analyze.c

来自「PostgreSQL7.4.6 for Linux」· C语言 代码 · 共 1,819 行 · 第 1/4 页

/*-------------------------------------------------------------------------
 *
 * analyze.c
 *	  the postgres statistics generator
 *
 * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
 * Portions Copyright (c) 1994, Regents of the University of California
 *
 *
 * IDENTIFICATION
 *	  $Header: /cvsroot/pgsql/src/backend/commands/analyze.c,v 1.64 2003/10/18 15:38:06 tgl Exp $
 *
 *-------------------------------------------------------------------------
 */
#include "postgres.h"

#include <math.h>

#include "access/heapam.h"
#include "access/tuptoaster.h"
#include "catalog/catalog.h"
#include "catalog/catname.h"
#include "catalog/indexing.h"
#include "catalog/namespace.h"
#include "catalog/pg_operator.h"
#include "catalog/pg_statistic.h"
#include "catalog/pg_type.h"
#include "commands/vacuum.h"
#include "miscadmin.h"
#include "parser/parse_oper.h"
#include "parser/parse_relation.h"
#include "utils/acl.h"
#include "utils/builtins.h"
#include "utils/datum.h"
#include "utils/fmgroids.h"
#include "utils/lsyscache.h"
#include "utils/syscache.h"
#include "utils/tuplesort.h"


/*
 * Analysis algorithms supported
 */
typedef enum
{
	ALG_MINIMAL = 1,			/* Compute only most-common-values */
	ALG_SCALAR					/* Compute MCV, histogram, sort
								 * correlation */
} AlgCode;

/*
 * To avoid consuming too much memory during analysis and/or too much space
 * in the resulting pg_statistic rows, we ignore varlena datums that are wider
 * than WIDTH_THRESHOLD (after detoasting!).  This is legitimate for MCV
 * and distinct-value calculations since a wide value is unlikely to be
 * duplicated at all, much less be a most-common value.  For the same reason,
 * ignoring wide values will not affect our estimates of histogram bin
 * boundaries very much.
 */
#define WIDTH_THRESHOLD  1024

/*
 * We build one of these structs for each attribute (column) that is to be
 * analyzed.  The struct and subsidiary data are in anl_context,
 * so they live until the end of the ANALYZE operation.
 */
typedef struct
{
	/* These fields are set up by examine_attribute */
	int			attnum;			/* attribute number */
	AlgCode		algcode;		/* Which algorithm to use for this column */
	int			minrows;		/* Minimum # of rows wanted for stats */
	Form_pg_attribute attr;		/* copy of pg_attribute row for column */
	Form_pg_type attrtype;		/* copy of pg_type row for column */
	Oid			eqopr;			/* '=' operator for datatype, if any */
	Oid			eqfunc;			/* and associated function */
	Oid			ltopr;			/* '<' operator for datatype, if any */

	/*
	 * These fields are filled in by the actual statistics-gathering
	 * routine
	 */
	bool		stats_valid;
	float4		stanullfrac;	/* fraction of entries that are NULL */
	int4		stawidth;		/* average width */
	float4		stadistinct;	/* # distinct values */
	int2		stakind[STATISTIC_NUM_SLOTS];
	Oid			staop[STATISTIC_NUM_SLOTS];
	int			numnumbers[STATISTIC_NUM_SLOTS];
	float4	   *stanumbers[STATISTIC_NUM_SLOTS];
	int			numvalues[STATISTIC_NUM_SLOTS];
	Datum	   *stavalues[STATISTIC_NUM_SLOTS];
} VacAttrStats;


typedef struct
{
	Datum		value;			/* a data value */
	int			tupno;			/* position index for tuple it came from */
} ScalarItem;

typedef struct
{
	int			count;			/* # of duplicates */
	int			first;			/* values[] index of first occurrence */
} ScalarMCVItem;


