mirror of https://github.com/postgres/postgres
Replace plain-memory ordered array by binary tree in ts_stat() function.
Performance is increased from 50% up to 10^3 times depending on data.
This commit is contained in:
Teodor Sigaev
parent
18004101ac
commit
25ca5a9a54
|
|
@ -7,7 +7,7 @@
|
|||
*
|
||||
*
|
||||
* IDENTIFICATION
|
||||
* $PostgreSQL: pgsql/src/backend/utils/adt/tsvector_op.c,v 1.17 2008/11/10 21:49:16 alvherre Exp $
|
||||
* $PostgreSQL: pgsql/src/backend/utils/adt/tsvector_op.c,v 1.18 2008/11/17 12:17:09 teodor Exp $
|
||||
*
|
||||
*-------------------------------------------------------------------------
|
||||
*/
|
||||
|
|
@ -34,34 +34,33 @@ typedef struct
|
|||
char *operand;
|
||||
} CHKVAL;
|
||||
|
||||
typedef struct
|
||||
{
|
||||
uint32 cur;
|
||||
TSVector stat;
|
||||
} StatStorage;
|
||||
|
||||
typedef struct
|
||||
typedef struct StatEntry
|
||||
{
|
||||
uint32 len;
|
||||
uint32 pos;
|
||||
uint32 ndoc;
|
||||
uint32 ndoc; /* zero indicates that we already was here while
|
||||
walking throug the tree */
|
||||
uint32 nentry;
|
||||
struct StatEntry *left;
|
||||
struct StatEntry *right;
|
||||
uint32 lenlexeme;
|
||||
char lexeme[1];
|
||||
} StatEntry;
|
||||
|
||||
#define STATENTRYHDRSZ (offsetof(StatEntry, lexeme))
|
||||
|
||||
typedef struct
|
||||
{
|
||||
int32 vl_len_; /* varlena header (do not touch directly!) */
|
||||
int4 size;
|
||||
int4 weight;
|
||||
char data[1];
|
||||
} tsstat;
|
||||
|
||||
#define STATHDRSIZE (sizeof(int4) * 4)
|
||||
#define CALCSTATSIZE(x, lenstr) ( (x) * sizeof(StatEntry) + STATHDRSIZE + (lenstr) )
|
||||
#define STATPTR(x) ( (StatEntry*) ( (char*)(x) + STATHDRSIZE ) )
|
||||
#define STATSTRPTR(x) ( (char*)(x) + STATHDRSIZE + ( sizeof(StatEntry) * ((TSVector)(x))->size ) )
|
||||
#define STATSTRSIZE(x) ( VARSIZE((TSVector)(x)) - STATHDRSIZE - ( sizeof(StatEntry) * ((TSVector)(x))->size ) )
|
||||
uint32 maxdepth;
|
||||
|
||||
StatEntry **stack;
|
||||
uint32 stackpos;
|
||||
|
||||
StatEntry* root;
|
||||
} TSVectorStat;
|
||||
|
||||
#define STATHDRSIZE (offsetof(TSVectorStat, data))
|
||||
|
||||
static Datum tsvector_update_trigger(PG_FUNCTION_ARGS, bool config_column);
|
||||
|
||||
|
|
@ -801,92 +800,95 @@ check_weight(TSVector txt, WordEntry *wptr, int8 weight)
|
|||
return num;
|
||||
}
|
||||
|
||||
#define compareStatWord(a,e,s,t) \
|
||||
tsCompareString(STATSTRPTR(s) + (a)->pos, (a)->len, \
|
||||
#define compareStatWord(a,e,t) \
|
||||
tsCompareString((a)->lexeme, (a)->lenlexeme, \
|
||||
STRPTR(t) + (e)->pos, (e)->len, \
|
||||
false)
|
||||
|
||||
typedef struct WordEntryMark
|
||||
static void
|
||||
insertStatEntry(MemoryContext persistentContext, TSVectorStat *stat, TSVector txt, uint32 off)
|
||||
{
|
||||
WordEntry *newentry;
|
||||
StatEntry *pos;
|
||||
} WordEntryMark;
|
||||
WordEntry *we = ARRPTR(txt) + off;
|
||||
StatEntry *node = stat->root,
|
||||
*pnode=NULL;
|
||||
int n,
|
||||
res;
|
||||
uint32 depth=1;
|
||||
|
||||
static tsstat *
|
||||
formstat(tsstat *stat, TSVector txt, List *entries)
|
||||
{
|
||||
tsstat *newstat;
|
||||
uint32 totallen,
|
||||
nentry,
|
||||
len = list_length(entries);
|
||||
uint32 slen = 0;
|
||||
WordEntry *ptr;
|
||||
char *curptr;
|
||||
StatEntry *sptr,
|
||||
*nptr;
|
||||
ListCell *entry;
|
||||
StatEntry *PosSE = STATPTR(stat),
|
||||
*prevPosSE;
|
||||
WordEntryMark *mark;
|
||||
if (stat->weight == 0)
|
||||
n = (we->haspos) ? POSDATALEN(txt, we) : 1;
|
||||
else
|
||||
n = (we->haspos) ? check_weight(txt, we, stat->weight) : 0;
|
||||
|
||||
foreach( entry, entries )
|
||||
if ( n == 0 )
|
||||
return; /* nothing to insert */
|
||||
|
||||
while( node )
|
||||
{
|
||||
mark = (WordEntryMark*)lfirst(entry);
|
||||
slen += mark->newentry->len;
|
||||
}
|
||||
res = compareStatWord(node, we, txt);
|
||||
|
||||
nentry = stat->size + len;
|
||||
slen += STATSTRSIZE(stat);
|
||||
totallen = CALCSTATSIZE(nentry, slen);
|
||||
newstat = palloc(totallen);
|
||||
SET_VARSIZE(newstat, totallen);
|
||||
newstat->weight = stat->weight;
|
||||
newstat->size = nentry;
|
||||
|
||||
memcpy(STATSTRPTR(newstat), STATSTRPTR(stat), STATSTRSIZE(stat));
|
||||
curptr = STATSTRPTR(newstat) + STATSTRSIZE(stat);
|
||||
|
||||
sptr = STATPTR(stat);
|
||||
nptr = STATPTR(newstat);
|
||||
|
||||
foreach(entry, entries)
|
||||
{
|
||||
prevPosSE = PosSE;
|
||||
|
||||
mark = (WordEntryMark*)lfirst(entry);
|
||||
ptr = mark->newentry;
|
||||
PosSE = mark->pos;
|
||||
|
||||
/*
|
||||
* Copy missed entries
|
||||
*/
|
||||
if ( PosSE > prevPosSE )
|
||||
if (res == 0)
|
||||
{
|
||||
memcpy( nptr, prevPosSE, sizeof(StatEntry) * (PosSE-prevPosSE) );
|
||||
nptr += PosSE-prevPosSE;
|
||||
break;
|
||||
}
|
||||
|
||||
/*
|
||||
* Copy new entry
|
||||
*/
|
||||
if (ptr->haspos)
|
||||
nptr->nentry = (stat->weight) ? check_weight(txt, ptr, stat->weight) : POSDATALEN(txt, ptr);
|
||||
else
|
||||
nptr->nentry = 1;
|
||||
nptr->ndoc = 1;
|
||||
nptr->len = ptr->len;
|
||||
memcpy(curptr, STRPTR(txt) + ptr->pos, nptr->len);
|
||||
nptr->pos = curptr - STATSTRPTR(newstat);
|
||||
curptr += nptr->len;
|
||||
nptr++;
|
||||
|
||||
pfree(mark);
|
||||
{
|
||||
pnode = node;
|
||||
node = ( res < 0 ) ? node->left : node->right;
|
||||
}
|
||||
depth++;
|
||||
}
|
||||
|
||||
if ( PosSE < (StatEntry *) STATSTRPTR(stat) )
|
||||
memcpy(nptr, PosSE, sizeof(StatEntry) * (stat->size - (PosSE - STATPTR(stat))));
