/*
* Add positions from src to dest after offsetting them by maxpos.
* Return the number added (might be less than expected due to overflow)
*/
static int32
add_pos(TSVector src, WordEntry *srcptr,
TSVector dest, WordEntry *destptr,
int32 maxpos)
{
uint16 *clen = &_POSVECPTR(dest, destptr)->npos;
int i;
uint16 slen = POSDATALEN(src, srcptr),
startlen;
WordEntryPos *spos = POSDATAPTR(src, srcptr),
*dpos = POSDATAPTR(dest, destptr);
if (!destptr->haspos)
*clen = 0;
startlen = *clen;
for (i = 0;
i < slen && *clen < MAXNUMPOS &&
(*clen == 0 || WEP_GETPOS(dpos[*clen - 1]) != MAXENTRYPOS - 1);
i++)
{
WEP_SETWEIGHT(dpos[*clen], WEP_GETWEIGHT(spos[i]));
WEP_SETPOS(dpos[*clen], LIMITPOS(WEP_GETPOS(spos[i]) + maxpos));
(*clen)++;
}
if (*clen != startlen)
destptr->haspos = 1;
return *clen - startlen;
}
/*
* Order: haspos, len, word, for all positions (pos, weight)
*/
static int
silly_cmp_tsvector(const TSVector a, const TSVector b)
{
if (VARSIZE(a) < VARSIZE(b))
return -1;
else if (VARSIZE(a) > VARSIZE(b))
return 1;
else if (a->size < b->size)
return -1;
else if (a->size > b->size)
return 1;
else
{
WordEntry *aptr = ARRPTR(a);
WordEntry *bptr = ARRPTR(b);
int i = 0;
int res;
for (i = 0; i < a->size; i++)
{
if (aptr->haspos != bptr->haspos)
{
return (aptr->haspos > bptr->haspos) ? -1 : 1;
}
else if ((res = tsCompareString(STRPTR(a) + aptr->pos, aptr->len, STRPTR(b) + bptr->pos, bptr->len, false)) != 0)
{
return res;
}
else if (aptr->haspos)
{
WordEntryPos *ap = POSDATAPTR(a, aptr);
WordEntryPos *bp = POSDATAPTR(b, bptr);
int j;
if (POSDATALEN(a, aptr) != POSDATALEN(b, bptr))
return (POSDATALEN(a, aptr) > POSDATALEN(b, bptr)) ? -1 : 1;
for (j = 0; j < POSDATALEN(a, aptr); j++)
{
if (WEP_GETPOS(*ap) != WEP_GETPOS(*bp))
{
return (WEP_GETPOS(*ap) > WEP_GETPOS(*bp)) ? -1 : 1;
}
else if (WEP_GETWEIGHT(*ap) != WEP_GETWEIGHT(*bp))
{
return (WEP_GETWEIGHT(*ap) > WEP_GETWEIGHT(*bp)) ? -1 : 1;
}
ap++, bp++;
}
}
aptr++;
bptr++;
}
}
return 0;
}
Datum
tsvector_setweight(PG_FUNCTION_ARGS)
{
TSVector in = PG_GETARG_TSVECTOR(0);
char cw = PG_GETARG_CHAR(1);
TSVector out;
int i,
j;
WordEntry *entry;
WordEntryPos *p;
int w = 0;
switch (cw)
{
case 'A':
case 'a':
w = 3;
break;
case 'B':
case 'b':
w = 2;
break;
case 'C':
case 'c':
w = 1;
break;
case 'D':
case 'd':
w = 0;
break;
default:
/* internal error */
elog(ERROR, "unrecognized weight: %d", cw);
}
out = (TSVector) palloc(VARSIZE(in));
memcpy(out, in, VARSIZE(in));
entry = ARRPTR(out);
i = out->size;
while (i--)
{
if ((j = POSDATALEN(out, entry)) != 0)
{
p = POSDATAPTR(out, entry);
while (j--)
{
WEP_SETWEIGHT(*p, w);
p++;
}
}
entry++;
}
PG_FREE_IF_COPY(in, 0);
PG_RETURN_POINTER(out);
}
/*
* Returns the number of positions in value 'wptr' within tsvector 'txt',
* that have a weight equal to one of the weights in 'weight' bitmask.
*/
static int
check_weight(TSVector txt, WordEntry *wptr, int8 weight)
{
int len = POSDATALEN(txt, wptr);
int num = 0;
WordEntryPos *ptr = POSDATAPTR(txt, wptr);
while (len--)
{
if (weight & (1 << WEP_GETWEIGHT(*ptr)))
num++;
ptr++;
}
return num;
}
Datum
tsvectorsend(PG_FUNCTION_ARGS)
{
TSVector vec = PG_GETARG_TSVECTOR(0);
StringInfoData buf;
int i,
j;
WordEntry *weptr = ARRPTR(vec);
pq_begintypsend(&buf);
pq_sendint(&buf, vec->size, sizeof(int32));
for (i = 0; i < vec->size; i++)
{
uint16 npos;
/*
* the strings in the TSVector array are not null-terminated, so we
* have to send the null-terminator separately
*/
pq_sendtext(&buf, STRPTR(vec) + weptr->pos, weptr->len);
pq_sendbyte(&buf, '\0');
npos = POSDATALEN(vec, weptr);
pq_sendint(&buf, npos, sizeof(uint16));
if (npos > 0)
{
WordEntryPos *wepptr = POSDATAPTR(vec, weptr);
for (j = 0; j < npos; j++)
pq_sendint(&buf, wepptr[j], sizeof(WordEntryPos));
}
weptr++;
}
PG_RETURN_BYTEA_P(pq_endtypsend(&buf));
}
Datum
tsvector_concat(PG_FUNCTION_ARGS)
{
TSVector in1 = PG_GETARG_TSVECTOR(0);
TSVector in2 = PG_GETARG_TSVECTOR(1);
TSVector out;
WordEntry *ptr;
WordEntry *ptr1,
*ptr2;
WordEntryPos *p;
int maxpos = 0,
i,
j,
i1,
i2,
dataoff,
output_bytes,
output_size;
char *data,
*data1,
*data2;
/* Get max position in in1; we'll need this to offset in2's positions */
ptr = ARRPTR(in1);
i = in1->size;
while (i--)
{
if ((j = POSDATALEN(in1, ptr)) != 0)
{
p = POSDATAPTR(in1, ptr);
while (j--)
{
if (WEP_GETPOS(*p) > maxpos)
maxpos = WEP_GETPOS(*p);
p++;
}
}
ptr++;
}
ptr1 = ARRPTR(in1);
ptr2 = ARRPTR(in2);
data1 = STRPTR(in1);
data2 = STRPTR(in2);
i1 = in1->size;
i2 = in2->size;
/*
* Conservative estimate of space needed. We might need all the data in
* both inputs, and conceivably add a pad byte before position data for
* each item where there was none before.
*/
output_bytes = VARSIZE(in1) + VARSIZE(in2) + i1 + i2;
out = (TSVector) palloc0(output_bytes);
SET_VARSIZE(out, output_bytes);
/*
* We must make out->size valid so that STRPTR(out) is sensible. We'll
* collapse out any unused space at the end.
*/
out->size = in1->size + in2->size;
ptr = ARRPTR(out);
data = STRPTR(out);
dataoff = 0;
while (i1 && i2)
{
int cmp = compareEntry(data1, ptr1, data2, ptr2);
if (cmp < 0)
{ /* in1 first */
ptr->haspos = ptr1->haspos;
ptr->len = ptr1->len;
memcpy(data + dataoff, data1 + ptr1->pos, ptr1->len);
ptr->pos = dataoff;
dataoff += ptr1->len;
if (ptr->haspos)
{
dataoff = SHORTALIGN(dataoff);
memcpy(data + dataoff, _POSVECPTR(in1, ptr1), POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16));
dataoff += POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16);
}
ptr++;
ptr1++;
i1--;
}
else if (cmp > 0)
{ /* in2 first */
ptr->haspos = ptr2->haspos;
ptr->len = ptr2->len;
memcpy(data + dataoff, data2 + ptr2->pos, ptr2->len);
ptr->pos = dataoff;
dataoff += ptr2->len;
if (ptr->haspos)
{
int addlen = add_pos(in2, ptr2, out, ptr, maxpos);
if (addlen == 0)
ptr->haspos = 0;
else
{
dataoff = SHORTALIGN(dataoff);
dataoff += addlen * sizeof(WordEntryPos) + sizeof(uint16);
}
}
ptr++;
ptr2++;
i2--;
}
else
{
ptr->haspos = ptr1->haspos | ptr2->haspos;
ptr->len = ptr1->len;
memcpy(data + dataoff, data1 + ptr1->pos, ptr1->len);
ptr->pos = dataoff;
dataoff += ptr1->len;
if (ptr->haspos)
{
if (ptr1->haspos)
{
dataoff = SHORTALIGN(dataoff);
memcpy(data + dataoff, _POSVECPTR(in1, ptr1), POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16));
dataoff += POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16);
if (ptr2->haspos)
dataoff += add_pos(in2, ptr2, out, ptr, maxpos) * sizeof(WordEntryPos);
}
else /* must have ptr2->haspos */
{
int addlen = add_pos(in2, ptr2, out, ptr, maxpos);
if (addlen == 0)
ptr->haspos = 0;
else
{
dataoff = SHORTALIGN(dataoff);
dataoff += addlen * sizeof(WordEntryPos) + sizeof(uint16);
}
}
}
ptr++;
ptr1++;
ptr2++;
i1--;
i2--;
}
}
while (i1)
{
ptr->haspos = ptr1->haspos;
ptr->len = ptr1->len;
memcpy(data + dataoff, data1 + ptr1->pos, ptr1->len);
ptr->pos = dataoff;
dataoff += ptr1->len;
if (ptr->haspos)
{
dataoff = SHORTALIGN(dataoff);
memcpy(data + dataoff, _POSVECPTR(in1, ptr1), POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16));
dataoff += POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16);
}
ptr++;
ptr1++;
i1--;
}
while (i2)
{
ptr->haspos = ptr2->haspos;
ptr->len = ptr2->len;
memcpy(data + dataoff, data2 + ptr2->pos, ptr2->len);
ptr->pos = dataoff;
dataoff += ptr2->len;
if (ptr->haspos)
{
int addlen = add_pos(in2, ptr2, out, ptr, maxpos);
if (addlen == 0)
ptr->haspos = 0;
else
{
dataoff = SHORTALIGN(dataoff);
dataoff += addlen * sizeof(WordEntryPos) + sizeof(uint16);
}
}
ptr++;
ptr2++;
i2--;
}
/*
* Instead of checking each offset individually, we check for overflow of
* pos fields once at the end.
