static int
CompareTSQ(TSQuery a, TSQuery b)
{
if (a->size != b->size)
{
return (a->size < b->size) ? -1 : 1;
}
else if (VARSIZE(a) != VARSIZE(b))
{
return (VARSIZE(a) < VARSIZE(b)) ? -1 : 1;
}
else if (a->size != 0)
{
QTNode *an = QT2QTN(GETQUERY(a), GETOPERAND(a));
QTNode *bn = QT2QTN(GETQUERY(b), GETOPERAND(b));
int res = QTNodeCompare(an, bn);
QTNFree(an);
QTNFree(bn);
return res;
}
return 0;
}
/*
* Remove QI_VALSTOP (stopword) nodes from TSQuery.
*/
TSQuery
cleanup_tsquery_stopwords(TSQuery in)
{
int32 len,
lenstr,
commonlen,
i;
NODE *root;
int ladd,
radd;
TSQuery out;
QueryItem *items;
char *operands;
if (in->size == 0)
return in;
/* eliminate stop words */
root = clean_stopword_intree(maketree(GETQUERY(in)), &ladd, &radd);
if (root == NULL)
{
ereport(NOTICE,
(errmsg("text-search query contains only stop words or doesn't contain lexemes, ignored")));
out = palloc(HDRSIZETQ);
out->size = 0;
SET_VARSIZE(out, HDRSIZETQ);
return out;
}
/*
* Build TSQuery from plain view
*/
lenstr = calcstrlen(root);
items = plaintree(root, &len);
commonlen = COMPUTESIZE(len, lenstr);
out = palloc(commonlen);
SET_VARSIZE(out, commonlen);
out->size = len;
memcpy(GETQUERY(out), items, len * sizeof(QueryItem));
items = GETQUERY(out);
operands = GETOPERAND(out);
for (i = 0; i < out->size; i++)
{
QueryOperand *op = (QueryOperand *) &items[i];
if (op->type != QI_VAL)
continue;
memcpy(operands, GETOPERAND(in) + op->distance, op->length);
operands[op->length] = '\0';
op->distance = operands - GETOPERAND(out);
operands += op->length + 1;
}
return out;
}
Datum
tsquery_rewrite(PG_FUNCTION_ARGS)
{
TSQuery query = PG_GETARG_TSQUERY_COPY(0);
TSQuery ex = PG_GETARG_TSQUERY(1);
TSQuery subst = PG_GETARG_TSQUERY(2);
TSQuery rewrited = query;
QTNode *tree,
*qex,
*subs = NULL;
if (query->size == 0 || ex->size == 0)
{
PG_FREE_IF_COPY(ex, 1);
PG_FREE_IF_COPY(subst, 2);
PG_RETURN_POINTER(rewrited);
}
tree = QT2QTN(GETQUERY(query), GETOPERAND(query));
QTNTernary(tree);
QTNSort(tree);
qex = QT2QTN(GETQUERY(ex), GETOPERAND(ex));
QTNTernary(qex);
QTNSort(qex);
if (subst->size)
subs = QT2QTN(GETQUERY(subst), GETOPERAND(subst));
tree = findsubquery(tree, qex, subs, NULL);
QTNFree(qex);
QTNFree(subs);
if (!tree)
{
SET_VARSIZE(rewrited, HDRSIZETQ);
rewrited->size = 0;
PG_FREE_IF_COPY(ex, 1);
PG_FREE_IF_COPY(subst, 2);
PG_RETURN_POINTER(rewrited);
}
else
{
QTNBinary(tree);
rewrited = QTN2QT(tree);
QTNFree(tree);
}
PG_FREE_IF_COPY(query, 0);
PG_FREE_IF_COPY(ex, 1);
PG_FREE_IF_COPY(subst, 2);
PG_RETURN_POINTER(rewrited);
}
Datum
tsquery_rewrite_query(PG_FUNCTION_ARGS)
{
QUERYTYPE *query = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(0)));
QUERYTYPE *ex = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_DATUM(1)));
QUERYTYPE *subst = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_DATUM(2)));
QUERYTYPE *rewrited = query;
QTNode *tree,
*qex,
*subs = NULL;
if (query->size == 0 || ex->size == 0)
{
PG_FREE_IF_COPY(ex, 1);
PG_FREE_IF_COPY(subst, 2);
PG_RETURN_POINTER(rewrited);
}
tree = QT2QTN(GETQUERY(query), GETOPERAND(query));
QTNTernary(tree);
QTNSort(tree);
qex = QT2QTN(GETQUERY(ex), GETOPERAND(ex));
