An identifier is a sequence of one or more characters. The first character must be a valid first character (letter, $, ) in an identifier of the Java programming language, hereafter in this chapter called simply "Java." Each subsequent character in the sequence must be a valid nonfirst character (letter, digit, $, ) in a Java identifier. (For details, see the Java SE API documentation of the isJavaIdentifierStart and isJavaIdentifierPart methods of the Character class.) The question mark (?) is a reserved character in the query language and cannot be used in an identifier.

A query language identifier is case-sensitive, with two exceptions:

An identifier cannot be the same as a query language keyword. Here is a list of query language keywords:

ABS
ALL
AND
ANY
AS
ASC
AVG
BETWEEN
BIT_LENGTH
BOTH
BY
CASE
CHAR_LENGTH
CHARACTER_LENGTH
CLASS
COALESCE
CONCAT
COUNT
CURRENT_DATE
CURRENT_TIMESTAMP
DELETE
DESC
DISTINCT
ELSE
EMPTY
END
ENTRY
ESCAPE
EXISTS
FALSE
FETCH
FROM
GROUP
HAVING
IN
INDEX
INNER
IS
JOIN
KEY
LEADING
LEFT
LENGTH
LIKE
LOCATE
LOWER
MAX
MEMBER
MIN
MOD
NEW
NOT
NULL
NULLIF
OBJECT
OF
OR
ORDER
OUTER
POSITION
SELECT
SET
SIZE
SOME
SQRT
SUBSTRING
SUM
THEN
TRAILING
TRIM
TRUE
TYPE
UNKNOWN
UPDATE
UPPER
VALUE
WHEN
WHERE

It is not recommended that you use an SQL keyword as an identifier, because the list of keywords may expand to include other reserved SQL words in the future.

An identification variable is an identifier declared in the FROM clause. Although they can reference identification variables, the SELECT and WHERE clauses cannot declare them. All identification variables must be declared in the FROM clause.

Because it is an identifier, an identification variable has the same naming conventions and restrictions as an identifier, with the exception that an identification variable is case-insensitive. For example, an identification variable cannot be the same as a query language keyword. (See Identifiers for more naming rules.) Also, within a given persistence unit, an identification variable name must not match the name of any entity or abstract schema.

The FROM clause can contain multiple declarations, separated by commas. A declaration can reference another identification variable that has been previously declared (to the left). In the following FROM clause, the variable t references the previously declared variable p:

Even if it is not used in the WHERE clause, an identification variable’s declaration can affect the results of the query. For example, compare the next two queries. The following query returns all players, whether or not they belong to a team:

In contrast, because it declares the t identification variable, the next query fetches all players who belong to a team:

The following query returns the same results as the preceding query, but the WHERE clause makes it easier to read:

An identification variable always designates a reference to a single value whose type is that of the expression used in the declaration. There are two kinds of declarations: range variable and collection member.

To declare an identification variable as an abstract schema type, you specify a range variable declaration. In other words, an identification variable can range over the abstract schema type of an entity. In the following example, an identification variable named p represents the abstract schema named Player:

A range variable declaration can include the optional AS operator:

To obtain objects, a query usually uses path expressions to navigate through the relationships. But for those objects that cannot be obtained by navigation, you can use a range variable declaration to designate a starting point, or query root.

If the query compares multiple values of the same abstract schema type, the FROM clause must declare multiple identification variables for the abstract schema:

For an example of such a query, see Comparison Operators.

In a one-to-many relationship, the multiple side consists of a collection of entities. An identification variable can represent a member of this collection. To access a collection member, the path expression in the variable’s declaration navigates through the relationships in the abstract schema. (For more information on path expressions, see Path Expressions.) Because a path expression can be based on another path expression, the navigation can traverse several relationships. See Traversing Multiple Relationships.

A collection member declaration must include the IN operator but can omit the optional AS operator.

In the following example, the entity represented by the abstract schema named Player has a relationship field called teams. The identification variable called t represents a single member of the teams collection:

The JOIN operator is used to traverse over relationships between entities and is functionally similar to the IN operator.