|
||||
if (depth > stat->maxdepth)
|
||||
stat->maxdepth = depth;
|
||||
|
||||
return newstat;
|
||||
if (node == NULL)
|
||||
{
|
||||
node = MemoryContextAlloc(persistentContext, STATENTRYHDRSZ + we->len );
|
||||
node->left = node->right = NULL;
|
||||
node->ndoc = 1;
|
||||
node->nentry = n;
|
||||
node->lenlexeme = we->len;
|
||||
memcpy(node->lexeme, STRPTR(txt) + we->pos, node->lenlexeme);
|
||||
|
||||
if ( pnode==NULL )
|
||||
{
|
||||
stat->root = node;
|
||||
}
|
||||
else
|
||||
{
|
||||
if (res < 0)
|
||||
pnode->left = node;
|
||||
else
|
||||
pnode->right = node;
|
||||
}
|
||||
|
||||
}
|
||||
else
|
||||
{
|
||||
node->ndoc++;
|
||||
node->nentry += n;
|
||||
}
|
||||
}
|
||||
|
||||
static void
|
||||
chooseNextStatEntry(MemoryContext persistentContext, TSVectorStat *stat, TSVector txt,
|
||||
uint32 low, uint32 high, uint32 offset)
|
||||
{
|
||||
uint32 pos;
|
||||
uint32 middle = (low + high) >> 1;
|
||||
|
||||
pos = (low + middle) >> 1;
|
||||
if (low != middle && pos >= offset && pos - offset < txt->size)
|
||||
insertStatEntry( persistentContext, stat, txt, pos - offset );
|
||||
pos = (high + middle + 1) >> 1;
|
||||
if (middle + 1 != high && pos >= offset && pos - offset < txt->size)
|
||||
insertStatEntry( persistentContext, stat, txt, pos - offset );
|
||||
|
||||
if (low != middle)
|
||||
chooseNextStatEntry(persistentContext, stat, txt, low, middle, offset);
|
||||
if (high != middle + 1)
|
||||
chooseNextStatEntry(persistentContext, stat, txt, middle + 1, high, offset);
|
||||
}
|
||||
|
||||
/*
|
||||
|
|
@ -901,115 +903,69 @@ formstat(tsstat *stat, TSVector txt, List *entries)
|
|||
* where vector_column is a tsvector-type column in vector_table.
|
||||
*/
|
||||
|
||||
static tsstat *
|
||||
ts_accum(tsstat *stat, Datum data)
|
||||
static TSVectorStat *
|
||||
ts_accum(MemoryContext persistentContext, TSVectorStat *stat, Datum data)
|
||||
{
|
||||
tsstat *newstat;
|
||||
TSVector txt = DatumGetTSVector(data);
|
||||
StatEntry *sptr;
|
||||
WordEntry *wptr;
|
||||
int n = 0;
|
||||
List *newentries=NIL;
|
||||
StatEntry *StopLow;
|
||||
TSVector txt = DatumGetTSVector(data);
|
||||
uint32 i,
|
||||
nbit = 0,
|
||||
offset;
|
||||
|
||||
if (stat == NULL)
|
||||
{ /* Init in first */
|
||||
stat = palloc(STATHDRSIZE);
|
||||
SET_VARSIZE(stat, STATHDRSIZE);
|
||||
stat->size = 0;
|
||||
stat->weight = 0;
|
||||
{ /* Init in first */
|
||||
stat = MemoryContextAllocZero(persistentContext, sizeof(TSVectorStat));
|
||||
stat->maxdepth = 1;
|
||||
}
|
||||
|
||||
/* simple check of correctness */
|
||||
if (txt == NULL || txt->size == 0)
|
||||
{
|
||||
if (txt != (TSVector) DatumGetPointer(data))
|
||||
if (txt && txt != (TSVector) DatumGetPointer(data))
|
||||
pfree(txt);
|
||||
return stat;
|
||||
}
|
||||
|
||||
sptr = STATPTR(stat);
|
||||
wptr = ARRPTR(txt);
|
||||
StopLow = STATPTR(stat);
|
||||
i = txt->size - 1;
|
||||
for (; i > 0; i >>= 1)
|
||||
nbit++;
|
||||
|
||||
while (wptr - ARRPTR(txt) < txt->size)
|
||||
{
|
||||
StatEntry *StopHigh = (StatEntry *) STATSTRPTR(stat);
|
||||
int cmp;
|
||||
nbit = 1 << nbit;
|
||||
offset = (nbit - txt->size) / 2;
|
||||
|
||||
/*
|
||||
* We do not set StopLow to begin of array because tsvector is ordered
|
||||
* with the sames rule, so we can search from last stopped position
|
||||
*/
|
||||
insertStatEntry( persistentContext, stat, txt, (nbit >> 1) - offset );
|
||||
chooseNextStatEntry(persistentContext, stat, txt, 0, nbit, offset);
|
||||
|
||||
while (StopLow < StopHigh)
|
||||
{
|
||||
sptr = StopLow + (StopHigh - StopLow) / 2;
|
||||
cmp = compareStatWord(sptr, wptr, stat, txt);
|
||||
if (cmp == 0)
|
||||
{
|
||||
if (stat->weight == 0)
|
||||
{
|
||||
sptr->ndoc++;
|
||||
sptr->nentry += (wptr->haspos) ? POSDATALEN(txt, wptr) : 1;
|
||||
}
|
||||
else if (wptr->haspos && (n = check_weight(txt, wptr, stat->weight)) != 0)
|
||||
{
|
||||
sptr->ndoc++;
|
||||
sptr->nentry += n;
|
||||
}
|
||||
break;
|
||||
}
|
||||
else if (cmp < 0)
|
||||
StopLow = sptr + 1;
|
||||
else
|
||||
StopHigh = sptr;
|
||||
}
|
||||
|
||||
if (StopLow >= StopHigh)
|
||||
{ /* not found */
|
||||
if (stat->weight == 0 || check_weight(txt, wptr, stat->weight) != 0)
|
||||
{
|
||||
WordEntryMark *mark = (WordEntryMark*)palloc(sizeof(WordEntryMark));
|
||||
|
||||
mark->newentry = wptr;
|
||||
mark->pos = StopLow;
|
||||
newentries = lappend( newentries, mark );
|
||||
|
||||
}
|
||||
}
|
||||
wptr++;
|
||||
}
|
||||
|
||||
if (list_length(newentries) == 0)
|
||||
{ /* no new words */
|
||||
if (txt != (TSVector) DatumGetPointer(data))
|
||||
pfree(txt);
|
||||
return stat;
|
||||
}
|
||||
|
||||
newstat = formstat(stat, txt, newentries);
|
||||
list_free(newentries);
|
||||
|
||||
if (txt != (TSVector) DatumGetPointer(data))
|
||||
pfree(txt);
|
||||
return newstat;
|
||||
return stat;
|
||||
}
|
||||
|
||||
static void
|
||||
ts_setup_firstcall(FunctionCallInfo fcinfo, FuncCallContext *funcctx,
|
||||
tsstat *stat)
|
||||
TSVectorStat *stat)
|
||||
{
|
||||
TupleDesc tupdesc;
|
||||
MemoryContext oldcontext;
|
||||
StatStorage *st;
|
||||
TupleDesc tupdesc;
|
||||
MemoryContext oldcontext;
|
||||
StatEntry *node;
|
||||
|
||||
funcctx->user_fctx = (void *) stat;
|
||||
|
||||
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
|
||||
st = palloc(sizeof(StatStorage));
|
||||
st->cur = 0;
|
||||
st->stat = palloc(VARSIZE(stat));
|
||||
memcpy(st->stat, stat, VARSIZE(stat));
|
||||
funcctx->user_fctx = (void *) st;
|
||||
|
||||
stat->stack = palloc0(sizeof(StatEntry *) * (stat->maxdepth + 1));
|
||||
stat->stackpos = 0;
|
||||
|
||||
node = stat->root;
|
||||
/* find leftmost value */
|
||||
for (;;)
|
||||
{
|
||||
stat->stack[ stat->stackpos ] = node;
|
||||
if (node->left)
|
||||
{
|
||||
stat->stackpos++;
|
||||
node = node->left;
|
||||
}
|
||||
else
|
||||
break;
|
||||
}
|
||||
|
||||
tupdesc = CreateTemplateTupleDesc(3, false);
|
||||
TupleDescInitEntry(tupdesc, (AttrNumber) 1, "word",
|
||||
|
|
@ -1024,26 +980,72 @@ ts_setup_firstcall(FunctionCallInfo fcinfo, FuncCallContext *funcctx,
|
|||
MemoryContextSwitchTo(oldcontext);
|
||||
}
|
||||
|
||||
static StatEntry *
|
||||
walkStatEntryTree(TSVectorStat *stat)
|
||||
{
|
||||
StatEntry *node = stat->stack[ stat->stackpos ];
|
||||
|
||||
if ( node == NULL )
|
||||
return NULL;
|
||||
|
||||
if ( node->ndoc != 0 )
|
||||
{
|
||||
/* return entry itself: we already was at left sublink */
|
||||
return node;
|
||||
}
|
||||
else if (node->right && node->right != stat->stack[stat->stackpos + 1])
|
||||
{
|
||||
/* go on right sublink */
|
||||
stat->stackpos++;
|
||||
node = node->right;
|
||||
|
||||
/* find most-left value */
|
||||
for (;;)
|
||||
{
|
||||
stat->stack[stat->stackpos] = node;
|
||||
if (node->left)
|
||||
{
|
||||
stat->stackpos++;
|
||||
node = node->left;
|
||||
}
|
||||
else
|
||||
break;
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
/* we already return all left subtree, itself and right subtree */
|
||||
if (stat->stackpos == 0)
|
||||
return NULL;
|
||||
|
||||
stat->stackpos--;
|
||||
return walkStatEntryTree(stat);
|
||||
}
|
||||
|
||||
return node;
|
||||
}
|
||||
|
||||
static Datum
|
||||
ts_process_call(FuncCallContext *funcctx)
|
||||
{
|
||||
StatStorage *st;
|
||||
TSVectorStat *st;
|
||||
StatEntry *entry;
|
||||
|
||||
st = (StatStorage *) funcctx->user_fctx;
|
||||
st = (TSVectorStat *) funcctx->user_fctx;
|
||||
|
||||
if (st->cur < st->stat->size)
|
||||
entry = walkStatEntryTree(st);
|
||||
|
||||
if (entry != NULL)
|
||||
{
|
||||
Datum result;
|
||||
char *values[3];
|
||||
char ndoc[16];
|
||||
char nentry[16];
|
||||
StatEntry *entry = STATPTR(st->stat) + st->cur;
|
||||
HeapTuple tuple;
|
||||
|
||||
values[0] = palloc(entry->len + 1);
|
||||
memcpy(values[0], STATSTRPTR(st->stat) + entry->pos, entry->len);
|
||||
(values[0])[entry->len] = '\0';
|
||||
values[0] = palloc(entry->lenlexeme + 1);
|
||||
memcpy(values[0], entry->lexeme, entry->lenlexeme);
|
||||
(values[0])[entry->lenlexeme] = '\0';
|
||||
sprintf(ndoc, "%d", entry->ndoc);
|
||||
values[1] = ndoc;
|
||||
sprintf(nentry, "%d", entry->nentry);
