1 ;;; wisent.el --- GNU Bison for Emacs - Runtime
3 ;; Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 David Ponce
5 ;; Author: David Ponce <david@dponce.com>
6 ;; Maintainer: David Ponce <david@dponce.com>
7 ;; Created: 30 January 2002
9 ;; X-RCS: $Id: wisent.el,v 1.1 2007-11-26 15:12:35 michaels Exp $
11 ;; This file is not part of GNU Emacs.
13 ;; This program is free software; you can redistribute it and/or
14 ;; modify it under the terms of the GNU General Public License as
15 ;; published by the Free Software Foundation; either version 2, or (at
16 ;; your option) any later version.
18 ;; This program is distributed in the hope that it will be useful, but
19 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
20 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
21 ;; General Public License for more details.
23 ;; You should have received a copy of the GNU General Public License
24 ;; along with this program; see the file COPYING. If not, write to
25 ;; the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
26 ;; Boston, MA 02110-1301, USA.
30 ;; Parser engine and runtime of Wisent.
32 ;; Wisent (the European Bison ;-) is an Elisp implementation of the
33 ;; GNU Compiler Compiler Bison. The Elisp code is a port of the C
34 ;; code of GNU Bison 1.28 & 1.31.
36 ;; For more details on the basic concepts for understanding Wisent,
37 ;; read the Bison manual ;)
39 ;; For more details on Wisent itself read the Wisent manual.
49 /\\_.-^^^-._/\\ The GNU
51 ( `o ` (European ;-) Bison
64 (if (fboundp 'char-valid-p)
65 (eval-and-compile (defalias 'wisent-char-p 'char-valid-p))
66 (eval-and-compile (defalias 'wisent-char-p 'char-or-char-int-p)))
68 ;;; Printed representation of terminals and nonterminals
69 (defconst wisent-escape-sequence-strings
72 (?\b . "'\\b'") ; backspace, BS, C-h
73 (?\t . "'\\t'") ; tab, TAB, C-i
74 (?\n . "'\\n'") ; newline, C-j
75 (?\v . "'\\v'") ; vertical tab, C-k
76 (?\f . "'\\f'") ; formfeed character, C-l
77 (?\r . "'\\r'") ; carriage return, RET, C-m
78 (?\e . "'\\e'") ; escape character, ESC, C-[
79 (?\\ . "'\\'") ; backslash character, \
80 (?\d . "'\\d'") ; delete character, DEL
82 "Printed representation of usual escape sequences.")
84 (defsubst wisent-item-to-string (item)
85 "Return a printed representation of ITEM.
86 ITEM can be a nonterminal or terminal symbol, or a character literal."
87 (if (wisent-char-p item)
88 (or (cdr (assq item wisent-escape-sequence-strings))
92 (defsubst wisent-token-to-string (token)
93 "Return a printed representation of lexical token TOKEN."
94 (format "%s%s(%S)" (wisent-item-to-string (car token))
95 (if (nth 2 token) (format "@%s" (nth 2 token)) "")
99 (defconst wisent-eoi-term '$EOI
100 "End Of Input token.")
102 (defconst wisent-error-term 'error
103 "Error recovery token.")
105 (defconst wisent-accept-tag 'accept
106 "Accept result after input successfully parsed.")
108 (defconst wisent-error-tag 'error
109 "Process a syntax error.")
111 ;;; Special functions
112 (defun wisent-automaton-p (obj)
113 "Return non-nil if OBJ is a LALR automaton.
114 If OBJ is a symbol check its value."
115 (and obj (symbolp obj) (boundp obj)
116 (setq obj (symbol-value obj)))
117 (and (vectorp obj) (= 4 (length obj))
118 (vectorp (aref obj 0)) (vectorp (aref obj 1))
119 (= (length (aref obj 0)) (length (aref obj 1)))
120 (listp (aref obj 2)) (vectorp (aref obj 3))))
122 (defsubst wisent-region (&rest positions)
123 "Return the start/end positions of the region including POSITIONS.
124 Each element of POSITIONS is a pair (START-POS . END-POS) or nil. The
125 returned value is the pair (MIN-START-POS . MAX-END-POS) or nil if no
126 POSITIONS are available."
127 (let ((pl (delq nil positions)))
129 (cons (apply #'min (mapcar #'car pl))
130 (apply #'max (mapcar #'cdr pl))))))
134 (defvar wisent-parse-verbose-flag nil
135 "*Non-nil means to issue more messages while parsing.")