*/
if (dataoff > MAXSTRPOS)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("string is too long for tsvector (%d bytes, max %d bytes)", dataoff, MAXSTRPOS)));
/*
* Adjust sizes (asserting that we didn't overrun the original estimates)
* and collapse out any unused array entries.
*/
output_size = ptr - ARRPTR(out);
Assert(output_size <= out->size);
out->size = output_size;
if (data != STRPTR(out))
memmove(STRPTR(out), data, dataoff);
output_bytes = CALCDATASIZE(out->size, dataoff);
Assert(output_bytes <= VARSIZE(out));
SET_VARSIZE(out, output_bytes);
PG_FREE_IF_COPY(in1, 0);
PG_FREE_IF_COPY(in2, 1);
PG_RETURN_POINTER(out);
}
Datum
tsvector_concat(PG_FUNCTION_ARGS)
{
TSVector in1 = PG_GETARG_TSVECTOR(0);
TSVector in2 = PG_GETARG_TSVECTOR(1);
TSVector out;
WordEntry *ptr;
WordEntry *ptr1,
*ptr2;
WordEntryPos *p;
int maxpos = 0,
i,
j,
i1,
i2,
dataoff;
char *data,
*data1,
*data2;
ptr = ARRPTR(in1);
i = in1->size;
while (i--)
{
if ((j = POSDATALEN(in1, ptr)) != 0)
{
p = POSDATAPTR(in1, ptr);
while (j--)
{
if (WEP_GETPOS(*p) > maxpos)
maxpos = WEP_GETPOS(*p);
p++;
}
}
ptr++;
}
ptr1 = ARRPTR(in1);
ptr2 = ARRPTR(in2);
data1 = STRPTR(in1);
data2 = STRPTR(in2);
i1 = in1->size;
i2 = in2->size;
/* conservative estimate of space needed */
out = (TSVector) palloc0(VARSIZE(in1) + VARSIZE(in2));
SET_VARSIZE(out, VARSIZE(in1) + VARSIZE(in2));
out->size = in1->size + in2->size;
ptr = ARRPTR(out);
data = STRPTR(out);
dataoff = 0;
while (i1 && i2)
{
int cmp = compareEntry(data1, ptr1, data2, ptr2);
if (cmp < 0)
{ /* in1 first */
ptr->haspos = ptr1->haspos;
ptr->len = ptr1->len;
memcpy(data + dataoff, data1 + ptr1->pos, ptr1->len);
ptr->pos = dataoff;
dataoff += ptr1->len;
if (ptr->haspos)
{
dataoff = SHORTALIGN(dataoff);
memcpy(data + dataoff, _POSVECPTR(in1, ptr1), POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16));
dataoff += POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16);
}
ptr++;
ptr1++;
i1--;
}
else if (cmp > 0)
{ /* in2 first */
ptr->haspos = ptr2->haspos;
ptr->len = ptr2->len;
memcpy(data + dataoff, data2 + ptr2->pos, ptr2->len);
ptr->pos = dataoff;
dataoff += ptr2->len;
if (ptr->haspos)
{
int addlen = add_pos(in2, ptr2, out, ptr, maxpos);
if (addlen == 0)
ptr->haspos = 0;
else
{
dataoff = SHORTALIGN(dataoff);
dataoff += addlen * sizeof(WordEntryPos) + sizeof(uint16);
}
}
ptr++;
ptr2++;
i2--;
}
else
{
ptr->haspos = ptr1->haspos | ptr2->haspos;
ptr->len = ptr1->len;
memcpy(data + dataoff, data1 + ptr1->pos, ptr1->len);
ptr->pos = dataoff;
dataoff += ptr1->len;
if (ptr->haspos)
{
if (ptr1->haspos)
{
dataoff = SHORTALIGN(dataoff);
memcpy(data + dataoff, _POSVECPTR(in1, ptr1), POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16));
dataoff += POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16);
if (ptr2->haspos)
dataoff += add_pos(in2, ptr2, out, ptr, maxpos) * sizeof(WordEntryPos);
}
else /* must have ptr2->haspos */
{
int addlen = add_pos(in2, ptr2, out, ptr, maxpos);
if (addlen == 0)
ptr->haspos = 0;
else
{
dataoff = SHORTALIGN(dataoff);
dataoff += addlen * sizeof(WordEntryPos) + sizeof(uint16);
}
}
}
ptr++;
ptr1++;
ptr2++;
i1--;
i2--;
}
}
while (i1)
{
ptr->haspos = ptr1->haspos;
ptr->len = ptr1->len;
memcpy(data + dataoff, data1 + ptr1->pos, ptr1->len);
ptr->pos = dataoff;
dataoff += ptr1->len;
if (ptr->haspos)
{
dataoff = SHORTALIGN(dataoff);
memcpy(data + dataoff, _POSVECPTR(in1, ptr1), POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16));
dataoff += POSDATALEN(in1, ptr1) * sizeof(WordEntryPos) + sizeof(uint16);
}
ptr++;
ptr1++;
i1--;
}
while (i2)
{
ptr->haspos = ptr2->haspos;
ptr->len = ptr2->len;
memcpy(data + dataoff, data2 + ptr2->pos, ptr2->len);
ptr->pos = dataoff;
dataoff += ptr2->len;
if (ptr->haspos)
{
int addlen = add_pos(in2, ptr2, out, ptr, maxpos);
if (addlen == 0)
ptr->haspos = 0;
else
{
dataoff = SHORTALIGN(dataoff);
dataoff += addlen * sizeof(WordEntryPos) + sizeof(uint16);
}
}
ptr++;
ptr2++;
i2--;
}
/*
* Instead of checking each offset individually, we check for overflow of
* pos fields once at the end.
*/
if (dataoff > MAXSTRPOS)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("string is too long for tsvector (%d bytes, max %d bytes)", dataoff, MAXSTRPOS)));
out->size = ptr - ARRPTR(out);
SET_VARSIZE(out, CALCDATASIZE(out->size, dataoff));
if (data != STRPTR(out))
memmove(STRPTR(out), data, dataoff);
PG_FREE_IF_COPY(in1, 0);
PG_FREE_IF_COPY(in2, 1);
PG_RETURN_POINTER(out);
}
Datum
get_covers(PG_FUNCTION_ARGS)
{
tsvector *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(1));
WordEntry *pptr = ARRPTR(txt);
int i,
dlen = 0,
j,
cur = 0,
len = 0,
rlen;
DocWord *dw,
*dwptr;
text *out;
char *cptr;
DocRepresentation *doc;
int olddwpos = 0;
int ncover = 1;
Extention ext;
doc = get_docrep(txt, query, &rlen);
if (!doc)
{
out = palloc(VARHDRSZ);
VARATT_SIZEP(out) = VARHDRSZ;
PG_FREE_IF_COPY(txt, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_POINTER(out);
}
for (i = 0; i < txt->size; i++)
{
if (!pptr[i].haspos)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("no pos info")));
dlen += POSDATALEN(txt, &(pptr[i]));
}
dwptr = dw = palloc(sizeof(DocWord) * dlen);
memset(dw, 0, sizeof(DocWord) * dlen);
for (i = 0; i < txt->size; i++)
{
WordEntryPos *posdata = POSDATAPTR(txt, &(pptr[i]));
for (j = 0; j < POSDATALEN(txt, &(pptr[i])); j++)
{
dw[cur].w = STRPTR(txt) + pptr[i].pos;
dw[cur].len = pptr[i].len;
dw[cur].pos = WEP_GETPOS(posdata[j]);
cur++;
}
len += (pptr[i].len + 1) * (int) POSDATALEN(txt, &(pptr[i]));
}
qsort((void *) dw, dlen, sizeof(DocWord), compareDocWord);
MemSet(&ext, 0, sizeof(Extention));
while (Cover(doc, rlen, query, &ext))
{
dwptr = dw + olddwpos;
while (dwptr->pos < ext.p && dwptr - dw < dlen)
dwptr++;
olddwpos = dwptr - dw;
dwptr->start = ncover;
while (dwptr->pos < ext.q + 1 && dwptr - dw < dlen)
dwptr++;
(dwptr - 1)->finish = ncover;
len += 4 /* {}+two spaces */ + 2 * 16 /* numbers */ ;
ncover++;
}
out = palloc(VARHDRSZ + len);
cptr = ((char *) out) + VARHDRSZ;
dwptr = dw;
while (dwptr - dw < dlen)
{
if (dwptr->start)
{
sprintf(cptr, "{%d ", dwptr->start);
cptr = strchr(cptr, '\0');
}
memcpy(cptr, dwptr->w, dwptr->len);
cptr += dwptr->len;
*cptr = ' ';
cptr++;
if (dwptr->finish)
{
sprintf(cptr, "}%d ", dwptr->finish);
cptr = strchr(cptr, '\0');
}
dwptr++;
}
VARATT_SIZEP(out) = cptr - ((char *) out);
pfree(dw);
for (i = 0; i < rlen; i++)
if (doc[i].needfree)
pfree(doc[i].item);
pfree(doc);
PG_FREE_IF_COPY(txt, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_POINTER(out);
}
static DocRepresentation *
get_docrep(tsvector * txt, QUERYTYPE * query, int *doclen)
{
ITEM *item = GETQUERY(query);
WordEntry *entry;
