mbutterick
/
brag
1
0
Fork 0
Code Issues 3 Pull Requests 1 Projects Releases Wiki Activity

allow blank eof rule

Browse Source
pull/6/head
Matthew Butterick 7 years ago
parent 73e59188a7
commit 5a316c669d
2 changed files with 342 additions and 385 deletions
Show Stats Download Patch File Download Diff File

391
br-parser-tools/br-parser-tools-lib/br-parser-tools/lex.rkt
Unescape Escape View File

@ -1,28 +1,30 @@
(module lex mzscheme #lang racket/base
;; Provides the syntax used to create lexers and the functions needed to ;; Provides the syntax used to create lexers and the functions needed to
;; create and use the buffer that the lexer reads from. See docs. ;; create and use the buffer that the lexer reads from. See docs.
(require-for-syntax mzlib/list (require (for-syntax mzlib/list
syntax/stx syntax/stx
syntax/define syntax/define
syntax/boundmap syntax/boundmap
"private-lex/util.rkt" "private-lex/util.rkt"
"private-lex/actions.rkt" "private-lex/actions.rkt"
"private-lex/front.rkt" "private-lex/front.rkt"
"private-lex/unicode-chars.rkt") "private-lex/unicode-chars.rkt"
racket/base
racket/promise))
(require mzlib/stxparam (require mzlib/stxparam
syntax/readerr syntax/readerr
"private-lex/token.rkt") "private-lex/token.rkt")
(provide lexer lexer-src-pos lexer-srcloc define-lex-abbrev define-lex-abbrevs define-lex-trans (provide lexer lexer-src-pos lexer-srcloc define-lex-abbrev define-lex-abbrevs define-lex-trans
;; Dealing with tokens and related structures ;; Dealing with tokens and related structures
define-tokens define-empty-tokens token-name token-value token? define-tokens define-empty-tokens token-name token-value token?
(struct position (offset line col)) (struct-out position)
(struct position-token (token start-pos end-pos)) (struct-out position-token)
(struct srcloc-token (token srcloc)) (struct-out srcloc-token)
;; File path for highlighting errors while lexing ;; File path for highlighting errors while lexing
file-path file-path
@ -35,9 +37,9 @@
;; A regular expression operator ;; A regular expression operator
char-set) char-set)
;; wrap-action: syntax-object src-pos? -> syntax-object ;; wrap-action: syntax-object src-pos? -> syntax-object
(define-for-syntax (wrap-action action src-loc-style) (define-for-syntax (wrap-action action src-loc-style)
(with-syntax ((action-stx (with-syntax ([action-stx
(cond (cond
[(eq? src-loc-style 'lexer-src-pos) [(eq? src-loc-style 'lexer-src-pos)
#`(let/ec ret #`(let/ec ret
@ -49,9 +51,9 @@
(syntax-parameterize (syntax-parameterize
([return-without-srcloc (make-rename-transformer #'ret)]) ([return-without-srcloc (make-rename-transformer #'ret)])
(make-srcloc-token #,action lexeme-srcloc)))] (make-srcloc-token #,action lexeme-srcloc)))]
[else action]))) [else action])])
(syntax/loc action (syntax/loc action
(lambda (start-pos-p end-pos-p lexeme-p input-port-p) (λ (start-pos-p end-pos-p lexeme-p input-port-p)
(define lexeme-srcloc-p (make-srcloc (object-name input-port-p) (define lexeme-srcloc-p (make-srcloc (object-name input-port-p)
(position-line start-pos-p) (position-line start-pos-p)
(position-col start-pos-p) (position-col start-pos-p)
@ -68,146 +70,111 @@
[lexeme-srcloc (make-rename-transformer #'lexeme-srcloc-p)]) [lexeme-srcloc (make-rename-transformer #'lexeme-srcloc-p)])
action-stx))))) action-stx)))))
(define-for-syntax (make-lexer-trans src-loc-style) (define-for-syntax (make-lexer-macro caller src-loc-style)
(lambda (stx) (λ (stx)
(syntax-case stx () (syntax-case stx ()
((_ re-act ...) [(_ . RE+ACTS)
(begin (let ()
(for-each (define spec/re-acts (syntax->list #'RE+ACTS))
(lambda (x) (for/and ([x (in-list spec/re-acts)])
(syntax-case x () (syntax-case x ()
((re act) (void)) [(RE ACT) #t]
(_ (raise-syntax-error #f [else (raise-syntax-error caller "not a regular expression / action pair" stx x)]))
"not a regular expression / action pair" (define eof-act (get-special-action spec/re-acts #'eof #'eof))
stx (define spec-act (get-special-action spec/re-acts #'special #'(void)))
x)))) (define spec-comment-act (get-special-action spec/re-acts #'special-comment #'#f))
(syntax->list (syntax (re-act ...)))) (define ids (list #'special #'special-comment #'eof))
(let* ((spec/re-act-lst (define re-acts (filter (λ (spec/re-act)
(syntax->list (syntax (re-act ...))))