QTNTernary(qex);
QTNSort(qex);
if (subst->size)
subs = QT2QTN(GETQUERY(subst), GETOPERAND(subst));
tree = findsubquery(tree, qex, PlainMemory, subs, NULL);
QTNFree(qex);
QTNFree(subs);
if (!tree)
{
rewrited->len = HDRSIZEQT;
rewrited->size = 0;
PG_FREE_IF_COPY(ex, 1);
PG_FREE_IF_COPY(subst, 2);
PG_RETURN_POINTER(rewrited);
}
else
{
QTNBinary(tree);
rewrited = QTN2QT(tree, PlainMemory);
QTNFree(tree);
}
PG_FREE_IF_COPY(query, 0);
PG_FREE_IF_COPY(ex, 1);
PG_FREE_IF_COPY(subst, 2);
PG_RETURN_POINTER(rewrited);
}
Datum
ts_match_vq(PG_FUNCTION_ARGS)
{
TSVector val = PG_GETARG_TSVECTOR(0);
TSQuery query = PG_GETARG_TSQUERY(1);
CHKVAL chkval;
bool result;
if (!val->size || !query->size)
{
PG_FREE_IF_COPY(val, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_BOOL(false);
}
chkval.arrb = ARRPTR(val);
chkval.arre = chkval.arrb + val->size;
chkval.values = STRPTR(val);
chkval.operand = GETOPERAND(query);
result = TS_execute(
GETQUERY(query),
&chkval,
true,
checkcondition_str
);
PG_FREE_IF_COPY(val, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_BOOL(result);
}
static void
hlfinditem(HeadlineParsedText *prs, TSQuery query, char *buf, int buflen)
{
int i;
QueryItem *item = GETQUERY(query);
HeadlineWordEntry *word;
while (prs->curwords + query->size >= prs->lenwords)
{
prs->lenwords *= 2;
prs->words = (HeadlineWordEntry *) repalloc((void *) prs->words, prs->lenwords * sizeof(HeadlineWordEntry));
}
word = &(prs->words[prs->curwords - 1]);
for (i = 0; i < query->size; i++)
{
if (item->type == QI_VAL &&
tsCompareString(GETOPERAND(query) + item->qoperand.distance, item->qoperand.length,
buf, buflen, item->qoperand.prefix) == 0)
{
if (word->item)
{
memcpy(&(prs->words[prs->curwords]), word, sizeof(HeadlineWordEntry));
prs->words[prs->curwords].item = &item->qoperand;
prs->words[prs->curwords].repeated = 1;
prs->curwords++;
}
else
word->item = &item->qoperand;
}
item++;
}
}
TSQuery
QTN2QT(QTNode *in)
{
TSQuery out;
int len;
int sumlen = 0,
nnode = 0;
QTN2QTState state;
cntsize(in, &sumlen, &nnode);
if (TSQUERY_TOO_BIG(nnode, sumlen))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("tsquery is too large")));
len = COMPUTESIZE(nnode, sumlen);
out = (TSQuery) palloc0(len);
SET_VARSIZE(out, len);
out->size = nnode;
state.curitem = GETQUERY(out);
state.operand = state.curoperand = GETOPERAND(out);
fillQT(&state, in);
return out;
}
Datum
tsquery_not(PG_FUNCTION_ARGS)
{
TSQuery a = PG_GETARG_TSQUERY_COPY(0);
QTNode *res;
TSQuery query;
if (a->size == 0)
PG_RETURN_POINTER(a);
res = (QTNode *) palloc0(sizeof(QTNode));
res->flags |= QTN_NEEDFREE;
res->valnode = (QueryItem *) palloc0(sizeof(QueryItem));
res->valnode->type = QI_OPR;
res->valnode->qoperator.oper = OP_NOT;
res->child = (QTNode **) palloc0(sizeof(QTNode *));
res->child[0] = QT2QTN(GETQUERY(a), GETOPERAND(a));
res->nchild = 1;
query = QTN2QT(res);
QTNFree(res);
PG_FREE_IF_COPY(a, 0);
PG_RETURN_POINTER(query);
}
Datum
exectsq(PG_FUNCTION_ARGS)
{
tsvector *val = (tsvector *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_DATUM(0)));
QUERYTYPE *query = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_DATUM(1)));
