1999-09-23 10:27:41 -04:00
|
|
|
;;; Module definitions for the scsh regexp system.
|
|
|
|
;;; This is a sleazy modularisation -- we just load everything into
|
|
|
|
;;; scsh-level-0, and export from there.
|
|
|
|
;;; -Olin <shivers@ai.mit.edu> 8/98
|
|
|
|
|
|
|
|
(define-interface basic-re-interface
|
|
|
|
(export (re-dsm? (proc (:value) :boolean))
|
|
|
|
(make-re-dsm (proc (:value :exact-integer :exact-integer) :value))
|
|
|
|
(re-dsm:body (proc (:value) :value))
|
|
|
|
(re-dsm:pre-dsm (proc (:value) :exact-integer))
|
|
|
|
(re-dsm:tsm (proc (:value) :exact-integer))
|
|
|
|
(re-dsm:posix (proc (:value) :value))
|
|
|
|
(set-re-dsm:posix (proc (:value :value) :unspecific))
|
|
|
|
((re-dsm:post-dsm re-dsm) (proc (:value) :exact-integer))
|
|
|
|
(open-dsm (proc (:value) (some-values :value :exact-integer)))
|
|
|
|
|
|
|
|
(re-seq? (proc (:value) :boolean))
|
|
|
|
(%%make-re-seq (proc (:value :exact-integer :value) :value))
|
|
|
|
(%make-re-seq (proc (:value :exact-integer) :value))
|
|
|
|
((re-seq make-re-seq) (proc (:value) :value))
|
|
|
|
(re-seq:elts (proc (:value) :value))
|
|
|
|
(re-seq:tsm (proc (:value) :exact-integer))
|
|
|
|
(re-seq:posix (proc (:value) :value))
|
|
|
|
(set-re-seq:posix (proc (:value :value) :unspecific))
|
|
|
|
|
|
|
|
(re-choice? (proc (:value) :boolean))
|
|
|
|
(%%make-re-choice (proc (:value :exact-integer :value) :value))
|
|
|
|
(%make-re-choice (proc (:value :exact-integer) :value))
|
|
|
|
((make-re-choice re-choice) (proc (:value) :value))
|
|
|
|
(re-choice:elts (proc (:value) :value))
|
|
|
|
(re-choice:tsm (proc (:value) :exact-integer))
|
|
|
|
(re-choice:posix (proc (:value) :value))
|
|
|
|
(set-re-choice:posix (proc (:value :value) :unspecific))
|
|
|
|
|
|
|
|
(re-repeat? (proc (:value) :boolean))
|
|
|
|
(%%make-re-repeat (proc (:exact-integer :value :value
|
|
|
|
:exact-integer :value)
|
|
|
|
:value))
|
|
|
|
(%make-re-repeat (proc (:exact-integer :value :value :exact-integer )
|
|
|
|
:value))
|
|
|
|
((re-repeat make-re-repeat)
|
|
|
|
(proc (:exact-integer :value :value) :value))
|
|
|
|
((re-repeat:from re-repeat:tsm)
|
|
|
|
(proc (:value) :exact-integer))
|
|
|
|
(re-repeat:to (proc (:value) :value))
|
|
|
|
((re-repeat:body re-repeat:posix)
|
|
|
|
(proc (:value) :value))
|
|
|
|
(set-re-repeat:posix (proc (:value :value) :unspecific))
|
|
|
|
|
|
|
|
(re-submatch? (proc (:value) :boolean))
|
|
|
|
(%%make-re-submatch (proc (:value :exact-integer :exact-integer :value)
|
|
|
|
:value))
|
|
|
|
(%make-re-submatch (proc (:value :exact-integer :exact-integer) :value))
|
|
|
|
((make-re-submatch re-submatch)
|
|
|
|
(proc (:value &opt :exact-integer :exact-integer) :value))
|
|
|
|
|
|
|
|
(re-submatch:body (proc (:value) :value))
|
|
|
|
((re-submatch:pre-dsm re-submatch:tsm re-submatch:post-dsm)
|
|
|
|
(proc (:value) :exact-integer))
|
|
|
|
(re-submatch:posix (proc (:value) :value))
|
|
|
|
(set-re-submatch:posix (proc (:value :value) :unspecific))
|
|
|
|
|
|
|
|
(re-string? (proc (:value) :boolean))
|
|
|
|
((make-re-string re-string) (proc (:string) :value))
|
|
|
|
(re-string:chars (proc (:value) :string))
|
|
|
|
(set-re-string:chars (proc (:value :string) :unspecific))
|
|
|
|
(re-string:posix (proc (:value) :value))
|
|
|
|
(set-re-string:posix (proc (:value :value) :unspecific))
|
|
|
|
|
|
|
|
re-trivial
|
|
|
|
(re-trivial? (proc (:value) :boolean))
|
|
|
|
|
|
|
|
(re-char-set? (proc (:value) :boolean))
|
|
|
|
((make-re-char-set re-char-set) (proc (:value) :value))
|
|
|
|
(re-char-set:cset (proc (:value) :value))
|
|
|
|
(set-re-char-set:cset (proc (:value :value) :unspecific))
|
|
|
|
(re-char-set:posix (proc (:value) :value))
|
|
|
|
(set-re-char-set:posix (proc (:value :value) :unspecific))
|
|
|
|
|
|
|
|
re-empty
|
|
|
|
(re-empty? (proc (:value) :boolean))
|
|
|
|
re-bos re-eos
|
|
|
|
re-bol re-eol
|
|
|
|
re-bow re-eow
|
|
|
|
|
|
|
|
((re-bos? re-eos? re-bol? re-eol? re-bow? re-eow? re-any?)