(eof-act
(get-special-action spec/re-act-lst #'eof #''eof))
(spec-act
(get-special-action spec/re-act-lst #'special #'(void)))
(spec-comment-act
(get-special-action spec/re-act-lst #'special-comment #'#f))
(ids (list #'special #'special-comment #'eof))
(re-act-lst
(filter
(lambda (spec/re-act)
(syntax-case spec/re-act () (syntax-case spec/re-act ()
(((special) act) [((special) act)
(not (ormap (not (ormap
(lambda (x) (λ (x)
(and (identifier? #'special) (and (identifier? #'special)
(module-or-top-identifier=? (syntax special) x))) (module-or-top-identifier=? #'special x)))
ids))) ids))]
(_ #t))) [_ #t])) spec/re-acts))
spec/re-act-lst)) (define names (map (λ (x) (datum->syntax #f (gensym))) re-acts))
(name-lst (map (lambda (x) (datum->syntax-object #f (gensym))) re-act-lst)) (define acts (map (λ (x) (stx-car (stx-cdr x))) re-acts))
(act-lst (map (lambda (x) (stx-car (stx-cdr x))) re-act-lst)) (define re-actnames (map (λ (re-act name) (list (stx-car re-act) name)) re-acts names))
(re-actname-lst (map (lambda (re-act name) (when (null? spec/re-acts)
(list (stx-car re-act) (raise-syntax-error caller "expected at least one action" stx))
name)) (define-values (trans start action-names no-look disappeared-uses) (build-lexer re-actnames))
re-act-lst
name-lst)))
(when (null? spec/re-act-lst)
(raise-syntax-error (or src-loc-style 'lexer) "expected at least one action" stx))
(let-values (((trans start action-names no-look disappeared-uses)
(build-lexer re-actname-lst)))
(when (vector-ref action-names start) ;; Start state is final (when (vector-ref action-names start) ;; Start state is final
(unless (and (unless (and
;; All the successor states are final ;; All the successor states are final
(andmap (lambda (x) (vector-ref action-names (vector-ref x 2))) (vector? (vector-ref trans start))
(andmap (λ (x) (vector-ref action-names (vector-ref x 2)))
(vector->list (vector-ref trans start))) (vector->list (vector-ref trans start)))
;; Each character has a successor state ;; Each character has a successor state
(let loop ((check 0) (let loop ([check 0]
(nexts (vector->list (vector-ref trans start)))) [nexts (vector->list (vector-ref trans start))])
(cond (cond
((null? nexts) #f) [(null? nexts) #f]
(else [else
(let ((next (car nexts))) (let ([next (car nexts)])
(and (= (vector-ref next 0) check) (and (= (vector-ref next 0) check)
(let ((next-check (vector-ref next 1))) (let ([next-check (vector-ref next 1)])
(or (>= next-check max-char-num) (or (>= next-check max-char-num)
(loop (add1 next-check) (cdr nexts)))))))))) (loop (add1 next-check) (cdr nexts))))))])))
(eprintf "Warning: lexer at ~a can accept the empty string.\n" stx))) (eprintf "warning: lexer at ~a can accept the empty string\n" stx)))
(with-syntax ((start-state-stx start) (with-syntax ([START-STATE-STX start]
(trans-table-stx trans) [TRANS-TABLE-STX trans]
(no-lookahead-stx no-look) [NO-LOOKAHEAD-STX no-look]
((name ...) name-lst) [(NAME ...) names]
