/*
* EnumValuesCreate
* Create an entry in pg_enum for each of the supplied enum values.
*
* vals is a list of Value strings.
*/
void
EnumValuesCreate(Oid enumTypeOid, List *vals)
{
Relation pg_enum;
NameData enumlabel;
Oid *oids;
int elemno,
num_elems;
Datum values[Natts_pg_enum];
bool nulls[Natts_pg_enum];
ListCell *lc;
HeapTuple tup;
num_elems = list_length(vals);
/*
* We do not bother to check the list of values for duplicates --- if
* you have any, you'll get a less-than-friendly unique-index violation.
* It is probably not worth trying harder.
*/
pg_enum = heap_open(EnumRelationId, RowExclusiveLock);
/*
* Allocate OIDs for the enum's members.
*
* While this method does not absolutely guarantee that we generate no
* duplicate OIDs (since we haven't entered each oid into the table
* before allocating the next), trouble could only occur if the OID
* counter wraps all the way around before we finish. Which seems
* unlikely.
*/
oids = (Oid *) palloc(num_elems * sizeof(Oid));
for (elemno = 0; elemno < num_elems; elemno++)
{
/*
* We assign even-numbered OIDs to all the new enum labels. This
* tells the comparison functions the OIDs are in the correct sort
* order and can be compared directly.
*/
Oid new_oid;
do {
new_oid = GetNewOid(pg_enum);
} while (new_oid & 1);
oids[elemno] = new_oid;
}
/* sort them, just in case OID counter wrapped from high to low */
qsort(oids, num_elems, sizeof(Oid), oid_cmp);
/* and make the entries */
memset(nulls, false, sizeof(nulls));
elemno = 0;
foreach(lc, vals)
{
char *lab = strVal(lfirst(lc));
/*
* labels are stored in a name field, for easier syscache lookup, so
* check the length to make sure it's within range.
*/
if (strlen(lab) > (NAMEDATALEN - 1))
ereport(ERROR,
(errcode(ERRCODE_INVALID_NAME),
errmsg("invalid enum label \"%s\"", lab),
errdetail("Labels must be %d characters or less.",
NAMEDATALEN - 1)));
values[Anum_pg_enum_enumtypid - 1] = ObjectIdGetDatum(enumTypeOid);
values[Anum_pg_enum_enumsortorder - 1] = Float4GetDatum(elemno + 1);
namestrcpy(&enumlabel, lab);
values[Anum_pg_enum_enumlabel - 1] = NameGetDatum(&enumlabel);
tup = heap_form_tuple(RelationGetDescr(pg_enum), values, nulls);
HeapTupleSetOid(tup, oids[elemno]);
simple_heap_insert(pg_enum, tup);
CatalogUpdateIndexes(pg_enum, tup);
heap_freetuple(tup);
elemno++;
}
void
worker_spi_main(Datum main_arg)
{
int index = DatumGetInt32(main_arg);
worktable *table;
StringInfoData buf;
char name[20];
table = palloc(sizeof(worktable));
sprintf(name, "schema%d", index);
table->schema = pstrdup(name);
table->name = pstrdup("counted");
/* Establish signal handlers before unblocking signals. */
pqsignal(SIGHUP, worker_spi_sighup);
pqsignal(SIGTERM, worker_spi_sigterm);
/* We're now ready to receive signals */
BackgroundWorkerUnblockSignals();
/* Connect to our database */
BackgroundWorkerInitializeConnection("postgres", NULL);
elog(LOG, "%s initialized with %s.%s",
MyBgworkerEntry->bgw_name, table->schema, table->name);
initialize_worker_spi(table);
/*
* Quote identifiers passed to us. Note that this must be done after
* initialize_worker_spi, because that routine assumes the names are not
* quoted.
*
* Note some memory might be leaked here.
*/
table->schema = quote_identifier(table->schema);
table->name = quote_identifier(table->name);
initStringInfo(&buf);
appendStringInfo(&buf,
"WITH deleted AS (DELETE "
"FROM %s.%s "
"WHERE type = 'delta' RETURNING value), "
"total AS (SELECT coalesce(sum(value), 0) as sum "
"FROM deleted) "
"UPDATE %s.%s "
"SET value = %s.value + total.sum "
"FROM total WHERE type = 'total' "
"RETURNING %s.value",
table->schema, table->name,
table->schema, table->name,
table->name,
table->name);
/*
* Main loop: do this until the SIGTERM handler tells us to terminate
*/
while (!got_sigterm)
{
int ret;
int rc;
/*
* Background workers mustn't call usleep() or any direct equivalent:
* instead, they may wait on their process latch, which sleeps as
* necessary, but is awakened if postmaster dies. That way the
* background process goes away immediately in an emergency.
*/
rc = WaitLatch(&MyProc->procLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
worker_spi_naptime * 1000L);
ResetLatch(&MyProc->procLatch);
/* emergency bailout if postmaster has died */
if (rc & WL_POSTMASTER_DEATH)
proc_exit(1);
/*
* In case of a SIGHUP, just reload the configuration.
*/
if (got_sighup)
{
got_sighup = false;
ProcessConfigFile(PGC_SIGHUP);
}
/*
* Start a transaction on which we can run queries. Note that each
* StartTransactionCommand() call should be preceded by a
* SetCurrentStatementStartTimestamp() call, which sets both the time
* for the statement we're about the run, and also the transaction
* start time. Also, each other query sent to SPI should probably be
* preceded by SetCurrentStatementStartTimestamp(), so that statement
* start time is always up to date.
*
* The SPI_connect() call lets us run queries through the SPI manager,
* and the PushActiveSnapshot() call creates an "active" snapshot
* which is necessary for queries to have MVCC data to work on.
*
* The pgstat_report_activity() call makes our activity visible
* through the pgstat views.
*/
SetCurrentStatementStartTimestamp();
StartTransactionCommand();
SPI_connect();
PushActiveSnapshot(GetTransactionSnapshot());
pgstat_report_activity(STATE_RUNNING, buf.data);
/* We can now execute queries via SPI */
ret = SPI_execute(buf.data, false, 0);
if (ret != SPI_OK_UPDATE_RETURNING)
elog(FATAL, "cannot select from table %s.%s: error code %d",
table->schema, table->name, ret);
if (SPI_processed > 0)
{
bool isnull;
int32 val;
val = DatumGetInt32(SPI_getbinval(SPI_tuptable->vals[0],
SPI_tuptable->tupdesc,
1, &isnull));
if (!isnull)
elog(LOG, "%s: count in %s.%s is now %d",
MyBgworkerEntry->bgw_name,
table->schema, table->name, val);
}
/*
* And finish our transaction.
*/
SPI_finish();
PopActiveSnapshot();
CommitTransactionCommand();
pgstat_report_activity(STATE_IDLE, NULL);
}
proc_exit(1);
}
Datum
palloc_bench(PG_FUNCTION_ARGS)
{
/* info for anyelement */
int i = 0;
int32 ncontexts = PG_GETARG_INT32(0);
int32 niterations = PG_GETARG_INT32(1);
int32 allocsize = PG_GETARG_INT32(2);
struct timeval start_time, end_time;
/* memory contexts */
MemoryContext initctx = CurrentMemoryContext;
MemoryContext ctx = initctx;
/* switch to the per-group hash-table memory context */
for (i = 0; i < ncontexts; i++) {
char name[256];
sprintf(name, "test context %d", i);
#ifdef TRACKING_FLAG
ctx = AllocSetContextCreateTracked(ctx,
name,
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE,
true);
#else
ctx = AllocSetContextCreate(ctx,
name,
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
#endif
}
MemoryContextSwitchTo(ctx);
gettimeofday(&start_time, NULL);
for (i = 0; i < niterations; i++) {
char * p = palloc(allocsize);
if (p != NULL)
pfree(p);
}
gettimeofday(&end_time, NULL);
MemoryContextSwitchTo(initctx);
elog(WARNING, "duration = %.2f ms", (end_time.tv_sec - start_time.tv_sec) * 1000 + (end_time.tv_usec - start_time.tv_usec) / 1000.0);
PG_RETURN_VOID();
}
Datum
xpath_table(PG_FUNCTION_ARGS)
{
/* Function parameters */
char *pkeyfield = text_to_cstring(PG_GETARG_TEXT_PP(0));
char *xmlfield = text_to_cstring(PG_GETARG_TEXT_PP(1));
char *relname = text_to_cstring(PG_GETARG_TEXT_PP(2));
char *xpathset = text_to_cstring(PG_GETARG_TEXT_PP(3));
char *condition = text_to_cstring(PG_GETARG_TEXT_PP(4));
/* SPI (input tuple) support */
SPITupleTable *tuptable;
HeapTuple spi_tuple;
TupleDesc spi_tupdesc;
/* Output tuple (tuplestore) support */
Tuplestorestate *tupstore = NULL;
TupleDesc ret_tupdesc;
HeapTuple ret_tuple;
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
AttInMetadata *attinmeta;
MemoryContext per_query_ctx;
MemoryContext oldcontext;
char **values;
xmlChar **xpaths;
char *pos;
const char *pathsep = "|";
int numpaths;
int ret;
int proc;
int i;
int j;
int rownr; /* For issuing multiple rows from one original
* document */
bool had_values; /* To determine end of nodeset results */
StringInfoData query_buf;
/* We only have a valid tuple description in table function mode */
if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("set-valued function called in context that cannot accept a set")));
if (rsinfo->expectedDesc == NULL)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("xpath_table must be called as a table function")));
/*
* We want to materialise because it means that we don't have to carry
* libxml2 parser state between invocations of this function
*/
if (!(rsinfo->allowedModes & SFRM_Materialize))
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("xpath_table requires Materialize mode, but it is not "
"allowed in this context")));
/*
* The tuplestore must exist in a higher context than this function call
* (per_query_ctx is used)
*/
per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
oldcontext = MemoryContextSwitchTo(per_query_ctx);
/*
* Create the tuplestore - work_mem is the max in-memory size before a
* file is created on disk to hold it.
*/
tupstore =
tuplestore_begin_heap(rsinfo->allowedModes & SFRM_Materialize_Random,
false, work_mem);
MemoryContextSwitchTo(oldcontext);
/* get the requested return tuple description */
ret_tupdesc = CreateTupleDescCopy(rsinfo->expectedDesc);
/* must have at least one output column (for the pkey) */
if (ret_tupdesc->natts < 1)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("xpath_table must have at least one output column")));
/*
* At the moment we assume that the returned attributes make sense for the
* XPath specififed (i.e. we trust the caller). It's not fatal if they get
* it wrong - the input function for the column type will raise an error
* if the path result can't be converted into the correct binary
* representation.
*/
attinmeta = TupleDescGetAttInMetadata(ret_tupdesc);
/* Set return mode and allocate value space. */
rsinfo->returnMode = SFRM_Materialize;
rsinfo->setDesc = ret_tupdesc;
values = (char **) palloc(ret_tupdesc->natts * sizeof(char *));
xpaths = (xmlChar **) palloc(ret_tupdesc->natts * sizeof(xmlChar *));
/*
* Split XPaths. xpathset is a writable CString.
*
* Note that we stop splitting once we've done all needed for tupdesc
*/
numpaths = 0;
pos = xpathset;
while (numpaths < (ret_tupdesc->natts - 1))
{
xpaths[numpaths++] = (xmlChar *) pos;
pos = strstr(pos, pathsep);
if (pos != NULL)
{
*pos = '\0';
pos++;
}
else
break;
}
/* Now build query */
initStringInfo(&query_buf);
/* Build initial sql statement */
appendStringInfo(&query_buf, "SELECT %s, %s FROM %s WHERE %s",
pkeyfield,
xmlfield,
relname,
condition);
if ((ret = SPI_connect()) < 0)
elog(ERROR, "xpath_table: SPI_connect returned %d", ret);
if ((ret = SPI_exec(query_buf.data, 0)) != SPI_OK_SELECT)
elog(ERROR, "xpath_table: SPI execution failed for query %s",
query_buf.data);
proc = SPI_processed;
/* elog(DEBUG1,"xpath_table: SPI returned %d rows",proc); */
tuptable = SPI_tuptable;
spi_tupdesc = tuptable->tupdesc;
/* Switch out of SPI context */
MemoryContextSwitchTo(oldcontext);
/*
* Check that SPI returned correct result. If you put a comma into one of
* the function parameters, this will catch it when the SPI query returns
* e.g. 3 columns.
*/
if (spi_tupdesc->natts != 2)
{
ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("expression returning multiple columns is not valid in parameter list"),
errdetail("Expected two columns in SPI result, got %d.", spi_tupdesc->natts)));
}
/*
* Setup the parser. This should happen after we are done evaluating the
* query, in case it calls functions that set up libxml differently.
*/
pgxml_parser_init();
/* For each row i.e. document returned from SPI */
for (i = 0; i < proc; i++)
{
char *pkey;
char *xmldoc;
xmlDocPtr doctree;
xmlXPathContextPtr ctxt;
xmlXPathObjectPtr res;
xmlChar *resstr;
xmlXPathCompExprPtr comppath;
/* Extract the row data as C Strings */
spi_tuple = tuptable->vals[i];
pkey = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
xmldoc = SPI_getvalue(spi_tuple, spi_tupdesc, 2);
/*
* Clear the values array, so that not-well-formed documents return
* NULL in all columns. Note that this also means that spare columns
* will be NULL.
*/
for (j = 0; j < ret_tupdesc->natts; j++)
values[j] = NULL;
/* Insert primary key */
values[0] = pkey;
/* Parse the document */
if (xmldoc)
doctree = xmlParseMemory(xmldoc, strlen(xmldoc));
else /* treat NULL as not well-formed */
doctree = NULL;
if (doctree == NULL)
{
/* not well-formed, so output all-NULL tuple */
ret_tuple = BuildTupleFromCStrings(attinmeta, values);
tuplestore_puttuple(tupstore, ret_tuple);
heap_freetuple(ret_tuple);
}
else
{
/* New loop here - we have to deal with nodeset results */
rownr = 0;
do
{
/* Now evaluate the set of xpaths. */
had_values = false;
for (j = 0; j < numpaths; j++)
{
ctxt = xmlXPathNewContext(doctree);
ctxt->node = xmlDocGetRootElement(doctree);
/* compile the path */
comppath = xmlXPathCompile(xpaths[j]);
if (comppath == NULL)
{
xmlFreeDoc(doctree);
xml_ereport(ERROR, ERRCODE_EXTERNAL_ROUTINE_EXCEPTION,
"XPath Syntax Error");
}
/* Now evaluate the path expression. */
res = xmlXPathCompiledEval(comppath, ctxt);
xmlXPathFreeCompExpr(comppath);
if (res != NULL)
{
switch (res->type)
{
case XPATH_NODESET:
/* We see if this nodeset has enough nodes */
if (res->nodesetval != NULL &&
rownr < res->nodesetval->nodeNr)
{
resstr =
xmlXPathCastNodeToString(res->nodesetval->nodeTab[rownr]);
had_values = true;
}
else
resstr = NULL;
break;
case XPATH_STRING:
resstr = xmlStrdup(res->stringval);
break;
default:
elog(NOTICE, "unsupported XQuery result: %d", res->type);
resstr = xmlStrdup((const xmlChar *) "<unsupported/>");
}
/*
* Insert this into the appropriate column in the
* result tuple.
*/
values[j + 1] = (char *) resstr;
}
xmlXPathFreeContext(ctxt);
}
/* Now add the tuple to the output, if there is one. */
if (had_values)
{
ret_tuple = BuildTupleFromCStrings(attinmeta, values);
tuplestore_puttuple(tupstore, ret_tuple);
heap_freetuple(ret_tuple);
}
rownr++;
} while (had_values);
}
xmlFreeDoc(doctree);
if (pkey)
pfree(pkey);
if (xmldoc)
pfree(xmldoc);
}
tuplestore_donestoring(tupstore);
SPI_finish();
rsinfo->setResult = tupstore;
/*
* SFRM_Materialize mode expects us to return a NULL Datum. The actual
* tuples are in our tuplestore and passed back through rsinfo->setResult.
* rsinfo->setDesc is set to the tuple description that we actually used
* to build our tuples with, so the caller can verify we did what it was
* expecting.
*/
return (Datum) 0;
}
static ArrayType *
enum_range_internal(Oid enumtypoid, Oid lower, Oid upper)
{
ArrayType *result;
Relation enum_rel;
Relation enum_idx;
SysScanDesc enum_scan;
HeapTuple enum_tuple;
ScanKeyData skey;
Datum *elems;
int max,
cnt;
bool left_found;
/*
* Scan the enum members in order using pg_enum_typid_sortorder_index.
* Note we must not use the syscache, and must use an MVCC snapshot here.
* See comments for RenumberEnumType in catalog/pg_enum.c for more info.
*/
ScanKeyInit(&skey,
Anum_pg_enum_enumtypid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(enumtypoid));
enum_rel = heap_open(EnumRelationId, AccessShareLock);
enum_idx = index_open(EnumTypIdSortOrderIndexId, AccessShareLock);
enum_scan = systable_beginscan_ordered(enum_rel, enum_idx,
GetTransactionSnapshot(),
1, &skey);
max = 64;
elems = (Datum *) palloc(max * sizeof(Datum));
cnt = 0;
left_found = !OidIsValid(lower);
while (HeapTupleIsValid(enum_tuple = systable_getnext_ordered(enum_scan, ForwardScanDirection)))
{
Oid enum_oid = HeapTupleGetOid(enum_tuple);
if (!left_found && lower == enum_oid)
left_found = true;
if (left_found)
{
if (cnt >= max)
{
max *= 2;
elems = (Datum *) repalloc(elems, max * sizeof(Datum));
}
elems[cnt++] = ObjectIdGetDatum(enum_oid);
}
if (OidIsValid(upper) && upper == enum_oid)
break;
}
systable_endscan_ordered(enum_scan);
index_close(enum_idx, AccessShareLock);
heap_close(enum_rel, AccessShareLock);
/* and build the result array */
/* note this hardwires some details about the representation of Oid */
result = construct_array(elems, cnt, enumtypoid, sizeof(Oid), true, 'i');
pfree(elems);
return result;
}
/*
* compute_array_stats() -- compute statistics for an array column
*
* This function computes statistics useful for determining selectivity of
* the array operators <@, &&, and @>. It is invoked by ANALYZE via the
* compute_stats hook after sample rows have been collected.
*
* We also invoke the standard compute_stats function, which will compute
* "scalar" statistics relevant to the btree-style array comparison operators.
* However, exact duplicates of an entire array may be rare despite many
* arrays sharing individual elements. This especially afflicts long arrays,
* which are also liable to lack all scalar statistics due to the low
* WIDTH_THRESHOLD used in analyze.c. So, in addition to the standard stats,
* we find the most common array elements and compute a histogram of distinct
* element counts.
*
* The algorithm used is Lossy Counting, as proposed in the paper "Approximate
* frequency counts over data streams" by G. S. Manku and R. Motwani, in
* Proceedings of the 28th International Conference on Very Large Data Bases,
* Hong Kong, China, August 2002, section 4.2. The paper is available at
* http://www.vldb.org/conf/2002/S10P03.pdf
*
* The Lossy Counting (aka LC) algorithm goes like this:
* Let s be the threshold frequency for an item (the minimum frequency we
* are interested in) and epsilon the error margin for the frequency. Let D
* be a set of triples (e, f, delta), where e is an element value, f is that
* element's frequency (actually, its current occurrence count) and delta is
* the maximum error in f. We start with D empty and process the elements in
* batches of size w. (The batch size is also known as "bucket size" and is
* equal to 1/epsilon.) Let the current batch number be b_current, starting
* with 1. For each element e we either increment its f count, if it's
* already in D, or insert a new triple into D with values (e, 1, b_current
* - 1). After processing each batch we prune D, by removing from it all
* elements with f + delta <= b_current. After the algorithm finishes we
* suppress all elements from D that do not satisfy f >= (s - epsilon) * N,
* where N is the total number of elements in the input. We emit the
* remaining elements with estimated frequency f/N. The LC paper proves
* that this algorithm finds all elements with true frequency at least s,
* and that no frequency is overestimated or is underestimated by more than
* epsilon. Furthermore, given reasonable assumptions about the input
* distribution, the required table size is no more than about 7 times w.
*
* In the absence of a principled basis for other particular values, we
* follow ts_typanalyze() and use parameters s = 0.07/K, epsilon = s/10.
* But we leave out the correction for stopwords, which do not apply to
* arrays. These parameters give bucket width w = K/0.007 and maximum
* expected hashtable size of about 1000 * K.
*
* Elements may repeat within an array. Since duplicates do not change the
* behavior of <@, && or @>, we want to count each element only once per
* array. Therefore, we store in the finished pg_statistic entry each
* element's frequency as the fraction of all non-null rows that contain it.
* We divide the raw counts by nonnull_cnt to get those figures.
*/
static void
compute_array_stats(VacAttrStats *stats, AnalyzeAttrFetchFunc fetchfunc,
int samplerows, double totalrows)
{
ArrayAnalyzeExtraData *extra_data;
int num_mcelem;
int null_cnt = 0;
int null_elem_cnt = 0;
int analyzed_rows = 0;
/* This is D from the LC algorithm. */
HTAB *elements_tab;
HASHCTL elem_hash_ctl;
HASH_SEQ_STATUS scan_status;
/* This is the current bucket number from the LC algorithm */
int b_current;
/* This is 'w' from the LC algorithm */
int bucket_width;
int array_no;
int64 element_no;
TrackItem *item;
int slot_idx;
HTAB *count_tab;
HASHCTL count_hash_ctl;
DECountItem *count_item;
extra_data = (ArrayAnalyzeExtraData *) stats->extra_data;
/*
* Invoke analyze.c's standard analysis function to create scalar-style
* stats for the column. It will expect its own extra_data pointer, so
* temporarily install that.
*/
stats->extra_data = extra_data->std_extra_data;
extra_data->std_compute_stats(stats, fetchfunc, samplerows, totalrows);
stats->extra_data = extra_data;
/*
* Set up static pointer for use by subroutines. We wait till here in
* case std_compute_stats somehow recursively invokes us (probably not
* possible, but ...)
*/
array_extra_data = extra_data;
/*
* We want statistics_target * 10 elements in the MCELEM array. This
* multiplier is pretty arbitrary, but is meant to reflect the fact that
* the number of individual elements tracked in pg_statistic ought to be
* more than the number of values for a simple scalar column.
*/
num_mcelem = stats->attr->attstattarget * 10;
/*
* We set bucket width equal to num_mcelem / 0.007 as per the comment
* above.
*/
bucket_width = num_mcelem * 1000 / 7;
/*
* Create the hashtable. It will be in local memory, so we don't need to
* worry about overflowing the initial size. Also we don't need to pay any
* attention to locking and memory management.
*/
MemSet(&elem_hash_ctl, 0, sizeof(elem_hash_ctl));
elem_hash_ctl.keysize = sizeof(Datum);
elem_hash_ctl.entrysize = sizeof(TrackItem);
elem_hash_ctl.hash = element_hash;
elem_hash_ctl.match = element_match;
elem_hash_ctl.hcxt = CurrentMemoryContext;
elements_tab = hash_create("Analyzed elements table",
num_mcelem,
&elem_hash_ctl,
HASH_ELEM | HASH_FUNCTION | HASH_COMPARE | HASH_CONTEXT);
/* hashtable for array distinct elements counts */
MemSet(&count_hash_ctl, 0, sizeof(count_hash_ctl));
count_hash_ctl.keysize = sizeof(int);
count_hash_ctl.entrysize = sizeof(DECountItem);
count_hash_ctl.hcxt = CurrentMemoryContext;
count_tab = hash_create("Array distinct element count table",
64,
&count_hash_ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
/* Initialize counters. */
b_current = 1;
element_no = 0;
/* Loop over the arrays. */
for (array_no = 0; array_no < samplerows; array_no++)
{
Datum value;
bool isnull;
ArrayType *array;
int num_elems;
Datum *elem_values;
bool *elem_nulls;
bool null_present;
int j;
int64 prev_element_no = element_no;
int distinct_count;
bool count_item_found;
vacuum_delay_point();
value = fetchfunc(stats, array_no, &isnull);
if (isnull)
{
/* array is null, just count that */
null_cnt++;
continue;
}
/* Skip too-large values. */
if (toast_raw_datum_size(value) > ARRAY_WIDTH_THRESHOLD)
continue;
else
analyzed_rows++;
/*
* Now detoast the array if needed, and deconstruct into datums.
*/
array = DatumGetArrayTypeP(value);
Assert(ARR_ELEMTYPE(array) == extra_data->type_id);
deconstruct_array(array,
extra_data->type_id,
extra_data->typlen,
extra_data->typbyval,
extra_data->typalign,
&elem_values, &elem_nulls, &num_elems);
/*
* We loop through the elements in the array and add them to our
* tracking hashtable.
*/
null_present = false;
for (j = 0; j < num_elems; j++)
{
Datum elem_value;
bool found;
/* No null element processing other than flag setting here */
if (elem_nulls[j])
{
null_present = true;
continue;
}
/* Lookup current element in hashtable, adding it if new */
elem_value = elem_values[j];
item = (TrackItem *) hash_search(elements_tab,
(const void *) &elem_value,
HASH_ENTER, &found);
if (found)
{
/* The element value is already on the tracking list */
/*
* The operators we assist ignore duplicate array elements, so
* count a given distinct element only once per array.
*/
if (item->last_container == array_no)
continue;
item->frequency++;
item->last_container = array_no;
}
else
{
/* Initialize new tracking list element */
/*
* If element type is pass-by-reference, we must copy it into
* palloc'd space, so that we can release the array below. (We
* do this so that the space needed for element values is
* limited by the size of the hashtable; if we kept all the
* array values around, it could be much more.)
*/
item->key = datumCopy(elem_value,
extra_data->typbyval,
extra_data->typlen);
item->frequency = 1;
item->delta = b_current - 1;
item->last_container = array_no;
}
/* element_no is the number of elements processed (ie N) */
element_no++;
/* We prune the D structure after processing each bucket */
if (element_no % bucket_width == 0)
{
prune_element_hashtable(elements_tab, b_current);
b_current++;
}
}
/* Count null element presence once per array. */
if (null_present)
null_elem_cnt++;
/* Update frequency of the particular array distinct element count. */
distinct_count = (int) (element_no - prev_element_no);
count_item = (DECountItem *) hash_search(count_tab, &distinct_count,
HASH_ENTER,
&count_item_found);
if (count_item_found)
count_item->frequency++;
else
count_item->frequency = 1;
/* Free memory allocated while detoasting. */
if (PointerGetDatum(array) != value)
pfree(array);
pfree(elem_values);
pfree(elem_nulls);
}
/* Skip pg_statistic slots occupied by standard statistics */
slot_idx = 0;
while (slot_idx < STATISTIC_NUM_SLOTS && stats->stakind[slot_idx] != 0)
slot_idx++;
if (slot_idx > STATISTIC_NUM_SLOTS - 2)
elog(ERROR, "insufficient pg_statistic slots for array stats");
/* We can only compute real stats if we found some non-null values. */
if (analyzed_rows > 0)
{
int nonnull_cnt = analyzed_rows;
int count_items_count;
int i;
TrackItem **sort_table;
int track_len;
int64 cutoff_freq;
int64 minfreq,
maxfreq;
/*
* We assume the standard stats code already took care of setting
* stats_valid, stanullfrac, stawidth, stadistinct. We'd have to
* re-compute those values if we wanted to not store the standard
* stats.
*/
/*
* Construct an array of the interesting hashtable items, that is,
* those meeting the cutoff frequency (s - epsilon)*N. Also identify
* the minimum and maximum frequencies among these items.
*
* Since epsilon = s/10 and bucket_width = 1/epsilon, the cutoff
* frequency is 9*N / bucket_width.
*/
cutoff_freq = 9 * element_no / bucket_width;
i = hash_get_num_entries(elements_tab); /* surely enough space */
sort_table = (TrackItem **) palloc(sizeof(TrackItem *) * i);
hash_seq_init(&scan_status, elements_tab);
track_len = 0;
minfreq = element_no;
maxfreq = 0;
while ((item = (TrackItem *) hash_seq_search(&scan_status)) != NULL)
{
if (item->frequency > cutoff_freq)
{
sort_table[track_len++] = item;
minfreq = Min(minfreq, item->frequency);
maxfreq = Max(maxfreq, item->frequency);
}
}
Assert(track_len <= i);
/* emit some statistics for debug purposes */
elog(DEBUG3, "compute_array_stats: target # mces = %d, "
"bucket width = %d, "
"# elements = " INT64_FORMAT ", hashtable size = %d, "
"usable entries = %d",
num_mcelem, bucket_width, element_no, i, track_len);
/*
* If we obtained more elements than we really want, get rid of those
* with least frequencies. The easiest way is to qsort the array into
* descending frequency order and truncate the array.
*/
if (num_mcelem < track_len)
{
qsort(sort_table, track_len, sizeof(TrackItem *),
trackitem_compare_frequencies_desc);
/* reset minfreq to the smallest frequency we're keeping */
minfreq = sort_table[num_mcelem - 1]->frequency;
}
else
num_mcelem = track_len;
/* Generate MCELEM slot entry */
if (num_mcelem > 0)
{
MemoryContext old_context;
Datum *mcelem_values;
float4 *mcelem_freqs;
/*
* We want to store statistics sorted on the element value using
* the element type's default comparison function. This permits
* fast binary searches in selectivity estimation functions.
*/
qsort(sort_table, num_mcelem, sizeof(TrackItem *),
trackitem_compare_element);
/* Must copy the target values into anl_context */
old_context = MemoryContextSwitchTo(stats->anl_context);
/*
* We sorted statistics on the element value, but we want to be
* able to find the minimal and maximal frequencies without going
* through all the values. We also want the frequency of null
* elements. Store these three values at the end of mcelem_freqs.
*/
mcelem_values = (Datum *) palloc(num_mcelem * sizeof(Datum));
mcelem_freqs = (float4 *) palloc((num_mcelem + 3) * sizeof(float4));
/*
* See comments above about use of nonnull_cnt as the divisor for
* the final frequency estimates.
