planner.c

来自「PostgreSQL7.4.6 for Linux」· C语言 代码 · 共 1,553 行 · 第 1/4 页

											numAggs,
											result_plan);
			/* Hashed aggregation produces randomly-ordered results */
			current_pathkeys = NIL;
		}
		else if (parse->hasAggs)
		{
			/* Plain aggregate plan --- sort if needed */
			AggStrategy aggstrategy;

			if (parse->groupClause)
			{
				if (!pathkeys_contained_in(group_pathkeys, current_pathkeys))
				{
					result_plan = (Plan *)
						make_sort_from_groupcols(parse,
												 parse->groupClause,
												 groupColIdx,
												 result_plan);
					current_pathkeys = group_pathkeys;
				}
				aggstrategy = AGG_SORTED;

				/*
				 * The AGG node will not change the sort ordering of its
				 * groups, so current_pathkeys describes the result too.
				 */
			}
			else
			{
				aggstrategy = AGG_PLAIN;
				/* Result will be only one row anyway; no sort order */
				current_pathkeys = NIL;
			}

			result_plan = (Plan *) make_agg(parse,
											tlist,
											(List *) parse->havingQual,
											aggstrategy,
											numGroupCols,
											groupColIdx,
											numGroups,
											numAggs,
											result_plan);
		}
		else
		{
			/*
			 * If there are no Aggs, we shouldn't have any HAVING qual
			 * anymore
			 */
			Assert(parse->havingQual == NULL);

			/*
			 * If we have a GROUP BY clause, insert a group node (plus the
			 * appropriate sort node, if necessary).
			 */
			if (parse->groupClause)
			{
				/*
				 * Add an explicit sort if we couldn't make the path come
				 * out the way the GROUP node needs it.
				 */
				if (!pathkeys_contained_in(group_pathkeys, current_pathkeys))
				{
					result_plan = (Plan *)
						make_sort_from_groupcols(parse,
												 parse->groupClause,
												 groupColIdx,
												 result_plan);
					current_pathkeys = group_pathkeys;
				}

				result_plan = (Plan *) make_group(parse,
												  tlist,
												  numGroupCols,
												  groupColIdx,
												  dNumGroups,
												  result_plan);
				/* The Group node won't change sort ordering */
			}
		}
	}							/* end of if (setOperations) */

	/*
	 * If we were not able to make the plan come out in the right order,
	 * add an explicit sort step.
	 */
	if (parse->sortClause)
	{
		if (!pathkeys_contained_in(sort_pathkeys, current_pathkeys))
		{
			result_plan = (Plan *)
				make_sort_from_sortclauses(parse,
										   tlist,
										   result_plan,
										   parse->sortClause);
			current_pathkeys = sort_pathkeys;
		}
	}

	/*
	 * If there is a DISTINCT clause, add the UNIQUE node.
	 */
	if (parse->distinctClause)
	{
		result_plan = (Plan *) make_unique(tlist, result_plan,
										   parse->distinctClause);

		/*
		 * If there was grouping or aggregation, leave plan_rows as-is
		 * (ie, assume the result was already mostly unique).  If not,
		 * it's reasonable to assume the UNIQUE filter has effects
		 * comparable to GROUP BY.
		 */
		if (!parse->groupClause && !parse->hasAggs)
		{
			List	   *distinctExprs;

			distinctExprs = get_sortgrouplist_exprs(parse->distinctClause,
													parse->targetList);
			result_plan->plan_rows = estimate_num_groups(parse,
														 distinctExprs,
												 result_plan->plan_rows);
		}
	}

	/*
	 * Finally, if there is a LIMIT/OFFSET clause, add the LIMIT node.
	 */
	if (parse->limitOffset || parse->limitCount)
	{
		result_plan = (Plan *) make_limit(tlist, result_plan,
										  parse->limitOffset,
										  parse->limitCount);
	}

	/*
	 * Return the actual output ordering in query_pathkeys for possible
	 * use by an outer query level.
	 */
	parse->query_pathkeys = current_pathkeys;

	return result_plan;
}

/*
 * hash_safe_grouping - are grouping operators hashable?
 *
 * We assume hashed aggregation will work if the datatype's equality operator
 * is marked hashjoinable.
 */
static bool
hash_safe_grouping(Query *parse)
{
	List	   *gl;

	foreach(gl, parse->groupClause)
	{
		GroupClause *grpcl = (GroupClause *) lfirst(gl);
		TargetEntry *tle = get_sortgroupclause_tle(grpcl, parse->targetList);
		Operator	optup;
		bool		oprcanhash;

		optup = equality_oper(tle->resdom->restype, true);
		if (!optup)
			return false;
		oprcanhash = ((Form_pg_operator) GETSTRUCT(optup))->oprcanhash;
		ReleaseSysCache(optup);
		if (!oprcanhash)
			return false;
	}
	return true;
}

/*---------------
 * make_subplanTargetList
 *	  Generate appropriate target list when grouping is required.
 *
 * When grouping_planner inserts Aggregate or Group plan nodes above
 * the result of query_planner, we typically want to pass a different
 * target list to query_planner than the outer plan nodes should have.
 * This routine generates the correct target list for the subplan.
 *
 * The initial target list passed from the parser already contains entries
 * for all ORDER BY and GROUP BY expressions, but it will not have entries
 * for variables used only in HAVING clauses; so we need to add those
 * variables to the subplan target list.  Also, if we are doing either
 * grouping or aggregation, we flatten all expressions except GROUP BY items
 * into their component variables; the other expressions will be computed by
 * the inserted nodes rather than by the subplan.  For example,
 * given a query like
 *		SELECT a+b,SUM(c+d) FROM table GROUP BY a+b;
 * we want to pass this targetlist to the subplan:
 *		a,b,c,d,a+b
 * where the a+b target will be used by the Sort/Group steps, and the
 * other targets will be used for computing the final results.	(In the
 * above example we could theoretically suppress the a and b targets and
 * pass down only c,d,a+b, but it's not really worth the trouble to
 * eliminate simple var references from the subplan.  We will avoid doing
 * the extra computation to recompute a+b at the outer level; see
 * replace_vars_with_subplan_refs() in setrefs.c.)
 *
 * If we are grouping or aggregating, *and* there are no non-Var grouping
 * expressions, then the returned tlist is effectively dummy; we do not
 * need to force it to be evaluated, because all the Vars it contains
 * should be present in the output of query_planner anyway.
 *
 * 'parse' is the query being processed.
 * 'tlist' is the query's target list.
 * 'groupColIdx' receives an array of column numbers for the GROUP BY
 *			expressions (if there are any) in the subplan's target list.
 * 'need_tlist_eval' is set true if we really need to evaluate the
 *			result tlist.
 *
 * The result is the targetlist to be passed to the subplan.
 *---------------
 */
static List *
make_subplanTargetList(Query *parse,
					   List *tlist,
					   AttrNumber **groupColIdx,
					   bool *need_tlist_eval)
{
	List	   *sub_tlist;
	List	   *extravars;
	int			numCols;

	*groupColIdx = NULL;

	/*
	 * If we're not grouping or aggregating, nothing to do here;
	 * query_planner should receive the unmodified target list.
	 */
	if (!parse->hasAggs && !parse->groupClause)
	{
		*need_tlist_eval = true;
		return tlist;
	}

	/*
	 * Otherwise, start with a "flattened" tlist (having just the vars
	 * mentioned in the targetlist and HAVING qual --- but not upper-
	 * level Vars; they will be replaced by Params later on).
	 */
	sub_tlist = flatten_tlist(tlist);
	extravars = pull_var_clause(parse->havingQual, false);
	sub_tlist = add_to_flat_tlist(sub_tlist, extravars);
	freeList(extravars);
	*need_tlist_eval = false;	/* only eval if not flat tlist */