In the following example, the query joins over the relationship between customers and orders:

The INNER keyword is optional:

These examples are equivalent to the following query, which uses the IN operator:

You can also join a single-valued relationship:

A LEFT JOIN or LEFT OUTER JOIN retrieves a set of entities where matching values in the join condition may be absent. The OUTER keyword is optional:

A FETCH JOIN is a join operation that returns associated entities as a side effect of running the query. In the following example, the query returns a set of departments and, as a side effect, the associated employees of the departments, even though the employees were not explicitly retrieved by the SELECT clause:

Here, the WHERE clause contains a single_valued_path_expression; the p is an identification variable, and salary is a persistent field of Player:

Here, the WHERE clause also contains a single_valued_path_expression; t is an identification variable, league is a single-valued relationship field, and sport is a persistent field of league:

Here, the WHERE clause contains a collection_valued_path_expression; p is an identification variable, and teams designates a collection-valued relationship field:

The type of a path expression is the type of the object represented by the ending element, which can be one of the following:

For example, the type of the expression p.salary is double because the terminating persistent field (salary) is a double.

In the expression p.teams, the terminating element is a collection-valued relationship field (teams). This expression’s type is a collection of the abstract schema type named Team. Because Team is the abstract schema name for the Team entity, this type maps to the entity. For more information on the type mapping of abstract schemas, see Return Types.

A path expression enables the query to navigate to related entities. The terminating elements of an expression determine whether navigation is allowed. If an expression contains a single-valued relationship field, the navigation can continue to an object that is related to the field. However, an expression cannot navigate beyond a persistent field or a collection-valued relationship field. For example, the expression p.teams.league.sport is illegal because teams is a collection-valued relationship field. To reach the sport field, the FROM clause could define an identification variable named t for the teams field:

There are four kinds of literals: string, numeric, Boolean, and enum.

An input parameter can be either a named parameter or a positional parameter.

The following rules apply to input parameters.

A WHERE clause consists of a conditional expression, which is evaluated from left to right within a precedence level. You can change the order of evaluation by using parentheses.

Table 42-2 lists the query language operators in order of decreasing precedence.

Table 42-2 Query Language Order Precedence

A BETWEEN expression determines whether an arithmetic expression falls within a range of values.

These two expressions are equivalent:

The following two expressions also are equivalent:

If an arithmetic expression has a NULL value, the value of the BETWEEN expression is unknown.

An IN expression determines whether a string belongs to a set of string literals or whether a number belongs to a set of number values.

The path expression must have a string or numeric value. If the path expression has a NULL value, the value of the IN expression is unknown.

In the following example, the expression is TRUE if the country is UK , but FALSE if the country is Peru:

You may also use input parameters:

A LIKE expression determines whether a wildcard pattern matches a string.

The path expression must have a string or numeric value. If this value is NULL, the value of the LIKE expression is unknown. The pattern value is a string literal that can contain wildcard characters. The underscore (_) wildcard character represents any single character. The percent (%) wildcard character represents zero or more characters. The ESCAPE clause specifies an escape character for the wildcard characters in the pattern value. Table 42-3 shows some sample LIKE expressions.

Table 42-3 LIKE Expression Examples

A NULL comparison expression tests whether a single-valued path expression or an input parameter has a NULL value. Usually, the NULL comparison expression is used to test whether a single-valued relationship has been set:

This query selects all teams where the league relationship is not set. Note that the following query is not equivalent:

The comparison with NULL using the equals operator (=) always returns an unknown value, even if the relationship is not set. The second query will always return an empty result.

The IS [NOT] EMPTY comparison expression tests whether a collection-valued path expression has no elements. In other words, it tests whether a collection-valued relationship has been set.

If the collection-valued path expression is NULL, the empty collection comparison expression has a NULL value.

Here is an example that finds all orders that do not have any line items:

The [NOT] MEMBER [OF] collection member expression determines whether a value is a member of a collection. The value and the collection members must have the same type.

If either the collection-valued or single-valued path expression is unknown, the collection member expression is unknown. If the collection-valued path expression designates an empty collection, the collection member expression is FALSE.

The OF keyword is optional.

The following example tests whether a line item is part of an order:

Subqueries may be used in the WHERE or HAVING clause of a query. Subqueries must be surrounded by parentheses.

The following example finds all customers who have placed more than ten orders:

Subqueries may contain EXISTS, ALL, and ANY expressions.

The query language includes several string, arithmetic, and date/time functions that may be used in the SELECT, WHERE, or HAVING clause of a query. The functions are listed in Table 42-4, Table 42-5, and Table 42-6.

In Table 42-4, the start and length arguments are of type int and designate positions in the String argument. The first position in a string is designated by 1.

Table 42-4 String Expressions

The CONCAT function concatenates two strings into one string.

The LENGTH function returns the length of a string in characters as an integer.

