parse_clause.c

PostgreSQL 8.1.4的源码 适用于Linux下的开源数据库系统

	/*
	 * We require user to supply an alias for a subselect, per SQL92. To relax
	 * this, we'd have to be prepared to gin up a unique alias for an
	 * unlabeled subselect.
	 */
	if (r->alias == NULL)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("subquery in FROM must have an alias")));

	/*
	 * Analyze and transform the subquery.
	 */
	parsetrees = parse_sub_analyze(r->subquery, pstate);

	/*
	 * Check that we got something reasonable.	Most of these conditions are
	 * probably impossible given restrictions of the grammar, but check 'em
	 * anyway.
	 */
	if (list_length(parsetrees) != 1)
		elog(ERROR, "unexpected parse analysis result for subquery in FROM");
	query = (Query *) linitial(parsetrees);
	if (query == NULL || !IsA(query, Query))
		elog(ERROR, "unexpected parse analysis result for subquery in FROM");

	if (query->commandType != CMD_SELECT)
		elog(ERROR, "expected SELECT query from subquery in FROM");
	if (query->resultRelation != 0 || query->into != NULL)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("subquery in FROM may not have SELECT INTO")));

	/*
	 * The subquery cannot make use of any variables from FROM items created
	 * earlier in the current query.  Per SQL92, the scope of a FROM item does
	 * not include other FROM items.  Formerly we hacked the namespace so that
	 * the other variables weren't even visible, but it seems more useful to
	 * leave them visible and give a specific error message.
	 *
	 * XXX this will need further work to support SQL99's LATERAL() feature,
	 * wherein such references would indeed be legal.
	 *
	 * We can skip groveling through the subquery if there's not anything
	 * visible in the current query.  Also note that outer references are OK.
	 */
	if (pstate->p_relnamespace || pstate->p_varnamespace)
	{
		if (contain_vars_of_level((Node *) query, 1))
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_COLUMN_REFERENCE),
					 errmsg("subquery in FROM may not refer to other relations of same query level")));
	}

	/*
	 * OK, build an RTE for the subquery.
	 */
	rte = addRangeTableEntryForSubquery(pstate, query, r->alias, true);

	return rte;
}


/*
 * transformRangeFunction --- transform a function call appearing in FROM
 */
static RangeTblEntry *
transformRangeFunction(ParseState *pstate, RangeFunction *r)
{
	Node	   *funcexpr;
	char	   *funcname;
	RangeTblEntry *rte;

	/*
	 * Get function name for possible use as alias.  We use the same
	 * transformation rules as for a SELECT output expression.	For a FuncCall
	 * node, the result will be the function name, but it is possible for the
	 * grammar to hand back other node types.
	 */
	funcname = FigureColname(r->funccallnode);

	/*
	 * Transform the raw expression.
	 */
	funcexpr = transformExpr(pstate, r->funccallnode);

	/*
	 * The function parameters cannot make use of any variables from other
	 * FROM items.	(Compare to transformRangeSubselect(); the coding is
	 * different though because we didn't parse as a sub-select with its own
	 * level of namespace.)
	 *
	 * XXX this will need further work to support SQL99's LATERAL() feature,
	 * wherein such references would indeed be legal.
	 */
	if (pstate->p_relnamespace || pstate->p_varnamespace)
	{
		if (contain_vars_of_level(funcexpr, 0))
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_COLUMN_REFERENCE),
					 errmsg("function expression in FROM may not refer to other relations of same query level")));
	}

	/*
	 * Disallow aggregate functions in the expression.	(No reason to postpone
	 * this check until parseCheckAggregates.)
	 */
	if (pstate->p_hasAggs)
	{
		if (checkExprHasAggs(funcexpr))
			ereport(ERROR,
					(errcode(ERRCODE_GROUPING_ERROR),
					 errmsg("cannot use aggregate function in function expression in FROM")));
	}

	/*
	 * If a coldeflist is supplied, ensure it defines a legal set of names (no
	 * duplicates) and datatypes (no pseudo-types, for instance).
	 */
	if (r->coldeflist)
	{
		TupleDesc	tupdesc;

		tupdesc = BuildDescForRelation(r->coldeflist);
		CheckAttributeNamesTypes(tupdesc, RELKIND_COMPOSITE_TYPE);
	}

