/*
* Make array of trigrams with sorting and removing duplicate items.
*
* str: source string, of length slen bytes.
*
* Returns the sorted array of unique trigrams.
*/
TRGM *
generate_trgm(char *str, int slen)
{
TRGM *trg;
int len;
protect_out_of_mem(slen);
trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) *3);
trg->flag = ARRKEY;
len = generate_trgm_only(GETARR(trg), str, slen);
SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
if (len == 0)
return trg;
/*
* Make trigrams unique.
*/
if (len > 1)
{
qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
len = unique_array(GETARR(trg), len);
}
SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
return trg;
}
Datum
ghstore_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval = entry;
if (entry->leafkey)
{
GISTTYPE *res = (GISTTYPE *) palloc0(CALCGTSIZE(0));
HStore *val = DatumGetHStoreP(entry->key);
HEntry *hsent = ARRPTR(val);
char *ptr = STRPTR(val);
int count = HS_COUNT(val);
int i;
SET_VARSIZE(res, CALCGTSIZE(0));
for (i = 0; i < count; ++i)
{
int h;
h = crc32_sz((char *) HSTORE_KEY(hsent, ptr, i),
HSTORE_KEYLEN(hsent, i));
HASH(GETSIGN(res), h);
if (!HSTORE_VALISNULL(hsent, i))
{
h = crc32_sz((char *) HSTORE_VAL(hsent, ptr, i),
HSTORE_VALLEN(hsent, i));
HASH(GETSIGN(res), h);
}
}
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset,
FALSE);
}
else if (!ISALLTRUE(DatumGetPointer(entry->key)))
{
int32 i;
GISTTYPE *res;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
LOOPBYTE
{
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(retval);
}
res = (GISTTYPE *) palloc(CALCGTSIZE(ALLISTRUE));
SET_VARSIZE(res, CALCGTSIZE(ALLISTRUE));
res->flag = ALLISTRUE;
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset,
FALSE);
}
datum_t
gtrgm_compress(PG_FUNC_ARGS)
{
struct gist_entry *entry = (struct gist_entry *) ARG_POINTER(0);
struct gist_entry *retval = entry;
if (entry->leafkey)
{ /* trgm */
TRGM *res;
text *val = D_TO_TEXT_P(entry->key);
res = generate_trgm(VLA_DATA(val), VLA_SZ(val) - VAR_HDR_SZ);
retval = (struct gist_entry *) palloc(sizeof(struct gist_entry));
gistentryinit(*retval, PTR_TO_D(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
else if (ISSIGNKEY(D_TO_PTR(entry->key)) &&
!ISALLTRUE(D_TO_PTR(entry->key)))
{
int4 i,
len;
TRGM *res;
BITVECP sign = GETSIGN(D_TO_PTR(entry->key));
LOOPBYTE
{
if ((sign[i] & 0xff) != 0xff)
RET_POINTER(retval);
}
len = CALCGTSIZE(SIGNKEY | ALLISTRUE, 0);
res = (TRGM *) palloc(len);
VLA_SET_SZ_STND(res, len);
res->flag = SIGNKEY | ALLISTRUE;
retval = (struct gist_entry *) palloc(sizeof(struct gist_entry));
gistentryinit(*retval, PTR_TO_D(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
Datum
gtrgm_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval = entry;
if (entry->leafkey)
{ /* trgm */
TRGM *res;
text *val = DatumGetTextP(entry->key);
res = generate_trgm(VARDATA(val), VARSIZE(val) - VARHDRSZ);
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
else if (ISSIGNKEY(DatumGetPointer(entry->key)) &&
!ISALLTRUE(DatumGetPointer(entry->key)))
{
int32 i,
len;
TRGM *res;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
LOOPBYTE
{
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(retval);
}
len = CALCGTSIZE(SIGNKEY | ALLISTRUE, 0);
res = (TRGM *) palloc(len);
SET_VARSIZE(res, len);
res->flag = SIGNKEY | ALLISTRUE;
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
/*
* Generates trigrams for wildcard search string.
*
* Returns array of trigrams that must occur in any string that matches the
* wildcard string. For example, given pattern "a%bcd%" the trigrams
* " a", "bcd" would be extracted.
*/
TRGM *
generate_wildcard_trgm(const char *str, int slen)
{
TRGM *trg;
char *buf,
*buf2;
trgm *tptr;
int len,
charlen,
bytelen;
const char *eword;
trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) *3);
trg->flag = ARRKEY;
SET_VARSIZE(trg, TRGMHDRSIZE);
if (slen + LPADDING + RPADDING < 3 || slen == 0)
return trg;
tptr = GETARR(trg);
buf = palloc(sizeof(char) * (slen + 4));
/*
* Extract trigrams from each substring extracted by get_wildcard_part.
