and index_strategy_get_limit().

◆ StrictWordDistanceStrategyNumber

#define StrictWordDistanceStrategyNumber   10

Definition at line 39 of file trgm.h.

Referenced by gtrgm_distance().

◆ StrictWordSimilarityStrategyNumber

#define StrictWordSimilarityStrategyNumber   9

Definition at line 38 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), gtrgm_consistent(), and index_strategy_get_limit().

◆ TRGMHDRSIZE

#define TRGMHDRSIZE   (VARHDRSZ + sizeof(uint8))

Definition at line 73 of file trgm.h.

Referenced by expandColorTrigrams(), generate_trgm(), and generate_wildcard_trgm().

◆ WordDistanceStrategyNumber

#define WordDistanceStrategyNumber   8

Definition at line 37 of file trgm.h.

Referenced by gtrgm_distance().

◆ WordSimilarityStrategyNumber

#define WordSimilarityStrategyNumber   7

Definition at line 36 of file trgm.h.

Referenced by gin_extract_query_trgm(), gin_trgm_consistent(), gin_trgm_triconsistent(), gtrgm_consistent(), and index_strategy_get_limit().

Typedef Documentation

◆ BITVEC

typedef char BITVEC[SIGLEN]

Definition at line 82 of file trgm.h.

◆ BITVECP

typedef char* BITVECP

Definition at line 83 of file trgm.h.

◆ trgm

typedef char trgm[3]

Definition at line 41 of file trgm.h.

◆ TrgmPackedGraph

typedef struct TrgmPackedGraph TrgmPackedGraph

Definition at line 122 of file trgm.h.

Function Documentation

◆ cnt_sml()

float4 cnt_sml ( TRGM *  trg1,
TRGM *  trg2,
bool  inexact 
)

Definition at line 1018 of file trgm_op.c.

References ARRNELEM, CALCSML, CMPTRGM, and GETARR.

Referenced by gtrgm_consistent(), gtrgm_distance(), and similarity().

1019 {
1020  trgm *ptr1,
1021  *ptr2;
1022  int count = 0;
1023  int len1,
1024  len2;
1025 
1026  ptr1 = GETARR(trg1);
1027  ptr2 = GETARR(trg2);
1028 
1029  len1 = ARRNELEM(trg1);
1030  len2 = ARRNELEM(trg2);
1031 
1032  /* explicit test is needed to avoid 0/0 division when both lengths are 0 */
1033  if (len1 <= 0 || len2 <= 0)
1034  return (float4) 0.0;
1035 
1036  while (ptr1 - GETARR(trg1) < len1 && ptr2 - GETARR(trg2) < len2)
1037  {
1038  int res = CMPTRGM(ptr1, ptr2);
1039 
1040  if (res < 0)
1041  ptr1++;
1042  else if (res > 0)
1043  ptr2++;
1044  else
1045  {
1046  ptr1++;
1047  ptr2++;
1048  count++;
1049  }
1050  }
1051 
1052  /*
1053  * If inexact then len2 is equal to count, because we don't know actual
1054  * length of second string in inexact search and we can assume that count
1055  * is a lower bound of len2.
1056  */
1057  return CALCSML(count, len1, inexact ? count : len2);
1058 }
CMPTRGM
#define CMPTRGM(a, b)
Definition: trgm.h:45
ARRNELEM
#define ARRNELEM(x)
Definition: trgm.h:108
GETARR
#define GETARR(x)
Definition: trgm.h:107
float4
float float4
Definition: c.h:457
trgm
char trgm[3]
Definition: trgm.h:41
CALCSML
#define CALCSML(count, len1, len2)
Definition: trgm.h:117

◆ compact_trigram()

void compact_trigram ( trgm *  tptr,
char *  str,
int  bytelen 
)

Definition at line 218 of file trgm_op.c.

References COMP_LEGACY_CRC32, CPTRGM, FIN_LEGACY_CRC32, and INIT_LEGACY_CRC32.

Referenced by fillTrgm(), and make_trigrams().

219 {
220  if (bytelen == 3)
221  {
222  CPTRGM(tptr, str);
223  }
224  else
225  {
226  pg_crc32 crc;
227 
228  INIT_LEGACY_CRC32(crc);
229  COMP_LEGACY_CRC32(crc, str, bytelen);
230  FIN_LEGACY_CRC32(crc);
231 
232  /*
233  * use only 3 upper bytes from crc, hope, it's good enough hashing
234  */
235  CPTRGM(tptr, &crc);
236  }
237 }
INIT_LEGACY_CRC32
#define INIT_LEGACY_CRC32(crc)
Definition: pg_crc.h:79
FIN_LEGACY_CRC32
#define FIN_LEGACY_CRC32(crc)
Definition: pg_crc.h:80
generate_unaccent_rules.str
str
Definition: generate_unaccent_rules.py:207
COMP_LEGACY_CRC32
#define COMP_LEGACY_CRC32(crc, data, len)
Definition: pg_crc.h:81
pg_crc32
uint32 pg_crc32
Definition: pg_crc.h:37
CPTRGM
#define CPTRGM(a, b)
Definition: trgm.h:47

◆ createTrgmNFA()

TRGM* createTrgmNFA ( text *  text_re,
Oid  collation,
TrgmPackedGraph **  graph,
MemoryContext  rcontext 
)

Definition at line 523 of file trgm_regexp.c.

References ALLOCSET_DEFAULT_SIZES, AllocSetContextCreate, createTrgmNFAInternal(), CurrentMemoryContext, MemoryContextDelete(), MemoryContextSwitchTo(), PG_CATCH, PG_END_TRY, PG_RE_THROW, pg_regfree(), PG_TRY, RE_compile(), REG_ADVANCED, and REG_ICASE.

Referenced by gin_extract_query_trgm(), and gtrgm_consistent().

525 {
526  TRGM *trg;
527  regex_t regex;
528  MemoryContext tmpcontext;
529  MemoryContext oldcontext;
530 
531  /*
532  * This processing generates a great deal of cruft, which we'd like to
533  * clean up before returning (since this function may be called in a
534  * query-lifespan memory context). Make a temp context we can work in so
535  * that cleanup is easy.
536  */
537  tmpcontext = AllocSetContextCreate(CurrentMemoryContext,
538  "createTrgmNFA temporary context",
539  ALLOCSET_DEFAULT_SIZES);
540  oldcontext = MemoryContextSwitchTo(tmpcontext);
541 
542  /*
543  * Stage 1: Compile the regexp into a NFA, using the regexp library.
544  */
545 #ifdef IGNORECASE
546  RE_compile(&regex, text_re, REG_ADVANCED | REG_ICASE, collation);
547 #else
548  RE_compile(&regex, text_re, REG_ADVANCED, collation);
549 #endif
550 
551  /*
552  * Since the regexp library allocates its internal data structures with
553  * malloc, we need to use a PG_TRY block to ensure that pg_regfree() gets
554  * done even if there's an error.
555  */
556  PG_TRY();
557  {
558  trg = createTrgmNFAInternal(&regex, graph, rcontext);
559  }
560  PG_CATCH();
561  {
562  pg_regfree(&regex);
563  PG_RE_THROW();
564  }
565  PG_END_TRY();
566 
567  pg_regfree(&regex);
568 
569  /* Clean up all the cruft we created */
570  MemoryContextSwitchTo(oldcontext);
571  MemoryContextDelete(tmpcontext);
572 
573  return trg;
574 }
MemoryContextDelete
void MemoryContextDelete(MemoryContext context)
Definition: mcxt.c:211
createTrgmNFAInternal
static TRGM * createTrgmNFAInternal(regex_t *regex, TrgmPackedGraph **graph, MemoryContext rcontext)
Definition: trgm_regexp.c:580
MemoryContextSwitchTo
static MemoryContext MemoryContextSwitchTo(MemoryContext context)
Definition: palloc.h:109
REG_ICASE
#define REG_ICASE
Definition: regex.h:106
TRGM
Definition: trgm.h:66
ALLOCSET_DEFAULT_SIZES
#define ALLOCSET_DEFAULT_SIZES
Definition: memutils.h:192
CurrentMemoryContext
MemoryContext CurrentMemoryContext
Definition: mcxt.c:38
REG_ADVANCED
#define REG_ADVANCED
Definition: regex.h:103
AllocSetContextCreate
#define AllocSetContextCreate(parent, name, allocparams)
Definition: memutils.h:170
PG_CATCH
#define PG_CATCH()
Definition: elog.h:293
RE_compile
static void RE_compile(regex_t *regex, text *text_re, int cflags, Oid collation)
Definition: trgm_regexp.c:734
PG_RE_THROW
#define PG_RE_THROW()
Definition: elog.h:314
PG_TRY
#define PG_TRY()
Definition: elog.h:284
pg_regfree
void pg_regfree(regex_t *re)
Definition: regfree.c:49
regex_t
Definition: regex.h:55
PG_END_TRY
#define PG_END_TRY()
Definition: elog.h:300

◆ generate_trgm()

TRGM* generate_trgm ( char *  str,
int  slen 
)

Definition at line 376 of file trgm_op.c.