*/
for (i = 0; i < num_mcelem; i++)
{
TrackItem *item = sort_table[i];
mcelem_values[i] = datumCopy(item->key,
extra_data->typbyval,
extra_data->typlen);
mcelem_freqs[i] = (double) item->frequency /
(double) nonnull_cnt;
}
mcelem_freqs[i++] = (double) minfreq / (double) nonnull_cnt;
mcelem_freqs[i++] = (double) maxfreq / (double) nonnull_cnt;
mcelem_freqs[i++] = (double) null_elem_cnt / (double) nonnull_cnt;
MemoryContextSwitchTo(old_context);
stats->stakind[slot_idx] = STATISTIC_KIND_MCELEM;
stats->staop[slot_idx] = extra_data->eq_opr;
stats->stanumbers[slot_idx] = mcelem_freqs;
/* See above comment about extra stanumber entries */
stats->numnumbers[slot_idx] = num_mcelem + 3;
stats->stavalues[slot_idx] = mcelem_values;
stats->numvalues[slot_idx] = num_mcelem;
/* We are storing values of element type */
stats->statypid[slot_idx] = extra_data->type_id;
stats->statyplen[slot_idx] = extra_data->typlen;
stats->statypbyval[slot_idx] = extra_data->typbyval;
stats->statypalign[slot_idx] = extra_data->typalign;
slot_idx++;
}
/* Generate DECHIST slot entry */
count_items_count = hash_get_num_entries(count_tab);
if (count_items_count > 0)
{
int num_hist = stats->attr->attstattarget;
DECountItem **sorted_count_items;
int j;
int delta;
int64 frac;
float4 *hist;
/* num_hist must be at least 2 for the loop below to work */
num_hist = Max(num_hist, 2);
/*
* Create an array of DECountItem pointers, and sort them into
* increasing count order.
*/
sorted_count_items = (DECountItem **)
palloc(sizeof(DECountItem *) * count_items_count);
hash_seq_init(&scan_status, count_tab);
j = 0;
while ((count_item = (DECountItem *) hash_seq_search(&scan_status)) != NULL)
{
sorted_count_items[j++] = count_item;
}
qsort(sorted_count_items, count_items_count,
sizeof(DECountItem *), countitem_compare_count);
/*
* Prepare to fill stanumbers with the histogram, followed by the
* average count. This array must be stored in anl_context.
*/
hist = (float4 *)
MemoryContextAlloc(stats->anl_context,
sizeof(float4) * (num_hist + 1));
hist[num_hist] = (double) element_no / (double) nonnull_cnt;
/*----------
* Construct the histogram of distinct-element counts (DECs).
*
* The object of this loop is to copy the min and max DECs to
* hist[0] and hist[num_hist - 1], along with evenly-spaced DECs
* in between (where "evenly-spaced" is with reference to the
* whole input population of arrays). If we had a complete sorted
* array of DECs, one per analyzed row, the i'th hist value would
* come from DECs[i * (analyzed_rows - 1) / (num_hist - 1)]
* (compare the histogram-making loop in compute_scalar_stats()).
* But instead of that we have the sorted_count_items[] array,
* which holds unique DEC values with their frequencies (that is,
* a run-length-compressed version of the full array). So we
* control advancing through sorted_count_items[] with the
* variable "frac", which is defined as (x - y) * (num_hist - 1),
* where x is the index in the notional DECs array corresponding
* to the start of the next sorted_count_items[] element's run,
* and y is the index in DECs from which we should take the next
* histogram value. We have to advance whenever x <= y, that is
* frac <= 0. The x component is the sum of the frequencies seen
* so far (up through the current sorted_count_items[] element),
* and of course y * (num_hist - 1) = i * (analyzed_rows - 1),
* per the subscript calculation above. (The subscript calculation
* implies dropping any fractional part of y; in this formulation
* that's handled by not advancing until frac reaches 1.)
*
* Even though frac has a bounded range, it could overflow int32
* when working with very large statistics targets, so we do that
* math in int64.
*----------
*/
delta = analyzed_rows - 1;
j = 0; /* current index in sorted_count_items */
/* Initialize frac for sorted_count_items[0]; y is initially 0 */
frac = (int64) sorted_count_items[0]->frequency * (num_hist - 1);
for (i = 0; i < num_hist; i++)
{
while (frac <= 0)
{
/* Advance, and update x component of frac */
j++;
frac += (int64) sorted_count_items[j]->frequency * (num_hist - 1);
}
hist[i] = sorted_count_items[j]->count;
frac -= delta; /* update y for upcoming i increment */
}
Assert(j == count_items_count - 1);
stats->stakind[slot_idx] = STATISTIC_KIND_DECHIST;
stats->staop[slot_idx] = extra_data->eq_opr;
stats->stanumbers[slot_idx] = hist;
stats->numnumbers[slot_idx] = num_hist + 1;
slot_idx++;
}
}
/*
* We don't need to bother cleaning up any of our temporary palloc's. The
* hashtable should also go away, as it used a child memory context.
*/
}
/*
* CreateConstraintEntry
* Create a constraint table entry.
*
* Subsidiary records (such as triggers or indexes to implement the
* constraint) are *not* created here. But we do make dependency links
* from the constraint to the things it depends on.
*/
Oid
CreateConstraintEntry(const char *constraintName,
Oid constraintNamespace,
char constraintType,
bool isDeferrable,
bool isDeferred,
Oid relId,
const int16 *constraintKey,
int constraintNKeys,
Oid domainId,
Oid indexRelId,
Oid foreignRelId,
const int16 *foreignKey,
const Oid *pfEqOp,
const Oid *ppEqOp,
const Oid *ffEqOp,
int foreignNKeys,
char foreignUpdateType,
char foreignDeleteType,
char foreignMatchType,
const Oid *exclOp,
Node *conExpr,
const char *conBin,
const char *conSrc,
bool conIsLocal,
int conInhCount)
{
Relation conDesc;
Oid conOid;
HeapTuple tup;
bool nulls[Natts_pg_constraint];
Datum values[Natts_pg_constraint];
ArrayType *conkeyArray;
ArrayType *confkeyArray;
ArrayType *conpfeqopArray;
ArrayType *conppeqopArray;
ArrayType *conffeqopArray;
ArrayType *conexclopArray;
NameData cname;
int i;
ObjectAddress conobject;
conDesc = heap_open(ConstraintRelationId, RowExclusiveLock);
Assert(constraintName);
namestrcpy(&cname, constraintName);
/*
* Convert C arrays into Postgres arrays.
*/
if (constraintNKeys > 0)
{
Datum *conkey;
conkey = (Datum *) palloc(constraintNKeys * sizeof(Datum));
for (i = 0; i < constraintNKeys; i++)
conkey[i] = Int16GetDatum(constraintKey[i]);
conkeyArray = construct_array(conkey, constraintNKeys,
INT2OID, 2, true, 's');
}
else
conkeyArray = NULL;
if (foreignNKeys > 0)
{
Datum *fkdatums;
fkdatums = (Datum *) palloc(foreignNKeys * sizeof(Datum));
for (i = 0; i < foreignNKeys; i++)
fkdatums[i] = Int16GetDatum(foreignKey[i]);
confkeyArray = construct_array(fkdatums, foreignNKeys,
INT2OID, 2, true, 's');
for (i = 0; i < foreignNKeys; i++)
fkdatums[i] = ObjectIdGetDatum(pfEqOp[i]);
conpfeqopArray = construct_array(fkdatums, foreignNKeys,
OIDOID, sizeof(Oid), true, 'i');
for (i = 0; i < foreignNKeys; i++)
fkdatums[i] = ObjectIdGetDatum(ppEqOp[i]);
conppeqopArray = construct_array(fkdatums, foreignNKeys,
OIDOID, sizeof(Oid), true, 'i');
for (i = 0; i < foreignNKeys; i++)
fkdatums[i] = ObjectIdGetDatum(ffEqOp[i]);
conffeqopArray = construct_array(fkdatums, foreignNKeys,
OIDOID, sizeof(Oid), true, 'i');
}
else
{
confkeyArray = NULL;
conpfeqopArray = NULL;
conppeqopArray = NULL;
conffeqopArray = NULL;
}
if (exclOp != NULL)
{
Datum *opdatums;
opdatums = (Datum *) palloc(constraintNKeys * sizeof(Datum));
for (i = 0; i < constraintNKeys; i++)
opdatums[i] = ObjectIdGetDatum(exclOp[i]);
conexclopArray = construct_array(opdatums, constraintNKeys,
OIDOID, sizeof(Oid), true, 'i');
}
else
conexclopArray = NULL;
/* initialize nulls and values */
for (i = 0; i < Natts_pg_constraint; i++)
{
nulls[i] = false;
values[i] = (Datum) NULL;
}
values[Anum_pg_constraint_conname - 1] = NameGetDatum(&cname);
values[Anum_pg_constraint_connamespace - 1] = ObjectIdGetDatum(constraintNamespace);
values[Anum_pg_constraint_contype - 1] = CharGetDatum(constraintType);
values[Anum_pg_constraint_condeferrable - 1] = BoolGetDatum(isDeferrable);
values[Anum_pg_constraint_condeferred - 1] = BoolGetDatum(isDeferred);
values[Anum_pg_constraint_conrelid - 1] = ObjectIdGetDatum(relId);
values[Anum_pg_constraint_contypid - 1] = ObjectIdGetDatum(domainId);
values[Anum_pg_constraint_conindid - 1] = ObjectIdGetDatum(indexRelId);
values[Anum_pg_constraint_confrelid - 1] = ObjectIdGetDatum(foreignRelId);
values[Anum_pg_constraint_confupdtype - 1] = CharGetDatum(foreignUpdateType);
values[Anum_pg_constraint_confdeltype - 1] = CharGetDatum(foreignDeleteType);
values[Anum_pg_constraint_confmatchtype - 1] = CharGetDatum(foreignMatchType);
values[Anum_pg_constraint_conislocal - 1] = BoolGetDatum(conIsLocal);
values[Anum_pg_constraint_coninhcount - 1] = Int32GetDatum(conInhCount);
if (conkeyArray)
values[Anum_pg_constraint_conkey - 1] = PointerGetDatum(conkeyArray);
else
nulls[Anum_pg_constraint_conkey - 1] = true;
if (confkeyArray)
values[Anum_pg_constraint_confkey - 1] = PointerGetDatum(confkeyArray);
else
nulls[Anum_pg_constraint_confkey - 1] = true;
if (conpfeqopArray)
values[Anum_pg_constraint_conpfeqop - 1] = PointerGetDatum(conpfeqopArray);
else
nulls[Anum_pg_constraint_conpfeqop - 1] = true;
if (conppeqopArray)
values[Anum_pg_constraint_conppeqop - 1] = PointerGetDatum(conppeqopArray);
else
nulls[Anum_pg_constraint_conppeqop - 1] = true;
if (conffeqopArray)
values[Anum_pg_constraint_conffeqop - 1] = PointerGetDatum(conffeqopArray);
else
nulls[Anum_pg_constraint_conffeqop - 1] = true;
if (conexclopArray)
values[Anum_pg_constraint_conexclop - 1] = PointerGetDatum(conexclopArray);
else
nulls[Anum_pg_constraint_conexclop - 1] = true;
/*
* initialize the binary form of the check constraint.
*/
if (conBin)
values[Anum_pg_constraint_conbin - 1] = CStringGetTextDatum(conBin);
else
nulls[Anum_pg_constraint_conbin - 1] = true;
/*
* initialize the text form of the check constraint
*/
if (conSrc)
values[Anum_pg_constraint_consrc - 1] = CStringGetTextDatum(conSrc);
else
nulls[Anum_pg_constraint_consrc - 1] = true;
tup = heap_form_tuple(RelationGetDescr(conDesc), values, nulls);
conOid = simple_heap_insert(conDesc, tup);
/* update catalog indexes */
CatalogUpdateIndexes(conDesc, tup);
conobject.classId = ConstraintRelationId;
conobject.objectId = conOid;
conobject.objectSubId = 0;
heap_close(conDesc, RowExclusiveLock);
if (OidIsValid(relId))
{
/*
* Register auto dependency from constraint to owning relation, or to
* specific column(s) if any are mentioned.
*/
ObjectAddress relobject;
relobject.classId = RelationRelationId;
relobject.objectId = relId;
if (constraintNKeys > 0)
{
for (i = 0; i < constraintNKeys; i++)
{
relobject.objectSubId = constraintKey[i];
recordDependencyOn(&conobject, &relobject, DEPENDENCY_AUTO);
}
}
else
{
relobject.objectSubId = 0;
recordDependencyOn(&conobject, &relobject, DEPENDENCY_AUTO);
}
}
if (OidIsValid(domainId))
{
/*
* Register auto dependency from constraint to owning domain
*/
ObjectAddress domobject;
domobject.classId = TypeRelationId;
domobject.objectId = domainId;
domobject.objectSubId = 0;
recordDependencyOn(&conobject, &domobject, DEPENDENCY_AUTO);
}
if (OidIsValid(foreignRelId))
{
/*
* Register normal dependency from constraint to foreign relation, or
* to specific column(s) if any are mentioned.
*/
ObjectAddress relobject;
relobject.classId = RelationRelationId;
relobject.objectId = foreignRelId;
if (foreignNKeys > 0)
{
for (i = 0; i < foreignNKeys; i++)
{
relobject.objectSubId = foreignKey[i];
recordDependencyOn(&conobject, &relobject, DEPENDENCY_NORMAL);
}
}
else
{
relobject.objectSubId = 0;
recordDependencyOn(&conobject, &relobject, DEPENDENCY_NORMAL);
}
}
if (OidIsValid(indexRelId) && constraintType == CONSTRAINT_FOREIGN)
{
/*
* Register normal dependency on the unique index that supports a
* foreign-key constraint. (Note: for indexes associated with unique
* or primary-key constraints, the dependency runs the other way, and
* is not made here.)
*/
ObjectAddress relobject;
relobject.classId = RelationRelationId;
relobject.objectId = indexRelId;
relobject.objectSubId = 0;
recordDependencyOn(&conobject, &relobject, DEPENDENCY_NORMAL);
}
if (foreignNKeys > 0)
{
/*
* Register normal dependencies on the equality operators that support
* a foreign-key constraint. If the PK and FK types are the same then
* all three operators for a column are the same; otherwise they are
* different.
*/
ObjectAddress oprobject;
oprobject.classId = OperatorRelationId;
oprobject.objectSubId = 0;
for (i = 0; i < foreignNKeys; i++)
{
oprobject.objectId = pfEqOp[i];
recordDependencyOn(&conobject, &oprobject, DEPENDENCY_NORMAL);
if (ppEqOp[i] != pfEqOp[i])
{
oprobject.objectId = ppEqOp[i];
recordDependencyOn(&conobject, &oprobject, DEPENDENCY_NORMAL);
}
if (ffEqOp[i] != pfEqOp[i])
{
oprobject.objectId = ffEqOp[i];
recordDependencyOn(&conobject, &oprobject, DEPENDENCY_NORMAL);
}
}
}
/*
* We don't bother to register dependencies on the exclusion operators of
* an exclusion constraint. We assume they are members of the opclass
* supporting the index, so there's an indirect dependency via that. (This
* would be pretty dicey for cross-type operators, but exclusion operators
* can never be cross-type.)
*/
if (conExpr != NULL)
{
/*
* Register dependencies from constraint to objects mentioned in CHECK
* expression.
*/
recordDependencyOnSingleRelExpr(&conobject, conExpr, relId,
DEPENDENCY_NORMAL,
DEPENDENCY_NORMAL);
}
return conOid;
}
/*
* nodeRead -
* Slightly higher-level reader.
*
* This routine applies some semantic knowledge on top of the purely
* lexical tokenizer pg_strtok(). It can read
* * Value token nodes (integers, floats, or strings);
* * General nodes (via parseNodeString() from readfuncs.c);
* * Lists of the above;
* * Lists of integers or OIDs.
* The return value is declared void *, not Node *, to avoid having to
* cast it explicitly in callers that assign to fields of different types.
*
* External callers should always pass NULL/0 for the arguments. Internally
* a non-NULL token may be passed when the upper recursion level has already
* scanned the first token of a node's representation.
*
* We assume pg_strtok is already initialized with a string to read (hence
* this should only be invoked from within a stringToNode operation).