*/
eword = str;
while ((eword = get_wildcard_part(eword, slen - (eword - str),
buf, &bytelen, &charlen)) != NULL)
{
#ifdef IGNORECASE
buf2 = lowerstr_with_len(buf, bytelen);
bytelen = strlen(buf2);
#else
buf2 = buf;
#endif
/*
* count trigrams
*/
tptr = make_trigrams(tptr, buf2, bytelen, charlen);
#ifdef IGNORECASE
pfree(buf2);
#endif
}
pfree(buf);
if ((len = tptr - GETARR(trg)) == 0)
return trg;
/*
* Make trigrams unique.
*/
if (len > 0)
{
qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
len = unique_array(GETARR(trg), len);
}
SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
return trg;
}
TRGM *
generate_trgm(char *str, int slen)
{
TRGM *trg;
char *buf;
trgm *tptr;
int len,
charlen,
bytelen;
char *bword,
*eword;
trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) *3);
trg->flag = ARRKEY;
SET_VARSIZE(trg, TRGMHDRSIZE);
if (slen + LPADDING + RPADDING < 3 || slen == 0)
return trg;
tptr = GETARR(trg);
buf = palloc(sizeof(char) * (slen + 4));
if (LPADDING > 0)
{
*buf = ' ';
if (LPADDING > 1)
*(buf + 1) = ' ';
}
eword = str;
while ((bword = find_word(eword, slen - (eword - str), &eword, &charlen)) != NULL)
{
#ifdef IGNORECASE
bword = lowerstr_with_len(bword, eword - bword);
bytelen = strlen(bword);
#else
bytelen = eword - bword;
#endif
memcpy(buf + LPADDING, bword, bytelen);
#ifdef IGNORECASE
pfree(bword);
#endif
buf[LPADDING + bytelen] = ' ';
buf[LPADDING + bytelen + 1] = ' ';
/*
* count trigrams
*/
tptr = make_trigrams(tptr, buf, bytelen + LPADDING + RPADDING,
charlen + LPADDING + RPADDING);
}
pfree(buf);
if ((len = tptr - GETARR(trg)) == 0)
return trg;
if (len > 0)
{
qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
len = unique_array(GETARR(trg), len);
}
SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
return trg;
}
/*
* Generates trigrams for wildcard search string.
*
* Returns array of trigrams that must occur in any string that matches the
* wildcard string. For example, given pattern "a%bcd%" the trigrams
* " a", "bcd" would be extracted.
*/
TRGM *
generate_wildcard_trgm(const char *str, int slen)
{
TRGM *trg;
char *buf,
*buf2;
trgm *tptr;
int len,
charlen,
bytelen;
const char *eword;
/*
* Guard against possible overflow in the palloc requests below. (We
* don't worry about the additive constants, since palloc can detect
* requests that are a little above MaxAllocSize --- we just need to
* prevent integer overflow in the multiplications.)
*/
if ((Size) (slen / 2) >= (MaxAllocSize / (sizeof(trgm) * 3)) ||
(Size) slen >= (MaxAllocSize / pg_database_encoding_max_length()))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("out of memory")));
trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) *3);
trg->flag = ARRKEY;
SET_VARSIZE(trg, TRGMHDRSIZE);
if (slen + LPADDING + RPADDING < 3 || slen == 0)
return trg;
tptr = GETARR(trg);
/* Allocate a buffer for blank-padded, but not yet case-folded, words */
buf = palloc(sizeof(char) * (slen + 4));
/*
* Extract trigrams from each substring extracted by get_wildcard_part.
*/
eword = str;
while ((eword = get_wildcard_part(eword, slen - (eword - str),
buf, &bytelen, &charlen)) != NULL)
{
#ifdef IGNORECASE
buf2 = lowerstr_with_len(buf, bytelen);
bytelen = strlen(buf2);
#else
buf2 = buf;
#endif
/*
* count trigrams
*/
tptr = make_trigrams(tptr, buf2, bytelen, charlen);
#ifdef IGNORECASE
pfree(buf2);
#endif
}
pfree(buf);
if ((len = tptr - GETARR(trg)) == 0)
return trg;
/*
* Make trigrams unique.
*/
if (len > 1)
{
qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
len = unique_array(GETARR(trg), len);
}
SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
return trg;
}
TRGM *
generate_trgm(char *str, int slen)
{
TRGM *trg;
char *buf;
trgm *tptr;
int len,
charlen,
bytelen;
char *bword,
*eword;
/*
* Guard against possible overflow in the palloc requests below. (We
* don't worry about the additive constants, since palloc can detect
* requests that are a little above MaxAllocSize --- we just need to
* prevent integer overflow in the multiplications.)
