pathnode.c

PostgreSQL 8.1.4的源码 适用于Linux下的开源数据库系统

	 * otherwise punt.
	 */
	if (sub_targetlist)
	{
		pathnode->rows = estimate_num_groups(root, sub_targetlist, rel->rows);
		numCols = list_length(sub_targetlist);
	}
	else
	{
		pathnode->rows = rel->rows;
		numCols = list_length(rel->reltargetlist);
	}

	/*
	 * Estimate cost for sort+unique implementation
	 */
	cost_sort(&sort_path, root, NIL,
			  subpath->total_cost,
			  rel->rows,
			  rel->width);

	/*
	 * Charge one cpu_operator_cost per comparison per input tuple. We assume
	 * all columns get compared at most of the tuples.	(XXX probably this is
	 * an overestimate.)  This should agree with make_unique.
	 */
	sort_path.total_cost += cpu_operator_cost * rel->rows * numCols;

	/*
	 * Is it safe to use a hashed implementation?  If so, estimate and compare
	 * costs.  We only try this if we know the targetlist for sure (else we
	 * can't be sure about the datatypes involved).
	 */
	pathnode->umethod = UNIQUE_PATH_SORT;
	if (enable_hashagg && sub_targetlist && hash_safe_tlist(sub_targetlist))
	{
		/*
		 * Estimate the overhead per hashtable entry at 64 bytes (same as in
		 * planner.c).
		 */
		int			hashentrysize = rel->width + 64;

		if (hashentrysize * pathnode->rows <= work_mem * 1024L)
		{
			cost_agg(&agg_path, root,
					 AGG_HASHED, 0,
					 numCols, pathnode->rows,
					 subpath->startup_cost,
					 subpath->total_cost,
					 rel->rows);
			if (agg_path.total_cost < sort_path.total_cost)
				pathnode->umethod = UNIQUE_PATH_HASH;
		}
	}

	if (pathnode->umethod == UNIQUE_PATH_HASH)
	{
		pathnode->path.startup_cost = agg_path.startup_cost;
		pathnode->path.total_cost = agg_path.total_cost;
	}
	else
	{
		pathnode->path.startup_cost = sort_path.startup_cost;
		pathnode->path.total_cost = sort_path.total_cost;
	}

	rel->cheapest_unique_path = (Path *) pathnode;

	return pathnode;
}

/*
 * translate_sub_tlist - get subquery column numbers represented by tlist
 *
 * The given targetlist should contain only Vars referencing the given relid.
 * Extract their varattnos (ie, the column numbers of the subquery) and return
 * as an integer List.
 *
 * If any of the tlist items is not a simple Var, we cannot determine whether
 * the subquery's uniqueness condition (if any) matches ours, so punt and
 * return NIL.
 */
static List *
translate_sub_tlist(List *tlist, int relid)
{
	List	   *result = NIL;
	ListCell   *l;

	foreach(l, tlist)
	{
		Var		   *var = (Var *) lfirst(l);

		if (!var || !IsA(var, Var) ||
			var->varno != relid)
			return NIL;			/* punt */

		result = lappend_int(result, var->varattno);
	}
	return result;
}

/*
 * query_is_distinct_for - does query never return duplicates of the
 *		specified columns?
 *
 * colnos is an integer list of output column numbers (resno's).  We are
 * interested in whether rows consisting of just these columns are certain
 * to be distinct.
 */
static bool
query_is_distinct_for(Query *query, List *colnos)
{
	ListCell   *l;

	/*
	 * DISTINCT (including DISTINCT ON) guarantees uniqueness if all the
	 * columns in the DISTINCT clause appear in colnos.
	 */
	if (query->distinctClause)
	{
		foreach(l, query->distinctClause)
		{
			SortClause *scl = (SortClause *) lfirst(l);
			TargetEntry *tle = get_sortgroupclause_tle(scl,
													   query->targetList);

			if (!list_member_int(colnos, tle->resno))
				break;			/* exit early if no match */
		}
		if (l == NULL)			/* had matches for all? */
			return true;
	}

	/*
	 * Similarly, GROUP BY guarantees uniqueness if all the grouped columns
	 * appear in colnos.
	 */
	if (query->groupClause)
	{
		foreach(l, query->groupClause)
		{
			GroupClause *grpcl = (GroupClause *) lfirst(l);
			TargetEntry *tle = get_sortgroupclause_tle(grpcl,
													   query->targetList);

			if (!list_member_int(colnos, tle->resno))
				break;			/* exit early if no match */
		}
		if (l == NULL)			/* had matches for all? */
			return true;
	}
	else
	{
		/*
		 * If we have no GROUP BY, but do have aggregates or HAVING, then the
		 * result is at most one row so it's surely unique.
		 */
		if (query->hasAggs || query->havingQual)
			return true;
	}

	/*
	 * UNION, INTERSECT, EXCEPT guarantee uniqueness of the whole output row,
	 * except with ALL
	 */
	if (query->setOperations)
	{
		SetOperationStmt *topop = (SetOperationStmt *) query->setOperations;

		Assert(IsA(topop, SetOperationStmt));
		Assert(topop->op != SETOP_NONE);

		if (!topop->all)
		{
			/* We're good if all the nonjunk output columns are in colnos */
			foreach(l, query->targetList)
			{
				TargetEntry *tle = (TargetEntry *) lfirst(l);

				if (!tle->resjunk &&
					!list_member_int(colnos, tle->resno))
					break;		/* exit early if no match */
			}
			if (l == NULL)		/* had matches for all? */
				return true;
		}
	}

	/*
	 * XXX Are there any other cases in which we can easily see the result
	 * must be distinct?
	 */

	return false;
}

/*
 * hash_safe_tlist - can datatypes of given tlist be hashed?
 *
 * We assume hashed aggregation will work if the datatype's equality operator
 * is marked hashjoinable.
 *
 * XXX this probably should be somewhere else.	See also hash_safe_grouping
 * in plan/planner.c.
 */
static bool
hash_safe_tlist(List *tlist)
{
	ListCell   *tl;

	foreach(tl, tlist)
	{
		Node	   *expr = (Node *) lfirst(tl);
		Operator	optup;
		bool		oprcanhash;

		optup = equality_oper(exprType(expr), true);
		if (!optup)
			return false;
		oprcanhash = ((Form_pg_operator) GETSTRUCT(optup))->oprcanhash;
		ReleaseSysCache(optup);
		if (!oprcanhash)
			return false;
	}
	return true;
}

/*
 * create_subqueryscan_path
 *	  Creates a path corresponding to a sequential scan of a subquery,
 *	  returning the pathnode.
 */
Path *
create_subqueryscan_path(RelOptInfo *rel, List *pathkeys)
{
	Path	   *pathnode = makeNode(Path);

	pathnode->pathtype = T_SubqueryScan;
	pathnode->parent = rel;
	pathnode->pathkeys = pathkeys;

	cost_subqueryscan(pathnode, rel);

	return pathnode;
}

/*
 * create_functionscan_path
 *	  Creates a path corresponding to a sequential scan of a function,
 *	  returning the pathnode.
 */
Path *
create_functionscan_path(PlannerInfo *root, RelOptInfo *rel)
{
	Path	   *pathnode = makeNode(Path);

	pathnode->pathtype = T_FunctionScan;
	pathnode->parent = rel;
	pathnode->pathkeys = NIL;	/* for now, assume unordered result */

	cost_functionscan(pathnode, root, rel);

	return pathnode;
}

/*
 * create_nestloop_path
 *	  Creates a pathnode corresponding to a nestloop join between two
 *	  relations.
 *
 * 'joinrel' is the join relation.
 * 'jointype' is the type of join required
 * 'outer_path' is the outer path
 * 'inner_path' is the inner path
 * 'restrict_clauses' are the RestrictInfo nodes to apply at the join
 * 'pathkeys' are the path keys of the new join path
 *
 * Returns the resulting path node.
 */
NestPath *
create_nestloop_path(PlannerInfo *root,
					 RelOptInfo *joinrel,
					 JoinType jointype,
					 Path *outer_path,
					 Path *inner_path,
					 List *restrict_clauses,
					 List *pathkeys)
{
	NestPath   *pathnode = makeNode(NestPath);

	pathnode->path.pathtype = T_NestLoop;
	pathnode->path.parent = joinrel;
	pathnode->jointype = jointype;
	pathnode->outerjoinpath = outer_path;
	pathnode->innerjoinpath = inner_path;
	pathnode->joinrestrictinfo = restrict_clauses;
	pathnode->path.pathkeys = pathkeys;