#define swapInt(a,b)	do {int _tmp; _tmp=a; a=b; b=_tmp;} while(0)
#define swapDatum(a,b)	do {Datum _tmp; _tmp=a; a=b; b=_tmp;} while(0)


/* Default statistics target (GUC parameter) */
int			default_statistics_target = 10;


static int	elevel = -1;

static MemoryContext anl_context = NULL;

/* context information for compare_scalars() */
static FmgrInfo *datumCmpFn;
static SortFunctionKind datumCmpFnKind;
static int *datumCmpTupnoLink;


static VacAttrStats *examine_attribute(Relation onerel, int attnum);
static int acquire_sample_rows(Relation onerel, HeapTuple *rows,
					int targrows, double *totalrows);
static double random_fract(void);
static double init_selection_state(int n);
static double select_next_random_record(double t, int n, double *stateptr);
static int	compare_rows(const void *a, const void *b);
static int	compare_scalars(const void *a, const void *b);
static int	compare_mcvs(const void *a, const void *b);
static void compute_minimal_stats(VacAttrStats *stats,
					  TupleDesc tupDesc, double totalrows,
					  HeapTuple *rows, int numrows);
static void compute_scalar_stats(VacAttrStats *stats,
					 TupleDesc tupDesc, double totalrows,
					 HeapTuple *rows, int numrows);
static void update_attstats(Oid relid, int natts, VacAttrStats **vacattrstats);


/*
 *	analyze_rel() -- analyze one relation
 */
void
analyze_rel(Oid relid, VacuumStmt *vacstmt)
{
	Relation	onerel;
	int			attr_cnt,
				tcnt,
				i;
	VacAttrStats **vacattrstats;
	int			targrows,
				numrows;
	double		totalrows;
	HeapTuple  *rows;

	if (vacstmt->verbose)
		elevel = INFO;
	else
		elevel = DEBUG2;

	/*
	 * Use the current context for storing analysis info.  vacuum.c
	 * ensures that this context will be cleared when I return, thus
	 * releasing the memory allocated here.
	 */
	anl_context = CurrentMemoryContext;

	/*
	 * Check for user-requested abort.	Note we want this to be inside a
	 * transaction, so xact.c doesn't issue useless WARNING.
	 */
	CHECK_FOR_INTERRUPTS();

	/*
	 * Race condition -- if the pg_class tuple has gone away since the
	 * last time we saw it, we don't need to process it.
	 */
	if (!SearchSysCacheExists(RELOID,
							  ObjectIdGetDatum(relid),
							  0, 0, 0))
		return;

	/*
	 * Open the class, getting only a read lock on it, and check
	 * permissions. Permissions check should match vacuum's check!
	 */
	onerel = relation_open(relid, AccessShareLock);

	if (!(pg_class_ownercheck(RelationGetRelid(onerel), GetUserId()) ||
		  (pg_database_ownercheck(MyDatabaseId, GetUserId()) && !onerel->rd_rel->relisshared)))
	{
		/* No need for a WARNING if we already complained during VACUUM */
		if (!vacstmt->vacuum)
			ereport(WARNING,
					(errmsg("skipping \"%s\" --- only table or database owner can analyze it",
							RelationGetRelationName(onerel))));
		relation_close(onerel, AccessShareLock);
		return;
	}

	/*
	 * Check that it's a plain table; we used to do this in getrels() but
	 * seems safer to check after we've locked the relation.
	 */
	if (onerel->rd_rel->relkind != RELKIND_RELATION)
	{
		/* No need for a WARNING if we already complained during VACUUM */
		if (!vacstmt->vacuum)
			ereport(WARNING,
					(errmsg("skipping \"%s\" --- cannot analyze indexes, views, or special system tables",
							RelationGetRelationName(onerel))));
		relation_close(onerel, AccessShareLock);
		return;
	}

	/*
	 * Silently ignore tables that are temp tables of other backends ---
	 * trying to analyze these is rather pointless, since their contents
	 * are probably not up-to-date on disk.  (We don't throw a warning
	 * here; it would just lead to chatter during a database-wide
	 * ANALYZE.)
	 */
	if (isOtherTempNamespace(RelationGetNamespace(onerel)))
	{
		relation_close(onerel, AccessShareLock);
		return;
	}

	/*
	 * We can ANALYZE any table except pg_statistic. See update_attstats
	 */
	if (IsSystemNamespace(RelationGetNamespace(onerel)) &&
	 strcmp(RelationGetRelationName(onerel), StatisticRelationName) == 0)
	{
		relation_close(onerel, AccessShareLock);
		return;
	}

	ereport(elevel,
			(errmsg("analyzing \"%s.%s\"",
					get_namespace_name(RelationGetNamespace(onerel)),
					RelationGetRelationName(onerel))));

	/*
	 * Determine which columns to analyze
	 *
	 * Note that system attributes are never analyzed.
	 */
	if (vacstmt->va_cols != NIL)
	{
		List	   *le;

		vacattrstats = (VacAttrStats **) palloc(length(vacstmt->va_cols) *
												sizeof(VacAttrStats *));
		tcnt = 0;
		foreach(le, vacstmt->va_cols)
		{
			char	   *col = strVal(lfirst(le));

			i = attnameAttNum(onerel, col, false);
			vacattrstats[tcnt] = examine_attribute(onerel, i);
			if (vacattrstats[tcnt] != NULL)
				tcnt++;
		}
		attr_cnt = tcnt;
	}
	else
	{
		attr_cnt = onerel->rd_att->natts;
		/* +1 here is just to avoid palloc(0) with zero-column table */
		vacattrstats = (VacAttrStats **) palloc((attr_cnt + 1) *
												sizeof(VacAttrStats *));
		tcnt = 0;
		for (i = 1; i <= attr_cnt; i++)
		{
			vacattrstats[tcnt] = examine_attribute(onerel, i);
			if (vacattrstats[tcnt] != NULL)
				tcnt++;
		}
		attr_cnt = tcnt;
	}

	/*
	 * Quit if no analyzable columns
	 */
	if (attr_cnt <= 0)
	{
		relation_close(onerel, AccessShareLock);
		return;
	}

	/*
	 * Determine how many rows we need to sample, using the worst case
	 * from all analyzable columns.  We use a lower bound of 100 rows to
	 * avoid possible overflow in Vitter's algorithm.
	 */
	targrows = 100;
	for (i = 0; i < attr_cnt; i++)
	{
		if (targrows < vacattrstats[i]->minrows)
			targrows = vacattrstats[i]->minrows;
	}

	/*
	 * Acquire the sample rows
	 */
	rows = (HeapTuple *) palloc(targrows * sizeof(HeapTuple));
	numrows = acquire_sample_rows(onerel, rows, targrows, &totalrows);

	/*
	 * If we are running a standalone ANALYZE, update pages/tuples stats
	 * in pg_class.  We have the accurate page count from heap_beginscan,
	 * but only an approximate number of tuples; therefore, if we are part
	 * of VACUUM ANALYZE do *not* overwrite the accurate count already
	 * inserted by VACUUM.
	 */
	if (!vacstmt->vacuum)
		vac_update_relstats(RelationGetRelid(onerel),
							onerel->rd_nblocks,
							totalrows,
							RelationGetForm(onerel)->relhasindex);

	/*
	 * Compute the statistics.	Temporary results during the calculations
	 * for each column are stored in a child context.  The calc routines
	 * are responsible to make sure that whatever they store into the
	 * VacAttrStats structure is allocated in anl_context.
	 */
	if (numrows > 0)
	{
		MemoryContext col_context,
					old_context;

		col_context = AllocSetContextCreate(anl_context,
											"Analyze Column",
											ALLOCSET_DEFAULT_MINSIZE,
											ALLOCSET_DEFAULT_INITSIZE,
											ALLOCSET_DEFAULT_MAXSIZE);
		old_context = MemoryContextSwitchTo(col_context);
		for (i = 0; i < attr_cnt; i++)
		{
			switch (vacattrstats[i]->algcode)
			{
				case ALG_MINIMAL:
					compute_minimal_stats(vacattrstats[i],
										  onerel->rd_att, totalrows,
										  rows, numrows);
					break;
				case ALG_SCALAR:
					compute_scalar_stats(vacattrstats[i],
										 onerel->rd_att, totalrows,
										 rows, numrows);
					break;
			}
			MemoryContextResetAndDeleteChildren(col_context);
		}
		MemoryContextSwitchTo(old_context);
		MemoryContextDelete(col_context);

		/*
		 * Emit the completed stats rows into pg_statistic, replacing any
		 * previous statistics for the target columns.	(If there are
		 * stats in pg_statistic for columns we didn't process, we leave
		 * them alone.)
		 */
		update_attstats(relid, attr_cnt, vacattrstats);
	}

	/*
	 * Close source relation now, but keep lock so that no one deletes it
	 * before we commit.  (If someone did, they'd fail to clean up the
	 * entries we made in pg_statistic.)
	 */
	relation_close(onerel, NoLock);
}

/*
 * examine_attribute -- pre-analysis of a single column
 *
 * Determine whether the column is analyzable; if so, create and initialize
 * a VacAttrStats struct for it.  If not, return NULL.
 */
static VacAttrStats *
examine_attribute(Relation onerel, int attnum)
{
	Form_pg_attribute attr = onerel->rd_att->attrs[attnum - 1];
	Operator	func_operator;
	HeapTuple	typtuple;
	Oid			eqopr = InvalidOid;
	Oid			eqfunc = InvalidOid;
	Oid			ltopr = InvalidOid;
	VacAttrStats *stats;

	/* Don't analyze dropped columns */
	if (attr->attisdropped)
		return NULL;

	/* Don't analyze column if user has specified not to */
	if (attr->attstattarget == 0)
		return NULL;

	/* If column has no "=" operator, we can't do much of anything */
	func_operator = equality_oper(attr->atttypid, true);
	if (func_operator != NULL)
	{
		eqopr = oprid(func_operator);
		eqfunc = oprfuncid(func_operator);
		ReleaseSysCache(func_operator);
	}
	if (!OidIsValid(eqfunc))
		return NULL;

	/*
	 * If we have "=" then we're at least able to do the minimal
	 * algorithm, so start filling in a VacAttrStats struct.
	 */
	stats = (VacAttrStats *) palloc0(sizeof(VacAttrStats));
	stats->attnum = attnum;
	stats->attr = (Form_pg_attribute) palloc(ATTRIBUTE_TUPLE_SIZE);
	memcpy(stats->attr, attr, ATTRIBUTE_TUPLE_SIZE);
	typtuple = SearchSysCache(TYPEOID,
							  ObjectIdGetDatum(attr->atttypid),
							  0, 0, 0);
	if (!HeapTupleIsValid(typtuple))
		elog(ERROR, "cache lookup failed for type %u", attr->atttypid);
	stats->attrtype = (Form_pg_type) palloc(sizeof(FormData_pg_type));
	memcpy(stats->attrtype, GETSTRUCT(typtuple), sizeof(FormData_pg_type));
	ReleaseSysCache(typtuple);
	stats->eqopr = eqopr;
	stats->eqfunc = eqfunc;

	/* If the attstattarget column is negative, use the default value */
	if (stats->attr->attstattarget < 0)
		stats->attr->attstattarget = default_statistics_target;

	/* Is there a "<" operator with suitable semantics? */
	func_operator = ordering_oper(attr->atttypid, true);
	if (func_operator != NULL)
	{
		ltopr = oprid(func_operator);
		ReleaseSysCache(func_operator);
	}
	stats->ltopr = ltopr;

	/*
	 * Determine the algorithm to use (this will get more complicated
	 * later)
	 */
	if (OidIsValid(ltopr))
	{