Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

Quantified LIKE ANY Predicates in jOOQ 3.12

DZone 's Guide to

Quantified LIKE ANY Predicates in jOOQ 3.12

Let's explore SQL's quantified comparison predicates.

· Integration Zone ·
Free Resource

Image title

SQL quantified comparison predicates

Quantified Comparison Predicates

Some of SQL's weirdest features are quantified comparison predicates. I've hardly ever seen these in the wild:

SELECT *
FROM t
WHERE id = ANY (1, 2, 3)

The above example is equivalent to using the much more readable IN predicate:

SELECT *
FROM t
WHERE id IN (1, 2, 3)

This equivalence is defined in the SQL standard. There are more esoteric cases that could be solved using such quantified comparison predicates more conveniently than otherwise, such as:

SELECT *
FROM t
WHERE (a, b) > ALL (
  SELECT x, y
  FROM u
)

This is the same thing as writing the more verbose, and in my opinion, a bit less readable:

SELECT *
FROM t
WHERE (a, b) > (
  SELECT x, y
  FROM u
  ORDER BY x, y
  FETCH FIRST ROW ONLY
)

Assuming, of course, that your RDBMS can compare row value expressions like that.

Quantified LIKE Predicate

Unfortunately, the SQL standard and most implementations support the above quantified comparison predicates only for the <, <=, >, >=, =, != comparison operators. Not for other predicate types. For example, the LIKE predicate would profit greatly from such a syntax:

SELECT *
FROM customers
WHERE last_name LIKE ANY ('A%', 'B%', 'C%')

The syntax is immediately understandable and translates to

SELECT *
FROM customers
WHERE last_name LIKE 'A%'
OR last_name LIKE 'B%'
OR last_name LIKE 'C%'

... which is much less convenient to write!

Furthermore, imagine producing such patterns from a subquery:

SELECT *
FROM customers
WHERE last_name LIKE ANY (
  SELECT pattern
  FROM patterns
  WHERE pattern.customer_type = customer.customer_type
)

This is a bit trickier to emulate in standard SQL. For example, in PostgreSQL, we could write:

SELECT *
FROM customers
WHERE true = ANY (
  SELECT last_name LIKE pattern
  FROM patterns
  WHERE pattern.customer_type = customer.customer_type
)

In this case, we can use boolean types. Oracle would make this a bit more difficult:

SELECT *
FROM customers
WHERE 1 = ANY (
  SELECT CASE
    WHEN last_name LIKE pattern THEN 1
    WHEN NOT(last_name LIKE pattern) THEN 0
    ELSE NULL
  END
  FROM patterns
  WHERE pattern.customer_type = customer.customer_type
)

Wouldn't this a useful SQL feature to support?

jOOQ 3.12 Support for This

jOOQ does support this syntax starting from jOOQ 3.12. You can now write:

ctx.selectFrom(CUSTOMERS)
   .where(CUSTOMERS.LAST_NAME.like(any("A%", "B%", "C%")))
   .fetch();

ctx.selectFrom(CUSTOMERS)
   .where(CUSTOMERS.LAST_NAME.like(any(
      select(PATTERNS.PATTERN)
      .from(PATTERNS)
      .where(PATTERN.CUSTOMER_TYPE.eq(CUSTOMER.CUSTOMER_TYPE))
   )))
   .fetch();

All the previously mentioned emulations are available. You can play around with it by downloading jOOQ:
https://www.jooq.org/download

Or directly on the website:
https://www.jooq.org/translate

Further Reading

Java Joy: Combining Predicates

Quantified Comparison Predicates — Some of SQL’s Rarest Species

Topics:
integration ,tutorial ,jooq ,predicates ,quantified like any predicates ,sql ,Quantified Comparison Predicates ,postgresql ,boolean types

Published at DZone with permission of

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}