CHKVAL chkval;
bool result;
SET_FUNCOID();
if (!val->size || !query->size)
{
PG_FREE_IF_COPY(val, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_BOOL(false);
}
chkval.arrb = ARRPTR(val);
chkval.arre = chkval.arrb + val->size;
chkval.values = STRPTR(val);
chkval.operand = GETOPERAND(query);
result = TS_execute(
GETQUERY(query),
&chkval,
true,
checkcondition_str
);
PG_FREE_IF_COPY(val, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_BOOL(result);
}
static void
hlfinditem(HLPRSTEXT * prs, QUERYTYPE * query, char *buf, int buflen)
{
int i;
ITEM *item = GETQUERY(query);
HLWORD *word;
while (prs->curwords + query->size >= prs->lenwords)
{
prs->lenwords *= 2;
prs->words = (HLWORD *) repalloc((void *) prs->words, prs->lenwords * sizeof(HLWORD));
}
word = &(prs->words[prs->curwords - 1]);
for (i = 0; i < query->size; i++)
{
if (item->type == VAL && item->length == buflen && strncmp(GETOPERAND(query) + item->distance, buf, buflen) == 0)
{
if (word->item)
{
memcpy(&(prs->words[prs->curwords]), word, sizeof(HLWORD));
prs->words[prs->curwords].item = item;
prs->words[prs->curwords].repeated = 1;
prs->curwords++;
}
else
word->item = item;
}
item++;
}
}
static QTNode *join_tsqueries(TSQuery a, TSQuery b, int8 operator)
{
QTNode *res = (QTNode *) pzalloc(sizeof(QTNode));
res->flags |= QTN_NEEDFREE;
res->valnode = (QueryItem *) pzalloc(sizeof(QueryItem));
res->valnode->type = QI_OPR;
res->valnode->qoperator.oper = operator;
res->child = (QTNode **) pzalloc(sizeof(QTNode *) * 2);
res->child[0] = QT2QTN(GETQUERY(b), GETOPERAND(b));
res->child[1] = QT2QTN(GETQUERY(a), GETOPERAND(a));
res->nchild = 2;
return res;
}
Datum
gin_extract_tsquery(PG_FUNCTION_ARGS)
{
TSQuery query = PG_GETARG_TSQUERY(0);
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
StrategyNumber strategy = PG_GETARG_UINT16(2);
Datum *entries = NULL;
*nentries = 0;
if (query->size > 0)
{
int4 i,
j = 0,
len;
QueryItem *item;
item = clean_NOT(GETQUERY(query), &len);
if (!item)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("query requires full scan, which is not supported by GIN indexes")));
item = GETQUERY(query);
for (i = 0; i < query->size; i++)
if (item[i].type == QI_VAL)
(*nentries)++;
entries = (Datum *) palloc(sizeof(Datum) * (*nentries));
for (i = 0; i < query->size; i++)
if (item[i].type == QI_VAL)
{
text *txt;
QueryOperand *val = &item[i].operand;
txt = (text *) palloc(VARHDRSZ + val->length);
SET_VARSIZE(txt, VARHDRSZ + val->length);
memcpy(VARDATA(txt), GETOPERAND(query) + val->distance, val->length);
entries[j++] = PointerGetDatum(txt);
if (strategy != TSearchWithClassStrategyNumber && val->weight != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("@@ operator does not support lexeme weight restrictions in GIN index searches"),
errhint("Use the @@@ operator instead.")));
}
}
else
*nentries = -1; /* nothing can be found */
PG_FREE_IF_COPY(query, 0);
PG_RETURN_POINTER(entries);
}
/*
* Returns a pointer to a WordEntry's array corresponding to 'item' from
* tsvector 't'. 'q' is the TSQuery containing 'item'.
* Returns NULL if not found.