References ARRKEY, CALCGTSIZE, comp_trgm(), TRGM::flag, generate_trgm_only(), GETARR, palloc(), protect_out_of_mem(), qsort, SET_VARSIZE, TRGMHDRSIZE, and unique_array().

Referenced by gin_extract_query_trgm(), gin_extract_value_trgm(), gtrgm_compress(), gtrgm_consistent(), gtrgm_distance(), show_trgm(), and similarity().

377 {
378  TRGM *trg;
379  int len;
380 
381  protect_out_of_mem(slen);
382 
383  trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) * 3);
384  trg->flag = ARRKEY;
385 
386  len = generate_trgm_only(GETARR(trg), str, slen, NULL);
387  SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
388 
389  if (len == 0)
390  return trg;
391 
392  /*
393  * Make trigrams unique.
394  */
395  if (len > 1)
396  {
397  qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
398  len = unique_array(GETARR(trg), len);
399  }
400 
401  SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
402 
403  return trg;
404 }
ARRKEY
#define ARRKEY
Definition: trgm.h:97
protect_out_of_mem
static void protect_out_of_mem(int slen)
Definition: trgm_op.c:359
GETARR
#define GETARR(x)
Definition: trgm.h:107
TRGM
Definition: trgm.h:66
generate_unaccent_rules.str
str
Definition: generate_unaccent_rules.py:207
TRGMHDRSIZE
#define TRGMHDRSIZE
Definition: trgm.h:73
TRGM::flag
uint8 flag
Definition: trgm.h:69
trgm
char trgm[3]
Definition: trgm.h:41
CALCGTSIZE
#define CALCGTSIZE(flag)
Definition: hstore_gist.c:52
unique_array
static int unique_array(trgm *a, int len)
Definition: trgm_op.c:167
palloc
void * palloc(Size size)
Definition: mcxt.c:924
SET_VARSIZE
#define SET_VARSIZE(PTR, len)
Definition: postgres.h:329
qsort
#define qsort(a, b, c, d)
Definition: port.h:421
generate_trgm_only
static int generate_trgm_only(trgm *trg, char *str, int slen, TrgmBound *bounds)
Definition: trgm_op.c:295
comp_trgm
static int comp_trgm(const void *a, const void *b)
Definition: trgm_op.c:161

◆ generate_wildcard_trgm()

TRGM* generate_wildcard_trgm ( const char *  str,
int  slen 
)

Definition at line 886 of file trgm_op.c.

References ARRKEY, buf, CALCGTSIZE, comp_trgm(), TRGM::flag, get_wildcard_part(), GETARR, lowerstr_with_len(), LPADDING, make_trigrams(), palloc(), pfree(), protect_out_of_mem(), qsort, RPADDING, SET_VARSIZE, generate_unaccent_rules::str, TRGMHDRSIZE, and unique_array().

Referenced by gin_extract_query_trgm(), and gtrgm_consistent().