138 (defun wisent-parse-toggle-verbose-flag ()
139 "Toggle whether to issue more messages while parsing."
141 (setq wisent-parse-verbose-flag (not wisent-parse-verbose-flag))
142 (when (interactive-p)
143 (message "More messages while parsing %sabled"
144 (if wisent-parse-verbose-flag "en" "dis"))))
146 (defsubst wisent-message (string &rest args)
147 "Print a one-line message if `wisent-parse-verbose-flag' is set.
148 Pass STRING and ARGS arguments to `message'."
149 (and wisent-parse-verbose-flag
150 (apply 'message string args)))
152 ;;;; --------------------
153 ;;;; The LR parser engine
154 ;;;; --------------------
156 (defcustom wisent-parse-max-stack-size 500
157 "The parser stack size."
161 (defcustom wisent-parse-max-recover 3
162 "Number of tokens to shift before turning off error status."
166 (defvar wisent-discarding-token-functions nil
167 "List of functions to be called when discarding a lexical token.
168 These functions receive the lexical token discarded.
169 When the parser encounters unexpected tokens, it can discards them,
170 based on what directed by error recovery rules. Either when the
171 parser reads tokens until one is found that can be shifted, or when an
172 semantic action calls the function `wisent-skip-token' or
174 For language specific hooks, make sure you define this as a local
177 (defvar wisent-pre-parse-hook nil
178 "Normal hook run just before entering the LR parser engine.")
180 (defvar wisent-post-parse-hook nil
181 "Normal hook run just after the LR parser engine terminated.")
183 (defvar wisent-loop nil
184 "The current parser action.
185 Stop parsing when set to nil.
186 This variable only has meaning in the scope of `wisent-parse'.")
188 (defvar wisent-nerrs nil
189 "The number of parse errors encountered so far.")
191 (defvar wisent-lookahead nil
192 "The lookahead lexical token.
193 This value is non-nil if the parser terminated because of an
194 unrecoverable error.")
196 ;; Variables and macros that are useful in semantic actions.
197 (defvar wisent-parse-lexer-function nil
198 "The user supplied lexer function.
199 This function don't have arguments.
200 This variable only has meaning in the scope of `wisent-parse'.")
202 (defvar wisent-parse-error-function nil
203 "The user supplied error function.
204 This function must accept one argument, a message string.
205 This variable only has meaning in the scope of `wisent-parse'.")
207 (defvar wisent-input nil
208 "The last token read.
209 This variable only has meaning in the scope of `wisent-parse'.")
211 (defvar wisent-recovering nil
212 "Non-nil means that the parser is recovering.
213 This variable only has meaning in the scope of `wisent-parse'.")
215 ;; Variables that only have meaning in the scope of a semantic action.
216 ;; These global definitions avoid byte-compiler warnings.
221 (defmacro wisent-lexer ()
222 "Obtain the next terminal in input."
223 '(funcall wisent-parse-lexer-function))
225 (defmacro wisent-error (msg)
226 "Call the user supplied error reporting function with message MSG."
227 `(funcall wisent-parse-error-function ,msg))
229 (defmacro wisent-errok ()
230 "Resume generating error messages immediately for subsequent syntax errors.
231 This is useful primarily in error recovery semantic actions."
232 '(setq wisent-recovering nil))
234 (defmacro wisent-clearin ()
235 "Discard the current lookahead token.
236 This will cause a new lexical token to be read.
237 This is useful primarily in error recovery semantic actions."
238 '(setq wisent-input nil))
240 (defmacro wisent-abort ()
241 "Abort parsing and save the lookahead token.
242 This is useful primarily in error recovery semantic actions."
243 '(setq wisent-lookahead wisent-input
246 (defmacro wisent-set-region (start end)
247 "Change the region of text matched by the current nonterminal.
248 START and END are respectively the beginning and end positions of the
249 region. If START or END values are not a valid positions the region
251 `(setq $region (and (number-or-marker-p ,start)
252 (number-or-marker-p ,end)
253 (cons ,start ,end))))
255 (defun wisent-skip-token ()
256 "Skip the lookahead token in order to resume parsing.
258 Must be used in error recovery semantic actions."
259 (if (eq (car wisent-input) wisent-eoi-term)
260 ;; Does nothing at EOI to avoid infinite recovery loop.
262 (wisent-message "%s: skip %s" $action
263 (wisent-token-to-string wisent-input))
265 'wisent-discarding-token-functions wisent-input)
269 (defun wisent-skip-block (&optional bounds)
270 "Safely skip a parenthesized block in order to resume parsing.
272 Must be used in error recovery semantic actions.
273 Optional argument BOUNDS is a pair (START . END) which indicates where
274 the parenthesized block starts. Typically the value of a `$regionN'
275 variable, where `N' is the the Nth element of the current rule
276 components that match the block beginning. It defaults to the value
277 of the `$region' variable."
278 (let ((start (car (or bounds $region)))
280 (if (not (number-or-marker-p start))
281 ;; No nonterminal region available, skip the lookahead token.
283 ;; Try to skip a block.
284 (if (not (setq end (save-excursion
286 (and (looking-at "\\s(")
288 (1- (scan-lists (point) 1 0))
290 ;; Not actually a block, skip the lookahead token.
292 ;; OK to safely skip the block, so read input until a matching
293 ;; close paren or EOI is encountered.
294 (setq input wisent-input)
295 (while (and (not (eq (car input) wisent-eoi-term))
296 (< (nth 2 input) end))
298 'wisent-discarding-token-functions input)
299 (setq input (wisent-lexer)))
300 (wisent-message "%s: in enclosing block, skip from %s to %s"
302 (wisent-token-to-string wisent-input)
303 (wisent-token-to-string input))
304 (if (eq (car wisent-input) wisent-eoi-term)
305 ;; Does nothing at EOI to avoid infinite recovery loop.
309 ;; Set end of $region to end of block.
310 (wisent-set-region (car $region) (1+ end))
313 ;;; Core parser engine
314 (defsubst wisent-production-bounds (stack i j)
315 "Determine the start and end locations of a production value.
316 Return a pair (START . END), where START is the first available start
317 location, and END the last available end location, in components
318 values of the rule currently reduced.
319 Return nil when no component location is available.
320 STACK is the parser stack.
321 I and J are the indices in STACK of respectively the value of the
322 first and last components of the current rule.
323 This function is for internal use by semantic actions' generated
325 (let ((f (cadr (aref stack i)))
326 (l (cddr (aref stack j))))
329 ((not f) (setq f (cadr (aref stack (setq i (+ i 2))))))
330 ((not l) (setq l (cddr (aref stack (setq j (- j 2))))))
332 (and f l (cons f l))))
334 (defmacro wisent-parse-action (i al)
335 "Return the next parser action.
336 I is a token item number and AL is the list of (item . action)
337 available at current state. The first element of AL contains the
338 default action for this state."
339 `(cdr (or (assq ,i ,al) (car ,al))))
341 (defsubst wisent-parse-start (start starts)
342 "Return the first lexical token to shift for START symbol.
343 STARTS is the table of allowed start symbols or nil if the LALR
344 automaton has only one entry point."
346 ;; Only one entry point, return the first lexical token
347 ;; available in input.
349 ;; Multiple start symbols defined, return the internal lexical
350 ;; token associated to START. By default START is the first
351 ;; nonterminal defined in STARTS.
352 (let ((token (cdr (if start (assq start starts) (car starts)))))
354 (list token (symbol-name token))
355 (error "Invalid start symbol %s" start)))))
357 (defun wisent-parse (automaton lexer &optional error start)
358 "Parse input using the automaton specified in AUTOMATON.
360 - AUTOMATON is an LALR(1) automaton generated by
361 `wisent-compile-grammar'.
363 - LEXER is a function with no argument called by the parser to obtain
364 the next terminal (token) in input.
366 - ERROR is an optional reporting function called when a parse error
367 occurs. It receives a message string to report. It defaults to the
368 function `wisent-message'.
370 - START specify the start symbol (nonterminal) used by the parser as
371 its goal. It defaults to the start symbol defined in the grammar
372 \(see also `wisent-compile-grammar')."
373 (run-hooks 'wisent-pre-parse-hook)
374 (let* ((actions (aref automaton 0))
375 (gotos (aref automaton 1))
376 (starts (aref automaton 2))
377 (stack (make-vector wisent-parse-max-stack-size nil))
380 (wisent-parse-error-function (or error 'wisent-message))
381 (wisent-parse-lexer-function lexer)
382 (wisent-recovering nil)
383 (wisent-input (wisent-parse-start start starts))
384 state tokid choices choice)
385 (setq wisent-nerrs 0 ;; Reset parse error counter
386 wisent-lookahead nil) ;; and lookahead token
387 (aset stack 0 0) ;; Initial state
389 (setq state (aref stack sp)
390 tokid (car wisent-input)
391 wisent-loop (wisent-parse-action tokid (aref actions state)))
394 ;; Input successfully parsed
395 ;; -------------------------
396 ((eq wisent-loop wisent-accept-tag)
397 (setq wisent-loop nil))
399 ;; Syntax error in input
400 ;; ---------------------
401 ((eq wisent-loop wisent-error-tag)
402 ;; Report this error if not already recovering from an error.
403 (setq choices (aref actions state))
404 (or wisent-recovering
406 (format "Syntax error, unexpected %s, expecting %s"
407 (wisent-token-to-string wisent-input)
408 (mapconcat 'wisent-item-to-string
409 (delq wisent-error-term
410 (mapcar 'car (cdr choices)))
412 ;; Increment the error counter
413 (setq wisent-nerrs (1+ wisent-nerrs))
414 ;; If just tried and failed to reuse lookahead token after an
415 ;; error, discard it.
416 (if (eq wisent-recovering wisent-parse-max-recover)
417 (if (eq tokid wisent-eoi-term)
418 (wisent-abort) ;; Terminate if at end of input.
419 (wisent-message "Error recovery: skip %s"
420 (wisent-token-to-string wisent-input))
422 'wisent-discarding-token-functions wisent-input)
423 (setq wisent-input (wisent-lexer)))
425 ;; Else will try to reuse lookahead token after shifting the
428 ;; Each real token shifted decrements this.
429 (setq wisent-recovering wisent-parse-max-recover)
430 ;; Pop the value/state stack to see if an action associated
431 ;; to special terminal symbol 'error exists.
432 (while (and (>= sp 0)
433 (not (and (setq state (aref stack sp)
434 choices (aref actions state)
435 choice (assq wisent-error-term choices))
436 (natnump (cdr choice)))))
440 ;; No 'error terminal was found. Just terminate.
442 ;; Try to recover and continue parsing.
443 ;; Shift the error terminal.
444 (setq state (cdr choice) ; new state
446 (aset stack (1- sp) nil) ; push value
447 (aset stack sp state) ; push new state
448 ;; Adjust input to error recovery state. Unless 'error
449 ;; triggers a reduction, eat the input stream until an
450 ;; expected terminal symbol is found, or EOI is reached.
451 (if (cdr (setq choices (aref actions state)))
452 (while (not (or (eq (car wisent-input) wisent-eoi-term)
453 (assq (car wisent-input) choices)))
454 (wisent-message "Error recovery: skip %s"
455 (wisent-token-to-string wisent-input))
457 'wisent-discarding-token-functions wisent-input)
458 (setq wisent-input (wisent-lexer)))))))
460 ;; Shift current token on top of the stack
461 ;; ---------------------------------------
462 ((natnump wisent-loop)
463 ;; Count tokens shifted since error; after
464 ;; `wisent-parse-max-recover', turn off error status.
465 (setq wisent-recovering (and (natnump wisent-recovering)
466 (> wisent-recovering 1)
467 (1- wisent-recovering)))
469 (aset stack (1- sp) (cdr wisent-input))
470 (aset stack sp wisent-loop)
471 (setq wisent-input (wisent-lexer)))
473 ;; Reduce by rule (call semantic action)
474 ;; -------------------------------------
476 (setq sp (funcall wisent-loop stack sp gotos))
477 (or wisent-input (setq wisent-input (wisent-lexer))))))
478 (run-hooks 'wisent-post-parse-hook)
479 (car (aref stack 1))))
481 ;;; wisent.el ends here