*/
void *
nodeRead(char *token, int tok_len)
{
Node *result;
NodeTag type;
if (token == NULL) /* need to read a token? */
{
token = pg_strtok(&tok_len);
if (token == NULL) /* end of input */
return NULL;
}
type = nodeTokenType(token, tok_len);
switch ((int) type)
{
case LEFT_BRACE:
result = parseNodeString();
token = pg_strtok(&tok_len);
if (token == NULL || token[0] != '}')
elog(ERROR, "did not find '}' at end of input node");
break;
case LEFT_PAREN:
{
List *l = NIL;
/*----------
* Could be an integer list: (i int int ...)
* or an OID list: (o int int ...)
* or a list of nodes/values: (node node ...)
*----------
*/
token = pg_strtok(&tok_len);
if (token == NULL)
elog(ERROR, "unterminated List structure");
if (tok_len == 1 && token[0] == 'i')
{
/* List of integers */
for (;;)
{
int val;
char *endptr;
token = pg_strtok(&tok_len);
if (token == NULL)
elog(ERROR, "unterminated List structure");
if (token[0] == ')')
break;
val = (int) strtol(token, &endptr, 10);
if (endptr != token + tok_len)
elog(ERROR, "unrecognized integer: \"%.*s\"",
tok_len, token);
l = lappend_int(l, val);
}
}
else if (tok_len == 1 && token[0] == 'o')
{
/* List of OIDs */
for (;;)
{
Oid val;
char *endptr;
token = pg_strtok(&tok_len);
if (token == NULL)
elog(ERROR, "unterminated List structure");
if (token[0] == ')')
break;
val = (Oid) strtoul(token, &endptr, 10);
if (endptr != token + tok_len)
elog(ERROR, "unrecognized OID: \"%.*s\"",
tok_len, token);
l = lappend_oid(l, val);
}
}
else
{
/* List of other node types */
for (;;)
{
/* We have already scanned next token... */
if (token[0] == ')')
break;
l = lappend(l, nodeRead(token, tok_len));
token = pg_strtok(&tok_len);
if (token == NULL)
elog(ERROR, "unterminated List structure");
}
}
result = (Node *) l;
break;
}
case RIGHT_PAREN:
elog(ERROR, "unexpected right parenthesis");
result = NULL; /* keep compiler happy */
break;
case OTHER_TOKEN:
if (tok_len == 0)
{
/* must be "<>" --- represents a null pointer */
result = NULL;
}
else
{
elog(ERROR, "unrecognized token: \"%.*s\"", tok_len, token);
result = NULL; /* keep compiler happy */
}
break;
case T_Integer:
/*
* we know that the token terminates on a char atol will stop at
*/
result = (Node *) makeInteger(atol(token));
break;
case T_Float:
{
char *fval = (char *) palloc(tok_len + 1);
memcpy(fval, token, tok_len);
fval[tok_len] = '\0';
result = (Node *) makeFloat(fval);
}
break;
case T_String:
/* need to remove leading and trailing quotes, and backslashes */
result = (Node *) makeString(debackslash(token + 1, tok_len - 2));
break;
case T_BitString:
{
char *val = palloc(tok_len);
/* skip leading 'b' */
memcpy(val, token + 1, tok_len - 1);
val[tok_len - 1] = '\0';
result = (Node *) makeBitString(val);
break;
}
default:
elog(ERROR, "unrecognized node type: %d", (int) type);
result = NULL; /* keep compiler happy */
break;
}
return (void *) result;
}
/*
* lowerstr_with_len --- fold string to lower case
*
* Input string need not be null-terminated.
*
* Returned string is palloc'd
*/
char *
lowerstr_with_len(const char *str, int len)
{
char *out;
if (len == 0)
return pstrdup("");
#ifdef USE_WIDE_UPPER_LOWER
/*
* Use wide char code only when max encoding length > 1 and ctype != C.
* Some operating systems fail with multi-byte encodings and a C locale.
* Also, for a C locale there is no need to process as multibyte. From
* backend/utils/adt/oracle_compat.c Teodor
*/
if (pg_database_encoding_max_length() > 1 && !lc_ctype_is_c())
{
wchar_t *wstr,
*wptr;
int wlen;
/*
* alloc number of wchar_t for worst case, len contains number of
* bytes >= number of characters and alloc 1 wchar_t for 0, because
* wchar2char wants zero-terminated string
*/
wptr = wstr = (wchar_t *) palloc(sizeof(wchar_t) * (len + 1));
wlen = char2wchar(wstr, len + 1, str, len);
Assert(wlen <= len);
while (*wptr)
{
*wptr = towlower((wint_t) *wptr);
wptr++;
}
/*
* Alloc result string for worst case + '\0'
*/
len = pg_database_encoding_max_length() * wlen + 1;
out = (char *) palloc(len);
wlen = wchar2char(out, wstr, len);
pfree(wstr);
if (wlen < 0)
ereport(ERROR,
(errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE),
errmsg("conversion from wchar_t to server encoding failed: %m")));
Assert(wlen < len);
}
else
#endif /* USE_WIDE_UPPER_LOWER */
{
const char *ptr = str;
char *outptr;
outptr = out = (char *) palloc(sizeof(char) * (len + 1));
while ((ptr - str) < len && *ptr)
{
*outptr++ = tolower(TOUCHAR(ptr));
ptr++;
}
*outptr = '\0';
}
return out;
}
/*
* create and populate the crosstab tuplestore using the provided source query
*/
static Tuplestorestate *
get_crosstab_tuplestore(char *sql,
HTAB *crosstab_hash,
TupleDesc tupdesc,
MemoryContext per_query_ctx,
bool randomAccess)
{
Tuplestorestate *tupstore;
int num_categories = hash_get_num_entries(crosstab_hash);
AttInMetadata *attinmeta = TupleDescGetAttInMetadata(tupdesc);
char **values;
HeapTuple tuple;
int ret;
int proc;
/* initialize our tuplestore (while still in query context!) */
tupstore = tuplestore_begin_heap(randomAccess, false, work_mem);
/* Connect to SPI manager */
if ((ret = SPI_connect()) < 0)
/* internal error */
elog(ERROR, "get_crosstab_tuplestore: SPI_connect returned %d", ret);
/* Now retrieve the crosstab source rows */
ret = SPI_execute(sql, true, 0);
proc = SPI_processed;
/* Check for qualifying tuples */
if ((ret == SPI_OK_SELECT) && (proc > 0))
{
SPITupleTable *spi_tuptable = SPI_tuptable;
TupleDesc spi_tupdesc = spi_tuptable->tupdesc;
int ncols = spi_tupdesc->natts;
char *rowid;
char *lastrowid = NULL;
bool firstpass = true;
int i,
j;
int result_ncols;
if (num_categories == 0)
{
/* no qualifying category tuples */
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("provided \"categories\" SQL must " \
"return 1 column of at least one row")));
}
/*
* The provided SQL query must always return at least three columns:
*
* 1. rowname the label for each row - column 1 in the final result
* 2. category the label for each value-column in the final result 3.
* value the values used to populate the value-columns
*
* If there are more than three columns, the last two are taken as
* "category" and "values". The first column is taken as "rowname".
* Additional columns (2 thru N-2) are assumed the same for the same
* "rowname", and are copied into the result tuple from the first time
* we encounter a particular rowname.
*/
if (ncols < 3)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid source data SQL statement"),
errdetail("The provided SQL must return 3 " \
" columns; rowid, category, and values.")));
result_ncols = (ncols - 2) + num_categories;
/* Recheck to make sure we tuple descriptor still looks reasonable */
if (tupdesc->natts != result_ncols)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("invalid return type"),
errdetail("Query-specified return " \
"tuple has %d columns but crosstab " \
"returns %d.", tupdesc->natts, result_ncols)));
/* allocate space */
values = (char **) palloc(result_ncols * sizeof(char *));
/* and make sure it's clear */
memset(values, '\0', result_ncols * sizeof(char *));
for (i = 0; i < proc; i++)
{
HeapTuple spi_tuple;
crosstab_cat_desc *catdesc;
char *catname;
/* get the next sql result tuple */
spi_tuple = spi_tuptable->vals[i];
/* get the rowid from the current sql result tuple */
rowid = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
/*
* if we're on a new output row, grab the column values up to
* column N-2 now
*/
if (firstpass || !xstreq(lastrowid, rowid))
{
/*
* a new row means we need to flush the old one first, unless
* we're on the very first row
*/
if (!firstpass)
{
/* rowid changed, flush the previous output row */
tuple = BuildTupleFromCStrings(attinmeta, values);
tuplestore_puttuple(tupstore, tuple);
for (j = 0; j < result_ncols; j++)
xpfree(values[j]);
}
values[0] = rowid;
for (j = 1; j < ncols - 2; j++)
values[j] = SPI_getvalue(spi_tuple, spi_tupdesc, j + 1);
/* we're no longer on the first pass */
firstpass = false;
}
/* look up the category and fill in the appropriate column */
catname = SPI_getvalue(spi_tuple, spi_tupdesc, ncols - 1);
if (catname != NULL)
{
crosstab_HashTableLookup(crosstab_hash, catname, catdesc);
if (catdesc)
values[catdesc->attidx + ncols - 2] =
SPI_getvalue(spi_tuple, spi_tupdesc, ncols);
}
xpfree(lastrowid);
xpstrdup(lastrowid, rowid);
}
/* flush the last output row */
tuple = BuildTupleFromCStrings(attinmeta, values);
tuplestore_puttuple(tupstore, tuple);
}
if (SPI_finish() != SPI_OK_FINISH)
/* internal error */
elog(ERROR, "get_crosstab_tuplestore: SPI_finish() failed");
tuplestore_donestoring(tupstore);
return tupstore;
}
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);
}
/*
* load up the categories hash table
*/
static HTAB *
load_categories_hash(char *cats_sql, MemoryContext per_query_ctx)
{
HTAB *crosstab_hash;
HASHCTL ctl;
int ret;
int proc;
MemoryContext SPIcontext;
/* initialize the category hash table */
MemSet(&ctl, 0, sizeof(ctl));
ctl.keysize = MAX_CATNAME_LEN;
ctl.entrysize = sizeof(crosstab_HashEnt);
ctl.hcxt = per_query_ctx;
/*
* use INIT_CATS, defined above as a guess of how many hash table entries
* to create, initially
*/
crosstab_hash = hash_create("crosstab hash",
INIT_CATS,
&ctl,
HASH_ELEM | HASH_CONTEXT);
/* Connect to SPI manager */
if ((ret = SPI_connect()) < 0)
/* internal error */
elog(ERROR, "load_categories_hash: SPI_connect returned %d", ret);
/* Retrieve the category name rows */
ret = SPI_execute(cats_sql, true, 0);
proc = SPI_processed;
/* Check for qualifying tuples */
if ((ret == SPI_OK_SELECT) && (proc > 0))
{
SPITupleTable *spi_tuptable = SPI_tuptable;
TupleDesc spi_tupdesc = spi_tuptable->tupdesc;
int i;
/*
* The provided categories SQL query must always return one column:
* category - the label or identifier for each column
*/
if (spi_tupdesc->natts != 1)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("provided \"categories\" SQL must " \
"return 1 column of at least one row")));
for (i = 0; i < proc; i++)
{
crosstab_cat_desc *catdesc;
char *catname;
HeapTuple spi_tuple;
/* get the next sql result tuple */
spi_tuple = spi_tuptable->vals[i];
/* get the category from the current sql result tuple */
catname = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
SPIcontext = MemoryContextSwitchTo(per_query_ctx);
catdesc = (crosstab_cat_desc *) palloc(sizeof(crosstab_cat_desc));
catdesc->catname = catname;
catdesc->attidx = i;
/* Add the proc description block to the hashtable */
crosstab_HashTableInsert(crosstab_hash, catdesc);
MemoryContextSwitchTo(SPIcontext);
}
}
if (SPI_finish() != SPI_OK_FINISH)
/* internal error */
elog(ERROR, "load_categories_hash: SPI_finish() failed");
return crosstab_hash;
}
Datum
normal_rand(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
int call_cntr;
int max_calls;
normal_rand_fctx *fctx;
float8 mean;
float8 stddev;
float8 carry_val;
bool use_carry;
MemoryContext oldcontext;
/* stuff done only on the first call of the function */
if (SRF_IS_FIRSTCALL())
{
/* create a function context for cross-call persistence */
funcctx = SRF_FIRSTCALL_INIT();
/*
* switch to memory context appropriate for multiple function calls
*/
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/* total number of tuples to be returned */
funcctx->max_calls = PG_GETARG_UINT32(0);
/* allocate memory for user context */
fctx = (normal_rand_fctx *) palloc(sizeof(normal_rand_fctx));
/*
* Use fctx to keep track of upper and lower bounds from call to call.
* It will also be used to carry over the spare value we get from the
* Box-Muller algorithm so that we only actually calculate a new value
* every other call.
