Commit | Line | Data |
---|---|---|
30764597 | 1 | ;;; syntax.el --- helper functions to find syntactic context |
88a05faf | 2 | |
3731a850 | 3 | ;; Copyright (C) 2000, 2001, 2002, 2003, 2004, |
8b72699e | 4 | ;; 2005, 2006, 2007, 2008 Free Software Foundation, Inc. |
88a05faf | 5 | |
30764597 PJ |
6 | ;; Maintainer: FSF |
7 | ;; Keywords: internal | |
8 | ||
88a05faf SM |
9 | ;; This file is part of GNU Emacs. |
10 | ||
d6cba7ae | 11 | ;; GNU Emacs is free software: you can redistribute it and/or modify |
88a05faf | 12 | ;; it under the terms of the GNU General Public License as published by |
d6cba7ae GM |
13 | ;; the Free Software Foundation, either version 3 of the License, or |
14 | ;; (at your option) any later version. | |
88a05faf SM |
15 | |
16 | ;; GNU Emacs is distributed in the hope that it will be useful, | |
17 | ;; but WITHOUT ANY WARRANTY; without even the implied warranty of | |
18 | ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
19 | ;; GNU General Public License for more details. | |
20 | ||
21 | ;; You should have received a copy of the GNU General Public License | |
d6cba7ae | 22 | ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. |
88a05faf SM |
23 | |
24 | ;;; Commentary: | |
25 | ||
26 | ;; The main exported function is `syntax-ppss'. You might also need | |
9c5b2653 | 27 | ;; to call `syntax-ppss-flush-cache' or to add it to |
ad8a840d | 28 | ;; before-change-functions'(although this is automatically done by |
88a05faf | 29 | ;; syntax-ppss when needed, but that might fail if syntax-ppss is |
ad8a840d | 30 | ;; called in a context where before-change-functions is temporarily |
88a05faf SM |
31 | ;; let-bound to nil). |
32 | ||
33 | ;;; Todo: | |
34 | ||
35 | ;; - do something about the case where the syntax-table is changed. | |
36 | ;; This typically happens with tex-mode and its `$' operator. | |
37 | ;; - move font-lock-syntactic-keywords in here. Then again, maybe not. | |
38 | ;; - new functions `syntax-state', ... to replace uses of parse-partial-state | |
39 | ;; with something higher-level (similar to syntax-ppss-context). | |
40 | ;; - interaction with mmm-mode. | |
88a05faf SM |
41 | |
42 | ;;; Code: | |
43 | ||
44 | ;; Note: PPSS stands for `parse-partial-sexp state' | |
45 | ||
46 | (eval-when-compile (require 'cl)) | |
47 | ||
746dca00 JB |
48 | (defvar font-lock-beginning-of-syntax-function) |
49 | ||
88a05faf SM |
50 | (defsubst syntax-ppss-depth (ppss) |
51 | (nth 0 ppss)) | |
52 | ||
0e6c966c | 53 | (defun syntax-ppss-toplevel-pos (ppss) |
be357779 SM |
54 | "Get the latest syntactically outermost position found in a syntactic scan. |
55 | PPSS is a scan state, as returned by `partial-parse-sexp' or `syntax-ppss'. | |
56 | An \"outermost position\" means one that it is outside of any syntactic entity: | |
57 | outside of any parentheses, comments, or strings encountered in the scan. | |
58 | If no such position is recorded in PPSS (because the end of the scan was | |
59 | itself at the outermost level), return nil." | |
6bddffd0 SM |
60 | ;; BEWARE! We rely on the undocumented 9th field. The 9th field currently |
61 | ;; contains the list of positions of the enclosing open-parens. | |
62 | ;; I.e. those positions are outside of any string/comment and the first of | |
63 | ;; those is outside of any paren (i.e. corresponds to a nil ppss). | |
64 | ;; If this list is empty but we are in a string or comment, then the 8th | |
65 | ;; field contains a similar "toplevel" position. | |
0e6c966c | 66 | (or (car (nth 9 ppss)) |
6bddffd0 | 67 | (nth 8 ppss))) |
0e6c966c | 68 | |
88a05faf SM |
69 | (defsubst syntax-ppss-context (ppss) |
70 | (cond | |