|
|
|
|
(proc (:value) :boolean))
|
|
|
|
|
|
|
|
re-any
|
|
|
|
re-nonl
|
|
|
|
re-word
|
|
|
|
|
|
|
|
(regexp? (proc (:value) :boolean))
|
|
|
|
(re-tsm (proc (:value) :exact-integer))
|
|
|
|
|
|
|
|
;; These guys can be in code produced by RX expander.
|
|
|
|
(flush-submatches (proc (:value) :value))
|
|
|
|
(uncase (proc (:value) :value))
|
|
|
|
(uncase-char-set (proc (:value) :value))
|
|
|
|
(uncase-string (proc (:string) :value))
|
|
|
|
))
|
|
|
|
|
|
|
|
|
|
|
|
;;; These guys were made obsolete by the new SRE package and exist for
|
|
|
|
;;; backwards compatibility only.
|
|
|
|
(define-interface re-old-funs-interface
|
|
|
|
(export
|
|
|
|
(string-match (proc (:value :string &opt :exact-integer) :value))
|
|
|
|
(make-regexp (proc (:string) :value))
|
|
|
|
(regexp-exec (proc (:value :string &opt :exact-integer) :value))
|
|
|
|
(->regexp (proc (:value) :value))
|
|
|
|
(regexp-quote (proc (:string) :value))))
|
|
|
|
|
|
|
|
|
|
|
|
(define-interface re-internals-interface
|
|
|
|
;; These are constructors for the Scheme unparser
|
|
|
|
(export
|
|
|
|
(make-re-string/posix (proc (:string :string :vector) :value))
|
|
|
|
((%make-re-seq/posix %make-re-choice/posix)
|
|
|
|
(proc (:value :exact-integer :string :vector) :value))
|
|
|
|
(make-re-char-set/posix (proc (:value :string :vector) :value))
|
|
|
|
(%make-re-repeat/posix (proc (:exact-integer :value :value :exact-integer :string :vector)
|
|
|
|
:value))
|
|
|
|
(%make-re-dsm/posix (proc (:value :exact-integer :exact-integer :string :vector)
|
|
|
|
:value))
|
|
|
|
(%make-re-submatch/posix (proc (:value :exact-integer :exact-integer :string :vector) :value))))
|
|
|
|
|
|
|
|
|
|
|
|
(define re-match-internals-interface
|
|
|
|
(export (regexp-match:string (proc (:value) :string))
|
|
|
|
(regexp-match:start (proc (:value) :vector))
|
|
|
|
(regexp-match:end (proc (:value) :vector))))
|
|
|
|
|
|
|
|
|
|
|
|
(define-interface posix-re-interface
|
|
|
|
(export (regexp->posix-string (proc (:value) :string)) ; posixstr.scm
|
|
|
|
(posix-string->regexp (proc (:string) :value)) ; spencer
|
|
|
|
))
|
|
|
|
|
|
|
|
(define-interface re-subst-interface
|
|
|
|
(export
|
|
|
|
(regexp-substitute (proc (:value :value &rest :value) :value))
|
|
|
|
(regexp-substitute/global (proc (:value :value :string &rest :value) :value))))
|
|
|
|
|
|
|
|
(define-interface re-folders-interface
|
|
|
|
(export
|
|
|
|
(regexp-fold (proc (:value (proc (:exact-integer :value :value) :value)
|
|
|
|
:value
|
|
|
|
:string
|
|
|
|
&opt (proc (:exact-integer :value) :value)
|
|
|
|
:exact-integer)
|
|
|
|
:value))
|
|
|
|