|
||||
|
|
@ -1053,25 +1055,22 @@ ts_process_call(FuncCallContext *funcctx)
|
|||
result = HeapTupleGetDatum(tuple);
|
||||
|
||||
pfree(values[0]);
|
||||
st->cur++;
|
||||
|
||||
/* mark entry as already visited */
|
||||
entry->ndoc = 0;
|
||||
|
||||
return result;
|
||||
}
|
||||
else
|
||||
{
|
||||
pfree(st->stat);
|
||||
pfree(st);
|
||||
}
|
||||
|
||||
return (Datum) 0;
|
||||
}
|
||||
|
||||
static tsstat *
|
||||
ts_stat_sql(text *txt, text *ws)
|
||||
static TSVectorStat *
|
||||
ts_stat_sql(MemoryContext persistentContext, text *txt, text *ws)
|
||||
{
|
||||
char *query = text_to_cstring(txt);
|
||||
int i;
|
||||
tsstat *newstat,
|
||||
*stat;
|
||||
TSVectorStat *stat;
|
||||
bool isnull;
|
||||
Portal portal;
|
||||
SPIPlanPtr plan;
|
||||
|
|
@ -1094,10 +1093,8 @@ ts_stat_sql(text *txt, text *ws)
|
|||
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
|
||||
errmsg("ts_stat query must return one tsvector column")));
|
||||
|
||||
stat = palloc(STATHDRSIZE);
|
||||
SET_VARSIZE(stat, STATHDRSIZE);
|
||||
stat->size = 0;
|
||||
stat->weight = 0;
|
||||
stat = MemoryContextAllocZero(persistentContext, sizeof(TSVectorStat));
|
||||
stat->maxdepth = 1;
|
||||
|
||||
if (ws)
|
||||
{
|
||||
|
|
@ -1141,12 +1138,7 @@ ts_stat_sql(text *txt, text *ws)
|
|||
Datum data = SPI_getbinval(SPI_tuptable->vals[i], SPI_tuptable->tupdesc, 1, &isnull);
|
||||
|
||||
if (!isnull)
|
||||
{
|
||||
newstat = ts_accum(stat, data);
|
||||
if (stat != newstat && stat)
|
||||
pfree(stat);
|
||||
stat = newstat;
|
||||
}
|
||||
stat = ts_accum(persistentContext, stat, data);
|
||||
}
|
||||
|
||||
SPI_freetuptable(SPI_tuptable);
|
||||
|
|
@ -1169,12 +1161,12 @@ ts_stat1(PG_FUNCTION_ARGS)
|
|||
|
||||
if (SRF_IS_FIRSTCALL())
|
||||
{
|
||||
tsstat *stat;
|
||||
TSVectorStat *stat;
|
||||
text *txt = PG_GETARG_TEXT_P(0);
|
||||
|
||||
funcctx = SRF_FIRSTCALL_INIT();
|
||||
SPI_connect();
|
||||
stat = ts_stat_sql(txt, NULL);
|
||||
stat = ts_stat_sql(funcctx->multi_call_memory_ctx, txt, NULL);
|
||||
PG_FREE_IF_COPY(txt, 0);
|
||||
ts_setup_firstcall(fcinfo, funcctx, stat);
|
||||
SPI_finish();
|
||||
|
|
@ -1194,13 +1186,13 @@ ts_stat2(PG_FUNCTION_ARGS)
|
|||
|
||||
if (SRF_IS_FIRSTCALL())
|
||||
{
|
||||
tsstat *stat;
|
||||
TSVectorStat *stat;
|
||||
text *txt = PG_GETARG_TEXT_P(0);
|
||||
text *ws = PG_GETARG_TEXT_P(1);
|
||||
|
||||
funcctx = SRF_FIRSTCALL_INIT();
|
||||
SPI_connect();
|
||||
stat = ts_stat_sql(txt, ws);
|
||||
stat = ts_stat_sql(funcctx->multi_call_memory_ctx, txt, ws);
|
||||
PG_FREE_IF_COPY(txt, 0);
|
||||
PG_FREE_IF_COPY(ws, 1);
|
||||
ts_setup_firstcall(fcinfo, funcctx, stat);
|
||||
|
|
|
|||
Loading…
Reference in New Issue