	/*
	 * OK, build an RTE for the function.
	 */
	rte = addRangeTableEntryForFunction(pstate, funcname, funcexpr,
										r, true);

	return rte;
}


/*
 * transformFromClauseItem -
 *	  Transform a FROM-clause item, adding any required entries to the
 *	  range table list being built in the ParseState, and return the
 *	  transformed item ready to include in the joinlist and namespaces.
 *	  This routine can recurse to handle SQL92 JOIN expressions.
 *
 * The function return value is the node to add to the jointree (a
 * RangeTblRef or JoinExpr).  Additional output parameters are:
 *
 * *top_rte: receives the RTE corresponding to the jointree item.
 * (We could extract this from the function return node, but it saves cycles
 * to pass it back separately.)
 *
 * *top_rti: receives the rangetable index of top_rte.	(Ditto.)
 *
 * *relnamespace: receives a List of the RTEs exposed as relation names
 * by this item.
 *
 * *containedRels: receives a bitmap set of the rangetable indexes
 * of all the base and join relations represented in this jointree item.
 * This is needed for checking JOIN/ON conditions in higher levels.
 *
 * We do not need to pass back an explicit varnamespace value, because
 * in all cases the varnamespace contribution is exactly top_rte.
 */
static Node *
transformFromClauseItem(ParseState *pstate, Node *n,
						RangeTblEntry **top_rte, int *top_rti,
						List **relnamespace,
						Relids *containedRels)
{
	if (IsA(n, RangeVar))
	{
		/* Plain relation reference */
		RangeTblRef *rtr;
		RangeTblEntry *rte;
		int			rtindex;

		rte = transformTableEntry(pstate, (RangeVar *) n);
		/* assume new rte is at end */
		rtindex = list_length(pstate->p_rtable);
		Assert(rte == rt_fetch(rtindex, pstate->p_rtable));
		*top_rte = rte;
		*top_rti = rtindex;
		*relnamespace = list_make1(rte);
		*containedRels = bms_make_singleton(rtindex);
		rtr = makeNode(RangeTblRef);
		rtr->rtindex = rtindex;
		return (Node *) rtr;
	}
	else if (IsA(n, RangeSubselect))
	{
		/* sub-SELECT is like a plain relation */
		RangeTblRef *rtr;
		RangeTblEntry *rte;
		int			rtindex;

		rte = transformRangeSubselect(pstate, (RangeSubselect *) n);
		/* assume new rte is at end */
		rtindex = list_length(pstate->p_rtable);
		Assert(rte == rt_fetch(rtindex, pstate->p_rtable));
		*top_rte = rte;
		*top_rti = rtindex;
		*relnamespace = list_make1(rte);
		*containedRels = bms_make_singleton(rtindex);
		rtr = makeNode(RangeTblRef);
		rtr->rtindex = rtindex;
		return (Node *) rtr;
	}
	else if (IsA(n, RangeFunction))
	{
		/* function is like a plain relation */
		RangeTblRef *rtr;
		RangeTblEntry *rte;
		int			rtindex;

		rte = transformRangeFunction(pstate, (RangeFunction *) n);
		/* assume new rte is at end */
		rtindex = list_length(pstate->p_rtable);
		Assert(rte == rt_fetch(rtindex, pstate->p_rtable));
		*top_rte = rte;
		*top_rti = rtindex;
		*relnamespace = list_make1(rte);
		*containedRels = bms_make_singleton(rtindex);
		rtr = makeNode(RangeTblRef);
		rtr->rtindex = rtindex;
		return (Node *) rtr;
	}
	else if (IsA(n, JoinExpr))
	{
		/* A newfangled join expression */
		JoinExpr   *j = (JoinExpr *) n;
		RangeTblEntry *l_rte;
		RangeTblEntry *r_rte;
		int			l_rtindex;
		int			r_rtindex;
		Relids		l_containedRels,
					r_containedRels,
					my_containedRels;
		List	   *l_relnamespace,
				   *r_relnamespace,
				   *my_relnamespace,
				   *l_colnames,
				   *r_colnames,
				   *res_colnames,
				   *l_colvars,
				   *r_colvars,
				   *res_colvars;
		RangeTblEntry *rte;

		/*
		 * Recursively process the left and right subtrees
		 */
		j->larg = transformFromClauseItem(pstate, j->larg,
										  &l_rte,
										  &l_rtindex,
										  &l_relnamespace,
										  &l_containedRels);
		j->rarg = transformFromClauseItem(pstate, j->rarg,
										  &r_rte,
										  &r_rtindex,
										  &r_relnamespace,
										  &r_containedRels);

		/*
		 * Check for conflicting refnames in left and right subtrees. Must do
		 * this because higher levels will assume I hand back a self-
		 * consistent namespace subtree.
		 */
		checkNameSpaceConflicts(pstate, l_relnamespace, r_relnamespace);