887 {
888  TRGM *trg;
889  char *buf,
890  *buf2;
891  trgm *tptr;
892  int len,
893  charlen,
894  bytelen;
895  const char *eword;
896 
897  protect_out_of_mem(slen);
898 
899  trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) * 3);
900  trg->flag = ARRKEY;
901  SET_VARSIZE(trg, TRGMHDRSIZE);
902 
903  if (slen + LPADDING + RPADDING < 3 || slen == 0)
904  return trg;
905 
906  tptr = GETARR(trg);
907 
908  /* Allocate a buffer for blank-padded, but not yet case-folded, words */
909  buf = palloc(sizeof(char) * (slen + 4));
910 
911  /*
912  * Extract trigrams from each substring extracted by get_wildcard_part.
913  */
914  eword = str;
915  while ((eword = get_wildcard_part(eword, slen - (eword - str),
916  buf, &bytelen, &charlen)) != NULL)
917  {
918 #ifdef IGNORECASE
919  buf2 = lowerstr_with_len(buf, bytelen);
920  bytelen = strlen(buf2);
921 #else
922  buf2 = buf;
923 #endif
924 
925  /*
926  * count trigrams
927  */
928  tptr = make_trigrams(tptr, buf2, bytelen, charlen);
929 
930 #ifdef IGNORECASE
931  pfree(buf2);
932 #endif
933  }
934 
935  pfree(buf);
936 
937  if ((len = tptr - GETARR(trg)) == 0)
938  return trg;
939 
940  /*
941  * Make trigrams unique.
942  */
943  if (len > 1)
944  {
945  qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
946  len = unique_array(GETARR(trg), len);
947  }
948 
949  SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
950 
951  return trg;
952 }
lowerstr_with_len
char * lowerstr_with_len(const char *str, int len)
Definition: ts_locale.c:241
ARRKEY
#define ARRKEY
Definition: trgm.h:97
protect_out_of_mem
static void protect_out_of_mem(int slen)
Definition: trgm_op.c:359
make_trigrams
static trgm * make_trigrams(trgm *tptr, char *str, int bytelen, int charlen)
Definition: trgm_op.c:243
GETARR
#define GETARR(x)
Definition: trgm.h:107
pfree
void pfree(void *pointer)
Definition: mcxt.c:1031
TRGM
Definition: trgm.h:66
generate_unaccent_rules.str
str
Definition: generate_unaccent_rules.py:207
buf
static char * buf
Definition: pg_test_fsync.c:67
TRGMHDRSIZE
#define TRGMHDRSIZE
Definition: trgm.h:73
TRGM::flag
uint8 flag
Definition: trgm.h:69
trgm
char trgm[3]
Definition: trgm.h:41
CALCGTSIZE
#define CALCGTSIZE(flag)
Definition: hstore_gist.c:52
RPADDING
#define RPADDING
Definition: trgm.h:17
unique_array
static int unique_array(trgm *a, int len)
Definition: trgm_op.c:167
palloc
void * palloc(Size size)
Definition: mcxt.c:924
LPADDING
#define LPADDING
Definition: trgm.h:16
get_wildcard_part
static const char * get_wildcard_part(const char *str, int lenstr, char *buf, int *bytelen, int *charlen)
Definition: trgm_op.c:741
SET_VARSIZE
#define SET_VARSIZE(PTR, len)
Definition: postgres.h:329
qsort
#define qsort(a, b, c, d)
Definition: port.h:421
comp_trgm
static int comp_trgm(const void *a, const void *b)
Definition: trgm_op.c:161

◆ index_strategy_get_limit()

double index_strategy_get_limit ( StrategyNumber  strategy)

Definition at line 132 of file trgm_op.c.

References elog, ERROR, similarity_threshold, SimilarityStrategyNumber, strict_word_similarity_threshold, StrictWordSimilarityStrategyNumber, word_similarity_threshold, and WordSimilarityStrategyNumber.

Referenced by gin_trgm_consistent(), gin_trgm_triconsistent(), and gtrgm_consistent().

133 {
134  switch (strategy)
135  {
136  case SimilarityStrategyNumber:
137  return similarity_threshold;
138  case WordSimilarityStrategyNumber:
139  return word_similarity_threshold;
140  case StrictWordSimilarityStrategyNumber:
141  return strict_word_similarity_threshold;
142  default:
143  elog(ERROR, "unrecognized strategy number: %d", strategy);
144  break;
145  }
146 
147  return 0.0; /* keep compiler quiet */
148 }
SimilarityStrategyNumber
#define SimilarityStrategyNumber
Definition: trgm.h:30
StrictWordSimilarityStrategyNumber
#define StrictWordSimilarityStrategyNumber
Definition: trgm.h:38
ERROR
#define ERROR
Definition: elog.h:43
strict_word_similarity_threshold
double strict_word_similarity_threshold
Definition: trgm_op.c:21
WordSimilarityStrategyNumber
#define WordSimilarityStrategyNumber
Definition: trgm.h:36
word_similarity_threshold
double word_similarity_threshold
Definition: trgm_op.c:20
elog
#define elog
Definition: elog.h:219
similarity_threshold
double similarity_threshold
Definition: trgm_op.c:19

◆ trgm2int()

uint32 trgm2int ( trgm *  ptr)

Definition at line 955 of file trgm_op.c.