	cost_nestloop(pathnode, root);

	return pathnode;
}

/*
 * create_mergejoin_path
 *	  Creates a pathnode corresponding to a mergejoin join between
 *	  two relations
 *
 * 'joinrel' is the join relation
 * 'jointype' is the type of join required
 * 'outer_path' is the outer path
 * 'inner_path' is the inner path
 * 'restrict_clauses' are the RestrictInfo nodes to apply at the join
 * 'pathkeys' are the path keys of the new join path
 * 'mergeclauses' are the RestrictInfo nodes to use as merge clauses
 *		(this should be a subset of the restrict_clauses list)
 * 'outersortkeys' are the sort varkeys for the outer relation
 * 'innersortkeys' are the sort varkeys for the inner relation
 */
MergePath *
create_mergejoin_path(PlannerInfo *root,
					  RelOptInfo *joinrel,
					  JoinType jointype,
					  Path *outer_path,
					  Path *inner_path,
					  List *restrict_clauses,
					  List *pathkeys,
					  List *mergeclauses,
					  List *outersortkeys,
					  List *innersortkeys)
{
	MergePath  *pathnode = makeNode(MergePath);

	/*
	 * If the given paths are already well enough ordered, we can skip doing
	 * an explicit sort.
	 */
	if (outersortkeys &&
		pathkeys_contained_in(outersortkeys, outer_path->pathkeys))
		outersortkeys = NIL;
	if (innersortkeys &&
		pathkeys_contained_in(innersortkeys, inner_path->pathkeys))
		innersortkeys = NIL;

	/*
	 * If we are not sorting the inner path, we may need a materialize node to
	 * ensure it can be marked/restored.  (Sort does support mark/restore, so
	 * no materialize is needed in that case.)
	 *
	 * Since the inner side must be ordered, and only Sorts and IndexScans can
	 * create order to begin with, you might think there's no problem --- but
	 * you'd be wrong.  Nestloop and merge joins can *preserve* the order of
	 * their inputs, so they can be selected as the input of a mergejoin, and
	 * they don't support mark/restore at present.
	 */
	if (innersortkeys == NIL &&
		!ExecSupportsMarkRestore(inner_path->pathtype))
		inner_path = (Path *)
			create_material_path(inner_path->parent, inner_path);

	pathnode->jpath.path.pathtype = T_MergeJoin;
	pathnode->jpath.path.parent = joinrel;
	pathnode->jpath.jointype = jointype;
	pathnode->jpath.outerjoinpath = outer_path;
	pathnode->jpath.innerjoinpath = inner_path;
	pathnode->jpath.joinrestrictinfo = restrict_clauses;
	pathnode->jpath.path.pathkeys = pathkeys;
	pathnode->path_mergeclauses = mergeclauses;
	pathnode->outersortkeys = outersortkeys;
	pathnode->innersortkeys = innersortkeys;

	cost_mergejoin(pathnode, root);

	return pathnode;
}

/*
 * create_hashjoin_path
 *	  Creates a pathnode corresponding to a hash join between two relations.
 *
 * 'joinrel' is the join relation
 * 'jointype' is the type of join required
 * 'outer_path' is the cheapest outer path
 * 'inner_path' is the cheapest inner path
 * 'restrict_clauses' are the RestrictInfo nodes to apply at the join
 * 'hashclauses' are the RestrictInfo nodes to use as hash clauses
 *		(this should be a subset of the restrict_clauses list)
 */
HashPath *
create_hashjoin_path(PlannerInfo *root,
					 RelOptInfo *joinrel,
					 JoinType jointype,
					 Path *outer_path,
					 Path *inner_path,
					 List *restrict_clauses,
					 List *hashclauses)
{
	HashPath   *pathnode = makeNode(HashPath);

	pathnode->jpath.path.pathtype = T_HashJoin;
	pathnode->jpath.path.parent = joinrel;
	pathnode->jpath.jointype = jointype;
	pathnode->jpath.outerjoinpath = outer_path;
	pathnode->jpath.innerjoinpath = inner_path;
	pathnode->jpath.joinrestrictinfo = restrict_clauses;
	/* A hashjoin never has pathkeys, since its ordering is unpredictable */
	pathnode->jpath.path.pathkeys = NIL;
	pathnode->path_hashclauses = hashclauses;

	cost_hashjoin(pathnode, root);

	return pathnode;
}