((act ...) (map (lambda (a) [(ACT ...) (map (λ (a) (wrap-action a src-loc-style)) acts)]
(wrap-action a src-loc-style)) [(ACT-NAME ...) (vector->list action-names)]
act-lst)) [SPEC-ACT-STX (wrap-action spec-act src-loc-style)]
((act-name ...) (vector->list action-names)) [HAS-COMMENT-ACT?-STX (if (syntax-e spec-comment-act) #t #f)]
(spec-act-stx [SPEC-COMMENT-ACT-STX (wrap-action spec-comment-act src-loc-style)]
(wrap-action spec-act src-loc-style)) [EOF-ACT-STX (wrap-action eof-act src-loc-style)])
(has-comment-act?-stx
(if (syntax-e spec-comment-act) #t #f))
(spec-comment-act-stx
(wrap-action spec-comment-act src-loc-style))
(eof-act-stx (wrap-action eof-act src-loc-style)))
(syntax-property (syntax-property
(syntax/loc stx (syntax/loc stx (let ([NAME ACT] ...)
(let ([name act] ...) (let ([proc (lexer-body START-STATE-STX
(let ([proc TRANS-TABLE-STX
(lexer-body start-state-stx (vector ACT-NAME ...)
trans-table-stx NO-LOOKAHEAD-STX
(vector act-name ...) SPEC-ACT-STX
no-lookahead-stx HAS-COMMENT-ACT?-STX
spec-act-stx SPEC-COMMENT-ACT-STX
has-comment-act?-stx EOF-ACT-STX)])
spec-comment-act-stx
eof-act-stx)])
;; reverse eta to get named procedures: ;; reverse eta to get named procedures:
(lambda (port) (proc port))))) (λ (port) (proc port)))))
'disappeared-use 'disappeared-use disappeared-uses)))])))
disappeared-uses)))))))))
(define-syntax lexer (make-lexer-trans #f)) (define-syntax lexer (make-lexer-macro 'lexer #f))
(define-syntax lexer-src-pos (make-lexer-trans 'lexer-src-pos)) (define-syntax lexer-src-pos (make-lexer-macro 'lexer-src-pos 'lexer-src-pos))
(define-syntax lexer-srcloc (make-lexer-trans 'lexer-srcloc)) (define-syntax lexer-srcloc (make-lexer-macro 'lexer-srcloc 'lexer-srcloc))
(define-syntax (define-lex-abbrev stx) (define-syntax (define-lex-abbrev stx)
(syntax-case stx () (syntax-case stx ()
((_ name re) [(_ NAME RE) (identifier? #'NAME)
(identifier? (syntax name))
(syntax/loc stx (syntax/loc stx
(define-syntax name (define-syntax NAME
(make-lex-abbrev (lambda () (quote-syntax re)))))) (make-lex-abbrev (λ () (quote-syntax RE)))))]
(_ [_ (raise-syntax-error 'define-lex-abbrev "form should be (define-lex-abbrev name re)" stx)]))
(raise-syntax-error
#f
"form should be (define-lex-abbrev name re)"
stx))))
(define-syntax (define-lex-abbrevs stx) (define-syntax (define-lex-abbrevs stx)
(syntax-case stx () (syntax-case stx ()
((_ x ...) [(_ . XS)
(with-syntax (((abbrev ...) (with-syntax ([(ABBREV ...) (map
(map (λ (a)
(lambda (a)
(syntax-case a () (syntax-case a ()
((name re) [(NAME RE) (identifier? #'NAME)
(identifier? (syntax name)) (syntax/loc a (define-lex-abbrev NAME RE))]
(syntax/loc a (define-lex-abbrev name re))) [_ (raise-syntax-error
(_ (raise-syntax-error
#f #f
"form should be (define-lex-abbrevs (name re) ...)" "form should be (define-lex-abbrevs (name re) ...)"
stx stx
a)))) a)]))
(syntax->list (syntax (x ...)))))) (syntax->list #'XS))])
(syntax/loc stx (begin abbrev ...)))) (syntax/loc stx (begin ABBREV ...)))]
(_ [_ (raise-syntax-error #f "form should be (define-lex-abbrevs (name re) ...)" stx)]))
(raise-syntax-error
#f
"form should be (define-lex-abbrevs (name re) ...)"
stx))))
(define-syntax (define-lex-trans stx) (define-syntax (define-lex-trans stx)
(syntax-case stx () (syntax-case stx ()
((_ name-form body-form) [(_ name-form body-form)
(let-values (((name body) (let-values (((name body)
(normalize-definition (syntax (define-syntax name-form body-form)) #'lambda))) (normalize-definition #'(define-syntax name-form body-form) #'λ)))
#`(define-syntax #,name #`(define-syntax #,name
(let ((func #,body)) (let ((func #,body))
@ -215,53 +182,52 @@
(raise-syntax-error 'define-lex-trans "expected a procedure as the transformer, got ~e" func)) (raise-syntax-error 'define-lex-trans "expected a procedure as the transformer, got ~e" func))
(unless (procedure-arity-includes? func 1) (unless (procedure-arity-includes? func 1)
(raise-syntax-error 'define-lex-trans "expected a procedure that accepts 1 argument as the transformer, got ~e" func)) (raise-syntax-error 'define-lex-trans "expected a procedure that accepts 1 argument as the transformer, got ~e" func))
(make-lex-trans func))))) (make-lex-trans func))))]
(_ [_
(raise-syntax-error (raise-syntax-error
#f #f
"form should be (define-lex-trans name transformer)" "form should be (define-lex-trans name transformer)"
stx)))) stx)]))
(define (get-next-state-helper char min max table) (define (get-next-state-helper char min max table)
(if (>= min max) (cond
#f [(>= min max) #f]
(let* ((try (quotient (+ min max) 2)) [else
(el (vector-ref table try)) (define try (quotient (+ min max) 2))
(r1 (vector-ref el 0)) (define el (vector-ref table try))
(r2 (vector-ref el 1))) (define r1 (vector-ref el 0))
(define r2 (vector-ref el 1))
(cond (cond
((and (>= char r1) (<= char r2)) (vector-ref el 2)) [(and (>= char r1) (<= char r2)) (vector-ref el 2)]
((< char r1) (get-next-state-helper char min try table)) [(< char r1) (get-next-state-helper char min try table)]
(else (get-next-state-helper char (add1 try) max table)))))) [else (get-next-state-helper char (add1 try) max table)])]))
(define (get-next-state char table) (define (get-next-state char table)
(if table (and table (get-next-state-helper char 0 (vector-length table) table)))
(get-next-state-helper char 0 (vector-length table) table)
#f))
(define (lexer-body start-state trans-table actions no-lookahead special-action (define (lexer-body start-state trans-table actions no-lookahead special-action
has-special-comment-action? special-comment-action eof-action) has-special-comment-action? special-comment-action eof-action)
(letrec ((lexer (letrec ([lexer
(lambda (ip) (λ (ip)
(let ((first-pos (get-position ip)) (let ((first-pos (get-position ip))
(first-char (peek-char-or-special ip 0))) (first-char (peek-char-or-special ip 0)))
;(printf "(peek-char-or-special port 0) = ~e\n" first-char) ;(printf "(peek-char-or-special port 0) = ~e\n" first-char)
(cond (cond
((eof-object? first-char) [(eof-object? first-char)
(do-match ip first-pos eof-action (read-char-or-special ip))) (do-match ip first-pos eof-action (read-char-or-special ip))]
((special-comment? first-char) [(special-comment? first-char)
(read-char-or-special ip) (read-char-or-special ip)
(cond (cond
(has-special-comment-action? (has-special-comment-action?
(do-match ip first-pos special-comment-action #f)) (do-match ip first-pos special-comment-action #f))
(else (lexer ip)))) (else (lexer ip)))]
((not (char? first-char)) [(not (char? first-char))
(do-match ip first-pos special-action (read-char-or-special ip))) (do-match ip first-pos special-action (read-char-or-special ip))]
(else [else
(let lexer-loop ( (let lexer-loop (
;; current-state ;; current-state
(state start-state) (state start-state)
@ -278,26 +244,26 @@
(length-chars 1) (length-chars 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))
(let ((next-state (let ([next-state
(cond (cond
((not (char? char)) #f) [(not (char? char)) #f]
(else (get-next-state (char->integer char) [else (get-next-state (char->integer char)
(vector-ref trans-table state)))))) (vector-ref trans-table state))])])
(cond (cond
((not next-state) [(not next-state)
(check-match ip first-pos longest-match-length (check-match ip first-pos longest-match-length
length-chars longest-match-action)) length-chars longest-match-action)]
((vector-ref no-lookahead next-state) [(vector-ref no-lookahead next-state)
(let ((act (vector-ref actions next-state))) (let ((act (vector-ref actions next-state)))
(check-match ip (check-match ip
first-pos first-pos
(if act length-chars longest-match-length) (if act length-chars longest-match-length)
length-chars length-chars
(if act act longest-match-action)))) (if act act longest-match-action)))]
(else [else
(let* ((act (vector-ref actions next-state)) (let* ([act (vector-ref actions next-state)]
(next-length-bytes (+ (char-utf-8-length char) length-bytes)) [next-length-bytes (+ (char-utf-8-length char) length-bytes)]
(next-char (peek-char-or-special ip next-length-bytes))) [next-char (peek-char-or-special ip next-length-bytes)])
#;(printf "(peek-char-or-special port ~e) = ~e\n" #;(printf "(peek-char-or-special port ~e) = ~e\n"
next-length-bytes next-char) next-length-bytes next-char)
(lexer-loop next-state (lexer-loop next-state
@ -309,20 +275,16 @@
(add1 length-chars) (add1 length-chars)
(if act (if act
length-chars length-chars
longest-match-length))))))))))))) longest-match-length)))])))])))])
(lambda (ip) (λ (ip)
(unless (input-port? ip) (unless (input-port? ip)
(raise-argument-error (raise-argument-error 'lexer "input-port?" 0 ip))
'lexer
"input-port?"
0
ip))
(lexer ip)))) (lexer ip))))
(define (check-match lb first-pos longest-match-length length longest-match-action) (define (check-match lb first-pos longest-match-length length longest-match-action)
(unless longest-match-action (unless longest-match-action
(let* ((match (read-string length lb)) (let* ([match (read-string length lb)]
(end-pos (get-position lb))) [end-pos (get-position lb)])
(raise-read-error (raise-read-error
(format "lexer: No match found in input starting with: ~a" match) (format "lexer: No match found in input starting with: ~a" match)
(file-path) (file-path)
@ -330,31 +292,26 @@
(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)))))
(let ((match (read-string longest-match-length lb))) (let ([match (read-string longest-match-length lb)])
;(printf "(read-string ~e port) = ~e\n" longest-match-length match) ;(printf "(read-string ~e port) = ~e\n" longest-match-length match)
(do-match lb first-pos longest-match-action match))) (do-match lb first-pos longest-match-action match)))
(define file-path (make-parameter #f)) (define file-path (make-parameter #f))
(define lexer-file-path file-path) (define lexer-file-path file-path)
(define (do-match ip first-pos action value) (define (do-match ip first-pos action value)
#;(printf "(action ~a ~a ~a ~a)\n" #;(printf "(action ~a ~a ~a ~a)\n"
(position-offset first-pos) (position-offset (get-position ip)) value ip) (position-offset first-pos) (position-offset (get-position ip)) value ip)
(action first-pos (get-position ip) value ip)) (action first-pos (get-position ip) value ip))
(define (get-position ip) (define (get-position ip)
(let-values (((line col off) (port-next-location ip))) (define-values (line col off) (port-next-location ip))
(make-position off line col))) (make-position off line col))
(define-syntax (create-unicode-abbrevs stx) (define-syntax (create-unicode-abbrevs stx)
(syntax-case stx () (syntax-case stx ()
((_ ctxt) [(_ CTXT)
(with-syntax (((ranges ...) (map (lambda (range) (with-syntax ([(RANGES ...) (for/list ([range (in-list (list (force alphabetic-ranges)
`(union ,@(map (lambda (x)
`(char-range ,(integer->char (car x))
,(integer->char (cdr x))))
range)))
(list (force alphabetic-ranges)
(force lower-case-ranges) (force lower-case-ranges)
(force upper-case-ranges) (force upper-case-ranges)
(force title-case-ranges) (force title-case-ranges)
@ -364,10 +321,12 @@
(force graphic-ranges) (force graphic-ranges)
(force whitespace-ranges) (force whitespace-ranges)
(force blank-ranges) (force blank-ranges)
(force iso-control-ranges)))) (force iso-control-ranges)))])
((names ...) (map (lambda (sym) `(union ,@(map (λ (x)
(datum->syntax-object (syntax ctxt) sym #f)) `(char-range ,(integer->char (car x))
'(alphabetic ,(integer->char (cdr x))))
range)))]
[(NAMES ...) (for/list ([sym (in-list '(alphabetic
lower-case lower-case
upper-case upper-case
title-case title-case
@ -377,36 +336,34 @@
graphic graphic
whitespace whitespace
blank blank
iso-control)))) iso-control))])
(syntax (define-lex-abbrevs (names ranges) ...)))))) (datum->syntax #'CTXT sym #f))])
#'(define-lex-abbrevs (NAMES RANGES) ...))]))
(define-lex-abbrev any-char (char-complement (union))) (define-lex-abbrev any-char (char-complement (union)))
(define-lex-abbrev any-string (intersection)) (define-lex-abbrev any-string (intersection))
(define-lex-abbrev nothing (union)) (define-lex-abbrev nothing (union))
(create-unicode-abbrevs #'here) (create-unicode-abbrevs #'here)
(define-lex-trans (char-set stx) (define-lex-trans (char-set stx)
(syntax-case stx () (syntax-case stx ()
((_ str) [(_ STR)
(string? (syntax-e (syntax str))) (string? (syntax-e #'STR))
(with-syntax (((char ...) (string->list (syntax-e (syntax str))))) (with-syntax ([(CHAR ...) (string->list (syntax-e #'STR))])
(syntax (union char ...)))))) #'(union CHAR ...))]))
(define-syntax provide-lex-keyword (define-syntax provide-lex-keyword
(syntax-rules () (syntax-rules ()
[(_ id ...) [(_ ID ...)
(begin (begin
(define-syntax-parameter id (define-syntax-parameter ID
(make-set!-transformer (make-set!-transformer
(lambda (stx) (λ (stx)
(raise-syntax-error (raise-syntax-error
#f 'provide-lex-keyword
(format "use of a lexer keyword (~a) is not in an appropriate lexer action" (format "use of a lexer keyword (~a) is not in an appropriate lexer action" 'ID)
'id)
stx)))) stx))))
... ...
(provide id ...))])) (provide ID ...))]))
(provide-lex-keyword start-pos end-pos lexeme lexeme-srcloc input-port return-without-pos return-without-srcloc)
) (provide-lex-keyword start-pos end-pos lexeme lexeme-srcloc input-port return-without-pos return-without-srcloc)

8
br-parser-tools/br-parser-tools-lib/br-parser-tools/private-lex/actions.rkt
Unescape Escape View File

@ -1,4 +1,4 @@
#lang scheme/base #lang racket/base
(provide (all-defined-out)) (provide (all-defined-out))
(require syntax/stx) (require syntax/stx)
@ -10,7 +10,7 @@
((null? rules) none) ((null? rules) none)
(else (else
(syntax-case (car rules) () (syntax-case (car rules) ()
(((special) act) [((special) ACT)
(and (identifier? #'special) (module-or-top-identifier=? (syntax special) which-special)) (and (identifier? #'special) (module-or-top-identifier=? (syntax special) which-special))
(syntax act)) #'ACT]
(_ (get-special-action (cdr rules) which-special none)))))) [_ (get-special-action (cdr rules) which-special none)]))))

Write Preview
Loading…
Cancel
Save