8 Syntax Objects, Patterns, and Templates🔗ℹ

See Syntax Objects for an overview of syntax objects.

A syntax object is a representation of source code. It is written as quoted term using single quotes, such as '1 + * 4 f()'. A syntax object is not a string, even though it uses quote marks. Instead, the usual shrubbery rules apply inside quotes, and shrubbery structure is preserved in a syntax object. That is, a string records a sequence of characters, but a syntax object represents structure that might be written in different ways; in particular, the printed form of a syntax object tends to use «» shrubbery notation instead of shrubbery notation’s whitespace-sensitive format.

When a syntax object is written using quotes, the #%quotes form is used implicitly, similar to the way that square brackets implicitly use #%brackets.

> 'apple'

- Syntax

'apple'

> #%quotes 'apple'

- Syntax

'apple'

> '1 + 2'

- Syntax

'1 + 2'

> 'fun (x):      x + 1'

- Syntax

'fun (x):« x + 1 »'

An expression written with quotes is more generally a template, because it can include $ as an escape and an ellipsis (written as ...) to instantiate a repetition. Syntax patterns in match are also written with quotes, $ as an escape, and an ellipsis to match a repetition.

Rules for $ and ... in patterns:

When $ appears in a pattern, then it must be followed by either (1) an identifier, or (2) parentheses containing an identifier followed by :: Identifier.

In the simple case that the identifier is not annotated with :: Identifier and the pattern has no ellipses (written as ...), then the identifier is bound to a non-empty sequence of terms from the corresponding part of the input syntax object. Sequences are matched greedily, meaning that as many terms are matched to an identifier as possible to find a match for the pattern overall.

> match '1 2 3 4'   | '1 $x 4': x

- Syntax

'2 3'

> match 'a b c d'   | '$x $y': [x, y]

- Listof(Syntax)

['a b c', 'd']

An identifier annotated with :: Identifier after $ can only match an individual identifier in the input.

> match 'a b c d'   | '$(x :: Identifier) $y': [x, y]

['a', 'b c d']

> match '1 2'   | '$(x :: Identifier) $y': [x, y]   | '$z': [z]

['1 2']

When an ellipsis ... appears in a pattern, then it greedily matches 0 or more repetitions of the preceding pattern element. If the preceding element contains a $ escape, then the escaped identifier is not bound to a single syntax object, but it is instead bound as a repetition that holds each matched term. A repetition can only be referenced through a corresponding escape in a template. Ellipses can be nested in a pattern, and a repettion must be used in a template with the same amount of nesting as in its pattern.

> match 'a a a b d'   | 'a ... b c ... d': "matches"

- String

"matches"

> match 'a a a b'   | '$x ... b': 'matches as $x ...'

- Syntax

'matches as a a a'

> match '(a 1) (b 2) (c 3)'   | '($x $y) ...': ['$x ...', '$y ...']

- Listof(Syntax)

['a b c', '1 2 3']

> match '(a: 1) (b: 2 3 4) (c: 5 6)'   | '($x: $y ...) ...': ['$x ...', '($y ...) ...']

- Listof(Syntax)

['a b c', '(1) (2 3 4) (5 6)']

When an ellipsis appears by itself in a shrubbery group, then the pattern matches 0 or more repetitions of the preceding group. If the group is an alternative written with |, then the pattern matches 0 or more alternatives.

> match 'a          b c          d'   | '$x      ...':       '{$x, ...}'

- Syntax

'{a, b c, d}'

> match 'cases          | a          | b c          | d'   | 'cases      | $x      | ...':       '{$x, ...}'

- Syntax

'{a, b c, d}'

Rules for $ and ... in templates:

When $ appears in a template, then it must be followed by an identifier or an expression that is written as a single term (e.g., a parenthesized expression).

If an identifier is provided, then it can refer to a repetition that is bound by a syntax-pattern match, as long as the $-escaped identifier in the template is under a number of ellipses that match the repetition binding. Each element of the repetition is put in the result syntax object in place of the escape. The rules for patterns above show several example templates that use repetitions.

If an escape does not refer to a repetition, then it must have an expression that produces a syntax object, and it must not be under any ellipses. The syntax object replaces the escape in the result syntax object.

> '1 $(if #true | '2' | 'oops') 3'

- Syntax

'1 2 3'

See above for information about $ escapes within the quotes for a syntax object.

The syntax_is_list function checks whether a syntax object would match the pattern '[$elem ..., ...]', returning #true if so and #false otherwise.

> syntax_is_integer('9')

- Boolean

#true

> syntax_is_integer('apple')

- Boolean

#false

> syntax_is_list('[w, x y, z]')

- Boolean

#true

> syntax_is_list('a b c')

- Boolean

#false

> syntax_to_integer('9')

- Int

9

> syntax_to_list('[w, x y, z]')

- Listof(Syntax)

['w', 'x y', 'z']

> integer_to_syntax(9)

- Syntax

'9'

> list_to_syntax(['w', 'x y', 'z'])

- Syntax

'[w, x y, z]'

The syntax_split function takes a single-group syntax object and splits it into a list of term syntax objects, or it takes a multi-group syntax object and splits it into a list of group syntax objects.

The syntax_join function combines a list of term syntax objects into one group syntax object, and it splices a list of multi-group syntax objects into a multi-group syntax object.

> syntax_split('1 2 3')

- Listof(Syntax)

['1', '2', '3']

> syntax_split('1 2 3                 4 5')

- Listof(Syntax)

['1 2 3', '4 5']

> syntax_join(['1', '2', '3'])

- Syntax

'1 2 3'

> syntax_join(['1', '2 3', '4'])

- Syntax

'1; 2 3; 4'

> syntax_join(['1', '2 3', '4                             5'])

- Syntax

'1; 2 3; 4; 5'