(comment-enter-backward): Be careful to restore point changed during narrowing.
[emacs.git] / lisp / sort.el
blob2fb6ec545085523ed04befdc09c39ec1b6cc3e3d
1 ;;; sort.el --- commands to sort text in an Emacs buffer
3 ;; Copyright (C) 1986, 1987, 1994, 1995, 2001, 2002, 2003,
4 ;; 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
6 ;; Author: Howie Kaye
7 ;; Maintainer: FSF
8 ;; Keywords: unix
10 ;; This file is part of GNU Emacs.
12 ;; GNU Emacs is free software; you can redistribute it and/or modify
13 ;; it under the terms of the GNU General Public License as published by
14 ;; the Free Software Foundation; either version 3, or (at your option)
15 ;; any later version.
17 ;; GNU Emacs is distributed in the hope that it will be useful,
18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 ;; GNU General Public License for more details.
22 ;; You should have received a copy of the GNU General Public License
23 ;; along with GNU Emacs; see the file COPYING. If not, write to the
24 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
25 ;; Boston, MA 02110-1301, USA.
27 ;;; Commentary:
29 ;; This package provides the sorting facilities documented in the Emacs
30 ;; user's manual.
32 ;;; Code:
34 (defgroup sort nil
35 "Commands to sort text in an Emacs buffer."
36 :group 'data)
38 (defcustom sort-fold-case nil
39 "*Non-nil if the buffer sort functions should ignore case."
40 :group 'sort
41 :type 'boolean)
42 ;;;###autoload(put 'sort-fold-case 'safe-local-variable 'booleanp)
44 ;;;###autoload
45 (defun sort-subr (reverse nextrecfun endrecfun
46 &optional startkeyfun endkeyfun predicate)
47 "General text sorting routine to divide buffer into records and sort them.
49 We divide the accessible portion of the buffer into disjoint pieces
50 called sort records. A portion of each sort record (perhaps all of
51 it) is designated as the sort key. The records are rearranged in the
52 buffer in order by their sort keys. The records may or may not be
53 contiguous.
55 Usually the records are rearranged in order of ascending sort key.
56 If REVERSE is non-nil, they are rearranged in order of descending sort key.
57 The variable `sort-fold-case' determines whether alphabetic case affects
58 the sort order.
60 The next four arguments are functions to be called to move point
61 across a sort record. They will be called many times from within sort-subr.
63 NEXTRECFUN is called with point at the end of the previous record.
64 It moves point to the start of the next record.
65 It should move point to the end of the buffer if there are no more records.
66 The first record is assumed to start at the position of point when sort-subr
67 is called.
69 ENDRECFUN is called with point within the record.
70 It should move point to the end of the record.
72 STARTKEYFUN moves from the start of the record to the start of the key.
73 It may return either a non-nil value to be used as the key, or
74 else the key is the substring between the values of point after
75 STARTKEYFUN and ENDKEYFUN are called. If STARTKEYFUN is nil, the key
76 starts at the beginning of the record.
78 ENDKEYFUN moves from the start of the sort key to the end of the sort key.
79 ENDKEYFUN may be nil if STARTKEYFUN returns a value or if it would be the
80 same as ENDRECFUN.
82 PREDICATE is the function to use to compare keys. If keys are numbers,
83 it defaults to `<', otherwise it defaults to `string<'."
84 ;; Heuristically try to avoid messages if sorting a small amt of text.
85 (let ((messages (> (- (point-max) (point-min)) 50000)))
86 (save-excursion
87 (if messages (message "Finding sort keys..."))
88 (let* ((sort-lists (sort-build-lists nextrecfun endrecfun
89 startkeyfun endkeyfun))
90 (old (reverse sort-lists))
91 (case-fold-search sort-fold-case))
92 (if (null sort-lists)
94 (or reverse (setq sort-lists (nreverse sort-lists)))
95 (if messages (message "Sorting records..."))
96 (setq sort-lists
97 (sort sort-lists
98 (cond (predicate
99 `(lambda (a b) (,predicate (car a) (car b))))
100 ((numberp (car (car sort-lists)))
101 'car-less-than-car)
102 ((consp (car (car sort-lists)))
103 (lambda (a b)
104 (> 0 (compare-buffer-substrings
105 nil (car (car a)) (cdr (car a))
106 nil (car (car b)) (cdr (car b))))))
108 (lambda (a b) (string< (car a) (car b)))))))
109 (if reverse (setq sort-lists (nreverse sort-lists)))
110 (if messages (message "Reordering buffer..."))
111 (sort-reorder-buffer sort-lists old)))
112 (if messages (message "Reordering buffer... Done"))))
113 nil)
115 ;; Parse buffer into records using the arguments as Lisp expressions;
116 ;; return a list of records. Each record looks like (KEY STARTPOS . ENDPOS)
117 ;; where KEY is the sort key (a number or string),
118 ;; and STARTPOS and ENDPOS are the bounds of this record in the buffer.
120 ;; The records appear in the list lastmost first!
122 (defun sort-build-lists (nextrecfun endrecfun startkeyfun endkeyfun)
123 (let ((sort-lists ())
124 (start-rec nil)
125 done key)
126 ;; Loop over sort records.
127 ;(goto-char (point-min)) -- it is the caller's responsibility to
128 ;arrange this if necessary
129 (while (not (eobp))
130 (setq start-rec (point)) ;save record start
131 (setq done nil)
132 ;; Get key value, or move to start of key.
133 (setq key (catch 'key
134 (or (and startkeyfun (funcall startkeyfun))
135 ;; If key was not returned as value,
136 ;; move to end of key and get key from the buffer.
137 (let ((start (point)))
138 (funcall (or endkeyfun
139 (prog1 endrecfun (setq done t))))
140 (cons start (point))))))
141 ;; Move to end of this record (start of next one, or end of buffer).
142 (cond ((prog1 done (setq done nil)))
143 (endrecfun (funcall endrecfun))
144 (nextrecfun (funcall nextrecfun) (setq done t)))
145 (if key (push
146 ;; consing optimization in case in which key is same as record.
147 (if (and (consp key)
148 (equal (car key) start-rec)
149 (equal (cdr key) (point)))
150 (cons key key)
151 (cons key (cons start-rec (point))))
152 sort-lists))
153 (and (not done) nextrecfun (funcall nextrecfun)))
154 sort-lists))
156 (defun sort-reorder-buffer (sort-lists old)
157 (let ((last (point-min))
158 (min (point-min)) (max (point-max))
159 (old-buffer (current-buffer))
160 (mb enable-multibyte-characters)
161 temp-buffer)
162 (with-temp-buffer
163 (set-buffer-multibyte mb)
164 ;; Record the temporary buffer.
165 (setq temp-buffer (current-buffer))
167 ;; Copy the sorted text into the temporary buffer.
168 (while sort-lists
169 (goto-char (point-max))
170 (insert-buffer-substring old-buffer
171 last
172 (nth 1 (car old)))
173 (goto-char (point-max))
174 (insert-buffer-substring old-buffer
175 (nth 1 (car sort-lists))
176 (cdr (cdr (car sort-lists))))
177 (setq last (cdr (cdr (car old)))
178 sort-lists (cdr sort-lists)
179 old (cdr old)))
180 (goto-char (point-max))
181 (insert-buffer-substring old-buffer last max)
183 ;; Copy the reordered text from the temporary buffer
184 ;; to the buffer we sorted (OLD-BUFFER).
185 (set-buffer old-buffer)
186 (let ((inhibit-quit t))
187 ;; Make sure insertions done for reordering
188 ;; saves any markers at the end of the sorted region,
189 ;; by leaving the last character of the region.
190 (delete-region min (1- max))
191 ;; Now replace the one remaining old character with the sorted text.
192 (goto-char (point-min))
193 (insert-buffer-substring temp-buffer)
194 (delete-region max (1+ max))))))
196 ;;;###autoload
197 (defun sort-lines (reverse beg end)
198 "Sort lines in region alphabetically; argument means descending order.
199 Called from a program, there are three arguments:
200 REVERSE (non-nil means reverse order), BEG and END (region to sort).
201 The variable `sort-fold-case' determines whether alphabetic case affects
202 the sort order."
203 (interactive "P\nr")
204 (save-excursion
205 (save-restriction
206 (narrow-to-region beg end)
207 (goto-char (point-min))
208 (let ;; To make `end-of-line' and etc. to ignore fields.
209 ((inhibit-field-text-motion t))
210 (sort-subr reverse 'forward-line 'end-of-line)))))
212 ;;;###autoload
213 (defun sort-paragraphs (reverse beg end)
214 "Sort paragraphs in region alphabetically; argument means descending order.
215 Called from a program, there are three arguments:
216 REVERSE (non-nil means reverse order), BEG and END (region to sort).
217 The variable `sort-fold-case' determines whether alphabetic case affects
218 the sort order."
219 (interactive "P\nr")
220 (save-excursion
221 (save-restriction
222 (narrow-to-region beg end)
223 (goto-char (point-min))
224 (sort-subr reverse
225 (function
226 (lambda ()
227 (while (and (not (eobp)) (looking-at paragraph-separate))
228 (forward-line 1))))
229 'forward-paragraph))))
231 ;;;###autoload
232 (defun sort-pages (reverse beg end)
233 "Sort pages in region alphabetically; argument means descending order.
234 Called from a program, there are three arguments:
235 REVERSE (non-nil means reverse order), BEG and END (region to sort).
236 The variable `sort-fold-case' determines whether alphabetic case affects
237 the sort order."
238 (interactive "P\nr")
239 (save-excursion
240 (save-restriction
241 (narrow-to-region beg end)
242 (goto-char (point-min))
243 (sort-subr reverse
244 (function (lambda () (skip-chars-forward "\n")))
245 'forward-page))))
247 (defvar sort-fields-syntax-table nil)
248 (if sort-fields-syntax-table nil
249 (let ((table (make-syntax-table))
250 (i 0))
251 (while (< i 256)
252 (modify-syntax-entry i "w" table)
253 (setq i (1+ i)))
254 (modify-syntax-entry ?\s " " table)
255 (modify-syntax-entry ?\t " " table)
256 (modify-syntax-entry ?\n " " table)
257 (modify-syntax-entry ?\. "_" table) ; for floating pt. numbers. -wsr
258 (setq sort-fields-syntax-table table)))
260 (defcustom sort-numeric-base 10
261 "*The default base used by `sort-numeric-fields'."
262 :group 'sort
263 :type 'integer)
264 ;;;###autoload(put 'sort-numeric-base 'safe-local-variable 'integerp)
266 ;;;###autoload
267 (defun sort-numeric-fields (field beg end)
268 "Sort lines in region numerically by the ARGth field of each line.
269 Fields are separated by whitespace and numbered from 1 up.
270 Specified field must contain a number in each line of the region,
271 which may begin with \"0x\" or \"0\" for hexadecimal and octal values.
272 Otherwise, the number is interpreted according to sort-numeric-base.
273 With a negative arg, sorts by the ARGth field counted from the right.
274 Called from a program, there are three arguments:
275 FIELD, BEG and END. BEG and END specify region to sort."
276 (interactive "p\nr")
277 (let ;; To make `end-of-line' and etc. to ignore fields.
278 ((inhibit-field-text-motion t))
279 (sort-fields-1 field beg end
280 (lambda ()
281 (sort-skip-fields field)
282 (let* ((case-fold-search t)
283 (base
284 (if (looking-at "\\(0x\\)[0-9a-f]\\|\\(0\\)[0-7]")
285 (cond ((match-beginning 1)
286 (goto-char (match-end 1))
288 ((match-beginning 2)
289 (goto-char (match-end 2))
291 (t nil)))))
292 (string-to-number (buffer-substring (point)
293 (save-excursion
294 (forward-sexp 1)
295 (point)))
296 (or base sort-numeric-base))))
297 nil)))
299 ;;;;;###autoload
300 ;;(defun sort-float-fields (field beg end)
301 ;; "Sort lines in region numerically by the ARGth field of each line.
302 ;;Fields are separated by whitespace and numbered from 1 up. Specified field
303 ;;must contain a floating point number in each line of the region. With a
304 ;;negative arg, sorts by the ARGth field counted from the right. Called from a
305 ;;program, there are three arguments: FIELD, BEG and END. BEG and END specify
306 ;;region to sort."
307 ;; (interactive "p\nr")
308 ;; (sort-fields-1 field beg end
309 ;; (function (lambda ()
310 ;; (sort-skip-fields field)
311 ;; (string-to-number
312 ;; (buffer-substring
313 ;; (point)
314 ;; (save-excursion
315 ;; (re-search-forward
316 ;; "[+-]?[0-9]*\.?[0-9]*\\([eE][+-]?[0-9]+\\)?")
317 ;; (point))))))
318 ;; nil))
320 ;;;###autoload
321 (defun sort-fields (field beg end)
322 "Sort lines in region lexicographically by the ARGth field of each line.
323 Fields are separated by whitespace and numbered from 1 up.
324 With a negative arg, sorts by the ARGth field counted from the right.
325 Called from a program, there are three arguments:
326 FIELD, BEG and END. BEG and END specify region to sort.
327 The variable `sort-fold-case' determines whether alphabetic case affects
328 the sort order."
329 (interactive "p\nr")
330 (let ;; To make `end-of-line' and etc. to ignore fields.
331 ((inhibit-field-text-motion t))
332 (sort-fields-1 field beg end
333 (function (lambda ()
334 (sort-skip-fields field)
335 nil))
336 (function (lambda () (skip-chars-forward "^ \t\n"))))))
338 (defun sort-fields-1 (field beg end startkeyfun endkeyfun)
339 (let ((tbl (syntax-table)))
340 (if (zerop field) (setq field 1))
341 (unwind-protect
342 (save-excursion
343 (save-restriction
344 (narrow-to-region beg end)
345 (goto-char (point-min))
346 (set-syntax-table sort-fields-syntax-table)
347 (sort-subr nil
348 'forward-line 'end-of-line
349 startkeyfun endkeyfun)))
350 (set-syntax-table tbl))))
352 ;; Position at the beginning of field N on the current line,
353 ;; assuming point is initially at the beginning of the line.
354 (defun sort-skip-fields (n)
355 (if (> n 0)
356 ;; Skip across N - 1 fields.
357 (let ((i (1- n)))
358 (while (> i 0)
359 (skip-chars-forward " \t")
360 (skip-chars-forward "^ \t\n")
361 (setq i (1- i)))
362 (skip-chars-forward " \t")
363 (if (eolp)
364 (error "Line has too few fields: %s"
365 (buffer-substring
366 (save-excursion (beginning-of-line) (point))
367 (save-excursion (end-of-line) (point))))))
368 (end-of-line)
369 ;; Skip back across - N - 1 fields.
370 (let ((i (1- (- n))))
371 (while (> i 0)
372 (skip-chars-backward " \t")
373 (skip-chars-backward "^ \t\n")
374 (setq i (1- i)))
375 (skip-chars-backward " \t"))
376 (if (bolp)
377 (error "Line has too few fields: %s"
378 (buffer-substring
379 (save-excursion (beginning-of-line) (point))
380 (save-excursion (end-of-line) (point)))))
381 ;; Position at the front of the field
382 ;; even if moving backwards.
383 (skip-chars-backward "^ \t\n")))
385 (defvar sort-regexp-fields-regexp)
386 (defvar sort-regexp-record-end)
388 ;; Move to the beginning of the next match for record-regexp,
389 ;; and set sort-regexp-record-end to the end of that match.
390 ;; If the next match is empty and does not advance point,
391 ;; skip one character and try again.
392 (defun sort-regexp-fields-next-record ()
393 (let ((oldpos (point)))
394 (and (re-search-forward sort-regexp-fields-regexp nil 'move)
395 (setq sort-regexp-record-end (match-end 0))
396 (if (= sort-regexp-record-end oldpos)
397 (progn
398 (forward-char 1)
399 (re-search-forward sort-regexp-fields-regexp nil 'move)
400 (setq sort-regexp-record-end (match-end 0)))
402 (goto-char (match-beginning 0)))))
404 ;;;###autoload
405 (defun sort-regexp-fields (reverse record-regexp key-regexp beg end)
406 "Sort the region lexicographically as specified by RECORD-REGEXP and KEY.
407 RECORD-REGEXP specifies the textual units which should be sorted.
408 For example, to sort lines RECORD-REGEXP would be \"^.*$\"
409 KEY specifies the part of each record (ie each match for RECORD-REGEXP)
410 is to be used for sorting.
411 If it is \"\\\\digit\" then the digit'th \"\\\\(...\\\\)\" match field from
412 RECORD-REGEXP is used.
413 If it is \"\\\\&\" then the whole record is used.
414 Otherwise, it is a regular-expression for which to search within the record.
415 If a match for KEY is not found within a record then that record is ignored.
417 With a negative prefix arg sorts in reverse order.
419 The variable `sort-fold-case' determines whether alphabetic case affects
420 the sort order.
422 For example: to sort lines in the region by the first word on each line
423 starting with the letter \"f\",
424 RECORD-REGEXP would be \"^.*$\" and KEY would be \"\\\\=\\<f\\\\w*\\\\>\""
425 ;; using negative prefix arg to mean "reverse" is now inconsistent with
426 ;; other sort-.*fields functions but then again this was before, since it
427 ;; didn't use the magnitude of the arg to specify anything.
428 (interactive "P\nsRegexp specifying records to sort:
429 sRegexp specifying key within record: \nr")
430 (cond ((or (equal key-regexp "") (equal key-regexp "\\&"))
431 (setq key-regexp 0))
432 ((string-match "\\`\\\\[1-9]\\'" key-regexp)
433 (setq key-regexp (- (aref key-regexp 1) ?0))))
434 (save-excursion
435 (save-restriction
436 (narrow-to-region beg end)
437 (goto-char (point-min))
438 (let (sort-regexp-record-end
439 (sort-regexp-fields-regexp record-regexp))
440 (re-search-forward sort-regexp-fields-regexp nil t)
441 (setq sort-regexp-record-end (point))
442 (goto-char (match-beginning 0))
443 (sort-subr reverse
444 'sort-regexp-fields-next-record
445 (function (lambda ()
446 (goto-char sort-regexp-record-end)))
447 (function (lambda ()
448 (let ((n 0))
449 (cond ((numberp key-regexp)
450 (setq n key-regexp))
451 ((re-search-forward
452 key-regexp sort-regexp-record-end t)
453 (setq n 0))
454 (t (throw 'key nil)))
455 (condition-case ()
456 (cons (match-beginning n)
457 (match-end n))
458 ;; if there was no such register
459 (error (throw 'key nil)))))))))))
462 (defvar sort-columns-subprocess t)
464 ;;;###autoload
465 (defun sort-columns (reverse &optional beg end)
466 "Sort lines in region alphabetically by a certain range of columns.
467 For the purpose of this command, the region BEG...END includes
468 the entire line that point is in and the entire line the mark is in.
469 The column positions of point and mark bound the range of columns to sort on.
470 A prefix argument means sort into REVERSE order.
471 The variable `sort-fold-case' determines whether alphabetic case affects
472 the sort order.
474 Note that `sort-columns' rejects text that contains tabs,
475 because tabs could be split across the specified columns
476 and it doesn't know how to handle that. Also, when possible,
477 it uses the `sort' utility program, which doesn't understand tabs.
478 Use \\[untabify] to convert tabs to spaces before sorting."
479 (interactive "P\nr")
480 (save-excursion
481 (let ;; To make `end-of-line' and etc. to ignore fields.
482 ((inhibit-field-text-motion t)
483 beg1 end1 col-beg1 col-end1 col-start col-end)
484 (goto-char (min beg end))
485 (setq col-beg1 (current-column))
486 (beginning-of-line)
487 (setq beg1 (point))
488 (goto-char (max beg end))
489 (setq col-end1 (current-column))
490 (forward-line)
491 (setq end1 (point))
492 (setq col-start (min col-beg1 col-end1))
493 (setq col-end (max col-beg1 col-end1))
494 (if (search-backward "\t" beg1 t)
495 (error "sort-columns does not work with tabs -- use M-x untabify"))
496 (if (not (or (memq system-type '(vax-vms windows-nt))
497 (let ((pos beg1) plist fontified)
498 (catch 'found
499 (while (< pos end1)
500 (setq plist (text-properties-at pos))
501 (setq fontified (plist-get plist 'fontified))
502 (while (consp plist)
503 (unless (or (eq (car plist) 'fontified)
504 (and (eq (car plist) 'face)
505 fontified))
506 (throw 'found t))
507 (setq plist (cddr plist)))
508 (setq pos (next-property-change pos nil end1)))))))
509 ;; Use the sort utility if we can; it is 4 times as fast.
510 ;; Do not use it if there are any non-font-lock properties
511 ;; in the region, since the sort utility would lose the
512 ;; properties. Tabs are used as field separator; on NetBSD,
513 ;; sort complains if "\n" is used as field separator.
514 (let ((sort-args (list (if reverse "-rt\t" "-t\t")
515 (format "-k1.%d,1.%d"
516 (1+ col-start)
517 (1+ col-end)))))
518 (when sort-fold-case
519 (push "-f" sort-args))
520 (apply #'call-process-region beg1 end1 "sort" t t nil sort-args))
521 ;; On VMS and ms-windows, use Emacs's own facilities.
522 (save-excursion
523 (save-restriction
524 (narrow-to-region beg1 end1)
525 (goto-char beg1)
526 (sort-subr reverse 'forward-line 'end-of-line
527 #'(lambda () (move-to-column col-start) nil)
528 #'(lambda () (move-to-column col-end) nil))))))))
530 ;;;###autoload
531 (defun reverse-region (beg end)
532 "Reverse the order of lines in a region.
533 From a program takes two point or marker arguments, BEG and END."
534 (interactive "r")
535 (if (> beg end)
536 (let (mid) (setq mid end end beg beg mid)))
537 (save-excursion
538 ;; put beg at the start of a line and end and the end of one --
539 ;; the largest possible region which fits this criteria
540 (goto-char beg)
541 (or (bolp) (forward-line 1))
542 (setq beg (point))
543 (goto-char end)
544 ;; the test for bolp is for those times when end is on an empty line;
545 ;; it is probably not the case that the line should be included in the
546 ;; reversal; it isn't difficult to add it afterward.
547 (or (and (eolp) (not (bolp))) (progn (forward-line -1) (end-of-line)))
548 (setq end (point-marker))
549 ;; the real work. this thing cranks through memory on large regions.
550 (let (ll (do t))
551 (while do
552 (goto-char beg)
553 (setq ll (cons (buffer-substring (point) (progn (end-of-line) (point)))
554 ll))
555 (setq do (/= (point) end))
556 (delete-region beg (if do (1+ (point)) (point))))
557 (while (cdr ll)
558 (insert (car ll) "\n")
559 (setq ll (cdr ll)))
560 (insert (car ll)))))
562 (provide 'sort)
564 ;;; arch-tag: fbac12be-2a7b-4c8a-9665-264d61f70bd9
565 ;;; sort.el ends here