References val.

Referenced by gin_extract_query_trgm(), gin_extract_value_trgm(), and show_trgm().

956 {
957  uint32 val = 0;
958 
959  val |= *(((unsigned char *) ptr));
960  val <<= 8;
961  val |= *(((unsigned char *) ptr) + 1);
962  val <<= 8;
963  val |= *(((unsigned char *) ptr) + 2);
964 
965  return val;
966 }
uint32
unsigned int uint32
Definition: c.h:325
val
long val
Definition: informix.c:689

◆ trgm_contained_by()

bool trgm_contained_by ( TRGM *  trg1,
TRGM *  trg2 
)

Definition at line 1066 of file trgm_op.c.

References ARRNELEM, CMPTRGM, and GETARR.

Referenced by gtrgm_consistent().

1067 {
1068  trgm *ptr1,
1069  *ptr2;
1070  int len1,
1071  len2;
1072 
1073  ptr1 = GETARR(trg1);
1074  ptr2 = GETARR(trg2);
1075 
1076  len1 = ARRNELEM(trg1);
1077  len2 = ARRNELEM(trg2);
1078 
1079  while (ptr1 - GETARR(trg1) < len1 && ptr2 - GETARR(trg2) < len2)
1080  {
1081  int res = CMPTRGM(ptr1, ptr2);
1082 
1083  if (res < 0)
1084  return false;
1085  else if (res > 0)
1086  ptr2++;
1087  else
1088  {
1089  ptr1++;
1090  ptr2++;
1091  }
1092  }
1093  if (ptr1 - GETARR(trg1) < len1)
1094  return false;
1095  else
1096  return true;
1097 }
CMPTRGM
#define CMPTRGM(a, b)
Definition: trgm.h:45
ARRNELEM
#define ARRNELEM(x)
Definition: trgm.h:108
GETARR
#define GETARR(x)
Definition: trgm.h:107
trgm
char trgm[3]
Definition: trgm.h:41

◆ trgm_presence_map()

bool* trgm_presence_map ( TRGM *  query,
TRGM *  key 
)

Definition at line 1105 of file trgm_op.c.

References ARRNELEM, CMPTRGM, GETARR, i, and palloc0().

Referenced by gtrgm_consistent().

1106 {
1107  bool *result;
1108  trgm *ptrq = GETARR(query),
1109  *ptrk = GETARR(key);
1110  int lenq = ARRNELEM(query),
1111  lenk = ARRNELEM(key),
1112  i;
1113 
1114  result = (bool *) palloc0(lenq * sizeof(bool));
1115 
1116  /* for each query trigram, do a binary search in the key array */
1117  for (i = 0; i < lenq; i++)
1118  {
1119  int lo = 0;
1120  int hi = lenk;
1121 
1122  while (lo < hi)
1123  {
1124  int mid = (lo + hi) / 2;
1125  int res = CMPTRGM(ptrq, ptrk + mid);
1126 
1127  if (res < 0)
1128  hi = mid;
1129  else if (res > 0)
1130  lo = mid + 1;
1131  else
1132  {
1133  result[i] = true;
1134  break;
1135  }
1136  }
1137  ptrq++;
1138  }
1139 
1140  return result;
1141 }
CMPTRGM
#define CMPTRGM(a, b)
Definition: trgm.h:45
ARRNELEM
#define ARRNELEM(x)
Definition: trgm.h:108
GETARR
#define GETARR(x)
Definition: trgm.h:107
palloc0
void * palloc0(Size size)
Definition: mcxt.c:955
trgm
char trgm[3]
Definition: trgm.h:41
i
int i
Definition: preproc-comment.c:23

◆ trigramsMatchGraph()

bool trigramsMatchGraph ( TrgmPackedGraph *  graph,
bool *  check 
)

Definition at line 641 of file trgm_regexp.c.

References TrgmPackedState::arcs, TrgmPackedState::arcsCount, TrgmPackedArc::colorTrgm, TrgmPackedGraph::colorTrigramGroups, TrgmPackedGraph::colorTrigramsActive, TrgmPackedGraph::colorTrigramsCount, i, TrgmPackedGraph::states, TrgmPackedGraph::statesActive, TrgmPackedGraph::statesCount, TrgmPackedGraph::statesQueue, and TrgmPackedArc::targetState.

Referenced by gin_trgm_consistent(), gin_trgm_triconsistent(), and gtrgm_consistent().

642 {
643  int i,
644  j,
645  k,
646  queueIn,
647  queueOut;
648 
649  /*
650  * Reset temporary working areas.
651  */
652  memset(graph->colorTrigramsActive, 0,
653  sizeof(bool) * graph->colorTrigramsCount);
654  memset(graph->statesActive, 0, sizeof(bool) * graph->statesCount);
655 
656  /*
657  * Check which color trigrams were matched. A match for any simple
658  * trigram associated with a color trigram counts as a match of the color
659  * trigram.
660  */
661  j = 0;
662  for (i = 0; i < graph->colorTrigramsCount; i++)
663  {
664  int cnt = graph->colorTrigramGroups[i];
665 
666  for (k = j; k < j + cnt; k++)
667  {
668  if (check[k])
669  {
670  /*
671  * Found one matched trigram in the group. Can skip the rest
672  * of them and go to the next group.
673  */
674  graph->colorTrigramsActive[i] = true;
675  break;
676  }
677  }
678  j = j + cnt;
679  }
680 
681  /*
682  * Initialize the statesQueue to hold just the initial state. Note:
683  * statesQueue has room for statesCount entries, which is certainly enough
684  * since no state will be put in the queue more than once. The
685  * statesActive array marks which states have been queued.
686  */
687  graph->statesActive[0] = true;
688  graph->statesQueue[0] = 0;
689  queueIn = 0;
690  queueOut = 1;
691 
692  /* Process queued states as long as there are any. */
693  while (queueIn < queueOut)
694  {
695  int stateno = graph->statesQueue[queueIn++];
696  TrgmPackedState *state = &graph->states[stateno];
697  int cnt = state->arcsCount;
698 
699  /* Loop over state's out-arcs */
700  for (i = 0; i < cnt; i++)
701  {
702  TrgmPackedArc *arc = &state->arcs[i];
703 
704  /*
705  * If corresponding color trigram is present then activate the
706  * corresponding state. We're done if that's the final state,
707  * otherwise queue the state if it's not been queued already.
708  */
709  if (graph->colorTrigramsActive[arc->colorTrgm])
710  {
711  int nextstate = arc->targetState;
712 
713  if (nextstate == 1)
714  return true; /* success: final state is reachable */
715 
716  if (!graph->statesActive[nextstate])
717  {
718  graph->statesActive[nextstate] = true;
719  graph->statesQueue[queueOut++] = nextstate;
720  }
721  }
722  }
723  }
724 
725  /* Queue is empty, so match fails. */
726  return false;
727 }
TrgmPackedGraph::colorTrigramsActive
bool * colorTrigramsActive
Definition: trgm_regexp.c:463
arc
Definition: regguts.h:276
TrgmPackedGraph::states
TrgmPackedState * states
Definition: trgm_regexp.c:460
TrgmPackedState::arcs
TrgmPackedArc * arcs
Definition: trgm_regexp.c:437
TrgmPackedGraph::statesActive
bool * statesActive
Definition: trgm_regexp.c:464
state
Definition: regguts.h:298
i
int i
Definition: preproc-comment.c:23
TrgmPackedGraph::colorTrigramGroups
int * colorTrigramGroups
Definition: trgm_regexp.c:453

Variable Documentation

◆ similarity_threshold

double similarity_threshold

Definition at line 19 of file trgm_op.c.

Referenced by _PG_init(), index_strategy_get_limit(), set_limit(), show_limit(), and similarity_op().

◆ strict_word_similarity_threshold

double strict_word_similarity_threshold

Definition at line 21 of file trgm_op.c.

Referenced by _PG_init(), index_strategy_get_limit(), iterate_word_similarity(), strict_word_similarity_commutator_op(), and strict_word_similarity_op().

◆ word_similarity_threshold

double word_similarity_threshold

Definition at line 20 of file trgm_op.c.

Referenced by _PG_init(), index_strategy_get_limit(), iterate_word_similarity(), word_similarity_commutator_op(), and word_similarity_op().