(regexp-fold (proc (:value (proc (:value :exact-integer :value) :value)
|
|
|
|
:value
|
|
|
|
:string
|
|
|
|
&opt (proc (:exact-integer :value) :value)
|
|
|
|
:exact-integer)
|
|
|
|
:value))
|
|
|
|
(regexp-for-each (proc (:value (proc (:value) :unspecific)
|
|
|
|
:string &opt :exact-integer)
|
|
|
|
:unspecific))))
|
|
|
|
|
|
|
|
(define-interface re-level-0-interface
|
|
|
|
(compound-interface posix-re-interface
|
|
|
|
basic-re-interface
|
|
|
|
(export (regexp-match? (proc (:value) :boolean))
|
|
|
|
(match:start (proc (:value :exact-integer) :value))
|
|
|
|
(match:end (proc (:value :exact-integer) :value))
|
|
|
|
(match:substring (proc (:value :exact-integer) :value))
|
|
|
|
(clean-up-cres (proc () :unspecific))
|
|
|
|
(regexp-search (proc (:value :string &opt :exact-integer)
|
|
|
|
:value))
|
|
|
|
(regexp-search? (proc (:value :string &opt :exact-integer)
|
|
|
|
:boolean))
|
|
|
|
(sre->regexp (proc (:value) :value))
|
|
|
|
(regexp->sre (proc (:value) :value))
|
2000-08-28 08:43:27 -04:00
|
|
|
(simplify-regexp (proc (:value) :value))
|
1999-09-23 10:27:41 -04:00
|
|
|
)))
|
|
|
|
|
|
|
|
|
|
|
|
(define-structures ((re-level-0 re-level-0-interface)
|
|
|
|
(re-match-internals re-match-internals-interface)
|
|
|
|
(re-internals re-internals-interface)
|
2000-08-28 08:43:27 -04:00
|
|
|
(sre-syntax-tools (export (if-sre-form :syntax)
|
1999-09-23 10:27:41 -04:00
|
|
|
sre-form?
|
|
|
|
parse-sre parse-sres
|
2000-08-28 08:43:27 -04:00
|
|
|
regexp->scheme
|
|
|
|
static-regexp?))
|
|
|
|
(sre-internal-syntax-tools (export expand-rx)))
|
1999-09-23 10:27:41 -04:00
|
|
|
(open scsh-utilities
|
|
|
|
defrec-package
|
|
|
|
define-foreign-syntax
|
|
|
|
weak
|
|
|
|
;re-posix-parsers ; regexp->posix-string
|
|
|
|
let-opt
|
|
|
|
sort ; Posix renderer
|
|
|
|
conditionals
|
|
|
|
define-record-types
|
|
|
|
defrec-package
|
|
|
|
receiving
|
|
|
|
char-set-package
|
|
|
|
error-package
|
|
|
|
ascii
|
1999-10-08 10:11:52 -04:00
|
|
|
primitives ; JMG add-finalizer!
|
|
|
|
define-record-types ; JMG debugging
|
1999-09-23 10:27:41 -04:00
|
|
|
string-lib ; string-fold
|
|
|
|
scheme)
|
2000-08-28 08:43:27 -04:00
|
|
|
|
1999-09-23 10:27:41 -04:00
|
|
|
(files re-low re simp re-high
|
|
|
|
parse posixstr spencer re-syntax)
|
2000-08-28 08:43:27 -04:00
|
|
|
|
|
|
|
(begin (define-syntax if-sre-form
|
|
|
|
(lambda (exp r c)
|
|
|
|
(if (sre-form? (cadr exp) r c)
|
|
|
|
(caddr exp)
|
|
|
|
(cadddr exp)))))
|
|
|
|
|
|
|
|
; (optimize auto-integrate)
|
1999-09-23 10:27:41 -04:00
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
;;; Stuff that could appear in code produced by (rx ...)