WordEntryPos *post;
int4 dimt,
j,
i;
int len = query->size * 4,
cur = 0;
DocRepresentation *doc;
char *operand;
*(uint16 *) POSNULL = lengthof(POSNULL) - 1;
doc = (DocRepresentation *) palloc(sizeof(DocRepresentation) * len);
operand = GETOPERAND(query);
reset_istrue_flag(query);
for (i = 0; i < query->size; i++)
{
if (item[i].type != VAL || item[i].istrue)
continue;
entry = find_wordentry(txt, query, &(item[i]));
if (!entry)
continue;
if (entry->haspos)
{
dimt = POSDATALEN(txt, entry);
post = POSDATAPTR(txt, entry);
}
else
{
dimt = *(uint16 *) POSNULL;
post = POSNULL + 1;
}
while (cur + dimt >= len)
{
len *= 2;
doc = (DocRepresentation *) repalloc(doc, sizeof(DocRepresentation) * len);
}
for (j = 0; j < dimt; j++)
{
if (j == 0)
{
ITEM *kptr,
*iptr = item + i;
int k;
doc[cur].needfree = false;
doc[cur].nitem = 0;
doc[cur].item = (ITEM **) palloc(sizeof(ITEM *) * query->size);
for (k = 0; k < query->size; k++)
{
kptr = item + k;
if (k == i ||
(item[k].type == VAL &&
compareITEM(&kptr, &iptr, operand) == 0))
{
doc[cur].item[doc[cur].nitem] = item + k;
doc[cur].nitem++;
kptr->istrue = 1;
}
}
}
else
{
doc[cur].needfree = false;
doc[cur].nitem = doc[cur - 1].nitem;
doc[cur].item = doc[cur - 1].item;
}
doc[cur].pos = WEP_GETPOS(post[j]);
doc[cur].wclass = WEP_GETWEIGHT(post[j]);
cur++;
}
}
*doclen = cur;
if (cur > 0)
{
if (cur > 1)
qsort((void *) doc, cur, sizeof(DocRepresentation), compareDocR);
return doc;
}
pfree(doc);
return NULL;
}
/*
* make value of tsvector
*/
static tsvector *
makevalue(PRSTEXT * prs)
{
int4 i,
j,
lenstr = 0,
totallen;
tsvector *in;
WordEntry *ptr;
char *str,
*cur;
prs->curwords = uniqueWORD(prs->words, prs->curwords);
for (i = 0; i < prs->curwords; i++)
{
lenstr += SHORTALIGN(prs->words[i].len);
if (prs->words[i].alen)
lenstr += sizeof(uint16) + prs->words[i].pos.apos[0] * sizeof(WordEntryPos);
}
totallen = CALCDATASIZE(prs->curwords, lenstr);
in = (tsvector *) palloc(totallen);
memset(in, 0, totallen);
in->len = totallen;
in->size = prs->curwords;
ptr = ARRPTR(in);
cur = str = STRPTR(in);
for (i = 0; i < prs->curwords; i++)
{
ptr->len = prs->words[i].len;
if (cur - str > MAXSTRPOS)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("value is too big")));
ptr->pos = cur - str;
memcpy((void *) cur, (void *) prs->words[i].word, prs->words[i].len);
pfree(prs->words[i].word);
cur += SHORTALIGN(prs->words[i].len);
if (prs->words[i].alen)
{
WordEntryPos *wptr;
ptr->haspos = 1;
*(uint16 *) cur = prs->words[i].pos.apos[0];
wptr = POSDATAPTR(in, ptr);
for (j = 0; j < *(uint16 *) cur; j++)
{
wptr[j].weight = 0;
wptr[j].pos = prs->words[i].pos.apos[j + 1];
}
cur += sizeof(uint16) + prs->words[i].pos.apos[0] * sizeof(WordEntryPos);
pfree(prs->words[i].pos.apos);
}
else
ptr->haspos = 0;
ptr++;
}
pfree(prs->words);
return in;
}
/*
* make value of tsvector, given parsed text
*/
TSVector
make_tsvector(ParsedText *prs)
{
int i,
j,
lenstr = 0,
totallen;
TSVector in;
WordEntry *ptr;
char *str;
int stroff;
prs->curwords = uniqueWORD(prs->words, prs->curwords);
for (i = 0; i < prs->curwords; i++)
{
lenstr += prs->words[i].len;
if (prs->words[i].alen)
{
lenstr = SHORTALIGN(lenstr);
lenstr += sizeof(uint16) + prs->words[i].pos.apos[0] * sizeof(WordEntryPos);
}
}
if (lenstr > MAXSTRPOS)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("string is too long for tsvector (%d bytes, max %d bytes)", lenstr, MAXSTRPOS)));
totallen = CALCDATASIZE(prs->curwords, lenstr);
in = (TSVector) palloc0(totallen);
SET_VARSIZE(in, totallen);
in->size = prs->curwords;
ptr = ARRPTR(in);
str = STRPTR(in);
stroff = 0;
for (i = 0; i < prs->curwords; i++)
{
ptr->len = prs->words[i].len;
ptr->pos = stroff;
memcpy(str + stroff, prs->words[i].word, prs->words[i].len);
stroff += prs->words[i].len;
pfree(prs->words[i].word);
if (prs->words[i].alen)
{
int k = prs->words[i].pos.apos[0];
WordEntryPos *wptr;
if (k > 0xFFFF)
elog(ERROR, "positions array too long");
ptr->haspos = 1;
stroff = SHORTALIGN(stroff);
*(uint16 *) (str + stroff) = (uint16) k;
wptr = POSDATAPTR(in, ptr);
for (j = 0; j < k; j++)
{
WEP_SETWEIGHT(wptr[j], 0);
WEP_SETPOS(wptr[j], prs->words[i].pos.apos[j + 1]);
}
stroff += sizeof(uint16) + k * sizeof(WordEntryPos);
pfree(prs->words[i].pos.apos);
}
else
ptr->haspos = 0;
ptr++;
}
pfree(prs->words);
return in;
}
static DocRepresentation *
get_docrep(TSVector txt, QueryRepresentation *qr, int *doclen)
{
QueryItem *item = GETQUERY(qr->query);
WordEntry *entry,
*firstentry;
WordEntryPos *post;
int32 dimt,
j,
i,
nitem;
int len = qr->query->size * 4,
cur = 0;
DocRepresentation *doc;
char *operand;
doc = (DocRepresentation *) palloc(sizeof(DocRepresentation) * len);
operand = GETOPERAND(qr->query);
for (i = 0; i < qr->query->size; i++)
{
QueryOperand *curoperand;
if (item[i].type != QI_VAL)
continue;
curoperand = &item[i].qoperand;
if (QR_GET_OPERAND_EXISTS(qr, &item[i]))
continue;
firstentry = entry = find_wordentry(txt, qr->query, curoperand, &nitem);
if (!entry)
continue;
while (entry - firstentry < nitem)
{
if (entry->haspos)
{
dimt = POSDATALEN(txt, entry);
post = POSDATAPTR(txt, entry);
}
else
{
dimt = POSNULL.npos;
post = POSNULL.pos;
}
while (cur + dimt >= len)
{
len *= 2;
doc = (DocRepresentation *) repalloc(doc, sizeof(DocRepresentation) * len);
}
for (j = 0; j < dimt; j++)
{
if (j == 0)
{
int k;
doc[cur].nitem = 0;
doc[cur].item = (QueryItem **) palloc(sizeof(QueryItem *) * qr->query->size);
for (k = 0; k < qr->query->size; k++)
{
QueryOperand *kptr = &item[k].qoperand;
QueryOperand *iptr = &item[i].qoperand;
if (k == i ||
(item[k].type == QI_VAL &&
compareQueryOperand(&kptr, &iptr, operand) == 0))
{
/*
* if k == i, we've already checked above that
* it's type == Q_VAL
*/
doc[cur].item[doc[cur].nitem] = item + k;
doc[cur].nitem++;
QR_SET_OPERAND_EXISTS(qr, item + k);
}
}
}
else
{
doc[cur].nitem = doc[cur - 1].nitem;
doc[cur].item = doc[cur - 1].item;
}
doc[cur].pos = WEP_GETPOS(post[j]);
doc[cur].wclass = WEP_GETWEIGHT(post[j]);
cur++;
}
entry++;
}
}
*doclen = cur;
if (cur > 0)
{
qsort((void *) doc, cur, sizeof(DocRepresentation), compareDocR);
return doc;
}
pfree(doc);
return NULL;
}
static float
calc_rank_or(float *w, TSVector t, TSQuery q)
{
WordEntry *entry,
*firstentry;
WordEntryPos *post;
int32 dimt,
j,
i,
nitem;
float res = 0.0;
QueryOperand **item;
int size = q->size;
item = SortAndUniqItems(q, &size);
for (i = 0; i < size; i++)
{
float resj,
wjm;
int32 jm;
firstentry = entry = find_wordentry(t, q, item[i], &nitem);
if (!entry)
continue;
while (entry - firstentry < nitem)
{
if (entry->haspos)
{
dimt = POSDATALEN(t, entry);
post = POSDATAPTR(t, entry);
}
else
{
dimt = POSNULL.npos;
post = POSNULL.pos;
}
resj = 0.0;
wjm = -1.0;
jm = 0;
for (j = 0; j < dimt; j++)
{
resj = resj + wpos(post[j]) / ((j + 1) * (j + 1));
if (wpos(post[j]) > wjm)
{
wjm = wpos(post[j]);
jm = j;
}
}
/*
limit (sum(i/i^2),i->inf) = pi^2/6
resj = sum(wi/i^2),i=1,noccurence,
wi - should be sorted desc,
don't sort for now, just choose maximum weight. This should be corrected
Oleg Bartunov
*/
res = res + (wjm + resj - wjm / ((jm + 1) * (jm + 1))) / 1.64493406685;
entry++;
}
}
if (size > 0)
res = res / size;
pfree(item);
return res;
}
Datum
tsvectorrecv(PG_FUNCTION_ARGS)
{
StringInfo buf = (StringInfo) PG_GETARG_POINTER(0);
TSVector vec;
int i;
int32 nentries;
int datalen; /* number of bytes used in the variable size
* area after fixed size TSVector header and
* WordEntries */
Size hdrlen;
Size len; /* allocated size of vec */
bool needSort = false;
nentries = pq_getmsgint(buf, sizeof(int32));
if (nentries < 0 || nentries > (MaxAllocSize / sizeof(WordEntry)))
elog(ERROR, "invalid size of tsvector");
hdrlen = DATAHDRSIZE + sizeof(WordEntry) * nentries;
len = hdrlen * 2; /* times two to make room for lexemes */
vec = (TSVector) palloc0(len);
vec->size = nentries;
datalen = 0;
for (i = 0; i < nentries; i++)
{
const char *lexeme;
uint16 npos;
size_t lex_len;
lexeme = pq_getmsgstring(buf);
npos = (uint16) pq_getmsgint(buf, sizeof(uint16));
/* sanity checks */
lex_len = strlen(lexeme);
if (lex_len > MAXSTRLEN)
elog(ERROR, "invalid tsvector: lexeme too long");
if (datalen > MAXSTRPOS)
elog(ERROR, "invalid tsvector: maximum total lexeme length exceeded");
if (npos > MAXNUMPOS)
elog(ERROR, "unexpected number of tsvector positions");
/*
* Looks valid. Fill the WordEntry struct, and copy lexeme.
*
* But make sure the buffer is large enough first.
*/
while (hdrlen + SHORTALIGN(datalen + lex_len) +
(npos + 1) * sizeof(WordEntryPos) >= len)
{
len *= 2;
vec = (TSVector) repalloc(vec, len);
}
vec->entries[i].haspos = (npos > 0) ? 1 : 0;
vec->entries[i].len = lex_len;
vec->entries[i].pos = datalen;
memcpy(STRPTR(vec) + datalen, lexeme, lex_len);
datalen += lex_len;
if (i > 0 && WordEntryCMP(&vec->entries[i],
&vec->entries[i - 1],
STRPTR(vec)) <= 0)
needSort = true;
/* Receive positions */
if (npos > 0)
{
uint16 j;
WordEntryPos *wepptr;
/*
* Pad to 2-byte alignment if necessary. Though we used palloc0
* for the initial allocation, subsequent repalloc'd memory areas
* are not initialized to zero.
*/
if (datalen != SHORTALIGN(datalen))
{
*(STRPTR(vec) + datalen) = '\0';
datalen = SHORTALIGN(datalen);
}
memcpy(STRPTR(vec) + datalen, &npos, sizeof(uint16));
wepptr = POSDATAPTR(vec, &vec->entries[i]);
for (j = 0; j < npos; j++)
{
wepptr[j] = (WordEntryPos) pq_getmsgint(buf, sizeof(WordEntryPos));
if (j > 0 && WEP_GETPOS(wepptr[j]) <= WEP_GETPOS(wepptr[j - 1]))
elog(ERROR, "position information is misordered");
}
datalen += (npos + 1) * sizeof(WordEntry);
}
}
SET_VARSIZE(vec, hdrlen + datalen);
if (needSort)
qsort_arg((void *) ARRPTR(vec), vec->size, sizeof(WordEntry),
compareentry, (void *) STRPTR(vec));
PG_RETURN_TSVECTOR(vec);
}
Datum
tsvectorout(PG_FUNCTION_ARGS)
{
TSVector out = PG_GETARG_TSVECTOR(0);
char *outbuf;
int32 i,
lenbuf = 0,
pp;
WordEntry *ptr = ARRPTR(out);
char *curbegin,
*curin,
*curout;
lenbuf = out->size * 2 /* '' */ + out->size - 1 /* space */ + 2 /* \0 */ ;
for (i = 0; i < out->size; i++)
{
lenbuf += ptr[i].len * 2 * pg_database_encoding_max_length() /* for escape */ ;
if (ptr[i].haspos)
lenbuf += 1 /* : */ + 7 /* int2 + , + weight */ * POSDATALEN(out, &(ptr[i]));
}
curout = outbuf = (char *) palloc(lenbuf);
for (i = 0; i < out->size; i++)
{
curbegin = curin = STRPTR(out) + ptr->pos;
if (i != 0)
*curout++ = ' ';
*curout++ = '\'';
while (curin - curbegin < ptr->len)
{
int len = pg_mblen(curin);
if (t_iseq(curin, '\''))
*curout++ = '\'';
else if (t_iseq(curin, '\\'))
*curout++ = '\\';
while (len--)
*curout++ = *curin++;
}
*curout++ = '\'';
if ((pp = POSDATALEN(out, ptr)) != 0)
{
WordEntryPos *wptr;
*curout++ = ':';
wptr = POSDATAPTR(out, ptr);
while (pp)
{
curout += sprintf(curout, "%d", WEP_GETPOS(*wptr));
switch (WEP_GETWEIGHT(*wptr))
{
case 3:
*curout++ = 'A';
break;
case 2:
*curout++ = 'B';
break;
case 1:
*curout++ = 'C';
break;
case 0:
default:
break;
}
if (pp > 1)
*curout++ = ',';
pp--;
wptr++;
}
}
ptr++;
}
*curout = '\0';
PG_FREE_IF_COPY(out, 0);
PG_RETURN_CSTRING(outbuf);
}
static float
calc_rank_or(float *w, tsvector * t, QUERYTYPE * q)
{
WordEntry *entry;
WordEntryPos *post;
int4 dimt,
j,
i;
float res = 0.0;
ITEM **item;
int size = q->size;
*(uint16 *) POSNULL = lengthof(POSNULL) - 1;
item = SortAndUniqItems(GETOPERAND(q), GETQUERY(q), &size);
for (i = 0; i < size; i++)
{
float resj,
wjm;
int4 jm;
entry = find_wordentry(t, q, item[i]);
if (!entry)
continue;
if (entry->haspos)
{
dimt = POSDATALEN(t, entry);
post = POSDATAPTR(t, entry);
}
else
{
dimt = *(uint16 *) POSNULL;
post = POSNULL + 1;
}
resj = 0.0;
wjm = -1.0;
jm = 0;
for (j = 0; j < dimt; j++)
{
resj = resj + wpos(post[j]) / ((j + 1) * (j + 1));
if (wpos(post[j]) > wjm)
{
wjm = wpos(post[j]);
jm = j;
}
}
/*
limit (sum(i/i^2),i->inf) = pi^2/6
resj = sum(wi/i^2),i=1,noccurence,
wi - should be sorted desc,
don't sort for now, just choose maximum weight. This should be corrected
Oleg Bartunov
*/
res = res + (wjm + resj - wjm / ((jm + 1) * (jm + 1))) / 1.64493406685;
}
if (size > 0)
res = res / size;
pfree(item);
return res;
}
Datum
tsvector_out(PG_FUNCTION_ARGS)
{
tsvector *out = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
char *outbuf;
int4 i,
j,
lenbuf = 0,
pp;
WordEntry *ptr = ARRPTR(out);
char *curin,
*curout;
lenbuf = out->size * 2 /* '' */ + out->size - 1 /* space */ + 2 /* \0 */ ;
for (i = 0; i < out->size; i++)
{
lenbuf += ptr[i].len * 2 /* for escape */ ;
if (ptr[i].haspos)
lenbuf += 7 * POSDATALEN(out, &(ptr[i]));
}
curout = outbuf = (char *) palloc(lenbuf);
for (i = 0; i < out->size; i++)
{
curin = STRPTR(out) + ptr->pos;
if (i != 0)
*curout++ = ' ';
*curout++ = '\'';
j = ptr->len;
while (j--)
{
if (*curin == '\'')
{
int4 pos = curout - outbuf;
outbuf = (char *) repalloc((void *) outbuf, ++lenbuf);
curout = outbuf + pos;
*curout++ = '\\';
}
*curout++ = *curin++;
}
*curout++ = '\'';
if ((pp = POSDATALEN(out, ptr)) != 0)
{
WordEntryPos *wptr;
*curout++ = ':';
wptr = POSDATAPTR(out, ptr);
while (pp)
{
sprintf(curout, "%d", wptr->pos);
curout = strchr(curout, '\0');
switch (wptr->weight)
{
case 3:
*curout++ = 'A';
break;
case 2:
*curout++ = 'B';
break;
case 1:
*curout++ = 'C';
break;
case 0:
default:
break;
}
if (pp > 1)
*curout++ = ',';
pp--;
wptr++;
}
}
ptr++;
}
*curout = '\0';
outbuf[lenbuf - 1] = '\0';
PG_FREE_IF_COPY(out, 0);
PG_RETURN_POINTER(outbuf);
}