The LOCATE function returns the position of a given string within a string. This function returns the first position at which the string was found as an integer. The first argument is the string to be located. The second argument is the string to be searched. The optional third argument is an integer that represents the starting string position. By default, LOCATE starts at the beginning of the string. The starting position of a string is 1. If the string cannot be located, LOCATE returns 0.

The SUBSTRING function returns a string that is a substring of the first argument based on the starting position and length.

The TRIM function trims the specified character from the beginning and/or end of a string. If no character is specified, TRIM removes spaces or blanks from the string. If the optional LEADING specification is used, TRIM removes only the leading characters from the string. If the optional TRAILING specification is used, TRIM removes only the trailing characters from the string. The default is BOTH, which removes the leading and trailing characters from the string.

The LOWER and UPPER functions convert a string to lowercase or uppercase, respectively.

In Table 42-5, the number argument can be an int, a float, or a double.

Table 42-5 Arithmetic Expressions

The ABS function takes a numeric expression and returns a number of the same type as the argument.

The MOD function returns the remainder of the first argument divided by the second.

The SQRT function returns the square root of a number.

The SIZE function returns an integer of the number of elements in the given collection.

In Table 42-6, the date/time functions return the date, time, or timestamp on the database server.

Table 42-6 Date/Time Expressions

Case expressions change based on a condition, similar to the case keyword of the Java programming language. The CASE keyword indicates the start of a case expression, and the expression is terminated by the END keyword. The WHEN and THEN keywords define individual conditions, and the ELSE keyword defines the default condition should none of the other conditions be satisfied.

The following query selects the name of a person and a conditional string, depending on the subtype of the Person entity. If the subtype is Student, the string kid is returned. If the subtype is Guardian or Staff, the string adult is returned. If the entity is some other subtype of Person, the string unknown is returned:

The following query sets a discount for various types of customers. Gold-level customers get a 20% discount, silver-level customers get a 15% discount, bronze-level customers get a 10% discount, and everyone else gets a 5% discount:

If the target of a reference is not in the persistent store, the target is NULL. For conditional expressions containing NULL, the query language uses the semantics defined by SQL92. Briefly, these semantics are as follows.

Table 42-7 AND Operator Logic

Table 42-8 OR Operator Logic

In the query language, only values of the same type can be compared. However, this rule has one exception: Exact and approximate numeric values can be compared. In such a comparison, the required type conversion adheres to the rules of Java numeric promotion.

The query language treats compared values as if they were Java types and not as if they represented types in the underlying data store. For example, a persistent field that could be either an integer or a NULL must be designated as an Integer object and not as an int primitive. This designation is required because a Java object can be NULL, but a primitive cannot.

Two strings are equal only if they contain the same sequence of characters. Trailing blanks are significant; for example, the strings 'abc' and 'abc ' are not equal.

Two entities of the same abstract schema type are equal only if their primary keys have the same value. Table 42-9 shows the operator logic of a negation, and Table 42-10 shows the truth values of conditional tests.

Table 42-9 NOT Operator Logic

Table 42-10 Conditional Test

The return type of the SELECT clause is defined by the result types of the select expressions contained within it. If multiple expressions are used, the result of the query is an Object[], and the elements in the array correspond to the order of the expressions in the SELECT clause and in type to the result types of each expression.

A SELECT clause cannot specify a collection-valued expression. For example, the SELECT clause p.teams is invalid because teams is a collection. However, the clause in the following query is valid because t is a single element of the teams collection:

The following query is an example of a query with multiple expressions in the SELECT clause:

This query returns a list of Object[] elements; the first array element is a string denoting the customer name, and the second array element is a string denoting the name of the customer’s country.

The result of a query may be the result of an aggregate function, listed in Table 42-11.

Table 42-11 Aggregate Functions in Select Statements

For select method queries with an aggregate function (AVG, COUNT, MAX, MIN, or SUM) in the SELECT clause, the following rules apply.

The following example returns the average order quantity:

The following example returns the total cost of the items ordered by Roxane Coss:

The following example returns the total number of orders:

The following example returns the total number of items that have prices in Hal Incandenza’s order:

The DISTINCT keyword eliminates duplicate return values. If a query returns a java.util.Collection, which allows duplicates, you must specify the DISTINCT keyword to eliminate duplicates.

Constructor expressions allow you to return Java instances that store a query result element instead of an Object[].

The following query creates a CustomerDetail instance per Customer matching the WHERE clause. A CustomerDetail stores the customer name and customer’s country name. So the query returns a List of CustomerDetail instances: