mbutterick
/
br-parser-tools
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

*** empty log message ***

Browse Source 
original commit: a17cdebeb65285162c71e3314daf8193ffa69de8
tokens
Scott Owens 23 years ago
parent a237798ba8
commit 8111e35427
1 changed files with 121 additions and 99 deletions
Show Stats Download Patch File Download Diff File

52
collects/parser-tools/lex.ss
Unescape Escape View File

@ -8,27 +8,30 @@
(require (lib "list.ss") (require (lib "list.ss")
(lib "readerr.ss" "syntax") (lib "readerr.ss" "syntax")
"private-lex/token.ss") "private-lex/token.ss")
(provide lexer define-lex-abbrev define-lex-abbrevs (provide lexer lexer-src-loc define-lex-abbrev define-lex-abbrevs
make-lex-buf make-lex-buf
get-position position-offset position-line position-col position? get-position position-offset position-line position-col position?
define-tokens define-empty-tokens) define-tokens define-empty-tokens)
(define-syntax lexer (define-syntaxes (lexer lexer-src-loc)
(let ((code (let ((code
(lambda (wrap)
`(letrec ((match `(letrec ((match
(lambda (lb first-pos end-pos longest-match-length longest-match-action length) (lambda (lb first-pos longest-match-length longest-match-action length)
(let ((match (let ((match
(push-back lb (- length longest-match-length)))) (push-back lb (- length longest-match-length)))
(end-pos (get-position lb)))
(if (not longest-match-action) (if (not longest-match-action)
(raise-read-error (raise-read-error
(format "lexer: No match found in input starting with: ~a" (format "lexer: No match found in input starting with: ~a"
(list->string (lex-buffer-from lb))) (list->string (filter char? (lex-buffer-from lb))))
#f #f
(position-line first-pos) (position-line first-pos)
(position-col first-pos) (position-col first-pos)
(position-offset first-pos) (position-offset first-pos)
(- (position-offset end-pos) (position-offset first-pos)))) (- (position-offset end-pos) (position-offset first-pos))))
(,wrap
(longest-match-action (longest-match-action
(lambda () (lambda ()
first-pos) first-pos)
@ -38,7 +41,7 @@
(if (char? (car match)) (if (char? (car match))
(list->string (reverse match)) (list->string (reverse match))
(list->string (reverse (cdr match))))) (list->string (reverse (cdr match)))))
lb))))) lb))))))
(lambda (lb) (lambda (lb)
(unless (lex-buffer? lb) (unless (lex-buffer? lb)
(raise-type-error (raise-type-error
@ -60,8 +63,7 @@
;; including the one just read ;; including the one just read
(length 1) (length 1)
;; how many characters are in the longest match ;; how many characters are in the longest match
(longest-match-length 0) (longest-match-length 0))
(end-pos first-pos))
(let ((next-state (let ((next-state
(cond (cond
((eof-object? char) ((eof-object? char)
@ -70,12 +72,10 @@
(vector-ref (vector-ref
trans-table trans-table
(bitwise-ior (char->integer char) (bitwise-ior (char->integer char)
(arithmetic-shift state 8)))))) (arithmetic-shift state 8)))))))
(pos (get-position lb)))
(cond (cond
((not next-state) (match lb ((not next-state) (match lb
first-pos first-pos
end-pos
longest-match-length longest-match-length
longest-match-action longest-match-action
length)) length))
@ -89,8 +89,8 @@
(add1 length) (add1 length)
(if act (if act
length length
longest-match-length) longest-match-length)))))))))))))
pos))))))))))) (values
(lambda (stx) (lambda (stx)
(syntax-case stx () (syntax-case stx ()
((_) ((_)
@ -111,8 +111,30 @@
(trans-table ,(table-trans table)) (trans-table ,(table-trans table))
(eof-table ,(table-eof table)) (eof-table ,(table-eof table))
(actions (vector ,@(vector->list (table-actions table))))) (actions (vector ,@(vector->list (table-actions table)))))
,code))) ,(code `(lambda (x) x)))))
(datum->syntax-object #'here code #f)))))))) (datum->syntax-object #'here code #f))))))
(lambda (stx)
(syntax-case stx ()
((_)
(raise-syntax-error #f "empty lexer is not allowed" stx))
((_ re-act ...)
(begin
(for-each
(lambda (x)
(syntax-case x ()
((re act) (void))
(_ (raise-syntax-error 'lexer
"expects regular expression / action pairs"
x))))
(syntax->list (syntax (re-act ...))))
(let* ((table (generate-table (syntax (re-act ...)) stx))
(code
`(let ((start-state ,(table-start table))
(trans-table ,(table-trans table))
(eof-table ,(table-eof table))
(actions (vector ,@(vector->list (table-actions table)))))
,(code `(lambda (x) (list x first-pos end-pos))))))
(datum->syntax-object #'here code #f)))))))))
(define-syntax (define-lex-abbrev stx) (define-syntax (define-lex-abbrev stx)

Write Preview
Loading…
Cancel
Save