		/*
		 * Generate combined relation membership info for possible use by
		 * transformJoinOnClause below.
		 */
		my_relnamespace = list_concat(l_relnamespace, r_relnamespace);
		my_containedRels = bms_join(l_containedRels, r_containedRels);

		pfree(r_relnamespace);	/* free unneeded list header */

		/*
		 * Extract column name and var lists from both subtrees
		 *
		 * Note: expandRTE returns new lists, safe for me to modify
		 */
		expandRTE(l_rte, l_rtindex, 0, false,
				  &l_colnames, &l_colvars);
		expandRTE(r_rte, r_rtindex, 0, false,
				  &r_colnames, &r_colvars);

		/*
		 * Natural join does not explicitly specify columns; must generate
		 * columns to join. Need to run through the list of columns from each
		 * table or join result and match up the column names. Use the first
		 * table, and check every column in the second table for a match.
		 * (We'll check that the matches were unique later on.) The result of
		 * this step is a list of column names just like an explicitly-written
		 * USING list.
		 */
		if (j->isNatural)
		{
			List	   *rlist = NIL;
			ListCell   *lx,
					   *rx;

			Assert(j->using == NIL);	/* shouldn't have USING() too */

			foreach(lx, l_colnames)
			{
				char	   *l_colname = strVal(lfirst(lx));
				Value	   *m_name = NULL;

				foreach(rx, r_colnames)
				{
					char	   *r_colname = strVal(lfirst(rx));

					if (strcmp(l_colname, r_colname) == 0)
					{
						m_name = makeString(l_colname);
						break;
					}
				}

				/* matched a right column? then keep as join column... */
				if (m_name != NULL)
					rlist = lappend(rlist, m_name);
			}

			j->using = rlist;
		}

		/*
		 * Now transform the join qualifications, if any.
		 */
		res_colnames = NIL;
		res_colvars = NIL;

		if (j->using)
		{
			/*
			 * JOIN/USING (or NATURAL JOIN, as transformed above). Transform
			 * the list into an explicit ON-condition, and generate a list of
			 * merged result columns.
			 */
			List	   *ucols = j->using;
			List	   *l_usingvars = NIL;
			List	   *r_usingvars = NIL;
			ListCell   *ucol;

			Assert(j->quals == NULL);	/* shouldn't have ON() too */

			foreach(ucol, ucols)
			{
				char	   *u_colname = strVal(lfirst(ucol));
				ListCell   *col;
				int			ndx;
				int			l_index = -1;
				int			r_index = -1;
				Var		   *l_colvar,
						   *r_colvar;

				/* Check for USING(foo,foo) */
				foreach(col, res_colnames)
				{
					char	   *res_colname = strVal(lfirst(col));

					if (strcmp(res_colname, u_colname) == 0)
						ereport(ERROR,
								(errcode(ERRCODE_DUPLICATE_COLUMN),
								 errmsg("column name \"%s\" appears more than once in USING clause",
										u_colname)));
				}

				/* Find it in left input */
				ndx = 0;
				foreach(col, l_colnames)
				{
					char	   *l_colname = strVal(lfirst(col));

					if (strcmp(l_colname, u_colname) == 0)
					{
						if (l_index >= 0)
							ereport(ERROR,
									(errcode(ERRCODE_AMBIGUOUS_COLUMN),
									 errmsg("common column name \"%s\" appears more than once in left table",
											u_colname)));
						l_index = ndx;
					}
					ndx++;
				}
				if (l_index < 0)
					ereport(ERROR,
							(errcode(ERRCODE_UNDEFINED_COLUMN),
							 errmsg("column \"%s\" specified in USING clause does not exist in left table",
									u_colname)));

				/* Find it in right input */
				ndx = 0;
				foreach(col, r_colnames)
				{
					char	   *r_colname = strVal(lfirst(col));

					if (strcmp(r_colname, u_colname) == 0)
					{
						if (r_index >= 0)
							ereport(ERROR,
									(errcode(ERRCODE_AMBIGUOUS_COLUMN),
									 errmsg("common column name \"%s\" appears more than once in right table",
											u_colname)));
						r_index = ndx;
					}
					ndx++;
				}
				if (r_index < 0)
					ereport(ERROR,
							(errcode(ERRCODE_UNDEFINED_COLUMN),
							 errmsg("column \"%s\" specified in USING clause does not exist in right table",
									u_colname)));

				l_colvar = list_nth(l_colvars, l_index);
				l_usingvars = lappend(l_usingvars, l_colvar);