71 | ((nth 3 ppss) 'string) | |
72 | ((nth 4 ppss) 'comment) | |
73 | (t nil))) | |
74 | ||
75 | (defvar syntax-ppss-max-span 20000 | |
76 | "Threshold below which cache info is deemed unnecessary. | |
77 | We try to make sure that cache entries are at least this far apart | |
78 | from each other, to avoid keeping too much useless info.") | |
79 | ||
80 | (defvar syntax-begin-function nil | |
81 | "Function to move back outside of any comment/string/paren. | |
82 | This function should move the cursor back to some syntactically safe | |
83 | point (where the PPSS is equivalent to nil).") | |
84 | ||
85 | (defvar syntax-ppss-cache nil | |
86 | "List of (POS . PPSS) pairs, in decreasing POS order.") | |
87 | (make-variable-buffer-local 'syntax-ppss-cache) | |
88 | (defvar syntax-ppss-last nil | |
89 | "Cache of (LAST-POS . LAST-PPSS).") | |
90 | (make-variable-buffer-local 'syntax-ppss-last) | |
91 | ||
9c5b2653 SM |
92 | (defalias 'syntax-ppss-after-change-function 'syntax-ppss-flush-cache) |
93 | (defun syntax-ppss-flush-cache (beg &rest ignored) | |
94 | "Flush the cache of `syntax-ppss' starting at position BEG." | |
88a05faf SM |
95 | ;; Flush invalid cache entries. |
96 | (while (and syntax-ppss-cache (> (caar syntax-ppss-cache) beg)) | |
97 | (setq syntax-ppss-cache (cdr syntax-ppss-cache))) | |
98 | ;; Throw away `last' value if made invalid. | |
99 | (when (< beg (or (car syntax-ppss-last) 0)) | |
889c5dad SM |
100 | ;; If syntax-begin-function jumped to BEG, then the old state at BEG can |
101 | ;; depend on the text after BEG (which is presumably changed). So if | |
102 | ;; BEG=(car (nth 10 syntax-ppss-last)) don't reuse that data because the | |
103 | ;; assumed nil state at BEG may not be valid any more. | |
6bddffd0 | 104 | (if (<= beg (or (syntax-ppss-toplevel-pos (cdr syntax-ppss-last)) |
8debde6e | 105 | (nth 3 syntax-ppss-last) |
889c5dad | 106 | 0)) |
88a05faf SM |
107 | (setq syntax-ppss-last nil) |
108 | (setcar syntax-ppss-last nil))) | |
109 | ;; Unregister if there's no cache left. Sadly this doesn't work | |
ad8a840d | 110 | ;; because `before-change-functions' is temporarily bound to nil here. |
88a05faf | 111 | ;; (unless syntax-ppss-cache |
ad8a840d | 112 | ;; (remove-hook 'before-change-functions 'syntax-ppss-flush-cache t)) |
88a05faf SM |
113 | ) |
114 | ||
115 | (defvar syntax-ppss-stats | |
116 | [(0 . 0.0) (0 . 0.0) (0 . 0.0) (0 . 0.0) (0 . 0.0) (1 . 2500.0)]) | |
117 | (defun syntax-ppss-stats () | |
e8ac59b8 SM |
118 | (mapcar (lambda (x) |
119 | (condition-case nil | |
120 | (cons (car x) (truncate (/ (cdr x) (car x)))) | |
121 | (error nil))) | |
88a05faf SM |
122 | syntax-ppss-stats)) |
123 | ||
124 | (defun syntax-ppss (&optional pos) | |
125 | "Parse-Partial-Sexp State at POS. | |
126 | The returned value is the same as `parse-partial-sexp' except that | |
127 | the 2nd and 6th values of the returned state cannot be relied upon. | |
88a05faf SM |
128 | Point is at POS when this function returns." |
129 | ;; Default values. | |
130 | (unless pos (setq pos (point))) | |
a1506d29 | 131 | ;; |
88a05faf SM |
132 | (let ((old-ppss (cdr syntax-ppss-last)) |
133 | (old-pos (car syntax-ppss-last)) | |
134 | (ppss nil) | |
135 | (pt-min (point-min))) | |
136 | (if (and old-pos (> old-pos pos)) (setq old-pos nil)) | |
137 | ;; Use the OLD-POS if usable and close. Don't update the `last' cache. | |
db14504a SM |
138 | (condition-case nil |
139 | (if (and old-pos (< (- pos old-pos) | |
140 | ;; The time to use syntax-begin-function and | |
141 | ;; find PPSS is assumed to be about 2 * distance. | |
142 | (* 2 (/ (cdr (aref syntax-ppss-stats 5)) | |
143 | (1+ (car (aref syntax-ppss-stats 5))))))) | |
144 | (progn | |
145 | (incf (car (aref syntax-ppss-stats 0))) | |
146 | (incf (cdr (aref syntax-ppss-stats 0)) (- pos old-pos)) | |
147 | (parse-partial-sexp old-pos pos nil nil old-ppss)) | |
88a05faf SM |
148 | |
149 | (cond | |
db14504a SM |
150 | ;; Use OLD-PPSS if possible and close enough. |
151 | ((and (not old-pos) old-ppss | |
6bddffd0 SM |
152 | ;; If `pt-min' is too far from `pos', we could try to use |
153 | ;; other positions in (nth 9 old-ppss), but that doesn't | |
154 | ;; seem to happen in practice and it would complicate this | |
155 | ;; code (and the before-change-function code even more). | |
156 | ;; But maybe it would be useful in "degenerate" cases such | |
157 | ;; as when the whole file is wrapped in a set | |
158 | ;; of parentheses. | |
159 | (setq pt-min (or (syntax-ppss-toplevel-pos old-ppss) | |
db14504a SM |
160 | (nth 2 old-ppss))) |
161 | (<= pt-min pos) (< (- pos pt-min) syntax-ppss-max-span)) | |
162 | (incf (car (aref syntax-ppss-stats 1))) | |
163 | (incf (cdr (aref syntax-ppss-stats 1)) (- pos pt-min)) | |
164 | (setq ppss (parse-partial-sexp pt-min pos))) | |
165 | ;; The OLD-* data can't be used. Consult the cache. | |
88a05faf | 166 | (t |
db14504a SM |
167 | (let ((cache-pred nil) |
168 | (cache syntax-ppss-cache) | |
169 | (pt-min (point-min)) | |
170 | ;; I differentiate between PT-MIN and PT-BEST because | |
171 | ;; I feel like it might be important to ensure that the | |
172 | ;; cache is only filled with 100% sure data (whereas | |
173 | ;; syntax-begin-function might return incorrect data). | |
174 | ;; Maybe that's just stupid. | |
175 | (pt-best (point-min)) | |
176 | (ppss-best nil)) | |
177 | ;; look for a usable cache entry. | |
178 | (while (and cache (< pos (caar cache))) | |
179 | (setq cache-pred cache) | |
180 | (setq cache (cdr cache))) | |
181 | (if cache (setq pt-min (caar cache) ppss (cdar cache))) | |
182 | ||
ad8a840d | 183 | ;; Setup the before-change function if necessary. |
db14504a | 184 | (unless (or syntax-ppss-cache syntax-ppss-last) |
ad8a840d SM |
185 | (add-hook 'before-change-functions |
186 | 'syntax-ppss-flush-cache t t)) | |
db14504a SM |
187 | |
188 | ;; Use the best of OLD-POS and CACHE. | |
189 | (if (or (not old-pos) (< old-pos pt-min)) | |
190 | (setq pt-best pt-min ppss-best ppss) | |
191 | (incf (car (aref syntax-ppss-stats 4))) | |
192 | (incf (cdr (aref syntax-ppss-stats 4)) (- pos old-pos)) | |
193 | (setq pt-best old-pos ppss-best old-ppss)) | |
194 | ||
195 | ;; Use the `syntax-begin-function' if available. | |
196 | ;; We could try using that function earlier, but: | |
197 | ;; - The result might not be 100% reliable, so it's better to use | |
198 | ;; the cache if available. | |
199 | ;; - The function might be slow. | |
200 | ;; - If this function almost always finds a safe nearby spot, | |
201 | ;; the cache won't be populated, so consulting it is cheap. | |
202 | (when (and (not syntax-begin-function) | |
203 | (boundp 'font-lock-beginning-of-syntax-function) | |
204 | font-lock-beginning-of-syntax-function) | |
205 | (set (make-local-variable 'syntax-begin-function) | |
206 | font-lock-beginning-of-syntax-function)) | |
207 | (when (and syntax-begin-function | |
208 | (progn (goto-char pos) | |
209 | (funcall syntax-begin-function) | |
210 | ;; Make sure it's better. | |
211 | (> (point) pt-best)) | |
212 | ;; Simple sanity check. | |
213 | (not (memq (get-text-property (point) 'face) | |
214 | '(font-lock-string-face font-lock-doc-face | |
215 | font-lock-comment-face)))) | |
216 | (incf (car (aref syntax-ppss-stats 5))) | |
217 | (incf (cdr (aref syntax-ppss-stats 5)) (- pos (point))) | |
218 | (setq pt-best (point) ppss-best nil)) | |
219 | ||
220 | (cond | |
221 | ;; Quick case when we found a nearby pos. | |
222 | ((< (- pos pt-best) syntax-ppss-max-span) | |
223 | (incf (car (aref syntax-ppss-stats 2))) | |
224 | (incf (cdr (aref syntax-ppss-stats 2)) (- pos pt-best)) | |
225 | (setq ppss (parse-partial-sexp pt-best pos nil nil ppss-best))) | |
226 | ;; Slow case: compute the state from some known position and | |
227 | ;; populate the cache so we won't need to do it again soon. | |
228 | (t | |
229 | (incf (car (aref syntax-ppss-stats 3))) | |
230 | (incf (cdr (aref syntax-ppss-stats 3)) (- pos pt-min)) | |
231 | ||
232 | ;; If `pt-min' is too far, add a few intermediate entries. | |
233 | (while (> (- pos pt-min) (* 2 syntax-ppss-max-span)) | |
234 | (setq ppss (parse-partial-sexp | |
235 | pt-min (setq pt-min (/ (+ pt-min pos) 2)) | |
236 | nil nil ppss)) | |
237 | (let ((pair (cons pt-min ppss))) | |
238 | (if cache-pred | |
239 | (push pair (cdr cache-pred)) | |
240 | (push pair syntax-ppss-cache)))) | |
241 | ||
242 | ;; Compute the actual return value. | |
243 | (setq ppss (parse-partial-sexp pt-min pos nil nil ppss)) | |
244 | ||
245 | ;; Debugging check. | |
246 | ;; (let ((real-ppss (parse-partial-sexp (point-min) pos))) | |
247 | ;; (setcar (last ppss 4) 0) | |
248 | ;; (setcar (last real-ppss 4) 0) | |
249 | ;; (setcar (last ppss 8) nil) | |
250 | ;; (setcar (last real-ppss 8) nil) | |
251 | ;; (unless (equal ppss real-ppss) | |
252 | ;; (message "!!Syntax: %s != %s" ppss real-ppss) | |
253 | ;; (setq ppss real-ppss))) | |
254 | ||
255 | ;; Store it in the cache. | |
256 | (let ((pair (cons pos ppss))) | |
257 | (if cache-pred | |
258 | (if (> (- (caar cache-pred) pos) syntax-ppss-max-span) | |
259 | (push pair (cdr cache-pred)) | |
260 | (setcar cache-pred pair)) | |
261 | (if (or (null syntax-ppss-cache) | |
262 | (> (- (caar syntax-ppss-cache) pos) | |
263 | syntax-ppss-max-span)) | |
264 | (push pair syntax-ppss-cache) | |
265 | (setcar syntax-ppss-cache pair))))))))) | |
266 | ||
267 | (setq syntax-ppss-last (cons pos ppss)) | |
268 | ppss) | |
269 | (args-out-of-range | |
270 | ;; If the buffer is more narrowed than when we built the cache, | |
271 | ;; we may end up calling parse-partial-sexp with a position before | |
272 | ;; point-min. In that case, just parse from point-min assuming | |
273 | ;; a nil state. | |
274 | (parse-partial-sexp (point-min) pos))))) | |
88a05faf SM |
275 | |
276 | ;; Debugging functions | |
277 | ||
278 | (defun syntax-ppss-debug () | |
279 | (let ((pt nil) | |
280 | (min-diffs nil)) | |
281 | (dolist (x (append syntax-ppss-cache (list (cons (point-min) nil)))) | |
282 | (when pt (push (- pt (car x)) min-diffs)) | |
283 | (setq pt (car x))) | |
284 | min-diffs)) | |
285 | ||
286 | ;; XEmacs compatibility functions | |
287 | ||
288 | ;; (defun buffer-syntactic-context (&optional buffer) | |
289 | ;; "Syntactic context at point in BUFFER. | |
290 | ;; Either of `string', `comment' or `nil'. | |
291 | ;; This is an XEmacs compatibility function." | |
292 | ;; (with-current-buffer (or buffer (current-buffer)) | |
293 | ;; (syntax-ppss-context (syntax-ppss)))) | |
294 | ||
295 | ;; (defun buffer-syntactic-context-depth (&optional buffer) | |
296 | ;; "Syntactic parenthesis depth at point in BUFFER. | |
297 | ;; This is an XEmacs compatibility function." | |
298 | ;; (with-current-buffer (or buffer (current-buffer)) | |
299 | ;; (syntax-ppss-depth (syntax-ppss)))) | |
300 | ||
88a05faf | 301 | (provide 'syntax) |
ab5796a9 | 302 | |
889c5dad | 303 | ;; arch-tag: 302f1eeb-e77c-4680-a8c5-c543e01161a5 |
88a05faf | 304 | ;;; syntax.el ends here |