	/*
	 * If grouping, create sub_tlist entries for all GROUP BY expressions
	 * (GROUP BY items that are simple Vars should be in the list
	 * already), and make an array showing where the group columns are in
	 * the sub_tlist.
	 */
	numCols = length(parse->groupClause);
	if (numCols > 0)
	{
		int			keyno = 0;
		AttrNumber *grpColIdx;
		List	   *gl;

		grpColIdx = (AttrNumber *) palloc(sizeof(AttrNumber) * numCols);
		*groupColIdx = grpColIdx;

		foreach(gl, parse->groupClause)
		{
			GroupClause *grpcl = (GroupClause *) lfirst(gl);
			Node	   *groupexpr = get_sortgroupclause_expr(grpcl, tlist);
			TargetEntry *te = NULL;
			List	   *sl;

			/* Find or make a matching sub_tlist entry */
			foreach(sl, sub_tlist)
			{
				te = (TargetEntry *) lfirst(sl);
				if (equal(groupexpr, te->expr))
					break;
			}
			if (!sl)
			{
				te = makeTargetEntry(makeResdom(length(sub_tlist) + 1,
												exprType(groupexpr),
												exprTypmod(groupexpr),
												NULL,
												false),
									 (Expr *) groupexpr);
				sub_tlist = lappend(sub_tlist, te);
				*need_tlist_eval = true;		/* it's not flat anymore */
			}

			/* and save its resno */
			grpColIdx[keyno++] = te->resdom->resno;
		}
	}

	return sub_tlist;
}

/*
 * locate_grouping_columns
 *		Locate grouping columns in the tlist chosen by query_planner.
 *
 * This is only needed if we don't use the sub_tlist chosen by
 * make_subplanTargetList.	We have to forget the column indexes found
 * by that routine and re-locate the grouping vars in the real sub_tlist.
 */
static void
locate_grouping_columns(Query *parse,
						List *tlist,
						List *sub_tlist,
						AttrNumber *groupColIdx)
{
	int			keyno = 0;
	List	   *gl;

	/*
	 * No work unless grouping.
	 */
	if (!parse->groupClause)
	{
		Assert(groupColIdx == NULL);
		return;
	}
	Assert(groupColIdx != NULL);

	foreach(gl, parse->groupClause)
	{
		GroupClause *grpcl = (GroupClause *) lfirst(gl);
		Node	   *groupexpr = get_sortgroupclause_expr(grpcl, tlist);
		TargetEntry *te = NULL;
		List	   *sl;

		foreach(sl, sub_tlist)
		{
			te = (TargetEntry *) lfirst(sl);
			if (equal(groupexpr, te->expr))
				break;
		}
		if (!sl)
			elog(ERROR, "failed to locate grouping columns");

		groupColIdx[keyno++] = te->resdom->resno;
	}
}

/*
 * postprocess_setop_tlist
 *	  Fix up targetlist returned by plan_set_operations().
 *
 * We need to transpose sort key info from the orig_tlist into new_tlist.
 * NOTE: this would not be good enough if we supported resjunk sort keys
 * for results of set operations --- then, we'd need to project a whole
 * new tlist to evaluate the resjunk columns.  For now, just ereport if we
 * find any resjunk columns in orig_tlist.
 */
static List *
postprocess_setop_tlist(List *new_tlist, List *orig_tlist)
{
	List	   *l;

	foreach(l, new_tlist)
	{
		TargetEntry *new_tle = (TargetEntry *) lfirst(l);
		TargetEntry *orig_tle;

		/* ignore resjunk columns in setop result */
		if (new_tle->resdom->resjunk)
			continue;

		Assert(orig_tlist != NIL);
		orig_tle = (TargetEntry *) lfirst(orig_tlist);
		orig_tlist = lnext(orig_tlist);
		if (orig_tle->resdom->resjunk)	/* should not happen */
			elog(ERROR, "resjunk output columns are not implemented");
		Assert(new_tle->resdom->resno == orig_tle->resdom->resno);
		Assert(new_tle->resdom->restype == orig_tle->resdom->restype);
		new_tle->resdom->ressortgroupref = orig_tle->resdom->ressortgroupref;
	}
	if (orig_tlist != NIL)
		elog(ERROR, "resjunk output columns are not implemented");
	return new_tlist;
}