*/
if ((Size) (slen / 2) >= (MaxAllocSize / (sizeof(trgm) * 3)) ||
(Size) slen >= (MaxAllocSize / pg_database_encoding_max_length()))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("out of memory")));
trg = (TRGM *) palloc(TRGMHDRSIZE + sizeof(trgm) * (slen / 2 + 1) *3);
trg->flag = ARRKEY;
SET_VARSIZE(trg, TRGMHDRSIZE);
if (slen + LPADDING + RPADDING < 3 || slen == 0)
return trg;
tptr = GETARR(trg);
/* Allocate a buffer for case-folded, blank-padded words */
buf = (char *) palloc(slen * pg_database_encoding_max_length() + 4);
if (LPADDING > 0)
{
*buf = ' ';
if (LPADDING > 1)
*(buf + 1) = ' ';
}
eword = str;
while ((bword = find_word(eword, slen - (eword - str), &eword, &charlen)) != NULL)
{
#ifdef IGNORECASE
bword = lowerstr_with_len(bword, eword - bword);
bytelen = strlen(bword);
#else
bytelen = eword - bword;
#endif
memcpy(buf + LPADDING, bword, bytelen);
#ifdef IGNORECASE
pfree(bword);
#endif
buf[LPADDING + bytelen] = ' ';
buf[LPADDING + bytelen + 1] = ' ';
/*
* count trigrams
*/
tptr = make_trigrams(tptr, buf, bytelen + LPADDING + RPADDING,
charlen + LPADDING + RPADDING);
}
pfree(buf);
if ((len = tptr - GETARR(trg)) == 0)
return trg;
/*
* Make trigrams unique.
*/
if (len > 1)
{
qsort((void *) GETARR(trg), len, sizeof(trgm), comp_trgm);
len = unique_array(GETARR(trg), len);
}
SET_VARSIZE(trg, CALCGTSIZE(ARRKEY, len));
return trg;
}
Datum
gtsvector_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval = entry;
if (entry->leafkey)
{ /* tsvector */
SignTSVector *res;
TSVector val = DatumGetTSVector(entry->key);
int32 len;
int32 *arr;
WordEntry *ptr = ARRPTR(val);
char *words = STRPTR(val);
len = CALCGTSIZE(ARRKEY, val->size);
res = (SignTSVector *) palloc(len);
SET_VARSIZE(res, len);
res->flag = ARRKEY;
arr = GETARR(res);
len = val->size;
while (len--)
{
pg_crc32 c;
INIT_LEGACY_CRC32(c);
COMP_LEGACY_CRC32(c, words + ptr->pos, ptr->len);
FIN_LEGACY_CRC32(c);
*arr = *(int32 *) &c;
arr++;
ptr++;
}
len = uniqueint(GETARR(res), val->size);
if (len != val->size)
{
/*
* there is a collision of hash-function; len is always less than
* val->size
*/
len = CALCGTSIZE(ARRKEY, len);
res = (SignTSVector *) repalloc((void *) res, len);
SET_VARSIZE(res, len);
}
/* make signature, if array is too long */
if (VARSIZE(res) > TOAST_INDEX_TARGET)
{
SignTSVector *ressign;
len = CALCGTSIZE(SIGNKEY, 0);
ressign = (SignTSVector *) palloc(len);
SET_VARSIZE(ressign, len);
ressign->flag = SIGNKEY;
makesign(GETSIGN(ressign), res);
res = ressign;
}
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
else if (ISSIGNKEY(DatumGetPointer(entry->key)) &&
!ISALLTRUE(DatumGetPointer(entry->key)))
{
int32 i,
len;
SignTSVector *res;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
LOOPBYTE
{
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(retval);
}
len = CALCGTSIZE(SIGNKEY | ALLISTRUE, 0);
res = (SignTSVector *) palloc(len);
SET_VARSIZE(res, len);
res->flag = SIGNKEY | ALLISTRUE;
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
Datum
g_intbig_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
if (entry->leafkey)
{
GISTENTRY *retval;
ArrayType *in = DatumGetArrayTypeP(entry->key);
int32 *ptr;
int num;
GISTTYPE *res = (GISTTYPE *) palloc0(CALCGTSIZE(0));
CHECKARRVALID(in);
if (ARRISEMPTY(in))
{
ptr = NULL;
num = 0;
}
else
{
ptr = ARRPTR(in);
num = ARRNELEMS(in);
}
SET_VARSIZE(res, CALCGTSIZE(0));
while (num--)
{
HASH(GETSIGN(res), *ptr);
ptr++;
}
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
if (in != DatumGetArrayTypeP(entry->key))
