Commit 7cd27061 authored by Robert Strandh's avatar Robert Strandh

A step on the way to factoring out the incremental lexer.

parent 9ac1172d
......@@ -41,6 +41,21 @@
(with-slots (start-mark size) tree
(+ (offset start-mark) size)))
(defclass lexer ()
((buffer :initarg :buffer :reader buffer)))
(defgeneric nb-lexemes (lexer))
(defgeneric lexeme (lexer pos))
(defclass incremental-lexer (lexer)
((lexemes :initform (make-instance 'standard-flexichain) :reader lexemes)))
(defmethod nb-lexemes ((lexer incremental-lexer))
(nb-elements (lexemes lexer)))
(defmethod lexeme ((lexer incremental-lexer) pos)
(element* (lexemes lexer) pos))
;;; grammar classes
......@@ -132,30 +147,23 @@
(defclass word (html-element) ())
(defclass delimiter (html-element) ())
(defun next-token (scan)
(let ((start-mark (clone-mark scan)))
(flet ((fo () (forward-object scan)))
(macrolet ((make-entry (type)
`(return-from next-token
(make-instance ,type :start-mark start-mark
:size (- (offset scan) (offset start-mark))))))
(loop with object = (object-after scan)
until (end-of-buffer-p scan)
do (case object
(#\< (fo) (make-entry 'tag-start))
(#\> (fo) (make-entry 'tag-end))
(#\/ (fo) (make-entry 'slash))
(t (cond ((alphanumericp object)
(loop until (end-of-buffer-p scan)
while (alphanumericp (object-after scan))
do (fo))
(make-entry 'word))
(fo) (make-entry 'delimiter))))))))))
(defun next-lexeme (scan)
(flet ((fo () (forward-object scan)))
(let ((object (object-after scan)))
(case object
(#\< (fo) (make-instance 'tag-start))
(#\> (fo) (make-instance 'tag-end))
(#\/ (fo) (make-instance 'slash))
(t (cond ((alphanumericp object)
(loop until (end-of-buffer-p scan)
while (alphanumericp (object-after scan))
do (fo))
(make-instance 'word))
(fo) (make-instance 'delimiter))))))))
(define-syntax html-syntax ("HTML" (basic-syntax))
((tokens :initform (make-instance 'standard-flexichain))
(guess-pos :initform 1)
((lexemes :initform (make-instance 'standard-flexichain))
(valid-parse :initform 1)
......@@ -264,11 +272,11 @@
(defmethod initialize-instance :after ((syntax html-syntax) &rest args)
(declare (ignore args))
(with-slots (parser tokens buffer) syntax
(with-slots (parser lexemes buffer) syntax
(setf parser (make-instance 'parser
:grammar *html-grammar*
:target 'html))
(insert* tokens 0 (make-instance 'start-element
(insert* lexemes 0 (make-instance 'start-element
:start-mark (make-instance 'standard-left-sticky-mark
:buffer buffer
:offset 0)
......@@ -280,52 +288,65 @@
;;; update syntax
(defmethod update-syntax-for-display (buffer (syntax html-syntax) top bot)
(with-slots (parser tokens valid-parse) syntax
(loop until (= valid-parse (nb-elements tokens))
while (mark<= (end-offset (element* tokens valid-parse)) bot)
do (let ((current-token (element* tokens (1- valid-parse)))
(next-token (element* tokens valid-parse)))
(setf (slot-value next-token 'state)
(advance-parse parser (list next-token) (slot-value current-token 'state))))
(with-slots (parser lexemes valid-parse) syntax
(loop until (= valid-parse (nb-elements lexemes))
while (mark<= (end-offset (element* lexemes valid-parse)) bot)
do (let ((current-token (element* lexemes (1- valid-parse)))
(next-lexeme (element* lexemes valid-parse)))
(setf (slot-value next-lexeme 'state)
(advance-parse parser (list next-lexeme) (slot-value current-token 'state))))
(incf valid-parse))))
(defun delete-invalid-lexemes (lexemes from to)
"delete all lexemes between FROM and TO and return the first invalid
position in LEXEMES"
(let ((start 1)
(end (nb-elements lexemes)))
;; use binary search to find the first lexeme to delete
(loop while (< start end)
do (let ((middle (floor (+ start end) 2)))
(if (mark< (end-offset (element* lexemes middle)) from)
(setf start (1+ middle))
(setf end middle))))
;; delete lexemes
(loop until (or (= start (nb-elements lexemes))
(mark> (start-mark (element* lexemes start)) to))
do (delete* lexemes start))
(defun inter-lexeme-object-p (lexemes object)
(declare (ignore lexemes))
(whitespacep object))
(defun skip-inter-lexeme-objects (lexemes scan)
(loop until (end-of-buffer-p scan)
while (inter-lexeme-object-p lexemes (object-after scan))
do (forward-object scan)))
(defun update-lex (lexemes start-pos end)
(let ((scan (make-instance 'standard-left-sticky-mark
:buffer (buffer end) ; FIXME, eventually use the buffer of the lexer
:offset (end-offset (element* lexemes (1- start-pos))))))
(loop do (skip-inter-lexeme-objects lexemes scan)
until (if (end-of-buffer-p end)
(end-of-buffer-p scan)
(mark> scan end))
do (let* ((start-mark (clone-mark scan))
(lexeme (next-lexeme scan))
(size (- (offset scan) (offset start-mark))))
(setf (slot-value lexeme 'start-mark) start-mark
(slot-value lexeme 'size) size)
(insert* lexemes start-pos lexeme))
(incf start-pos))))
(defmethod update-syntax (buffer (syntax html-syntax))
(let ((low-mark (low-mark buffer))
(high-mark (high-mark buffer))
(with-slots (tokens guess-pos valid-parse) syntax
(when (mark<= low-mark high-mark)
;; go back to a position before low-mark
(loop until (or (= guess-pos 1)
(mark< (end-offset (element* tokens (1- guess-pos))) low-mark))
do (decf guess-pos))
;; go forward to the last position before low-mark
(loop with nb-elements = (nb-elements tokens)
until (or (= guess-pos nb-elements)
(mark>= (end-offset (element* tokens guess-pos)) low-mark))
do (incf guess-pos))
;; mark valid parse
(setf valid-parse guess-pos)
;; delete entries that must be reparsed
(loop until (or (= guess-pos (nb-elements tokens))
(mark> (start-mark (element* tokens guess-pos)) high-mark))
do (delete* tokens guess-pos))
(setf scan (make-instance 'standard-left-sticky-mark
:buffer buffer
:offset (if (zerop guess-pos)
(end-offset (element* tokens (1- guess-pos))))))
;; scan
(loop with start-mark = nil
do (loop until (end-of-buffer-p scan)
while (whitespacep (object-after scan))
do (forward-object scan))
until (if (end-of-buffer-p high-mark)
(end-of-buffer-p scan)
(mark> scan high-mark))
do (setf start-mark (clone-mark scan))
(insert* tokens guess-pos (next-token scan))
(incf guess-pos))))))
(with-slots (lexemes valid-parse) syntax
(let* ((low-mark (low-mark buffer))
(high-mark (high-mark buffer))
(first-invalid-position (delete-invalid-lexemes lexemes low-mark high-mark)))
(setf valid-parse first-invalid-position)
(update-lex lexemes first-invalid-position high-mark))))
......@@ -451,35 +472,35 @@
(setf *cursor-positions* (make-array (1+ (number-of-lines-in-region top bot)))
*current-line* 0
(aref *cursor-positions* 0) (stream-cursor-position pane))
(with-slots (tokens) syntax
(let ((average-token-size (max (float (/ (size (buffer pane)) (nb-elements tokens)))
(with-slots (lexemes) syntax
(let ((average-token-size (max (float (/ (size (buffer pane)) (nb-elements lexemes)))
;; find the last token before bot
(let ((end-token-index (max (floor (/ (offset bot) average-token-size)) 1)))
;; go back to a token before bot
(loop until (mark<= (end-offset (element* tokens (1- end-token-index))) bot)
(loop until (mark<= (end-offset (element* lexemes (1- end-token-index))) bot)
do (decf end-token-index))
;; go forward to the last token before bot
(loop until (or (= end-token-index (nb-elements tokens))
(mark> (start-offset (element* tokens end-token-index)) bot))
(loop until (or (= end-token-index (nb-elements lexemes))
(mark> (start-offset (element* lexemes end-token-index)) bot))
do (incf end-token-index))
(let ((start-token-index end-token-index))
;; go back to the first token after top, or until the previous token
;; contains a valid parser state
(loop until (or (mark<= (end-offset (element* tokens (1- start-token-index))) top)
(loop until (or (mark<= (end-offset (element* lexemes (1- start-token-index))) top)
(not (parse-state-empty-p
(slot-value (element* tokens (1- start-token-index)) 'state))))
(slot-value (element* lexemes (1- start-token-index)) 'state))))
do (decf start-token-index))
(let ((*white-space-start* (offset top)))
;; display the parse tree if any
(unless (parse-state-empty-p (slot-value (element* tokens (1- start-token-index)) 'state))
(display-parse-state (slot-value (element* tokens (1- start-token-index)) 'state)
(unless (parse-state-empty-p (slot-value (element* lexemes (1- start-token-index)) 'state))
(display-parse-state (slot-value (element* lexemes (1- start-token-index)) 'state)
;; display the tokens
;; display the lexemes
(with-drawing-options (pane :ink +red+)
(loop while (< start-token-index end-token-index)
do (let ((token (element* tokens start-token-index)))
do (let ((token (element* lexemes start-token-index)))
(display-parse-tree token syntax pane))
(incf start-token-index))))))))
(let* ((cursor-line (number-of-lines-in-region top (point pane)))
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment