You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
beautiful-racket/beautiful-racket-ragg/br/ragg/codegen/sexp-based-lang.rkt

96 lines
2.5 KiB
Racket

9 years ago
#lang racket/base
;; A language level for automatically generating parsers out of BNF grammars.
;;
;; Danny Yoo (dyoo@hashcollision.org)
;;
;; Intent: make it trivial to generate languages for Racket. At the
;; moment, I find it painful to use parser-tools. This library is
;; meant to make it less agonizing.
;;
;; The intended use of this language is as follows:
;;
;;;;; s-exp-grammar.rkt ;;;;;;;;;
;; #lang br/ragg
;; s-exp : "(" s-exp* ")" | ATOM
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; What this generates is:
;;
;; * parse: a function that consumes a source and a
;; position-aware lexer, and produces a syntax object.
;;
;; * make-rule-parser: a custom parser given a provided start rule.
;;
;; You'll still need to do a little work, by providing a lexer that
;; defines what the uppercased tokens mean. For example, you can
;; use the parser-tools/lex lexer tools:
;;
;; (require ragg/support
;; parser-tools/lex
;; parser-tools/lex-sre)
;;
;; (define tokenize
;; (lexer-src-pos
;; [(:+ alphabetic)
;; (token 'ATOM lexeme)]
;; [whitespace
;; (return-without-pos (tokenize/1 input-port))]
;; [(:or "(" ")")
;; (token lexeme lexeme)]))
;;
;; However, that should be all you need. The output of an
;; generated grammar is an honest-to-goodness syntax
;; object with source locations, fully-labeled by the rules.
;;
;; (parse (tokenize an-input-port))
;;
;;
;; The first rule is treated as the start rule; any successful parse
;; must finish with end-of-file.
;; Terminology:
;;
;; A rule is a rule identifier, followed by a colon ":", followed by a
;; pattern.
;; A rule identifier is an identifier that is not in upper case.
;; A rule identifier should follow the Racket rules for identifiers,
;; except that it can't contain * or +.
;;
;; A token is a rule identifier that is all in upper case.
;; A pattern may either be
;;
;; * an implicit sequence of patterns,
;;
;; * a literal string,
;;
;; * a rule identifier,
;;
;; * a quanitifed pattern, either with "*" or "+",
;;
;; * an optional pattern: a pattern surrounded by "[" and "]", or
;;
;; * a grouped sequence: a pattern surrounded by "(" and ")".
(require (for-syntax racket/base
"codegen.rkt"))
(provide rules
(rename-out [#%plain-module-begin #%module-begin]))
(define-syntax (rules stx)
(rules-codegen #:parser-provider-module 'br/ragg/cfg-parser/cfg-parser ;; 'parser-tools/yacc
#:parser-provider-form 'cfg-parser ;; 'parser
stx))