|
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
|
|
|
|
|
(define-interface rx-lib-interface
|
|
|
|
(compound-interface (export coerce-dynamic-regexp
|
|
|
|
coerce-dynamic-charset
|
|
|
|
spec->char-set
|
|
|
|
flush-submatches
|
|
|
|
uncase
|
|
|
|
uncase-char-set
|
|
|
|
uncase-string)
|
|
|
|
re-internals-interface))
|
|
|
|
|
|
|
|
(define-structure rx-lib rx-lib-interface
|
|
|
|
(open re-internals
|
|
|
|
conditionals
|
|
|
|
re-level-0
|
|
|
|
char-set-package
|
|
|
|
scsh-utilities ; fold
|
|
|
|
error-package
|
|
|
|
ascii
|
|
|
|
scheme)
|
|
|
|
(files rx-lib)
|
1999-10-08 10:11:52 -04:00
|
|
|
; (optimize auto-integrate)
|
|
|
|
)
|
1999-09-23 10:27:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
|
|
2000-08-28 08:43:27 -04:00
|
|
|
(define-interface rx-syntax-interface (export (rx :syntax)))
|
1999-09-23 10:27:41 -04:00
|
|
|
|
|
|
|
(define-structure rx-syntax rx-syntax-interface
|
|
|
|
(open re-level-0
|
|
|
|
char-set-package
|
|
|
|
rx-lib
|
|
|
|
scheme)
|
2000-08-28 08:43:27 -04:00
|
|
|
(for-syntax (open sre-internal-syntax-tools scheme))
|
|
|
|
(begin (define-syntax rx expand-rx))
|
1999-10-08 10:11:52 -04:00
|
|
|
; (optimize auto-integrate)
|
|
|
|
)
|
1999-09-23 10:27:41 -04:00
|
|
|
|
2000-08-28 08:43:27 -04:00
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
|
|
|
|
|
(define-interface re-match-syntax-interface
|
|
|
|
(export (let-match :syntax)
|
|
|
|
(if-match :syntax)
|
|
|
|
(match-cond :syntax)))
|
|
|
|
|
|
|
|
(define-structure re-match-syntax re-match-syntax-interface
|
|
|
|
(for-syntax (open scheme
|
|
|
|
signals)) ; For ERROR
|
|
|
|
(open re-level-0 scheme)
|
|
|
|
(access signals) ; for ERROR
|
|
|
|
(files re-match-syntax))
|
|
|
|
|
|
|
|
|
|
|
|
|
1999-09-23 10:27:41 -04:00
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
|
|
|
|
|
(define-structure re-old-funs re-old-funs-interface
|
|
|
|
(open re-level-0 error-package receiving scheme)
|
|
|
|
(files oldfuns)
|
1999-10-08 10:11:52 -04:00
|
|
|
; (optimize auto-integrate)
|
|
|
|
)
|
1999-09-23 10:27:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
(define-structure re-subst re-subst-interface
|
|
|
|
(open re-level-0
|
|
|
|
re-match-internals
|
|
|
|
scsh-utilities ; fold & some string utilities that need to be moved.
|
|
|
|
scsh-level-0 ; write-string
|
|
|
|
string-lib ; string-copy!
|
|
|
|
scheme)
|
|
|
|
(files re-subst)
|
1999-10-08 10:11:52 -04:00
|
|
|
; (optimize auto-integrate)
|
|
|
|
)
|
1999-09-23 10:27:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
(define-structure re-folders re-folders-interface
|
|
|
|
(open re-level-0 let-opt conditionals error-package scheme)
|
|
|
|
(files re-fold)
|
1999-10-08 10:11:52 -04:00
|
|
|
; (optimize auto-integrate)
|
|
|
|
)
|
1999-09-23 10:27:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
(define-interface re-exports-interface
|
|
|
|
(compound-interface re-level-0-interface
|
|
|
|
rx-syntax-interface
|
|
|
|
re-subst-interface
|
2000-08-28 08:43:27 -04:00
|
|
|
re-match-syntax-interface
|
1999-09-23 10:27:41 -04:00
|
|
|
re-folders-interface))
|
|
|
|
|
|
|
|
(define-structure re-exports re-exports-interface
|
2000-08-28 08:43:27 -04:00
|
|
|
(open rx-syntax
|
|
|
|
re-level-0
|
|
|
|
re-subst
|
|
|
|
re-match-syntax
|
|
|
|
re-folders)
|
1999-10-08 10:11:52 -04:00
|
|
|
; (optimize auto-integrate)
|
|
|
|
)
|
1999-09-23 10:27:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
;;; File Exports
|
|
|
|
;;; ---- -------
|
|
|
|
;;; parse sre->regexp regexp->sre
|
|
|
|
;;; parse-sre parse-sres regexp->scheme
|
|
|
|
;;; char-set->in-pair static-regexp?
|
|
|
|
;;; posixstr regexp->posix-string
|
|
|
|
;;; re-high compile-regexp regexp-search regexp-search?
|
|
|
|
;;; re-subst regexp-substitute regexp-substitute/global
|
|
|
|
;;; re-low match:start match:end match:substring
|
|
|
|
;;; CRE record, new-cre, compile-posix-re->c-struct
|
|
|
|
;;; cre-search cre-search? clean-up-cres
|
|
|
|
;;; re-syntax sre-form? if-sre-form expand-rx
|
|
|
|
;;; re.scm The ADT. flush-submatches uncase uncase-char-set
|
|
|
|
;;; char-set-full? char-set-empty?
|
|
|
|
;;; re-char-class? static-char-class?
|
|
|
|
;;; rx-lib coerce-dynamic-regexp coerce-dynamic-charset spec->char-set
|
|
|
|
;;; simp simplify-regexp
|
|
|
|
;;; spencer posix-string->regexp
|