*/
static WordEntry *
find_wordentry(TSVector t, TSQuery q, QueryOperand *item, int32 *nitem)
{
WordEntry *StopLow = ARRPTR(t);
WordEntry *StopHigh = (WordEntry *) STRPTR(t);
WordEntry *StopMiddle = StopHigh;
int difference;
*nitem = 0;
/* Loop invariant: StopLow <= item < StopHigh */
while (StopLow < StopHigh)
{
StopMiddle = StopLow + (StopHigh - StopLow) / 2;
difference = WordECompareQueryItem(STRPTR(t), GETOPERAND(q), StopMiddle, item, false);
if (difference == 0)
{
StopHigh = StopMiddle;
*nitem = 1;
break;
}
else if (difference > 0)
StopLow = StopMiddle + 1;
else
StopHigh = StopMiddle;
}
if (item->prefix)
{
if (StopLow >= StopHigh)
StopMiddle = StopHigh;
*nitem = 0;
while (StopMiddle < (WordEntry *) STRPTR(t) &&
WordECompareQueryItem(STRPTR(t), GETOPERAND(q), StopMiddle, item, true) == 0)
{
(*nitem)++;
StopMiddle++;
}
}
return (*nitem > 0) ? StopHigh : NULL;
}
Datum
to_tsquery_byid(PG_FUNCTION_ARGS)
{
Oid cfgid = PG_GETARG_OID(0);
text *in = PG_GETARG_TEXT_P(1);
TSQuery query;
QueryItem *res;
int4 len;
query = parse_tsquery(text_to_cstring(in), pushval_morph, ObjectIdGetDatum(cfgid), false);
if (query->size == 0)
PG_RETURN_TSQUERY(query);
/* clean out any stopword placeholders from the tree */
res = clean_fakeval(GETQUERY(query), &len);
if (!res)
{
SET_VARSIZE(query, HDRSIZETQ);
query->size = 0;
PG_RETURN_POINTER(query);
}
memcpy((void *) GETQUERY(query), (void *) res, len * sizeof(QueryItem));
/*
* Removing the stopword placeholders might've resulted in fewer
* QueryItems. If so, move the operands up accordingly.
*/
if (len != query->size)
{
char *oldoperand = GETOPERAND(query);
int4 lenoperand = VARSIZE(query) - (oldoperand - (char *) query);
Assert(len < query->size);
query->size = len;
memmove((void *) GETOPERAND(query), oldoperand, VARSIZE(query) - (oldoperand - (char *) query));
SET_VARSIZE(query, COMPUTESIZE(len, lenoperand));
}
pfree(res);
PG_RETURN_TSQUERY(query);
}
static QTNode *
join_tsqueries(TSQuery a, TSQuery b, int8 operator, uint16 distance)
{
QTNode *res = (QTNode *) palloc0(sizeof(QTNode));
res->flags |= QTN_NEEDFREE;
res->valnode = (QueryItem *) palloc0(sizeof(QueryItem));
res->valnode->type = QI_OPR;
res->valnode->qoperator.oper = operator;
if (operator == OP_PHRASE)
res->valnode->qoperator.distance = distance;
res->child = (QTNode **) palloc0(sizeof(QTNode *) * 2);
res->child[0] = QT2QTN(GETQUERY(b), GETOPERAND(b));
res->child[1] = QT2QTN(GETQUERY(a), GETOPERAND(a));
res->nchild = 2;
return res;
}
Datum
plainto_tsquery_byid(PG_FUNCTION_ARGS)
{
Oid cfgid = PG_GETARG_OID(0);
text *in = PG_GETARG_TEXT_P(1);
TSQuery query;
QueryItem *res;
int4 len;
query = parse_tsquery(text_to_cstring(in), pushval_morph, ObjectIdGetDatum(cfgid), true);
if (query->size == 0)
PG_RETURN_TSQUERY(query);
res = clean_fakeval(GETQUERY(query), &len);
if (!res)
{
SET_VARSIZE(query, HDRSIZETQ);
query->size = 0;
PG_RETURN_POINTER(query);
}
memcpy((void *) GETQUERY(query), (void *) res, len * sizeof(QueryItem));
if (len != query->size)
{
char *oldoperand = GETOPERAND(query);
int4 lenoperand = VARSIZE(query) - (oldoperand - (char *) query);
Assert(len < query->size);
query->size = len;
memcpy((void *) GETOPERAND(query), oldoperand, lenoperand);
SET_VARSIZE(query, COMPUTESIZE(len, lenoperand));
}
pfree(res);
PG_RETURN_POINTER(query);
}
Datum
gin_extract_tsquery(PG_FUNCTION_ARGS)
{
QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
uint32 *nentries = (uint32 *) PG_GETARG_POINTER(1);
StrategyNumber strategy = DatumGetUInt16(PG_GETARG_DATUM(2));
Datum *entries = NULL;
*nentries = 0;
if (query->size > 0)
{
int4 i,
j = 0,
len;
ITEM *item;
item = clean_NOT_v2(GETQUERY(query), &len);
if (!item)
elog(ERROR, "Query requires full scan, GIN doesn't support it");
item = GETQUERY(query);
for (i = 0; i < query->size; i++)
if (item[i].type == VAL)
(*nentries)++;
entries = (Datum *) palloc(sizeof(Datum) * (*nentries));
for (i = 0; i < query->size; i++)
if (item[i].type == VAL)
{
text *txt;
txt = (text *) palloc(VARHDRSZ + item[i].length);
VARATT_SIZEP(txt) = VARHDRSZ + item[i].length;
memcpy(VARDATA(txt), GETOPERAND(query) + item[i].distance, item[i].length);
entries[j++] = PointerGetDatum(txt);
if (strategy == 1 && item[i].weight != 0)
elog(ERROR, "With class of lexeme restrictions use @@@ operation");
}
}
PG_FREE_IF_COPY(query, 0);
PG_RETURN_POINTER(entries);
}
/*
* Extract data from the pg_statistic arrays into useful format.
*/
static Selectivity
mcelem_tsquery_selec(TSQuery query, Datum *mcelem, int nmcelem,
float4 *numbers, int nnumbers)
{
float4 minfreq;
TextFreq *lookup;
Selectivity selec;
int i;
/*
* There should be two more Numbers than Values, because the last two
* cells are taken for minimal and maximal frequency. Punt if not.
*
* (Note: the MCELEM statistics slot definition allows for a third extra
* number containing the frequency of nulls, but we're not expecting that
* to appear for a tsvector column.)
*/
if (nnumbers != nmcelem + 2)
return tsquery_opr_selec_no_stats(query);
/*
* Transpose the data into a single array so we can use bsearch().
*/
lookup = (TextFreq *) palloc(sizeof(TextFreq) * nmcelem);
for (i = 0; i < nmcelem; i++)
{
/*
* The text Datums came from an array, so it cannot be compressed or
* stored out-of-line -- it's safe to use VARSIZE_ANY*.
*/
Assert(!VARATT_IS_COMPRESSED(mcelem[i]) && !VARATT_IS_EXTERNAL(mcelem[i]));
lookup[i].element = (text *) DatumGetPointer(mcelem[i]);
lookup[i].frequency = numbers[i];
}
/*
* Grab the lowest frequency. compute_tsvector_stats() stored it for us in
* the one before the last cell of the Numbers array. See ts_typanalyze.c
*/
minfreq = numbers[nnumbers - 2];
selec = tsquery_opr_selec(GETQUERY(query), GETOPERAND(query), lookup,
nmcelem, minfreq);
pfree(lookup);
return selec;
}
/*
* debug function, used only for view query
* which will be executed in non-leaf pages in index
*/
Datum
tsquerytree(PG_FUNCTION_ARGS)
{
QUERYTYPE *query = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_DATUM(0)));
INFIX nrm;
text *res;
ITEM *q;
int4 len;
if (query->size == 0)
{
res = (text *) palloc(VARHDRSZ);
VARATT_SIZEP(res) = VARHDRSZ;
PG_RETURN_POINTER(res);
}
q = clean_NOT_v2(GETQUERY(query), &len);
if (!q)
{
res = (text *) palloc(1 + VARHDRSZ);
VARATT_SIZEP(res) = 1 + VARHDRSZ;
*((char *) VARDATA(res)) = 'T';
}
else
{
nrm.curpol = q;
nrm.buflen = 32;
nrm.cur = nrm.buf = (char *) palloc(sizeof(char) * nrm.buflen);
*(nrm.cur) = '\0';
nrm.op = GETOPERAND(query);
infix(&nrm, true);
res = (text *) palloc(nrm.cur - nrm.buf + VARHDRSZ);
VARATT_SIZEP(res) = nrm.cur - nrm.buf + VARHDRSZ;
strncpy(VARDATA(res), nrm.buf, nrm.cur - nrm.buf);
pfree(q);
}
PG_FREE_IF_COPY(query, 0);
PG_RETURN_POINTER(res);
}
Datum
ltxtq_out(PG_FUNCTION_ARGS)
{
ltxtquery *query = PG_GETARG_LTXTQUERY(0);
INFIX nrm;
if (query->size == 0)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("syntax error"),
errdetail("Empty query.")));
nrm.curpol = GETQUERY(query);
nrm.buflen = 32;
nrm.cur = nrm.buf = (char *) palloc(sizeof(char) * nrm.buflen);
*(nrm.cur) = '\0';
nrm.op = GETOPERAND(query);
infix(&nrm, true);
PG_FREE_IF_COPY(query, 0);
PG_RETURN_POINTER(nrm.buf);
}
TSQuery
QTN2QT(QTNode *in)
{
TSQuery out;
int len;
int sumlen = 0,
nnode = 0;
QTN2QTState state;
cntsize(in, &sumlen, &nnode);
len = COMPUTESIZE(nnode, sumlen);
out = (TSQuery) palloc0(len);
SET_VARSIZE(out, len);
out->size = nnode;
state.curitem = GETQUERY(out);
state.operand = state.curoperand = GETOPERAND(out);
fillQT(&state, in);
return out;
}
Datum
ltxtq_exec(PG_FUNCTION_ARGS)
{
ltree *val = PG_GETARG_LTREE(0);
ltxtquery *query = PG_GETARG_LTXTQUERY(1);
CHKVAL chkval;
bool result;
chkval.node = val;
chkval.operand = GETOPERAND(query);
