Subqueries

In PartiQL, as is the case with SQL as well, expressions may involve SFW subqueries. PartiQL SFW subqueries are enclosed in parentheses (i.e., identical to SQL). For compatibility with SQL, a SFW subquery starting with a SELECT clause (as opposed to a subquery starting with SELECT VALUE or PIVOT) coerces into a scalar or into an array, depending on the context. The following cases replicate SQL’s coercing behavior and analyze in which cases the result of a subquery coerces into scalar and in which cases they coerce into arrays.

An PartiQL extension with respect to SQL is that, in the permissive mode, subqueries that fail to coerce to the required type (scalar or tuple) still run, as opposed to failing. They simply omit from the results the data that correspond to the coercion failures.

Subquery to Scalar Coercion

In each of the following cases a SFW subquery coerces into a scalar

  • if it appears as the rhs of a comparison operator (=, >, etc) where the lhs is not an array literal. And, vice versa, if it appears as the lhs of a comparison operator where the rhs is not an array literal. (If it is the lhs of a comparison operator where the lhs is an array literal, it coerces into array.)

  • if it is an SFW subquery expression that (a) is not the collection expression of a FROM clause item and (b) is not the rhs of an IN. (If it is the rhs of an IN then it should not be coerced.)

Essentially, a subquery that is coerced may appear in all clauses except the FROM. For example, it may be a SELECT subquery \$s\$ that appears as an item of a SELECT, SELECT VALUE, or PIVOT clause. Or it may be a subexpression of an expression that appears in SELECT, SELECT VALUE , or PIVOT clause. Or it may be a subexpression of the WHERE clause expression, as long as it is not the rhs of an IN. In any of these cases the result of the subquery \$s\$ is cast into a scalar.

Technically, the subquery \$s\$ (which uses SELECT) is rewritten into an equivalent subquery \$s'\$ that utilizes SELECT VALUE. Then the result of \$s'\$ is cast into a scalar by applying the function \$"COLL_TO_SCALAR"(s')\$.

Example 1.  

The SQL query

SELECT v.foo,
       (SELECT w.bar
        FROM someDataSet w
        WHERE w.sth = v.sthelse) AS bar
FROM anotherDataSet v

is rewritten into

SELECT VALUE {
    'foo':  v.foo
    'bar':  COLL_TO_SCALAR(SELECT VALUE {'bar':  w.bar}
                           FROM someDataSet w
                           WHERE w.sth = v.sthelse)
}
FROM anotherDataSet v

As is the common semantics of PartiQL in the permissive mode, when COLL_TO_SCALAR fails to cast the subquery into a scalar, it outputs MISSING. The inputs that are coerced into scalars are the ones that SQL prescribes: When the input is a collection consisting of a single tuple with a single attribute, the input is coerced into a scalar. All other inputs to COLL_TO_SCALAR lead to MISSING.

Example 2.  

In this example, in one instance the inner SELECT evaluates to a collection with more than one element. Because the COLL_TO_SCALAR function produces a MISSING instead of failing, the query works.

Consider the tables

customers :  [
    {'id':1, 'name':'Mary'},
    {'id':2, 'name':'Helen'},
    {'id':1, 'name':'John'}
]
orders :  [
    {'custId':1, 'name':'foo'},
    {'custId':2, 'name':'bar'}
]

The following query would fail in SQL, because there are two customer tuples with the same id. Of course, in a well-designed SQL database that has a primary key or uniqueness constraint on the id, there would not be two customers with the same id. However, lack of constraints is typical in the data targeted by PartiQL. This query runs in the permissive mode of PartiQL.

SELECT o.name AS orderName,
       (SELECT c.name FROM customers c WHERE c.id=o.custId) AS customerName
FROM orders o

The result is

<< {'orderName':'foo'}, {'orderName':'bar', 'customerName':'Helen'} >>

Notice the missing 'customerName' in the first tuple.

As in SQL, an implementation with static type checks will be able to detect and warn that, in certain cases, a coercion will always fail and produce MISSING.

Example 3.  

The following SELECT clause is guaranteed to produce tuples with bar and bar2. Thus it cannot coerce into scalar.

SELECT w.bar1 AS bar1, w.bar2 AS bar2
FROM someDataSet w

Static type analysis can infer that the nested query above will deliver tuples consisting of bar and bar2. Thus, even before accessing any data, it can warn the user that this query is erroneous.

Subquery to Array Coercion

An SELECT SFW subquery coerces into an array when it is the rhs (respectively, lhs) of a comparison operator whose other argument is an array. [1]

The reduction of a SELECT subquery to the PartiQL is exhibited by the following example.

Example 4.  

The SQL query

SELECT v.foo
FROM anotherDataSet v
WHERE (v.a, v.b) = (SELECT w.c, w,d
                    FROM someDataSet w
                    WHERE w.sth = v.sthelse)

is rewritten into

SELECT VALUE {'foo':  v.foo}
FROM anotherDataSet v
WHERE (v.a, v.b) = (SELECT VALUE [w.c, w,d]
                    FROM someDataSet w
                    WHERE w.sth = v.sthelse)
1. Recall, in the interest of compatibility to SQL, PartiQL allows array literals to be denoted with parentheses instead of brackets.