A set expression consists of a sequence of tokens. White space or punctuation delimits tokens. A token may be punctuation, a constant, a name, or a reserved word. Names and reserved words consist of a string of alphanumeric characters or underscores starting with a letter or underscore. Reserved words will always start with a lower-case letter.

The only permissable punctuation within a set expression are parentheses, which are used for grouping in set expressions, filters or transforms; commas, which are used with certain operators; the value operators, and the regular expression matching operator ~. Punctuation tokens are recognized as tokens in themselves and don't have to be set off with whitespace.

There are two kinds of constants, integers and strings. Integer constants are a string of decimal digits and must specify a number between 0 and 2^32 -1 . There are two types of string constants:

Certain operators (~, ~=, substitute()) take regular expressions. Regular expressions can be contained in either kind of string constant, and are Java regular expressions as defined in the documentation for java.util.regex.Pattern.

A sequence of one or more reserved words represents a set operator, a logical operator, a simple set expression, a filter or a transform.

Names are words that must not conflict with any of the reserved words, and must reference an expression or transform previously defined with the set command.

A valid set expression consists of one of the set expressions defined in the language, a name that had been defined as a set expression, a scalar value expression, a set expression with a transform prepended, two set expressions joined by one of the set operators, or a set expression enclosed within parentheses.

A transform is one of the transforms defined by the language, a name that has been defined as a transform, a transform with another transform appended, two transforms joined by one of the set operators, a filter, a count-preserving or group value expression, or a transform enclosed within parentheses.

A filter is one of the filters defined in the language, a filter with the reserved word not prepended, two filters joined by the logical operators and or or, two value expressions related by one of the relational operators, or a filter enclosed within parentheses.

A value expression is a constant, one of the value expression operators defined in the language, two value expressions with compatible context combined with value operators, or a value expression enclosed within parentheses.

Reserved Words:

Reserved words all start with a lowercase letter. Therefore, it is a convention of the language that user defined operators (for set expressions or transforms) start with an uppercase letter; this will distinguish them from current and future reserved words.