result = ltree_execute(
GETQUERY(query),
&chkval,
true,
checkcondition_str
);
PG_FREE_IF_COPY(val, 0);
PG_FREE_IF_COPY(query, 1);
PG_RETURN_BOOL(result);
}
Datum
tsquery_out(PG_FUNCTION_ARGS)
{
QUERYTYPE *query = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_DATUM(0)));
INFIX nrm;
if (query->size == 0)
{
char *b = palloc(1);
*b = '\0';
PG_RETURN_POINTER(b);
}
nrm.curpol = GETQUERY(query);
nrm.buflen = 32;
nrm.cur = nrm.buf = (char *) palloc(sizeof(char) * nrm.buflen);
*(nrm.cur) = '\0';
nrm.op = GETOPERAND(query);
infix(&nrm, true);
PG_FREE_IF_COPY(query, 0);
PG_RETURN_POINTER(nrm.buf);
}
static WordEntry *
find_wordentry(tsvector * t, QUERYTYPE * q, ITEM * item)
{
WordEntry *StopLow = ARRPTR(t);
WordEntry *StopHigh = (WordEntry *) STRPTR(t);
WordEntry *StopMiddle;
int difference;
/* Loop invariant: StopLow <= item < StopHigh */
while (StopLow < StopHigh)
{
StopMiddle = StopLow + (StopHigh - StopLow) / 2;
difference = WordECompareITEM(STRPTR(t), GETOPERAND(q), StopMiddle, item);
if (difference == 0)
return StopMiddle;
else if (difference < 0)
StopLow = StopMiddle + 1;
else
StopHigh = StopMiddle;
}
return NULL;
}
Datum
tsquery_rewrite(PG_FUNCTION_ARGS)
{
QUERYTYPE *query = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(0)));
text *in = PG_GETARG_TEXT_P(1);
QUERYTYPE *rewrited = query;
QTNode *tree;
char *buf;
void *plan;
Portal portal;
bool isnull;
int i;
if (query->size == 0)
{
PG_FREE_IF_COPY(in, 1);
PG_RETURN_POINTER(rewrited);
}
tree = QT2QTN(GETQUERY(query), GETOPERAND(query));
QTNTernary(tree);
QTNSort(tree);
buf = (char *) palloc(VARSIZE(in));
memcpy(buf, VARDATA(in), VARSIZE(in) - VARHDRSZ);
buf[VARSIZE(in) - VARHDRSZ] = '\0';
SPI_connect();
if (tsqOid == InvalidOid)
get_tsq_Oid();
if ((plan = SPI_prepare(buf, 0, NULL)) == NULL)
elog(ERROR, "SPI_prepare('%s') returns NULL", buf);
if ((portal = SPI_cursor_open(NULL, plan, NULL, NULL, false)) == NULL)
elog(ERROR, "SPI_cursor_open('%s') returns NULL", buf);
SPI_cursor_fetch(portal, true, 100);
if (SPI_tuptable->tupdesc->natts != 2)
elog(ERROR, "number of fields doesn't equal to 2");
if (SPI_gettypeid(SPI_tuptable->tupdesc, 1) != tsqOid)
elog(ERROR, "column #1 isn't of tsquery type");
if (SPI_gettypeid(SPI_tuptable->tupdesc, 2) != tsqOid)
elog(ERROR, "column #2 isn't of tsquery type");
while (SPI_processed > 0 && tree)
{
for (i = 0; i < SPI_processed && tree; i++)
{
Datum qdata = SPI_getbinval(SPI_tuptable->vals[i], SPI_tuptable->tupdesc, 1, &isnull);
Datum sdata;
if (isnull)
continue;
sdata = SPI_getbinval(SPI_tuptable->vals[i], SPI_tuptable->tupdesc, 2, &isnull);
if (!isnull)
{
QUERYTYPE *qtex = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(qdata));
QUERYTYPE *qtsubs = (QUERYTYPE *) DatumGetPointer(PG_DETOAST_DATUM(sdata));
QTNode *qex,
*qsubs = NULL;
if (qtex->size == 0)
{
if (qtex != (QUERYTYPE *) DatumGetPointer(qdata))
pfree(qtex);
if (qtsubs != (QUERYTYPE *) DatumGetPointer(sdata))
pfree(qtsubs);
continue;
}
qex = QT2QTN(GETQUERY(qtex), GETOPERAND(qtex));
QTNTernary(qex);
QTNSort(qex);
if (qtsubs->size)
qsubs = QT2QTN(GETQUERY(qtsubs), GETOPERAND(qtsubs));
tree = findsubquery(tree, qex, SPIMemory, qsubs, NULL);
QTNFree(qex);
if (qtex != (QUERYTYPE *) DatumGetPointer(qdata))
pfree(qtex);
QTNFree(qsubs);
if (qtsubs != (QUERYTYPE *) DatumGetPointer(sdata))
pfree(qtsubs);
}
}
SPI_freetuptable(SPI_tuptable);
SPI_cursor_fetch(portal, true, 100);
}
SPI_freetuptable(SPI_tuptable);
SPI_cursor_close(portal);
SPI_freeplan(plan);
SPI_finish();
if (tree)
{
QTNBinary(tree);
rewrited = QTN2QT(tree, PlainMemory);
QTNFree(tree);
PG_FREE_IF_COPY(query, 0);
}
else
{
rewrited->len = HDRSIZEQT;
rewrited->size = 0;
}
pfree(buf);
PG_FREE_IF_COPY(in, 1);
PG_RETURN_POINTER(rewrited);
}
Datum
rewrite_accum(PG_FUNCTION_ARGS)
{
QUERYTYPE *acc = (QUERYTYPE *) PG_GETARG_POINTER(0);
ArrayType *qa = (ArrayType *) DatumGetPointer(PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(1)));
QUERYTYPE *q;
QTNode *qex,
*subs = NULL,
*acctree;
bool isfind = false;
Datum *elemsp;
int nelemsp;
AggregateContext = ((AggState *) fcinfo->context)->aggcontext;
if (acc == NULL || PG_ARGISNULL(0))
{
acc = (QUERYTYPE *) MEMALLOC(AggMemory, sizeof(QUERYTYPE));
acc->len = HDRSIZEQT;
acc->size = 0;
}
if (qa == NULL || PG_ARGISNULL(1))
{
PG_FREE_IF_COPY(qa, 1);
PG_RETURN_POINTER(acc);
}
if (ARR_NDIM(qa) != 1)
elog(ERROR, "array must be one-dimensional, not %d dimension", ARR_NDIM(qa));
if (ArrayGetNItems(ARR_NDIM(qa), ARR_DIMS(qa)) != 3)
elog(ERROR, "array should have only three elements");
if (tsqOid == InvalidOid)
{
SPI_connect();
get_tsq_Oid();
SPI_finish();
}
if (ARR_ELEMTYPE(qa) != tsqOid)
elog(ERROR, "array should contain tsquery type");
deconstruct_array(qa, tsqOid, -1, false, 'i', &elemsp, NULL, &nelemsp);
q = (QUERYTYPE *) DatumGetPointer(elemsp[0]);
if (q->size == 0)
{
pfree(elemsp);
PG_RETURN_POINTER(acc);
}
if (!acc->size)
{
if (acc->len > HDRSIZEQT)
{
pfree(elemsp);
PG_RETURN_POINTER(acc);
}
else
acctree = QT2QTN(GETQUERY(q), GETOPERAND(q));
}
else
acctree = QT2QTN(GETQUERY(acc), GETOPERAND(acc));
QTNTernary(acctree);
QTNSort(acctree);
q = (QUERYTYPE *) DatumGetPointer(elemsp[1]);
if (q->size == 0)
{
pfree(elemsp);
PG_RETURN_POINTER(acc);
}
qex = QT2QTN(GETQUERY(q), GETOPERAND(q));
QTNTernary(qex);
QTNSort(qex);
q = (QUERYTYPE *) DatumGetPointer(elemsp[2]);
if (q->size)
subs = QT2QTN(GETQUERY(q), GETOPERAND(q));
acctree = findsubquery(acctree, qex, PlainMemory, subs, &isfind);
if (isfind || !acc->size)
{
/* pfree( acc ); do not pfree(p), because nodeAgg.c will */
if (acctree)
{
QTNBinary(acctree);
acc = QTN2QT(acctree, AggMemory);
}
else
{
acc = (QUERYTYPE *) MEMALLOC(AggMemory, HDRSIZEQT * 2);
acc->len = HDRSIZEQT * 2;
acc->size = 0;
}
}
pfree(elemsp);
QTNFree(qex);
QTNFree(subs);
QTNFree(acctree);
PG_RETURN_POINTER(acc);
}
Datum
gin_extract_tsquery(PG_FUNCTION_ARGS)
{
TSQuery query = PG_GETARG_TSQUERY(0);
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
/* StrategyNumber strategy = PG_GETARG_UINT16(2); */
bool **ptr_partialmatch = (bool **) PG_GETARG_POINTER(3);
Pointer **extra_data = (Pointer **) PG_GETARG_POINTER(4);
/* bool **nullFlags = (bool **) PG_GETARG_POINTER(5); */
int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
Datum *entries = NULL;
*nentries = 0;
if (query->size > 0)
{
QueryItem *item = GETQUERY(query);
int32 i,
j;
bool *partialmatch;
int *map_item_operand;
/*
* If the query doesn't have any required positive matches (for
* instance, it's something like '! foo'), we have to do a full index
* scan.
*/
if (tsquery_requires_match(item))
*searchMode = GIN_SEARCH_MODE_DEFAULT;
else
*searchMode = GIN_SEARCH_MODE_ALL;
/* count number of VAL items */
j = 0;
for (i = 0; i < query->size; i++)
{
if (item[i].type == QI_VAL)
j++;
}
*nentries = j;
entries = (Datum *) palloc(sizeof(Datum) * j);
partialmatch = *ptr_partialmatch = (bool *) palloc(sizeof(bool) * j);
/*
* Make map to convert item's number to corresponding operand's (the
* same, entry's) number. Entry's number is used in check array in
* consistent method. We use the same map for each entry.
*/
*extra_data = (Pointer *) palloc(sizeof(Pointer) * j);
map_item_operand = (int *) palloc0(sizeof(int) * query->size);
/* Now rescan the VAL items and fill in the arrays */
j = 0;
for (i = 0; i < query->size; i++)
{
if (item[i].type == QI_VAL)
{
QueryOperand *val = &item[i].qoperand;
text *txt;
txt = cstring_to_text_with_len(GETOPERAND(query) + val->distance,
val->length);
entries[j] = PointerGetDatum(txt);
partialmatch[j] = val->prefix;
(*extra_data)[j] = (Pointer) map_item_operand;
map_item_operand[i] = j;
j++;
}
}
}
PG_FREE_IF_COPY(query, 0);
PG_RETURN_POINTER(entries);
}
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;
}
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;
}
static float
calc_rank_and(float *w, tsvector * t, QUERYTYPE * q)
{
uint16 **pos;
int i,
k,
l,
p;
WordEntry *entry;
WordEntryPos *post,
*ct;
int4 dimt,
lenct,
dist;
float res = -1.0;
ITEM **item;
int size = q->size;
item = SortAndUniqItems(GETOPERAND(q), GETQUERY(q), &size);
if (size < 2)
{
pfree(item);
return calc_rank_or(w, t, q);
}
pos = (uint16 **) palloc(sizeof(uint16 *) * q->size);
memset(pos, 0, sizeof(uint16 *) * q->size);
*(uint16 *) POSNULL = lengthof(POSNULL) - 1;
WEP_SETPOS(POSNULL[1], MAXENTRYPOS - 1);
for (i = 0; i < size; i++)
{
entry = find_wordentry(t, q, item[i]);
if (!entry)
continue;
if (entry->haspos)
pos[i] = (uint16 *) _POSDATAPTR(t, entry);
else
pos[i] = (uint16 *) POSNULL;
dimt = *(uint16 *) (pos[i]);
post = (WordEntryPos *) (pos[i] + 1);
for (k = 0; k < i; k++)
{
if (!pos[k])
continue;
lenct = *(uint16 *) (pos[k]);
ct = (WordEntryPos *) (pos[k] + 1);
for (l = 0; l < dimt; l++)
{
for (p = 0; p < lenct; p++)
{
dist = Abs((int) WEP_GETPOS(post[l]) - (int) WEP_GETPOS(ct[p]));
if (dist || (dist == 0 && (pos[i] == (uint16 *) POSNULL || pos[k] == (uint16 *) POSNULL)))
{
float curw;
if (!dist)
dist = MAXENTRYPOS;
curw = sqrt(wpos(post[l]) * wpos(ct[p]) * word_distance(dist));
res = (res < 0) ? curw : 1.0 - (1.0 - res) * (1.0 - curw);
}
}
}
}
}
pfree(pos);
pfree(item);
return res;
}