pfree(in);
PG_RETURN_POINTER(retval);
}
else if (!ISALLTRUE(DatumGetPointer(entry->key)))
{
GISTENTRY *retval;
int i;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
GISTTYPE *res;
LOOPBYTE
{
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(entry);
}
res = (GISTTYPE *) palloc(CALCGTSIZE(ALLISTRUE));
SET_VARSIZE(res, CALCGTSIZE(ALLISTRUE));
res->flag = ALLISTRUE;
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
PG_RETURN_POINTER(retval);
}
Datum
gtsvector_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval = entry;
if (entry->leafkey)
{ /* tsvector */
GISTTYPE *res;
tsvector *val = (tsvector *) PG_DETOAST_DATUM(entry->key);
int4 len;
int4 *arr;
WordEntry *ptr = ARRPTR(val);
char *words = STRPTR(val);
len = CALCGTSIZE(ARRKEY, val->size);
res = (GISTTYPE *) palloc(len);
SET_VARSIZE(res, len);
res->flag = ARRKEY;
arr = GETARR(res);
len = val->size;
while (len--)
{
*arr = crc32_sz(&words[ptr->pos], ptr->len);
arr++;
ptr++;
}
len = uniqueint(GETARR(res), val->size);
if (len != val->size)
{
/*
* there is a collision of hash-function; len is always less than
* val->size
*/
len = CALCGTSIZE(ARRKEY, len);
res = (GISTTYPE *) repalloc((void *) res, len);
SET_VARSIZE(res, len);
}
/* make signature, if array is too long */
if (VARSIZE(res) > TOAST_INDEX_TARGET)
{
GISTTYPE *ressign;
len = CALCGTSIZE(SIGNKEY, 0);
ressign = (GISTTYPE *) palloc(len);
SET_VARSIZE(ressign, len);
ressign->flag = SIGNKEY;
makesign(GETSIGN(ressign), res);
res = ressign;
}
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
else if (ISSIGNKEY(DatumGetPointer(entry->key)) &&
!ISALLTRUE(DatumGetPointer(entry->key)))
{
int4 i,
len;
GISTTYPE *res;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
LOOPBYTE(
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(retval);
);
entry->offset, FALSE);
}
else if (ISSIGNKEY(DatumGetPointer(entry->key)) &&
!ISALLTRUE(DatumGetPointer(entry->key)))
{
int4 i,
len;
GISTTYPE *res;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
LOOPBYTE(
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(retval);
);
len = CALCGTSIZE(SIGNKEY | ALLISTRUE, 0);
res = (GISTTYPE *) palloc(len);
SET_VARSIZE(res, len);
res->flag = SIGNKEY | ALLISTRUE;
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset, FALSE);
}
PG_RETURN_POINTER(retval);
}
Datum
gtsvector_decompress(PG_FUNCTION_ARGS)
{
Datum
ghstore_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval = entry;
if (entry->leafkey)
{
GISTTYPE *res = (GISTTYPE *) palloc0(CALCGTSIZE(0));
HStore *toastedval = (HStore *) DatumGetPointer(entry->key);
HStore *val = (HStore *) DatumGetPointer(PG_DETOAST_DATUM(entry->key));
HEntry *ptr = ARRPTR(val);
char *words = STRPTR(val);
SET_VARSIZE(res, CALCGTSIZE(0));
while (ptr - ARRPTR(val) < val->size)
{
int h;
h = crc32_sz((char *) (words + ptr->pos), ptr->keylen);
HASH(GETSIGN(res), h);
if (!ptr->valisnull)
{
h = crc32_sz((char *) (words + ptr->pos + ptr->keylen), ptr->vallen);
HASH(GETSIGN(res), h);
}
ptr++;
}
if (val != toastedval)
pfree(val);
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset,
FALSE);
}
else if (!ISALLTRUE(DatumGetPointer(entry->key)))
{
int4 i;
GISTTYPE *res;
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
LOOPBYTE
{
if ((sign[i] & 0xff) != 0xff)
PG_RETURN_POINTER(retval);
}
res = (GISTTYPE *) palloc(CALCGTSIZE(ALLISTRUE));
SET_VARSIZE(res, CALCGTSIZE(ALLISTRUE));
res->flag = ALLISTRUE;
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(res),
entry->rel, entry->page,
entry->offset,
FALSE);
}