*/
fctx->mean = PG_GETARG_FLOAT8(1);
fctx->stddev = PG_GETARG_FLOAT8(2);
fctx->carry_val = 0;
fctx->use_carry = false;
funcctx->user_fctx = fctx;
MemoryContextSwitchTo(oldcontext);
}
/* stuff done on every call of the function */
funcctx = SRF_PERCALL_SETUP();
call_cntr = funcctx->call_cntr;
max_calls = funcctx->max_calls;
fctx = funcctx->user_fctx;
mean = fctx->mean;
stddev = fctx->stddev;
carry_val = fctx->carry_val;
use_carry = fctx->use_carry;
if (call_cntr < max_calls) /* do when there is more left to send */
{
float8 result;
if (use_carry)
{
/*
* reset use_carry and use second value obtained on last pass
*/
fctx->use_carry = false;
result = carry_val;
}
else
{
float8 normval_1;
float8 normval_2;
/* Get the next two normal values */
get_normal_pair(&normval_1, &normval_2);
/* use the first */
result = mean + (stddev * normval_1);
/* and save the second */
fctx->carry_val = mean + (stddev * normval_2);
fctx->use_carry = true;
}
/* send the result */
SRF_RETURN_NEXT(funcctx, Float8GetDatum(result));
}
else
/* do when there is no more left */
SRF_RETURN_DONE(funcctx);
}
static Tuplestorestate *
build_tuplestore_recursively(char *key_fld,
char *parent_key_fld,
char *relname,
char *orderby_fld,
char *branch_delim,
char *start_with,
char *branch,
int level,
int *serial,
int max_depth,
bool show_branch,
bool show_serial,
MemoryContext per_query_ctx,
AttInMetadata *attinmeta,
Tuplestorestate *tupstore)
{
TupleDesc tupdesc = attinmeta->tupdesc;
int ret;
int proc;
int serial_column;
StringInfoData sql;
char **values;
char *current_key;
char *current_key_parent;
char current_level[INT32_STRLEN];
char serial_str[INT32_STRLEN];
char *current_branch;
HeapTuple tuple;
if (max_depth > 0 && level > max_depth)
return tupstore;
initStringInfo(&sql);
/* Build initial sql statement */
if (!show_serial)
{
appendStringInfo(&sql, "SELECT %s, %s FROM %s WHERE %s = %s AND %s IS NOT NULL AND %s <> %s",
key_fld,
parent_key_fld,
relname,
parent_key_fld,
quote_literal_cstr(start_with),
key_fld, key_fld, parent_key_fld);
serial_column = 0;
}
else
{
appendStringInfo(&sql, "SELECT %s, %s FROM %s WHERE %s = %s AND %s IS NOT NULL AND %s <> %s ORDER BY %s",
key_fld,
parent_key_fld,
relname,
parent_key_fld,
quote_literal_cstr(start_with),
key_fld, key_fld, parent_key_fld,
orderby_fld);
serial_column = 1;
}
if (show_branch)
values = (char **) palloc((CONNECTBY_NCOLS + serial_column) * sizeof(char *));
else
values = (char **) palloc((CONNECTBY_NCOLS_NOBRANCH + serial_column) * sizeof(char *));
/* First time through, do a little setup */
if (level == 0)
{
/* root value is the one we initially start with */
values[0] = start_with;
/* root value has no parent */
values[1] = NULL;
/* root level is 0 */
sprintf(current_level, "%d", level);
values[2] = current_level;
/* root branch is just starting root value */
if (show_branch)
values[3] = start_with;
/* root starts the serial with 1 */
if (show_serial)
{
sprintf(serial_str, "%d", (*serial)++);
if (show_branch)
values[4] = serial_str;
else
values[3] = serial_str;
}
/* construct the tuple */
tuple = BuildTupleFromCStrings(attinmeta, values);
/* now store it */
tuplestore_puttuple(tupstore, tuple);
/* increment level */
level++;
}
/* Retrieve the desired rows */
ret = SPI_execute(sql.data, true, 0);
proc = SPI_processed;
/* Check for qualifying tuples */
if ((ret == SPI_OK_SELECT) && (proc > 0))
{
HeapTuple spi_tuple;
SPITupleTable *tuptable = SPI_tuptable;
TupleDesc spi_tupdesc = tuptable->tupdesc;
int i;
StringInfoData branchstr;
StringInfoData chk_branchstr;
StringInfoData chk_current_key;
/* First time through, do a little more setup */
if (level == 0)
{
/*
* Check that return tupdesc is compatible with the one we got
* from the query, but only at level 0 -- no need to check more
* than once
*/
if (!compatConnectbyTupleDescs(tupdesc, spi_tupdesc))
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("invalid return type"),
errdetail("Return and SQL tuple descriptions are " \
"incompatible.")));
}
initStringInfo(&branchstr);
initStringInfo(&chk_branchstr);
initStringInfo(&chk_current_key);
for (i = 0; i < proc; i++)
{
/* initialize branch for this pass */
appendStringInfo(&branchstr, "%s", branch);
appendStringInfo(&chk_branchstr, "%s%s%s", branch_delim, branch, branch_delim);
/* get the next sql result tuple */
spi_tuple = tuptable->vals[i];
/* get the current key and parent */
current_key = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
appendStringInfo(&chk_current_key, "%s%s%s", branch_delim, current_key, branch_delim);
current_key_parent = pstrdup(SPI_getvalue(spi_tuple, spi_tupdesc, 2));
/* get the current level */
sprintf(current_level, "%d", level);
/* check to see if this key is also an ancestor */
if (strstr(chk_branchstr.data, chk_current_key.data))
elog(ERROR, "infinite recursion detected");
/* OK, extend the branch */
appendStringInfo(&branchstr, "%s%s", branch_delim, current_key);
current_branch = branchstr.data;
/* build a tuple */
values[0] = pstrdup(current_key);
values[1] = current_key_parent;
values[2] = current_level;
if (show_branch)
values[3] = current_branch;
if (show_serial)
{
sprintf(serial_str, "%d", (*serial)++);
if (show_branch)
values[4] = serial_str;
else
values[3] = serial_str;
}
tuple = BuildTupleFromCStrings(attinmeta, values);
xpfree(current_key);
xpfree(current_key_parent);
/* store the tuple for later use */
tuplestore_puttuple(tupstore, tuple);
heap_freetuple(tuple);
/* recurse using current_key_parent as the new start_with */
tupstore = build_tuplestore_recursively(key_fld,
parent_key_fld,
relname,
orderby_fld,
branch_delim,
values[0],
current_branch,
level + 1,
serial,
max_depth,
show_branch,
show_serial,
per_query_ctx,
attinmeta,
tupstore);
/* reset branch for next pass */
resetStringInfo(&branchstr);
resetStringInfo(&chk_branchstr);
resetStringInfo(&chk_current_key);
}
xpfree(branchstr.data);
xpfree(chk_branchstr.data);
xpfree(chk_current_key.data);
}
return tupstore;
}
/*
** The GiST PickSplit method for _intments
** We use Guttman's poly time split algorithm
*/
Datum
g_int_picksplit(PG_FUNCTION_ARGS)
{
bytea *entryvec = (bytea *) PG_GETARG_POINTER(0);
GIST_SPLITVEC *v = (GIST_SPLITVEC *) PG_GETARG_POINTER(1);
OffsetNumber i,
j;
ArrayType *datum_alpha,
*datum_beta;
ArrayType *datum_l,
*datum_r;
ArrayType *union_d,
*union_dl,
*union_dr;
ArrayType *inter_d;
bool firsttime;
float size_alpha,
size_beta,
size_union,
size_inter;
float size_waste,
waste;
float size_l,
size_r;
int nbytes;
OffsetNumber seed_1 = 0,
seed_2 = 0;
OffsetNumber *left,
*right;
OffsetNumber maxoff;
SPLITCOST *costvector;
#ifdef GIST_DEBUG
elog(DEBUG3, "--------picksplit %d", (VARSIZE(entryvec) - VARHDRSZ) / sizeof(GISTENTRY));
#endif
maxoff = ((VARSIZE(entryvec) - VARHDRSZ) / sizeof(GISTENTRY)) - 2;
nbytes = (maxoff + 2) * sizeof(OffsetNumber);
v->spl_left = (OffsetNumber *) palloc(nbytes);
v->spl_right = (OffsetNumber *) palloc(nbytes);
firsttime = true;
waste = 0.0;
for (i = FirstOffsetNumber; i < maxoff; i = OffsetNumberNext(i))
{
datum_alpha = (ArrayType *) DatumGetPointer(((GISTENTRY *) VARDATA(entryvec))[i].key);
for (j = OffsetNumberNext(i); j <= maxoff; j = OffsetNumberNext(j))
{
datum_beta = (ArrayType *) DatumGetPointer(((GISTENTRY *) VARDATA(entryvec))[j].key);
/* compute the wasted space by unioning these guys */
/* size_waste = size_union - size_inter; */
union_d = inner_int_union(datum_alpha, datum_beta);
rt__int_size(union_d, &size_union);
inter_d = inner_int_inter(datum_alpha, datum_beta);
rt__int_size(inter_d, &size_inter);
size_waste = size_union - size_inter;
pfree(union_d);
if (inter_d != (ArrayType *) NULL)
pfree(inter_d);
/*
* are these a more promising split that what we've already
* seen?
*/
if (size_waste > waste || firsttime)
{
waste = size_waste;
seed_1 = i;
seed_2 = j;
firsttime = false;
}
}
}
left = v->spl_left;
v->spl_nleft = 0;
right = v->spl_right;
v->spl_nright = 0;
if (seed_1 == 0 || seed_2 == 0)
{
seed_1 = 1;
seed_2 = 2;
}
datum_alpha = (ArrayType *) DatumGetPointer(((GISTENTRY *) VARDATA(entryvec))[seed_1].key);
datum_l = copy_intArrayType(datum_alpha);
rt__int_size(datum_l, &size_l);
datum_beta = (ArrayType *) DatumGetPointer(((GISTENTRY *) VARDATA(entryvec))[seed_2].key);
datum_r = copy_intArrayType(datum_beta);
rt__int_size(datum_r, &size_r);
maxoff = OffsetNumberNext(maxoff);
/*
* sort entries
*/
costvector = (SPLITCOST *) palloc(sizeof(SPLITCOST) * maxoff);
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
{
costvector[i - 1].pos = i;
datum_alpha = (ArrayType *) DatumGetPointer(((GISTENTRY *) VARDATA(entryvec))[i].key);
union_d = inner_int_union(datum_l, datum_alpha);
rt__int_size(union_d, &size_alpha);
pfree(union_d);
union_d = inner_int_union(datum_r, datum_alpha);
rt__int_size(union_d, &size_beta);
pfree(union_d);
costvector[i - 1].cost = abs((size_alpha - size_l) - (size_beta - size_r));
}
qsort((void *) costvector, maxoff, sizeof(SPLITCOST), comparecost);
/*
* Now split up the regions between the two seeds. An important
* property of this split algorithm is that the split vector v has the
* indices of items to be split in order in its left and right
* vectors. We exploit this property by doing a merge in the code
* that actually splits the page.
*
* For efficiency, we also place the new index tuple in this loop. This
* is handled at the very end, when we have placed all the existing
* tuples and i == maxoff + 1.
*/
for (j = 0; j < maxoff; j++)
{
i = costvector[j].pos;
/*
* If we've already decided where to place this item, just put it
* on the right list. Otherwise, we need to figure out which page
* needs the least enlargement in order to store the item.
*/
if (i == seed_1)
{
*left++ = i;
v->spl_nleft++;
continue;
}
else if (i == seed_2)
{
*right++ = i;
v->spl_nright++;
continue;
}
/* okay, which page needs least enlargement? */
datum_alpha = (ArrayType *) DatumGetPointer(((GISTENTRY *) VARDATA(entryvec))[i].key);
union_dl = inner_int_union(datum_l, datum_alpha);
union_dr = inner_int_union(datum_r, datum_alpha);
rt__int_size(union_dl, &size_alpha);
rt__int_size(union_dr, &size_beta);
/* pick which page to add it to */
if (size_alpha - size_l < size_beta - size_r + WISH_F(v->spl_nleft, v->spl_nright, 0.01))
{
if (datum_l)
pfree(datum_l);
if (union_dr)
pfree(union_dr);
datum_l = union_dl;
size_l = size_alpha;
*left++ = i;
v->spl_nleft++;
}
else
{
if (datum_r)
pfree(datum_r);
if (union_dl)
pfree(union_dl);
datum_r = union_dr;
size_r = size_beta;
*right++ = i;
v->spl_nright++;
}
}
pfree(costvector);
*right = *left = FirstOffsetNumber;
datum_l->flags &= ~LEAFKEY;
datum_r->flags &= ~LEAFKEY;
v->spl_ldatum = PointerGetDatum(datum_l);
v->spl_rdatum = PointerGetDatum(datum_r);
PG_RETURN_POINTER(v);
}
/*
* Check whether supplied input is valid JSON.
*/
static void
json_validate_cstring(char *input)
{
JsonLexContext lex;
JsonParseStack *stack,
*stacktop;
int stacksize;
/* Set up lexing context. */
lex.input = input;
lex.token_terminator = lex.input;
lex.line_number = 1;
lex.line_start = input;
/* Set up parse stack. */
stacksize = 32;
stacktop = palloc(sizeof(JsonParseStack) * stacksize);
stack = stacktop;
stack->state = JSON_PARSE_VALUE;
/* Main parsing loop. */
for (;;)
{
JsonStackOp op;
/* Fetch next token. */
json_lex(&lex);
/* Check for unexpected end of input. */
if (lex.token_start == NULL)
report_parse_error(stack, &lex);
redo:
/* Figure out what to do with this token. */
op = JSON_STACKOP_NONE;
switch (stack->state)
{
case JSON_PARSE_VALUE:
if (lex.token_type != JSON_VALUE_INVALID)
op = JSON_STACKOP_POP;
else if (lex.token_start[0] == '[')
stack->state = JSON_PARSE_ARRAY_START;
else if (lex.token_start[0] == '{')
stack->state = JSON_PARSE_OBJECT_START;
else
report_parse_error(stack, &lex);
break;
case JSON_PARSE_ARRAY_START:
if (lex.token_type != JSON_VALUE_INVALID)
stack->state = JSON_PARSE_ARRAY_NEXT;
else if (lex.token_start[0] == ']')
op = JSON_STACKOP_POP;
else if (lex.token_start[0] == '['
|| lex.token_start[0] == '{')
{
stack->state = JSON_PARSE_ARRAY_NEXT;
op = JSON_STACKOP_PUSH_WITH_PUSHBACK;
}
else
report_parse_error(stack, &lex);
break;
case JSON_PARSE_ARRAY_NEXT:
if (lex.token_type != JSON_VALUE_INVALID)
report_parse_error(stack, &lex);
else if (lex.token_start[0] == ']')
op = JSON_STACKOP_POP;
else if (lex.token_start[0] == ',')
op = JSON_STACKOP_PUSH;
else
report_parse_error(stack, &lex);
break;
case JSON_PARSE_OBJECT_START:
if (lex.token_type == JSON_VALUE_STRING)
stack->state = JSON_PARSE_OBJECT_LABEL;
else if (lex.token_type == JSON_VALUE_INVALID
&& lex.token_start[0] == '}')
op = JSON_STACKOP_POP;
else
report_parse_error(stack, &lex);
break;
case JSON_PARSE_OBJECT_LABEL:
if (lex.token_type == JSON_VALUE_INVALID
&& lex.token_start[0] == ':')
{
stack->state = JSON_PARSE_OBJECT_NEXT;
op = JSON_STACKOP_PUSH;
}
else
report_parse_error(stack, &lex);
break;
case JSON_PARSE_OBJECT_NEXT:
if (lex.token_type != JSON_VALUE_INVALID)
report_parse_error(stack, &lex);
else if (lex.token_start[0] == '}')
op = JSON_STACKOP_POP;
else if (lex.token_start[0] == ',')
stack->state = JSON_PARSE_OBJECT_COMMA;
else
report_parse_error(stack, &lex);
break;
case JSON_PARSE_OBJECT_COMMA:
if (lex.token_type == JSON_VALUE_STRING)
stack->state = JSON_PARSE_OBJECT_LABEL;
else
report_parse_error(stack, &lex);
break;
default:
elog(ERROR, "unexpected json parse state: %d",
(int) stack->state);
}
/* Push or pop the stack, if needed. */
switch (op)
{
case JSON_STACKOP_PUSH:
case JSON_STACKOP_PUSH_WITH_PUSHBACK:
++stack;
if (stack >= &stacktop[stacksize])
{
int stackoffset = stack - stacktop;
stacksize = stacksize + 32;
stacktop = repalloc(stacktop,
sizeof(JsonParseStack) * stacksize);
stack = stacktop + stackoffset;
}
stack->state = JSON_PARSE_VALUE;
if (op == JSON_STACKOP_PUSH_WITH_PUSHBACK)
goto redo;
break;
case JSON_STACKOP_POP:
if (stack == stacktop)
{
/* Expect end of input. */
json_lex(&lex);
if (lex.token_start != NULL)
report_parse_error(NULL, &lex);
return;
}
--stack;
break;
case JSON_STACKOP_NONE:
/* nothing to do */
break;
}
}
}
/*
* AggregateCreate
*/
Oid
AggregateCreate(const char *aggName,
Oid aggNamespace,
int numArgs,
oidvector *parameterTypes,
Datum allParameterTypes,
Datum parameterModes,
Datum parameterNames,
List *parameterDefaults,
List *aggtransfnName,
List *aggfinalfnName,
List *aggsortopName,
Oid aggTransType,
const char *agginitval)
{
Relation aggdesc;
HeapTuple tup;
bool nulls[Natts_pg_aggregate];
Datum values[Natts_pg_aggregate];
Form_pg_proc proc;
Oid transfn;
Oid finalfn = InvalidOid; /* can be omitted */
Oid sortop = InvalidOid; /* can be omitted */
Oid *aggArgTypes = parameterTypes->values;
bool hasPolyArg;
bool hasInternalArg;
Oid rettype;
Oid finaltype;
Oid *fnArgs;
int nargs_transfn;
Oid procOid;
TupleDesc tupDesc;
int i;
ObjectAddress myself,
referenced;
AclResult aclresult;
/* sanity checks (caller should have caught these) */
if (!aggName)
elog(ERROR, "no aggregate name supplied");
if (!aggtransfnName)
elog(ERROR, "aggregate must have a transition function");
/* check for polymorphic and INTERNAL arguments */
hasPolyArg = false;
hasInternalArg = false;
for (i = 0; i < numArgs; i++)
{
if (IsPolymorphicType(aggArgTypes[i]))
hasPolyArg = true;
else if (aggArgTypes[i] == INTERNALOID)
hasInternalArg = true;
}
/*
* If transtype is polymorphic, must have polymorphic argument also; else
* we will have no way to deduce the actual transtype.
*/
if (IsPolymorphicType(aggTransType) && !hasPolyArg)
ereport(ERROR,
(errcode(ERRCODE_INVALID_FUNCTION_DEFINITION),
errmsg("cannot determine transition data type"),
errdetail("An aggregate using a polymorphic transition type must have at least one polymorphic argument.")));
/* find the transfn */
nargs_transfn = numArgs + 1;
fnArgs = (Oid *) palloc(nargs_transfn * sizeof(Oid));
fnArgs[0] = aggTransType;
memcpy(fnArgs + 1, aggArgTypes, numArgs * sizeof(Oid));
transfn = lookup_agg_function(aggtransfnName, nargs_transfn, fnArgs,
&rettype);
/*
* Return type of transfn (possibly after refinement by
* enforce_generic_type_consistency, if transtype isn't polymorphic) must
* exactly match declared transtype.
*
* In the non-polymorphic-transtype case, it might be okay to allow a
* rettype that's binary-coercible to transtype, but I'm not quite
* convinced that it's either safe or useful. When transtype is
* polymorphic we *must* demand exact equality.
*/
if (rettype != aggTransType)
ereport(ERROR,
(errcode(ERRCODE_DATATYPE_MISMATCH),
errmsg("return type of transition function %s is not %s",
NameListToString(aggtransfnName),
format_type_be(aggTransType))));
tup = SearchSysCache1(PROCOID, ObjectIdGetDatum(transfn));
if (!HeapTupleIsValid(tup))
elog(ERROR, "cache lookup failed for function %u", transfn);
proc = (Form_pg_proc) GETSTRUCT(tup);
/*
* If the transfn is strict and the initval is NULL, make sure first input
* type and transtype are the same (or at least binary-compatible), so
* that it's OK to use the first input value as the initial transValue.
*/
if (proc->proisstrict && agginitval == NULL)
{
if (numArgs < 1 ||
!IsBinaryCoercible(aggArgTypes[0], aggTransType))
ereport(ERROR,
(errcode(ERRCODE_INVALID_FUNCTION_DEFINITION),
errmsg("must not omit initial value when transition function is strict and transition type is not compatible with input type")));
}
ReleaseSysCache(tup);
/* handle finalfn, if supplied */
if (aggfinalfnName)
{
fnArgs[0] = aggTransType;
finalfn = lookup_agg_function(aggfinalfnName, 1, fnArgs,
&finaltype);
}
else
{
/*
* If no finalfn, aggregate result type is type of the state value
*/
finaltype = aggTransType;
}
Assert(OidIsValid(finaltype));
/*
* If finaltype (i.e. aggregate return type) is polymorphic, inputs must
* be polymorphic also, else parser will fail to deduce result type.
* (Note: given the previous test on transtype and inputs, this cannot
* happen, unless someone has snuck a finalfn definition into the catalogs
* that itself violates the rule against polymorphic result with no
* polymorphic input.)
*/
if (IsPolymorphicType(finaltype) && !hasPolyArg)
ereport(ERROR,
(errcode(ERRCODE_DATATYPE_MISMATCH),
errmsg("cannot determine result data type"),
errdetail("An aggregate returning a polymorphic type "
"must have at least one polymorphic argument.")));
/*
* Also, the return type can't be INTERNAL unless there's at least one
* INTERNAL argument. This is the same type-safety restriction we enforce
* for regular functions, but at the level of aggregates. We must test
* this explicitly because we allow INTERNAL as the transtype.
*/
if (finaltype == INTERNALOID && !hasInternalArg)
ereport(ERROR,
(errcode(ERRCODE_INVALID_FUNCTION_DEFINITION),
errmsg("unsafe use of pseudo-type \"internal\""),
errdetail("A function returning \"internal\" must have at least one \"internal\" argument.")));
/* handle sortop, if supplied */
if (aggsortopName)
{
if (numArgs != 1)
ereport(ERROR,
(errcode(ERRCODE_INVALID_FUNCTION_DEFINITION),
errmsg("sort operator can only be specified for single-argument aggregates")));
sortop = LookupOperName(NULL, aggsortopName,
aggArgTypes[0], aggArgTypes[0],
false, -1);
}
/*
* permission checks on used types
*/
for (i = 0; i < numArgs; i++)
{
aclresult = pg_type_aclcheck(aggArgTypes[i], GetUserId(), ACL_USAGE);
if (aclresult != ACLCHECK_OK)
aclcheck_error_type(aclresult, aggArgTypes[i]);
}
aclresult = pg_type_aclcheck(aggTransType, GetUserId(), ACL_USAGE);
if (aclresult != ACLCHECK_OK)
aclcheck_error_type(aclresult, aggTransType);
aclresult = pg_type_aclcheck(finaltype, GetUserId(), ACL_USAGE);
if (aclresult != ACLCHECK_OK)
aclcheck_error_type(aclresult, finaltype);
/*
* Everything looks okay. Try to create the pg_proc entry for the
* aggregate. (This could fail if there's already a conflicting entry.)
*/
procOid = ProcedureCreate(aggName,
aggNamespace,
false, /* no replacement */
false, /* doesn't return a set */
finaltype, /* returnType */
GetUserId(), /* proowner */
INTERNALlanguageId, /* languageObjectId */
InvalidOid, /* no validator */
"aggregate_dummy", /* placeholder proc */
NULL, /* probin */
true, /* isAgg */
false, /* isWindowFunc */
false, /* security invoker (currently not
* definable for agg) */
false, /* isLeakProof */
false, /* isStrict (not needed for agg) */
PROVOLATILE_IMMUTABLE, /* volatility (not
* needed for agg) */
parameterTypes, /* paramTypes */
allParameterTypes, /* allParamTypes */
parameterModes, /* parameterModes */
parameterNames, /* parameterNames */
parameterDefaults, /* parameterDefaults */
PointerGetDatum(NULL), /* proconfig */
1, /* procost */
0); /* prorows */
/*
* Okay to create the pg_aggregate entry.
*/
/* initialize nulls and values */
for (i = 0; i < Natts_pg_aggregate; i++)
{
nulls[i] = false;
values[i] = (Datum) NULL;
}
values[Anum_pg_aggregate_aggfnoid - 1] = ObjectIdGetDatum(procOid);
values[Anum_pg_aggregate_aggtransfn - 1] = ObjectIdGetDatum(transfn);
values[Anum_pg_aggregate_aggfinalfn - 1] = ObjectIdGetDatum(finalfn);
values[Anum_pg_aggregate_aggsortop - 1] = ObjectIdGetDatum(sortop);
values[Anum_pg_aggregate_aggtranstype - 1] = ObjectIdGetDatum(aggTransType);
if (agginitval)
values[Anum_pg_aggregate_agginitval - 1] = CStringGetTextDatum(agginitval);
else
nulls[Anum_pg_aggregate_agginitval - 1] = true;
aggdesc = heap_open(AggregateRelationId, RowExclusiveLock);
tupDesc = aggdesc->rd_att;
tup = heap_form_tuple(tupDesc, values, nulls);
simple_heap_insert(aggdesc, tup);
CatalogUpdateIndexes(aggdesc, tup);
heap_close(aggdesc, RowExclusiveLock);
/*
* Create dependencies for the aggregate (above and beyond those already
* made by ProcedureCreate). Note: we don't need an explicit dependency
* on aggTransType since we depend on it indirectly through transfn.
*/
myself.classId = ProcedureRelationId;
myself.objectId = procOid;
myself.objectSubId = 0;
/* Depends on transition function */
referenced.classId = ProcedureRelationId;
referenced.objectId = transfn;
referenced.objectSubId = 0;
recordDependencyOn(&myself, &referenced, DEPENDENCY_NORMAL);
/* Depends on final function, if any */
if (OidIsValid(finalfn))
{
referenced.classId = ProcedureRelationId;
referenced.objectId = finalfn;
referenced.objectSubId = 0;
recordDependencyOn(&myself, &referenced, DEPENDENCY_NORMAL);
}
/* Depends on sort operator, if any */
if (OidIsValid(sortop))
{
referenced.classId = OperatorRelationId;
referenced.objectId = sortop;
referenced.objectSubId = 0;
recordDependencyOn(&myself, &referenced, DEPENDENCY_NORMAL);
}
return procOid;
}
Datum
pg_freespacemap_relations(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
Datum result;
MemoryContext oldcontext;
FreeSpaceRelationsContext *fctx; /* User function context. */
TupleDesc tupledesc;
HeapTuple tuple;
FSMHeader *FreeSpaceMap; /* FSM main structure. */
FSMRelation *fsmrel; /* Individual relation. */
if (SRF_IS_FIRSTCALL())
{
int i;
int numRelations; /* Max no. of Relations in map. */
/*
* Get the free space map data structure.
*/
FreeSpaceMap = GetFreeSpaceMap();
numRelations = MaxFSMRelations;
funcctx = SRF_FIRSTCALL_INIT();
/* Switch context when allocating stuff to be used in later calls */
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/*
* Create a function context for cross-call persistence.
*/
fctx = (FreeSpaceRelationsContext *) palloc(sizeof(FreeSpaceRelationsContext));
funcctx->user_fctx = fctx;
/* Construct a tuple descriptor for the result rows. */
tupledesc = CreateTemplateTupleDesc(NUM_FREESPACE_RELATIONS_ELEM, false);
TupleDescInitEntry(tupledesc, (AttrNumber) 1, "reltablespace",
OIDOID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 2, "reldatabase",
OIDOID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 3, "relfilenode",
OIDOID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 4, "avgrequest",
INT4OID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 5, "interestingpages",
INT4OID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 6, "storedpages",
INT4OID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 7, "nextpage",
INT4OID, -1, 0);
fctx->tupdesc = BlessTupleDesc(tupledesc);
/*
* Allocate numRelations worth of FreeSpaceRelationsRec records, this
* is also an upper bound.
*/
fctx->record = (FreeSpaceRelationsRec *) palloc(sizeof(FreeSpaceRelationsRec) * numRelations);
/* Return to original context when allocating transient memory */
MemoryContextSwitchTo(oldcontext);
/*
* Lock free space map and scan though all the relations.
*/
LWLockAcquire(FreeSpaceLock, LW_EXCLUSIVE);
i = 0;
for (fsmrel = FreeSpaceMap->usageList; fsmrel; fsmrel = fsmrel->nextUsage)
{
fctx->record[i].reltablespace = fsmrel->key.spcNode;
fctx->record[i].reldatabase = fsmrel->key.dbNode;
fctx->record[i].relfilenode = fsmrel->key.relNode;
fctx->record[i].avgrequest = (int64) fsmrel->avgRequest;
fctx->record[i].interestingpages = fsmrel->interestingPages;
fctx->record[i].storedpages = fsmrel->storedPages;
fctx->record[i].nextpage = fsmrel->nextPage;
fctx->record[i].isindex = fsmrel->isIndex;
i++;
}
/* Release free space map. */
LWLockRelease(FreeSpaceLock);
/* Set the real no. of calls as we know it now! */
Assert(i <= numRelations);
funcctx->max_calls = i;
}
funcctx = SRF_PERCALL_SETUP();
/* Get the saved state */
fctx = funcctx->user_fctx;
if (funcctx->call_cntr < funcctx->max_calls)
{
int i = funcctx->call_cntr;
FreeSpaceRelationsRec *record = &fctx->record[i];
Datum values[NUM_FREESPACE_RELATIONS_ELEM];
bool nulls[NUM_FREESPACE_RELATIONS_ELEM];
values[0] = ObjectIdGetDatum(record->reltablespace);
nulls[0] = false;
values[1] = ObjectIdGetDatum(record->reldatabase);
nulls[1] = false;
values[2] = ObjectIdGetDatum(record->relfilenode);
nulls[2] = false;
/*
* avgrequest isn't meaningful for an index
*/
if (record->isindex)
{
nulls[3] = true;
}
else
{
values[3] = UInt32GetDatum(record->avgrequest);
nulls[3] = false;
}
values[4] = Int32GetDatum(record->interestingpages);
nulls[4] = false;
values[5] = Int32GetDatum(record->storedpages);
nulls[5] = false;
values[6] = Int32GetDatum(record->nextpage);
nulls[6] = false;
/* Build and return the tuple. */
tuple = heap_form_tuple(fctx->tupdesc, values, nulls);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
}
else
SRF_RETURN_DONE(funcctx);
}
/*
* array_typanalyze -- typanalyze function for array columns
*/
Datum
array_typanalyze(PG_FUNCTION_ARGS)
{
VacAttrStats *stats = (VacAttrStats *) PG_GETARG_POINTER(0);
Oid element_typeid;
TypeCacheEntry *typentry;
ArrayAnalyzeExtraData *extra_data;
/*
* Call the standard typanalyze function. It may fail to find needed
* operators, in which case we also can't do anything, so just fail.
*/
if (!std_typanalyze(stats))
PG_RETURN_BOOL(false);
/*
* Check attribute data type is a varlena array (or a domain over one).
*/
element_typeid = get_base_element_type(stats->attrtypid);
if (!OidIsValid(element_typeid))
elog(ERROR, "array_typanalyze was invoked for non-array type %u",
stats->attrtypid);
/*
* Gather information about the element type. If we fail to find
* something, return leaving the state from std_typanalyze() in place.
*/
typentry = lookup_type_cache(element_typeid,
TYPECACHE_EQ_OPR |
TYPECACHE_CMP_PROC_FINFO |
TYPECACHE_HASH_PROC_FINFO);
if (!OidIsValid(typentry->eq_opr) ||
!OidIsValid(typentry->cmp_proc_finfo.fn_oid) ||
!OidIsValid(typentry->hash_proc_finfo.fn_oid))
PG_RETURN_BOOL(true);
/* Store our findings for use by compute_array_stats() */
extra_data = (ArrayAnalyzeExtraData *) palloc(sizeof(ArrayAnalyzeExtraData));
extra_data->type_id = typentry->type_id;
extra_data->eq_opr = typentry->eq_opr;
extra_data->typbyval = typentry->typbyval;
extra_data->typlen = typentry->typlen;
extra_data->typalign = typentry->typalign;
extra_data->cmp = &typentry->cmp_proc_finfo;
extra_data->hash = &typentry->hash_proc_finfo;
/* Save old compute_stats and extra_data for scalar statistics ... */
extra_data->std_compute_stats = stats->compute_stats;
extra_data->std_extra_data = stats->extra_data;
/* ... and replace with our info */
stats->compute_stats = compute_array_stats;
stats->extra_data = extra_data;
/*
* Note we leave stats->minrows set as std_typanalyze set it. Should it
* be increased for array analysis purposes?
*/
PG_RETURN_BOOL(true);
}
Datum
pg_freespacemap_pages(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
Datum result;
MemoryContext oldcontext;
FreeSpacePagesContext *fctx; /* User function context. */
TupleDesc tupledesc;
HeapTuple tuple;
FSMHeader *FreeSpaceMap; /* FSM main structure. */
FSMRelation *fsmrel; /* Individual relation. */
if (SRF_IS_FIRSTCALL())
{
int i;
int nchunks; /* Size of freespace.c's arena. */
int numPages; /* Max possible no. of pages in map. */
int nPages; /* Mapped pages for a relation. */
/*
* Get the free space map data structure.
*/
FreeSpaceMap = GetFreeSpaceMap();
/* this must match calculation in InitFreeSpaceMap(): */
nchunks = (MaxFSMPages - 1) / CHUNKPAGES + 1;
/* Worst case (lots of indexes) could have this many pages: */
numPages = nchunks * INDEXCHUNKPAGES;
funcctx = SRF_FIRSTCALL_INIT();
/* Switch context when allocating stuff to be used in later calls */
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/*
* Create a function context for cross-call persistence.
*/
fctx = (FreeSpacePagesContext *) palloc(sizeof(FreeSpacePagesContext));
funcctx->user_fctx = fctx;
/* Construct a tuple descriptor for the result rows. */
tupledesc = CreateTemplateTupleDesc(NUM_FREESPACE_PAGES_ELEM, false);
TupleDescInitEntry(tupledesc, (AttrNumber) 1, "reltablespace",
OIDOID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 2, "reldatabase",
OIDOID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 3, "relfilenode",
OIDOID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 4, "relblocknumber",
INT8OID, -1, 0);
TupleDescInitEntry(tupledesc, (AttrNumber) 5, "bytes",
INT4OID, -1, 0);
fctx->tupdesc = BlessTupleDesc(tupledesc);
/*
* Allocate numPages worth of FreeSpacePagesRec records, this is an
* upper bound.
*/
fctx->record = (FreeSpacePagesRec *) palloc(sizeof(FreeSpacePagesRec) * numPages);
/* Return to original context when allocating transient memory */
MemoryContextSwitchTo(oldcontext);
/*
* Lock free space map and scan though all the relations. For each
* relation, gets all its mapped pages.
*/
LWLockAcquire(FreeSpaceLock, LW_EXCLUSIVE);
i = 0;
for (fsmrel = FreeSpaceMap->usageList; fsmrel; fsmrel = fsmrel->nextUsage)
{
if (fsmrel->isIndex)
{
/* Index relation. */
IndexFSMPageData *page;
page = (IndexFSMPageData *)
(FreeSpaceMap->arena + fsmrel->firstChunk * CHUNKBYTES);
for (nPages = 0; nPages < fsmrel->storedPages; nPages++)
{
fctx->record[i].reltablespace = fsmrel->key.spcNode;
fctx->record[i].reldatabase = fsmrel->key.dbNode;
fctx->record[i].relfilenode = fsmrel->key.relNode;
fctx->record[i].relblocknumber = IndexFSMPageGetPageNum(page);
fctx->record[i].bytes = 0;
fctx->record[i].isindex = true;
page++;
i++;
}
}
else
{
/* Heap relation. */
FSMPageData *page;
page = (FSMPageData *)
(FreeSpaceMap->arena + fsmrel->firstChunk * CHUNKBYTES);
for (nPages = 0; nPages < fsmrel->storedPages; nPages++)
{
fctx->record[i].reltablespace = fsmrel->key.spcNode;
fctx->record[i].reldatabase = fsmrel->key.dbNode;
fctx->record[i].relfilenode = fsmrel->key.relNode;
fctx->record[i].relblocknumber = FSMPageGetPageNum(page);
fctx->record[i].bytes = FSMPageGetSpace(page);
fctx->record[i].isindex = false;
page++;
i++;
}
}
}
/* Release free space map. */
LWLockRelease(FreeSpaceLock);
/* Set the real no. of calls as we know it now! */
Assert(i <= numPages);
funcctx->max_calls = i;
}
funcctx = SRF_PERCALL_SETUP();
/* Get the saved state */
fctx = funcctx->user_fctx;
if (funcctx->call_cntr < funcctx->max_calls)
{
int i = funcctx->call_cntr;
FreeSpacePagesRec *record = &fctx->record[i];
Datum values[NUM_FREESPACE_PAGES_ELEM];
bool nulls[NUM_FREESPACE_PAGES_ELEM];
values[0] = ObjectIdGetDatum(record->reltablespace);
nulls[0] = false;
values[1] = ObjectIdGetDatum(record->reldatabase);
nulls[1] = false;
values[2] = ObjectIdGetDatum(record->relfilenode);
nulls[2] = false;
values[3] = Int64GetDatum((int64) record->relblocknumber);
nulls[3] = false;
/*
* Set (free) bytes to NULL for an index relation.
*/
if (record->isindex)
{
nulls[4] = true;
}
else
{
values[4] = UInt32GetDatum(record->bytes);
nulls[4] = false;
}
/* Build and return the tuple. */
tuple = heap_form_tuple(fctx->tupdesc, values, nulls);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
}
else
SRF_RETURN_DONE(funcctx);
}
/* ----------------------------------------------------------------
* ExecGather(node)
*
* Scans the relation via multiple workers and returns
* the next qualifying tuple.
* ----------------------------------------------------------------
*/
static TupleTableSlot *
ExecGather(PlanState *pstate)
{
GatherState *node = castNode(GatherState, pstate);
TupleTableSlot *slot;
ExprContext *econtext;
CHECK_FOR_INTERRUPTS();
/*
* Initialize the parallel context and workers on first execution. We do
* this on first execution rather than during node initialization, as it
* needs to allocate a large dynamic segment, so it is better to do it
* only if it is really needed.
*/
if (!node->initialized)
{
EState *estate = node->ps.state;
Gather *gather = (Gather *) node->ps.plan;
/*
* Sometimes we might have to run without parallelism; but if parallel
* mode is active then we can try to fire up some workers.
*/
if (gather->num_workers > 0 && estate->es_use_parallel_mode)
{
ParallelContext *pcxt;
/* Initialize, or re-initialize, shared state needed by workers. */
if (!node->pei)
node->pei = ExecInitParallelPlan(node->ps.lefttree,
estate,
gather->initParam,
gather->num_workers,
node->tuples_needed);
else
ExecParallelReinitialize(node->ps.lefttree,
node->pei,
gather->initParam);
/*
* Register backend workers. We might not get as many as we
* requested, or indeed any at all.
*/
pcxt = node->pei->pcxt;
LaunchParallelWorkers(pcxt);
/* We save # workers launched for the benefit of EXPLAIN */
node->nworkers_launched = pcxt->nworkers_launched;
/* Set up tuple queue readers to read the results. */
if (pcxt->nworkers_launched > 0)
{
ExecParallelCreateReaders(node->pei);
/* Make a working array showing the active readers */
node->nreaders = pcxt->nworkers_launched;
node->reader = (TupleQueueReader **)
palloc(node->nreaders * sizeof(TupleQueueReader *));
memcpy(node->reader, node->pei->reader,
node->nreaders * sizeof(TupleQueueReader *));
}
else
{
/* No workers? Then never mind. */
node->nreaders = 0;
node->reader = NULL;
}
node->nextreader = 0;
}
/* Run plan locally if no workers or enabled and not single-copy. */
node->need_to_scan_locally = (node->nreaders == 0)
|| (!gather->single_copy && parallel_leader_participation);
node->initialized = true;
}
/*
* Reset per-tuple memory context to free any expression evaluation
* storage allocated in the previous tuple cycle.
*/
econtext = node->ps.ps_ExprContext;
ResetExprContext(econtext);
/*
* Get next tuple, either from one of our workers, or by running the plan
* ourselves.
*/
slot = gather_getnext(node);
if (TupIsNull(slot))
return NULL;
/* If no projection is required, we're done. */
if (node->ps.ps_ProjInfo == NULL)
return slot;
/*
* Form the result tuple using ExecProject(), and return it.
*/
econtext->ecxt_outertuple = slot;
return ExecProject(node->ps.ps_ProjInfo);
}
/*
* Write WAL record of a page split.
*/
XLogRecPtr
gistXLogSplit(RelFileNode node, BlockNumber blkno, bool page_is_leaf,
SplitedPageLayout *dist,
BlockNumber origrlink, GistNSN orignsn,
Buffer leftchildbuf, bool markfollowright)
{
XLogRecData *rdata;
gistxlogPageSplit xlrec;
SplitedPageLayout *ptr;
int npage = 0,
cur;
XLogRecPtr recptr;
for (ptr = dist; ptr; ptr = ptr->next)
npage++;
rdata = (XLogRecData *) palloc(sizeof(XLogRecData) * (npage * 2 + 2));
xlrec.node = node;
xlrec.origblkno = blkno;
xlrec.origrlink = origrlink;
xlrec.orignsn = orignsn;
xlrec.origleaf = page_is_leaf;
xlrec.npage = (uint16) npage;
xlrec.leftchild =
BufferIsValid(leftchildbuf) ? BufferGetBlockNumber(leftchildbuf) : InvalidBlockNumber;
xlrec.markfollowright = markfollowright;
rdata[0].data = (char *) &xlrec;
rdata[0].len = sizeof(gistxlogPageSplit);
rdata[0].buffer = InvalidBuffer;
cur = 1;
/*
* Include a full page image of the child buf. (only necessary if a
* checkpoint happened since the child page was split)
*/
if (BufferIsValid(leftchildbuf))
{
rdata[cur - 1].next = &(rdata[cur]);
rdata[cur].data = NULL;
rdata[cur].len = 0;
rdata[cur].buffer = leftchildbuf;
rdata[cur].buffer_std = true;
cur++;
}
for (ptr = dist; ptr; ptr = ptr->next)
{
rdata[cur - 1].next = &(rdata[cur]);
rdata[cur].buffer = InvalidBuffer;
rdata[cur].data = (char *) &(ptr->block);
rdata[cur].len = sizeof(gistxlogPage);
cur++;
rdata[cur - 1].next = &(rdata[cur]);
rdata[cur].buffer = InvalidBuffer;
rdata[cur].data = (char *) (ptr->list);
rdata[cur].len = ptr->lenlist;
cur++;
}
rdata[cur - 1].next = NULL;
recptr = XLogInsert(RM_GIST_ID, XLOG_GIST_PAGE_SPLIT, rdata);
pfree(rdata);
return recptr;
}
/*
* Connect to remote server using specified server and user mapping properties.
*/
static PGconn *
connect_pg_server(ForeignServer *server, UserMapping *user)
{
PGconn *volatile conn = NULL;
/*
* Use PG_TRY block to ensure closing connection on error.
*/
PG_TRY();
{
const char **keywords;
const char **values;
int n;
/*
* Construct connection params from generic options of ForeignServer
* and UserMapping. (Some of them might not be libpq options, in
* which case we'll just waste a few array slots.) Add 3 extra slots
* for fallback_application_name, client_encoding, end marker.
*/
n = list_length(server->options) + list_length(user->options) + 3;
keywords = (const char **) palloc(n * sizeof(char *));
values = (const char **) palloc(n * sizeof(char *));
n = 0;
n += ExtractConnectionOptions(server->options,
keywords + n, values + n);
n += ExtractConnectionOptions(user->options,
keywords + n, values + n);
/* Use "postgres_fdw" as fallback_application_name. */
keywords[n] = "fallback_application_name";
values[n] = "postgres_fdw";
n++;
/* Set client_encoding so that libpq can convert encoding properly. */
keywords[n] = "client_encoding";
values[n] = GetDatabaseEncodingName();
n++;
keywords[n] = values[n] = NULL;
/* verify connection parameters and make connection */
check_conn_params(keywords, values);
conn = PQconnectdbParams(keywords, values, false);
if (!conn || PQstatus(conn) != CONNECTION_OK)
ereport(ERROR,
(errcode(ERRCODE_SQLCLIENT_UNABLE_TO_ESTABLISH_SQLCONNECTION),
errmsg("could not connect to server \"%s\"",
server->servername),
errdetail_internal("%s", pchomp(PQerrorMessage(conn)))));
/*
* Check that non-superuser has used password to establish connection;
* otherwise, he's piggybacking on the postgres server's user
* identity. See also dblink_security_check() in contrib/dblink.
*/
if (!superuser() && !PQconnectionUsedPassword(conn))
ereport(ERROR,
(errcode(ERRCODE_S_R_E_PROHIBITED_SQL_STATEMENT_ATTEMPTED),
errmsg("password is required"),
errdetail("Non-superuser cannot connect if the server does not request a password."),
errhint("Target server's authentication method must be changed.")));
/* Prepare new session for use */
configure_remote_session(conn);
pfree(keywords);
pfree(values);
}
PG_CATCH();
{
/* Release PGconn data structure if we managed to create one */
if (conn)
PQfinish(conn);
PG_RE_THROW();
}
PG_END_TRY();
return conn;
}
/*
* Write XLOG record describing a page update. The update can include any
* number of deletions and/or insertions of tuples on a single index page.
*
* If this update inserts a downlink for a split page, also record that
* the F_FOLLOW_RIGHT flag on the child page is cleared and NSN set.
*
* Note that both the todelete array and the tuples are marked as belonging
* to the target buffer; they need not be stored in XLOG if XLogInsert decides
* to log the whole buffer contents instead. Also, we take care that there's
* at least one rdata item referencing the buffer, even when ntodelete and
* ituplen are both zero; this ensures that XLogInsert knows about the buffer.
*/
XLogRecPtr
gistXLogUpdate(RelFileNode node, Buffer buffer,
OffsetNumber *todelete, int ntodelete,
IndexTuple *itup, int ituplen,
Buffer leftchildbuf)
{
XLogRecData *rdata;
gistxlogPageUpdate xlrec;
int cur,
i;
XLogRecPtr recptr;
rdata = (XLogRecData *) palloc(sizeof(XLogRecData) * (3 + ituplen));
xlrec.node = node;
xlrec.blkno = BufferGetBlockNumber(buffer);
xlrec.ntodelete = ntodelete;
xlrec.leftchild =
BufferIsValid(leftchildbuf) ? BufferGetBlockNumber(leftchildbuf) : InvalidBlockNumber;
rdata[0].data = (char *) &xlrec;
rdata[0].len = sizeof(gistxlogPageUpdate);
rdata[0].buffer = InvalidBuffer;
rdata[0].next = &(rdata[1]);
rdata[1].data = (char *) todelete;
rdata[1].len = sizeof(OffsetNumber) * ntodelete;
rdata[1].buffer = buffer;
rdata[1].buffer_std = true;
cur = 2;
/* new tuples */
for (i = 0; i < ituplen; i++)
{
rdata[cur - 1].next = &(rdata[cur]);
rdata[cur].data = (char *) (itup[i]);
rdata[cur].len = IndexTupleSize(itup[i]);
rdata[cur].buffer = buffer;
rdata[cur].buffer_std = true;
cur++;
}
/*
* Include a full page image of the child buf. (only necessary if a
* checkpoint happened since the child page was split)
*/
if (BufferIsValid(leftchildbuf))
{
rdata[cur - 1].next = &(rdata[cur]);
rdata[cur].data = NULL;
rdata[cur].len = 0;
rdata[cur].buffer = leftchildbuf;
rdata[cur].buffer_std = true;
cur++;
}
rdata[cur - 1].next = NULL;
recptr = XLogInsert(RM_GIST_ID, XLOG_GIST_PAGE_UPDATE, rdata);
pfree(rdata);
return recptr;
}
Datum
xslt_process(PG_FUNCTION_ARGS)
{
#ifdef USE_LIBXSLT
text *doct = PG_GETARG_TEXT_P(0);
text *ssheet = PG_GETARG_TEXT_P(1);
text *paramstr;
const char **params;
xsltStylesheetPtr stylesheet = NULL;
xmlDocPtr doctree;
xmlDocPtr restree;
xmlDocPtr ssdoc = NULL;
xmlChar *resstr;
int resstat;
int reslen;
if (fcinfo->nargs == 3)
{
paramstr = PG_GETARG_TEXT_P(2);
params = parse_params(paramstr);
}
else
{
/* No parameters */
params = (const char **) palloc(sizeof(char *));
params[0] = NULL;
}
/* Setup parser */
pgxml_parser_init();
/* Check to see if document is a file or a literal */
if (VARDATA(doct)[0] == '<')
doctree = xmlParseMemory((char *) VARDATA(doct), VARSIZE(doct) - VARHDRSZ);
else
doctree = xmlParseFile(text_to_cstring(doct));
if (doctree == NULL)
xml_ereport(ERROR, ERRCODE_EXTERNAL_ROUTINE_EXCEPTION,
"error parsing XML document");
/* Same for stylesheet */
if (VARDATA(ssheet)[0] == '<')
{
ssdoc = xmlParseMemory((char *) VARDATA(ssheet),
VARSIZE(ssheet) - VARHDRSZ);
if (ssdoc == NULL)
{
xmlFreeDoc(doctree);
xml_ereport(ERROR, ERRCODE_EXTERNAL_ROUTINE_EXCEPTION,
"error parsing stylesheet as XML document");
}
stylesheet = xsltParseStylesheetDoc(ssdoc);
}
else
stylesheet = xsltParseStylesheetFile((xmlChar *) text_to_cstring(ssheet));
if (stylesheet == NULL)
{
xmlFreeDoc(doctree);
xsltCleanupGlobals();
xml_ereport(ERROR, ERRCODE_EXTERNAL_ROUTINE_EXCEPTION,
"failed to parse stylesheet");
}
restree = xsltApplyStylesheet(stylesheet, doctree, params);
resstat = xsltSaveResultToString(&resstr, &reslen, restree, stylesheet);
xsltFreeStylesheet(stylesheet);
xmlFreeDoc(restree);
xmlFreeDoc(doctree);
xsltCleanupGlobals();
if (resstat < 0)
PG_RETURN_NULL();
PG_RETURN_TEXT_P(cstring_to_text_with_len((char *) resstr, reslen));
#else /* !USE_LIBXSLT */
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("xslt_process() is not available without libxslt")));
PG_RETURN_NULL();
#endif /* USE_LIBXSLT */
}
/* Either all the labels must be NULL, or none. */
node = SGITNODEPTR(innerTuple);
if (IndexTupleHasNulls(node))
{
SGITITERATE(innerTuple, i, node)
{
if (!IndexTupleHasNulls(node))
elog(ERROR, "some but not all node labels are null in SPGiST inner tuple");
}
/* They're all null, so just return NULL */
return NULL;
}
else
{
nodeLabels = (Datum *) palloc(sizeof(Datum) * innerTuple->nNodes);
SGITITERATE(innerTuple, i, node)
{
if (IndexTupleHasNulls(node))
elog(ERROR, "some but not all node labels are null in SPGiST inner tuple");
nodeLabels[i] = SGNTDATUM(node, state);
}
return nodeLabels;
}
}
/*
* Add a new item to the page, replacing a PLACEHOLDER item if possible.
* Return the location it's inserted at, or InvalidOffsetNumber on failure.
*
* If startOffset isn't NULL, we start searching for placeholders at
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* fdwroutine if it's a foreign table, the FDW function pointers
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*
* If inhparent is true, all we need to do is set up the attr arrays:
* the RelOptInfo actually represents the appendrel formed by an inheritance
* tree, and so the parent rel's physical size and index information isn't
* important for it.
*/
void
get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
/*
* We need not lock the relation since it was already locked, either by
* the rewriter or when expand_inherited_rtentry() added it to the query's
* rangetable.
*/
relation = heap_open(relationObjectId, NoLock);
/* Temporary and unlogged relations are inaccessible during recovery. */
if (!RelationNeedsWAL(relation) && RecoveryInProgress())
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot access temporary or unlogged relations during recovery")));
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
rel->reltablespace = RelationGetForm(relation)->reltablespace;
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* Estimate relation size --- unless it's an inheritance parent, in which
* case the size will be computed later in set_append_rel_pathlist, and we
* must leave it zero for now to avoid bollixing the total_table_pages
* calculation.
*/
if (!inhparent)
estimate_rel_size(relation, rel->attr_widths - rel->min_attr,
&rel->pages, &rel->tuples, &rel->allvisfrac);
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
* Don't bother with indexes for an inheritance parent, either.
*/
if (inhparent ||
(IgnoreSystemIndexes && IsSystemRelation(relation)))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
LOCKMODE lmode;
indexoidlist = RelationGetIndexList(relation);
/*
* For each index, we get the same type of lock that the executor will
* need, and do not release it. This saves a couple of trips to the
* shared lock manager while not creating any real loss of
* concurrency, because no schema changes could be happening on the
* index while we hold lock on the parent rel, and neither lock type
* blocks any other kind of index operation.
*/
if (rel->relid == root->parse->resultRelation)
lmode = RowExclusiveLock;
else
lmode = AccessShareLock;
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
/*
* Extract info from the relation descriptor for the index.
*/
indexRelation = index_open(indexoid, lmode);
index = indexRelation->rd_index;
/*
* Ignore invalid indexes, since they can't safely be used for
* queries. Note that this is OK because the data structure we
* are constructing is only used by the planner --- the executor
* still needs to insert into "invalid" indexes, if they're marked
* IndexIsReady.
*/
if (!IndexIsValid(index))
{
index_close(indexRelation, NoLock);
continue;
}
/*
* If the index is valid, but cannot yet be used, ignore it; but
* mark the plan we are generating as transient. See
* src/backend/access/heap/README.HOT for discussion.
*/
if (index->indcheckxmin &&
!TransactionIdPrecedes(HeapTupleHeaderGetXmin(indexRelation->rd_indextuple->t_data),
TransactionXmin))
{
root->glob->transientPlan = true;
index_close(indexRelation, NoLock);
continue;
}
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->reltablespace =
RelationGetForm(indexRelation)->reltablespace;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->indexcollations = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->opfamily = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->opcintype = (Oid *) palloc(sizeof(Oid) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
info->indexkeys[i] = index->indkey.values[i];
info->indexcollations[i] = indexRelation->rd_indcollation[i];
info->opfamily[i] = indexRelation->rd_opfamily[i];
info->opcintype[i] = indexRelation->rd_opcintype[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->canreturn = index_can_return(indexRelation);
info->amcanorderbyop = indexRelation->rd_am->amcanorderbyop;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
info->amsearcharray = indexRelation->rd_am->amsearcharray;
info->amsearchnulls = indexRelation->rd_am->amsearchnulls;
info->amhasgettuple = OidIsValid(indexRelation->rd_am->amgettuple);
info->amhasgetbitmap = OidIsValid(indexRelation->rd_am->amgetbitmap);
/*
* Fetch the ordering information for the index, if any.
*/
if (info->relam == BTREE_AM_OID)
{
/*
* If it's a btree index, we can use its opfamily OIDs
* directly as the sort ordering opfamily OIDs.
*/
Assert(indexRelation->rd_am->amcanorder);
info->sortopfamily = info->opfamily;
info->reverse_sort = (bool *) palloc(sizeof(bool) * ncolumns);
info->nulls_first = (bool *) palloc(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
info->reverse_sort[i] = (opt & INDOPTION_DESC) != 0;
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
}
}
else if (indexRelation->rd_am->amcanorder)
{
/*
* Otherwise, identify the corresponding btree opfamilies by
* trying to map this index's "<" operators into btree. Since
* "<" uniquely defines the behavior of a sort order, this is
* a sufficient test.
*
* XXX This method is rather slow and also requires the
* undesirable assumption that the other index AM numbers its
* strategies the same as btree. It'd be better to have a way
* to explicitly declare the corresponding btree opfamily for
* each opfamily of the other index type. But given the lack
* of current or foreseeable amcanorder index types, it's not
* worth expending more effort on now.
*/
info->sortopfamily = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->reverse_sort = (bool *) palloc(sizeof(bool) * ncolumns);
info->nulls_first = (bool *) palloc(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
Oid ltopr;
Oid btopfamily;
Oid btopcintype;
int16 btstrategy;
info->reverse_sort[i] = (opt & INDOPTION_DESC) != 0;
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
ltopr = get_opfamily_member(info->opfamily[i],
info->opcintype[i],
info->opcintype[i],
BTLessStrategyNumber);
if (OidIsValid(ltopr) &&
get_ordering_op_properties(ltopr,
&btopfamily,
&btopcintype,
&btstrategy) &&
btopcintype == info->opcintype[i] &&
btstrategy == BTLessStrategyNumber)
{
/* Successful mapping */
info->sortopfamily[i] = btopfamily;
}
else
{
/* Fail ... quietly treat index as unordered */
info->sortopfamily = NULL;
info->reverse_sort = NULL;
info->nulls_first = NULL;
break;
}
}
}
else
{
info->sortopfamily = NULL;
info->reverse_sort = NULL;
info->nulls_first = NULL;
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
/* Build targetlist using the completed indexprs data */
info->indextlist = build_index_tlist(root, info, relation);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
info->immediate = index->indimmediate;
info->hypothetical = false;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
}
else
{
double allvisfrac; /* dummy */
estimate_rel_size(indexRelation, NULL,
&info->pages, &info->tuples, &allvisfrac);
if (info->tuples > rel->tuples)
info->tuples = rel->tuples;
}
if (info->relam == BTREE_AM_OID)
{
/* For btrees, get tree height while we have the index open */
info->tree_height = _bt_getrootheight(indexRelation);
}
else
{
/* For other index types, just set it to "unknown" for now */
info->tree_height = -1;
}
index_close(indexRelation, NoLock);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
/*
** GiST Compress and Decompress methods
*/
Datum
g_int_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval;
ArrayType *r;
int len;
int *dr;
int i,
min,
cand;
if (entry->leafkey)
{
r = (ArrayType *) PG_DETOAST_DATUM_COPY(entry->key);
PREPAREARR(r);
r->flags |= LEAFKEY;
retval = palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(r),
entry->rel, entry->page, entry->offset, VARSIZE(r), FALSE);
PG_RETURN_POINTER(retval);
}
r = (ArrayType *) PG_DETOAST_DATUM(entry->key);
if (ISLEAFKEY(r) || ARRISVOID(r))
{
if (r != (ArrayType *) DatumGetPointer(entry->key))
pfree(r);
PG_RETURN_POINTER(entry);
}
if ((len = ARRNELEMS(r)) >= 2 * MAXNUMRANGE)
{ /* compress */
if (r == (ArrayType *) DatumGetPointer(entry->key))
r = (ArrayType *) PG_DETOAST_DATUM_COPY(entry->key);
r = resize_intArrayType(r, 2 * (len));
dr = ARRPTR(r);
for (i = len - 1; i >= 0; i--)
dr[2 * i] = dr[2 * i + 1] = dr[i];
len *= 2;
cand = 1;
while (len > MAXNUMRANGE * 2)
{
min = 0x7fffffff;
for (i = 2; i < len; i += 2)
if (min > (dr[i] - dr[i - 1]))
{
min = (dr[i] - dr[i - 1]);
cand = i;
}
memmove((void *) &dr[cand - 1], (void *) &dr[cand + 1], (len - cand - 1) * sizeof(int));
len -= 2;
}
r = resize_intArrayType(r, len);
retval = palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(r),
entry->rel, entry->page, entry->offset, VARSIZE(r), FALSE);
PG_RETURN_POINTER(retval);
}
else
PG_RETURN_POINTER(entry);
PG_RETURN_POINTER(entry);
}
PGDLLEXPORT Datum
kshortest_path(PG_FUNCTION_ARGS) {
FuncCallContext *funcctx;
TupleDesc tuple_desc;
General_path_element_t *path = NULL;
size_t result_count = 0;
/* stuff done only on the first call of the function */
if (SRF_IS_FIRSTCALL()) {
MemoryContext oldcontext;
funcctx = SRF_FIRSTCALL_INIT();
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/*
CREATE OR REPLACE FUNCTION _pgr_ksp(
sql text,
start_vid bigint,
end_vid bigint,
k integer,
directed boolean,
heap_paths boolean
*/
PGR_DBG("Calling process");
compute(
text_to_cstring(PG_GETARG_TEXT_P(0)), /* SQL */
PG_GETARG_INT64(1), /* start_vid */
PG_GETARG_INT64(2), /* end_vid */
PG_GETARG_INT32(3), /* k */
PG_GETARG_BOOL(4), /* directed */
PG_GETARG_BOOL(5), /* heap_paths */
&path,
&result_count);
PGR_DBG("Total number of tuples to be returned %ld \n", result_count);
/* */
/**********************************************************************/
#if PGSQL_VERSION > 95
funcctx->max_calls = result_count;
#else
funcctx->max_calls = (uint32_t)result_count;
#endif
funcctx->user_fctx = path;
if (get_call_result_type(fcinfo, NULL, &tuple_desc) != TYPEFUNC_COMPOSITE)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("function returning record called in context "
"that cannot accept type record\n")));
funcctx->tuple_desc = tuple_desc;
MemoryContextSwitchTo(oldcontext);
}
funcctx = SRF_PERCALL_SETUP();
tuple_desc = funcctx->tuple_desc;
path = (General_path_element_t*) funcctx->user_fctx;
if (funcctx->call_cntr < funcctx->max_calls) { /* do when there is more left to send */
HeapTuple tuple;
Datum result;
Datum *values;
bool* nulls;
values = palloc(7 * sizeof(Datum));
nulls = palloc(7 * sizeof(bool));
size_t i;
for (i = 0; i < 7; ++i) {
nulls[i] = false;
}
values[0] = Int32GetDatum(funcctx->call_cntr + 1);
values[1] = Int32GetDatum(path[funcctx->call_cntr].start_id + 1);
values[2] = Int32GetDatum(path[funcctx->call_cntr].seq);
values[3] = Int64GetDatum(path[funcctx->call_cntr].node);
values[4] = Int64GetDatum(path[funcctx->call_cntr].edge);
values[5] = Float8GetDatum(path[funcctx->call_cntr].cost);
values[6] = Float8GetDatum(path[funcctx->call_cntr].agg_cost);
tuple = heap_form_tuple(tuple_desc, values, nulls);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
} else { /* do when there is no more left */
SRF_RETURN_DONE(funcctx);
}
}
#endif
/* Set up the elfpart structure and allocate an empty symbol table.
*/
static void new(elfpart *part)
{
static Elf32_Sym blanksym = { 0, 0, 0, 0, 0, SHN_UNDEF };
Elf32_Sym *sym;
part->shtype = SHT_SYMTAB;
part->shname = ".symtab";
part->info = 1;
part->entsize = sizeof(Elf32_Sym);
part->size = part->entsize;
part->count = 1;
sym = palloc(part);
*sym = blanksym;
}
/* Translate the symbols' shndx fields from a part index to a section
* header index.
*/
static void complete(elfpart *part, blueprint const *bp)
{
Elf32_Sym *sym;
int i, n;
for (i = 0, sym = part->part ; i < part->count ; ++i, ++sym) {
if (sym->st_shndx > 0 && sym->st_shndx < bp->partcount) {
n = sym->st_shndx;
sym->st_shndx = 1;
