1.3 Parsing Syntax

This section describes syntax-parse, the syntax/parse library’s facility for parsing syntax. Both syntax-parse and the specification facility, syntax classes, use a common language of syntax patterns, which is described in detail in Syntax Patterns.

Two parsing forms are provided: syntax-parse and syntax-parser.

Each clause consists of a syntax pattern, an optional sequence of pattern directives, and a non-empty sequence of body forms.

If the syntax object fails to match any of the patterns (or all matches fail the corresponding clauses’ side conditions), a syntax error is raised.

The following options are supported:

#:context context-expr

context-expr   :   (or/c syntax? symbol? #f       (list/c symbol? syntax?))

When present, context-expr is used in reporting parse failures; otherwise stx-expr is used. If context-expr evaluates to (list who context-stx), then who appears in the error message as the form raising the error, and context-stx is used as the term. If context-expr evaluates to a symbol, it is used as who and stx-expr (the syntax to be destructured) is used as context-stx. If context-expr evaluates to a syntax object, it is used as context-stx and who is inferred as with raise-syntax-error.

The current-syntax-context parameter is also set to the syntax object context-stx.

Examples:

> (syntax-parse #'(a b 3)     [(x:id ...) 'ok])

a: expected identifier

at: 3

in: (a b 3)

> (syntax-parse #'(a b 3)     #:context #'(lambda (a b 3) (+ a b))     [(x:id ...) 'ok])

lambda: expected identifier

at: 3

in: (lambda (a b 3) (+ a b))

> (syntax-parse #'(a b 3)     #:context 'check-id-list     [(x:id ...) 'ok])

check-id-list: expected identifier

at: 3

in: (a b 3)

#:literals (literal ...)

literal   =   literal-id     |   (pattern-id literal-id)     |   (pattern-id literal-id #:phase phase-expr)

phase-expr   :   (or/c exact-integer? #f)

Unlike syntax-case, syntax-parse requires all literals to have a binding. To match identifiers by their symbolic names, use #:datum-literals or the ~datum pattern form instead. The #:literals option specifies identifiers that should be treated as literals rather than pattern variables. An entry in the literals list has two components: the identifier used within the pattern to signify the positions to be matched (pattern-id), and the identifier expected to occur in those positions (literal-id). If the entry is a single identifier, that identifier is used for both purposes.

If the #:phase option is given, then the literal is compared at phase phase-expr. Specifically, the binding of the literal-id at phase phase-expr must match the input’s binding at phase phase-expr.

In other words, the syntax-patterns are interpreted as if each occurrence of pattern-id were replaced with the following pattern:

(~literal literal-id #:phase phase-expr)

#:datum-literals (datum-literal ...)

datum-literal   =   literal-id     |   (pattern-id literal-id)

Like #:literals, but the literals are matched as symbols instead of as identifiers.

In other words, the syntax-patterns are interpreted as if each occurrence of pattern-id were replaced with the following pattern:

(~datum literal-id)

#:literal-sets (literal-set ...)

literal-set   =   literal-set-id     |   (literal-set-id literal-set-option ...)           literal-set-option   =   #:at lctx     |   #:phase phase-expr

phase-expr   :   (or/c exact-integer? #f)

Many literals can be declared at once via one or more literal sets, imported with the #:literal-sets option. See literal sets for more information.

If the #:at keyword is given, the lexical context of the lctx term is used to determine which identifiers in the patterns are treated as literals; this option is useful primarily for macros that generate syntax-parse expressions.

#:track-literals

If specified, each final body expression is further constrained to produce a single value, which must be a syntax object, and its 'disappeared-use syntax property is automatically extended to include literals matched as part of pattern-matching. Literals are automatically tracked from uses of #:literals, #:literal-sets, or ~literal, but they can also be manually tracked using syntax-parse-state-cons!. The property is added or extended in the same way as a property added by syntax-parse-track-literals.

Due to the way the body forms are wrapped, specifying this option means the final body form will no longer be in tail position with respect to the enclosing syntax-parse form.

Added in version 6.90.0.29 of package base.

#:conventions (conventions-id ...)

Imports conventions that give default syntax classes to pattern variables that do not explicitly specify a syntax class.

#:local-conventions (convention-rule ...)

Uses the conventions specified. The advantage of #:local-conventions over #:conventions is that local conventions can be in the scope of syntax-class parameter bindings. See the section on conventions for examples.

#:disable-colon-notation

Suppresses the “colon notation” for annotated pattern variables.

Examples:

> (syntax-parse #'(a b c)     [(x:y ...) 'ok])

syntax-parse: not defined as syntax class

at: y

in: (syntax-parse (syntax (a b c)) ((x:y ...) (quote ok)))

> (syntax-parse #'(a b c) #:disable-colon-notation     [(x:y ...) 'ok])

'ok

Compare with define/with-syntax, a similar definition form that uses the simpler syntax-case patterns.