(struct sound_device): Function pointer field "write"
[bpt/emacs.git] / lisp / format.el
CommitLineData
a8581027 1;;; format.el --- read and save files in multiple formats
b578f267 2
823139fb 3;; Copyright (c) 1994, 1995, 1997, 1999 Free Software Foundation
029894b9 4
823139fb 5;; Author: Boris Goldowsky <boris@gnu.org>
029894b9
BG
6
7;; This file is part of GNU Emacs.
8
9;; GNU Emacs is free software; you can redistribute it and/or modify
10;; it under the terms of the GNU General Public License as published by
11;; the Free Software Foundation; either version 2, or (at your option)
12;; any later version.
b578f267 13
029894b9
BG
14;; GNU Emacs is distributed in the hope that it will be useful,
15;; but WITHOUT ANY WARRANTY; without even the implied warranty of
16;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17;; GNU General Public License for more details.
b578f267 18
029894b9 19;; You should have received a copy of the GNU General Public License
b578f267
EN
20;; along with GNU Emacs; see the file COPYING. If not, write to the
21;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
22;; Boston, MA 02111-1307, USA.
029894b9
BG
23
24;;; Commentary:
b578f267
EN
25
26;; This file defines a unified mechanism for saving & loading files stored
27;; in different formats. `format-alist' contains information that directs
029894b9 28;; Emacs to call an encoding or decoding function when reading or writing
823139fb 29;; files that match certain conditions.
029894b9 30;;
b578f267
EN
31;; When a file is visited, its format is determined by matching the
32;; beginning of the file against regular expressions stored in
33;; `format-alist'. If this fails, you can manually translate the buffer
34;; using `format-decode-buffer'. In either case, the formats used are
35;; listed in the variable `buffer-file-format', and become the default
36;; format for saving the buffer. To save a buffer in a different format,
37;; change this variable, or use `format-write-file'.
029894b9
BG
38;;
39;; Auto-save files are normally created in the same format as the visited
b578f267
EN
40;; file, but the variable `auto-save-file-format' can be set to a
41;; particularly fast or otherwise preferred format to be used for
42;; auto-saving (or nil to do no encoding on auto-save files, but then you
43;; risk losing any text-properties in the buffer).
029894b9 44;;
b578f267
EN
45;; You can manually translate a buffer into or out of a particular format
46;; with the functions `format-encode-buffer' and `format-decode-buffer'.
47;; To translate just the region use the functions `format-encode-region'
823139fb 48;; and `format-decode-region'.
029894b9 49;;
b578f267
EN
50;; You can define a new format by writing the encoding and decoding
51;; functions, and adding an entry to `format-alist'. See enriched.el for
52;; an example of how to implement a file format. There are various
53;; functions defined in this file that may be useful for writing the
54;; encoding and decoding functions:
55;; * `format-annotate-region' and `format-deannotate-region' allow a
56;; single alist of information to be used for encoding and decoding.
57;; The alist defines a correspondence between strings in the file
58;; ("annotations") and text-properties in the buffer.
029894b9
BG
59;; * `format-replace-strings' is similarly useful for doing simple
60;; string->string translations in a reversible manner.
61
b578f267
EN
62;;; Code:
63
029894b9
BG
64(put 'buffer-file-format 'permanent-local t)
65
823139fb 66(defvar format-alist
8d0dd9d2 67 '((text/enriched "Extended MIME text/enriched format."
029894b9
BG
68 "Content-[Tt]ype:[ \t]*text/enriched"
69 enriched-decode enriched-encode t enriched-mode)
c84cf2bf 70 (plain "ISO 8859-1 standard format, no text properties."
029894b9
BG
71 ;; Plain only exists so that there is an obvious neutral choice in
72 ;; the completion list.
c84cf2bf 73 nil nil nil nil nil)
823139fb
DL
74 (ibm "IBM Code Page 850 (DOS)"
75 nil ; The original "1\\(^\\)" is obscure.
271af233 76 "recode -f ibm-pc:latin1" "recode -f latin1:ibm-pc" t nil)
823139fb
DL
77 (mac "Apple Macintosh"
78 nil
271af233 79 "recode -f mac:latin1" "recode -f latin1:mac" t nil)
823139fb
DL
80 (hp "HP Roman8"
81 nil
271af233 82 "recode -f roman8:latin1" "recode -f latin1:roman8" t nil)
823139fb
DL
83 (TeX "TeX (encoding)"
84 nil
c84cf2bf 85 iso-tex2iso iso-iso2tex t nil)
823139fb
DL
86 (gtex "German TeX (encoding)"
87 nil
c84cf2bf 88 iso-gtex2iso iso-iso2gtex t nil)
f1d6fe69 89 (html "HTML/SGML \"ISO 8879:1986//ENTITIES Added Latin 1//EN\" (encoding)"
823139fb 90 nil
f1d6fe69 91 iso-sgml2iso iso-iso2sgml t nil)
823139fb
DL
92 (rot13 "rot13"
93 nil
c84cf2bf 94 "tr a-mn-z n-za-m" "tr a-mn-z n-za-m" t nil)
823139fb
DL
95 (duden "Duden Ersatzdarstellung"
96 nil
97 "diac" iso-iso2duden t nil)
98 (de646 "German ASCII (ISO 646)"
99 nil
271af233 100 "recode -f iso646-ge:latin1" "recode -f latin1:iso646-ge" t nil)
823139fb
DL
101 (denet "net German"
102 nil
c84cf2bf 103 iso-german iso-cvt-read-only t nil)
823139fb
DL
104 (esnet "net Spanish"
105 nil
8d0dd9d2 106 iso-spanish iso-cvt-read-only t nil))
029894b9
BG
107 "List of information about understood file formats.
108Elements are of the form \(NAME DOC-STR REGEXP FROM-FN TO-FN MODIFY MODE-FN).
d1ae737b 109
029894b9 110NAME is a symbol, which is stored in `buffer-file-format'.
d1ae737b 111
029894b9
BG
112DOC-STR should be a single line providing more information about the
113 format. It is currently unused, but in the future will be shown to
114 the user if they ask for more information.
d1ae737b 115
029894b9 116REGEXP is a regular expression to match against the beginning of the file;
823139fb
DL
117 it should match only files in that format. Use nil to avoid
118 matching at all for formats for which this isn't appropriate to
119 require explicit encoding/decoding.
d1ae737b 120
823139fb 121FROM-FN is called to decode files in that format; it gets two args, BEGIN
029894b9
BG
122 and END, and can make any modifications it likes, returning the new
123 end. It must make sure that the beginning of the file no longer
124 matches REGEXP, or else it will get called again.
d1ae737b
RS
125 Alternatively, FROM-FN can be a string, which specifies a shell command
126 (including options) to be used as a filter to perform the conversion.
127
d1abb42f
RS
128TO-FN is called to encode a region into that format; it is passed three
129 arguments: BEGIN, END, and BUFFER. BUFFER is the original buffer that
130 the data being written came from, which the function could use, for
131 example, to find the values of local variables. TO-FN should either
132 return a list of annotations like `write-region-annotate-functions',
133 or modify the region and return the new end.
d1ae737b
RS
134 Alternatively, TO-FN can be a string, which specifies a shell command
135 (including options) to be used as a filter to perform the conversion.
136
d1abb42f
RS
137MODIFY, if non-nil, means the TO-FN wants to modify the region. If nil,
138 TO-FN will not make any changes but will instead return a list of
823139fb 139 annotations.
d1ae737b 140
3e224645
EZ
141MODE-FN, if specified, is called when visiting a file with that format.
142 It is called with a single positive argument, on the assumption
34f30f79
RS
143 that it turns on some Emacs mode.
144
145PRESERVE, if non-nil, means that `format-write-file' should not remove
146 this format from `buffer-file-formats'.")
029894b9
BG
147
148;;; Basic Functions (called from Lisp)
149
c84cf2bf
RS
150(defun format-encode-run-method (method from to &optional buffer)
151 "Translate using function or shell script METHOD the text from FROM to TO.
152If METHOD is a string, it is a shell command;
153otherwise, it should be a Lisp function.
154BUFFER should be the buffer that the output originally came from."
155 (if (stringp method)
823139fb 156 (let ((error-buff (get-buffer-create "*Format Errors*"))
7cc45a35 157 (coding-system-for-read 'no-conversion)
823139fb
DL
158 format-alist)
159 (with-current-buffer error-buff
160 (widen)
161 (erase-buffer))
10fa0073
GM
162 (if (and (zerop (save-window-excursion
163 (shell-command-on-region from to method t t
164 error-buff)))
823139fb
DL
165 ;; gzip gives zero exit status with bad args, for instance.
166 (zerop (with-current-buffer error-buff
167 (buffer-size))))
168 (bury-buffer error-buff)
169 (switch-to-buffer-other-window error-buff)
10fa0073 170 (error "Format encoding failed")))
c84cf2bf
RS
171 (funcall method from to buffer)))
172
173(defun format-decode-run-method (method from to &optional buffer)
174 "Decode using function or shell script METHOD the text from FROM to TO.
823139fb
DL
175If METHOD is a string, it is a shell command; otherwise, it should be
176a Lisp function. Decoding is done for the given BUFFER."
c84cf2bf 177 (if (stringp method)
823139fb 178 (let ((error-buff (get-buffer-create "*Format Errors*"))
7cc45a35 179 (coding-system-for-write 'no-conversion)
823139fb
DL
180 format-alist)
181 (with-current-buffer error-buff
182 (widen)
183 (erase-buffer))
184 ;; We should perhaps go via a temporary buffer and copy it
185 ;; back, in case of errors.
186 (if (and (zerop (save-window-excursion
187 (shell-command-on-region (point-min) (point-max)
188 method t t
189 error-buff)))
190 ;; gzip gives zero exit status with bad args, for instance.
191 (zerop (with-current-buffer error-buff
192 (buffer-size))))
193 (bury-buffer error-buff)
194 (switch-to-buffer-other-window error-buff)
195 (error "Format decoding failed"))
c84cf2bf
RS
196 (point))
197 (funcall method from to)))
198
10fa0073 199(defun format-annotate-function (format from to orig-buf format-count)
823139fb 200 "Return annotations for writing region as FORMAT.
029894b9
BG
201FORMAT is a symbol naming one of the formats defined in `format-alist',
202it must be a single symbol, not a list like `buffer-file-format'.
d1abb42f
RS
203FROM and TO delimit the region to be operated on in the current buffer.
204ORIG-BUF is the original buffer that the data came from.
10fa0073
GM
205
206FORMAT-COUNT is an integer specifying how many times this function has
207been called in the process of decoding ORIG-BUF.
208
029894b9
BG
209This function works like a function on `write-region-annotate-functions':
210it either returns a list of annotations, or returns with a different buffer
10fa0073
GM
211current, which contains the modified text to write. In the latter case,
212this function's value is nil.
029894b9
BG
213
214For most purposes, consider using `format-encode-region' instead."
10fa0073
GM
215 ;; This function is called by write-region (actually
216 ;; build_annotations) for each element of buffer-file-format.
029894b9
BG
217 (let* ((info (assq format format-alist))
218 (to-fn (nth 4 info))
219 (modify (nth 5 info)))
220 (if to-fn
221 (if modify
222 ;; To-function wants to modify region. Copy to safe place.
10fa0073 223 (let ((copy-buf (get-buffer-create (format " *Format Temp %d*"
8a2f1176
RS
224 format-count)))
225 (sel-disp selective-display)
226 (multibyte enable-multibyte-characters))
227 (with-current-buffer copy-buf
228 (setq selective-display sel-disp)
229 (set-buffer-multibyte multibyte))
029894b9
BG
230 (copy-to-buffer copy-buf from to)
231 (set-buffer copy-buf)
232 (format-insert-annotations write-region-annotations-so-far from)
c84cf2bf 233 (format-encode-run-method to-fn (point-min) (point-max) orig-buf)
029894b9
BG
234 nil)
235 ;; Otherwise just call function, it will return annotations.
d1abb42f 236 (funcall to-fn from to orig-buf)))))
029894b9
BG
237
238(defun format-decode (format length &optional visit-flag)
239 ;; This function is called by insert-file-contents whenever a file is read.
240 "Decode text from any known FORMAT.
823139fb 241FORMAT is a symbol appearing in `format-alist' or a list of such symbols,
029894b9
BG
242or nil, in which case this function tries to guess the format of the data by
243matching against the regular expressions in `format-alist'. After a match is
244found and the region decoded, the alist is searched again from the beginning
245for another match.
246
247Second arg LENGTH is the number of characters following point to operate on.
248If optional third arg VISIT-FLAG is true, set `buffer-file-format'
249to the list of formats used, and call any mode functions defined for those
250formats.
251
252Returns the new length of the decoded region.
253
254For most purposes, consider using `format-decode-region' instead."
255 (let ((mod (buffer-modified-p))
10fa0073 256 (begin (point))
029894b9 257 (end (+ (point) length)))
10fa0073
GM
258 (unwind-protect
259 (progn
260 ;; Don't record undo information for the decoding.
10fa0073
GM
261
262 (if (null format)
263 ;; Figure out which format it is in, remember list in `format'.
264 (let ((try format-alist))
265 (while try
266 (let* ((f (car try))
267 (regexp (nth 2 f))
268 (p (point)))
269 (if (and regexp (looking-at regexp)
270 (< (match-end 0) (+ begin length)))
271 (progn
272 (setq format (cons (car f) format))
273 ;; Decode it
274 (if (nth 3 f)
275 (setq end (format-decode-run-method (nth 3 f) begin end)))
276 ;; Call visit function if required
277 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
278 ;; Safeguard against either of the functions changing pt.
279 (goto-char p)
280 ;; Rewind list to look for another format
281 (setq try format-alist))
282 (setq try (cdr try))))))
283 ;; Deal with given format(s)
284 (or (listp format) (setq format (list format)))
285 (let ((do format) f)
286 (while do
287 (or (setq f (assq (car do) format-alist))
288 (error "Unknown format" (car do)))
289 ;; Decode:
290 (if (nth 3 f)
291 (setq end (format-decode-run-method (nth 3 f) begin end)))
292 ;; Call visit function if required
293 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
294 (setq do (cdr do)))))
295 (if visit-flag
296 (setq buffer-file-format format)))
297
021cfbea
GM
298 (set-buffer-modified-p mod))
299
10fa0073 300 ;; Return new length of region
029894b9
BG
301 (- end begin)))
302
303;;;
304;;; Interactive functions & entry points
305;;;
306
307(defun format-decode-buffer (&optional format)
308 "Translate the buffer from some FORMAT.
309If the format is not specified, this function attempts to guess.
823139fb 310`buffer-file-format' is set to the format used, and any mode-functions
029894b9
BG
311for the format are called."
312 (interactive
313 (list (format-read "Translate buffer from format (default: guess): ")))
314 (save-excursion
315 (goto-char (point-min))
316 (format-decode format (buffer-size) t)))
317
318(defun format-decode-region (from to &optional format)
319 "Decode the region from some format.
320Arg FORMAT is optional; if omitted the format will be determined by looking
321for identifying regular expressions at the beginning of the region."
322 (interactive
823139fb 323 (list (region-beginning) (region-end)
029894b9
BG
324 (format-read "Translate region from format (default: guess): ")))
325 (save-excursion
326 (goto-char from)
327 (format-decode format (- to from) nil)))
328
329(defun format-encode-buffer (&optional format)
330 "Translate the buffer into FORMAT.
331FORMAT defaults to `buffer-file-format'. It is a symbol naming one of the
332formats defined in `format-alist', or a list of such symbols."
333 (interactive
334 (list (format-read (format "Translate buffer to format (default %s): "
335 buffer-file-format))))
336 (format-encode-region (point-min) (point-max) format))
337
338(defun format-encode-region (beg end &optional format)
823139fb 339 "Translate the region into some FORMAT.
029894b9
BG
340FORMAT defaults to `buffer-file-format', it is a symbol naming
341one of the formats defined in `format-alist', or a list of such symbols."
823139fb
DL
342 (interactive
343 (list (region-beginning) (region-end)
344 (format-read (format "Translate region to format (default %s): "
345 buffer-file-format))))
346 (if (null format) (setq format buffer-file-format))
347 (if (symbolp format) (setq format (list format)))
348 (save-excursion
349 (goto-char end)
350 (let ((cur-buf (current-buffer))
351 (end (point-marker)))
352 (while format
353 (let* ((info (assq (car format) format-alist))
354 (to-fn (nth 4 info))
355 (modify (nth 5 info))
356 result)
357 (if to-fn
358 (if modify
359 (setq end (format-encode-run-method to-fn beg end
360 (current-buffer)))
361 (format-insert-annotations
362 (funcall to-fn beg end (current-buffer)))))
363 (setq format (cdr format)))))))
029894b9
BG
364
365(defun format-write-file (filename format)
823139fb 366 "Write current buffer into file FILENAME using some FORMAT.
029894b9
BG
367Makes buffer visit that file and sets the format as the default for future
368saves. If the buffer is already visiting a file, you can specify a directory
823139fb 369name as FILENAME, to write a file of the same old name in that directory."
029894b9
BG
370 (interactive
371 ;; Same interactive spec as write-file, plus format question.
372 (let* ((file (if buffer-file-name
373 (read-file-name "Write file: "
374 nil nil nil nil)
375 (read-file-name "Write file: "
376 (cdr (assq 'default-directory
377 (buffer-local-variables)))
378 nil nil (buffer-name))))
823139fb 379 (fmt (format-read (format "Write file `%s' in format: "
029894b9
BG
380 (file-name-nondirectory file)))))
381 (list file fmt)))
34f30f79
RS
382 (let ((old-formats buffer-file-format)
383 preserve-formats)
384 (dolist (fmt old-formats)
385 (let ((aelt (assq fmt format-alist)))
386 (if (nth 7 aelt)
387 (push fmt preserve-formats))))
388 (setq buffer-file-format format)
389 (dolist (fmt preserve-formats)
390 (unless (memq fmt buffer-file-format)
391 (setq buffer-file-format (append buffer-file-format (list fmt))))))
029894b9
BG
392 (write-file filename))
393
ce6af52b 394(defun format-find-file (filename format)
823139fb 395 "Find the file FILENAME using data format FORMAT.
ce6af52b
KH
396If FORMAT is nil then do not do any format conversion."
397 (interactive
398 ;; Same interactive spec as write-file, plus format question.
399 (let* ((file (read-file-name "Find file: "))
823139fb 400 (fmt (format-read (format "Read file `%s' in format: "
ce6af52b
KH
401 (file-name-nondirectory file)))))
402 (list file fmt)))
403 (let ((format-alist nil))
404 (find-file filename))
405 (if format
406 (format-decode-buffer format)))
407
408(defun format-insert-file (filename format &optional beg end)
823139fb 409 "Insert the contents of file FILENAME using data format FORMAT.
ce6af52b
KH
410If FORMAT is nil then do not do any format conversion.
411The optional third and fourth arguments BEG and END specify
412the part of the file to read.
413
414The return value is like the value of `insert-file-contents':
415a list (ABSOLUTE-FILE-NAME . SIZE)."
416 (interactive
417 ;; Same interactive spec as write-file, plus format question.
418 (let* ((file (read-file-name "Find file: "))
823139fb 419 (fmt (format-read (format "Read file `%s' in format: "
ce6af52b
KH
420 (file-name-nondirectory file)))))
421 (list file fmt)))
422 (let (value size)
423 (let ((format-alist nil))
424 (setq value (insert-file-contents filename nil beg end))
425 (setq size (nth 1 value)))
426 (if format
c84cf2bf 427 (setq size (format-decode format size)
ce6af52b
KH
428 value (cons (car value) size)))
429 value))
430
029894b9
BG
431(defun format-read (&optional prompt)
432 "Read and return the name of a format.
433Return value is a list, like `buffer-file-format'; it may be nil.
434Formats are defined in `format-alist'. Optional arg is the PROMPT to use."
435 (let* ((table (mapcar (lambda (x) (list (symbol-name (car x))))
436 format-alist))
437 (ans (completing-read (or prompt "Format: ") table nil t)))
438 (if (not (equal "" ans)) (list (intern ans)))))
439
440
441;;;
442;;; Below are some functions that may be useful in writing encoding and
443;;; decoding functions for use in format-alist.
444;;;
445
446(defun format-replace-strings (alist &optional reverse beg end)
447 "Do multiple replacements on the buffer.
d0fc7e3b 448ALIST is a list of (FROM . TO) pairs, which should be proper arguments to
029894b9 449`search-forward' and `replace-match' respectively.
d0fc7e3b 450Optional 2nd arg REVERSE, if non-nil, means the pairs are (TO . FROM), so that
029894b9 451you can use the same list in both directions if it contains only literal
823139fb
DL
452strings.
453Optional args BEG and END specify a region of the buffer on which to operate."
029894b9
BG
454 (save-excursion
455 (save-restriction
456 (or beg (setq beg (point-min)))
457 (if end (narrow-to-region (point-min) end))
458 (while alist
459 (let ((from (if reverse (cdr (car alist)) (car (car alist))))
2c3ac81d 460 (to (if reverse (car (car alist)) (cdr (car alist)))))
029894b9
BG
461 (goto-char beg)
462 (while (search-forward from nil t)
463 (goto-char (match-beginning 0))
464 (insert to)
465 (set-text-properties (- (point) (length to)) (point)
466 (text-properties-at (point)))
467 (delete-region (point) (+ (point) (- (match-end 0)
468 (match-beginning 0)))))
469 (setq alist (cdr alist)))))))
470
471;;; Some list-manipulation functions that we need.
472
473(defun format-delq-cons (cons list)
823139fb
DL
474 "Remove the given CONS from LIST by side effect and return the new LIST.
475Since CONS could be the first element of LIST, write
476`\(setq foo \(format-delq-cons element foo))' to be sure of changing
477the value of `foo'."
029894b9
BG
478 (if (eq cons list)
479 (cdr list)
480 (let ((p list))
481 (while (not (eq (cdr p) cons))
e8af40ee 482 (if (null p) (error "format-delq-cons: not an element"))
029894b9
BG
483 (setq p (cdr p)))
484 ;; Now (cdr p) is the cons to delete
485 (setcdr p (cdr cons))
486 list)))
487
488(defun format-make-relatively-unique (a b)
489 "Delete common elements of lists A and B, return as pair.
490Compares using `equal'."
491 (let* ((acopy (copy-sequence a))
492 (bcopy (copy-sequence b))
493 (tail acopy))
494 (while tail
495 (let ((dup (member (car tail) bcopy))
496 (next (cdr tail)))
497 (if dup (setq acopy (format-delq-cons tail acopy)
498 bcopy (format-delq-cons dup bcopy)))
499 (setq tail next)))
500 (cons acopy bcopy)))
501
502(defun format-common-tail (a b)
503 "Given two lists that have a common tail, return it.
504Compares with `equal', and returns the part of A that is equal to the
505equivalent part of B. If even the last items of the two are not equal,
506returns nil."
507 (let ((la (length a))
508 (lb (length b)))
509 ;; Make sure they are the same length
823139fb 510 (if (> la lb)
029894b9
BG
511 (setq a (nthcdr (- la lb) a))
512 (setq b (nthcdr (- lb la) b))))
513 (while (not (equal a b))
514 (setq a (cdr a)
515 b (cdr b)))
516 a)
517
f0a6c717
GM
518(defun format-proper-list-p (list)
519 "Return t if LIST is a proper list.
520A proper list is a list ending with a nil cdr, not with an atom "
521 (when (listp list)
522 (while (consp list)
523 (setq list (cdr list)))
524 (null list)))
525
029894b9
BG
526(defun format-reorder (items order)
527 "Arrange ITEMS to following partial ORDER.
528Elements of ITEMS equal to elements of ORDER will be rearranged to follow the
529ORDER. Unmatched items will go last."
530 (if order
531 (let ((item (member (car order) items)))
532 (if item
823139fb 533 (cons (car item)
029894b9
BG
534 (format-reorder (format-delq-cons item items)
535 (cdr order)))
536 (format-reorder items (cdr order))))
537 items))
538
539(put 'face 'format-list-valued t) ; These text-properties take values
540(put 'unknown 'format-list-valued t) ; that are lists, the elements of which
541 ; should be considered separately.
542 ; See format-deannotate-region and
543 ; format-annotate-region.
544
f3bbef87
GM
545;; This text property has list values, but they are treated atomically.
546
547(put 'display 'format-list-atomic-p t)
548
029894b9
BG
549;;;
550;;; Decoding
551;;;
552
553(defun format-deannotate-region (from to translations next-fn)
554 "Translate annotations in the region into text properties.
823139fb 555This sets text properties between FROM to TO as directed by the
029894b9
BG
556TRANSLATIONS and NEXT-FN arguments.
557
558NEXT-FN is a function that searches forward from point for an annotation.
559It should return a list of 4 elements: \(BEGIN END NAME POSITIVE). BEGIN and
560END are buffer positions bounding the annotation, NAME is the name searched
561for in TRANSLATIONS, and POSITIVE should be non-nil if this annotation marks
562the beginning of a region with some property, or nil if it ends the region.
563NEXT-FN should return nil if there are no annotations after point.
564
565The basic format of the TRANSLATIONS argument is described in the
566documentation for the `format-annotate-region' function. There are some
567additional things to keep in mind for decoding, though:
568
569When an annotation is found, the TRANSLATIONS list is searched for a
570text-property name and value that corresponds to that annotation. If the
571text-property has several annotations associated with it, it will be used only
572if the other annotations are also in effect at that point. The first match
573found whose annotations are all present is used.
574
575The text property thus determined is set to the value over the region between
576the opening and closing annotations. However, if the text-property name has a
577non-nil `format-list-valued' property, then the value will be consed onto the
578surrounding value of the property, rather than replacing that value.
579
580There are some special symbols that can be used in the \"property\" slot of
581the TRANSLATIONS list: PARAMETER and FUNCTION \(spelled in uppercase).
582Annotations listed under the pseudo-property PARAMETER are considered to be
583arguments of the immediately surrounding annotation; the text between the
584opening and closing parameter annotations is deleted from the buffer but saved
92308d3a
RS
585as a string.
586
587The surrounding annotation should be listed under the pseudo-property
588FUNCTION. Instead of inserting a text-property for this annotation,
589the function listed in the VALUE slot is called to make whatever
590changes are appropriate. It can also return a list of the form
591\(START LOC PROP VALUE) which specifies a property to put on. The
592function's first two arguments are the START and END locations, and
593the rest of the arguments are any PARAMETERs found in that region.
029894b9
BG
594
595Any annotations that are found by NEXT-FN but not defined by TRANSLATIONS
596are saved as values of the `unknown' text-property \(which is list-valued).
597The TRANSLATIONS list should usually contain an entry of the form
598 \(unknown \(nil format-annotate-value))
599to write these unknown annotations back into the file."
600 (save-excursion
601 (save-restriction
602 (narrow-to-region (point-min) to)
603 (goto-char from)
604 (let (next open-ans todo loc unknown-ans)
605 (while (setq next (funcall next-fn))
606 (let* ((loc (nth 0 next))
607 (end (nth 1 next))
608 (name (nth 2 next))
609 (positive (nth 3 next))
610 (found nil))
611
612 ;; Delete the annotation
613 (delete-region loc end)
22828206
RS
614 (cond
615 ;; Positive annotations are stacked, remembering location
616 (positive (setq open-ans (cons `(,name ((,loc . nil))) open-ans)))
617 ;; It is a negative annotation:
618 ;; Close the top annotation & add its text property.
619 ;; If the file's nesting is messed up, the close might not match
620 ;; the top thing on the open-annotations stack.
621 ;; If no matching annotation is open, just ignore the close.
622 ((not (assoc name open-ans))
623 (message "Extra closing annotation (%s) in file" name))
624 ;; If one is open, but not on the top of the stack, close
625 ;; the things in between as well. Set `found' when the real
626 ;; one is closed.
627 (t
628 (while (not found)
629 (let* ((top (car open-ans)) ; first on stack: should match.
630 (top-name (car top)) ; text property name
631 (top-extents (nth 1 top)) ; property regions
632 (params (cdr (cdr top))) ; parameters
633 (aalist translations)
634 (matched nil))
635 (if (equal name top-name)
636 (setq found t)
637 (message "Improper nesting in file."))
638 ;; Look through property names in TRANSLATIONS
639 (while aalist
640 (let ((prop (car (car aalist)))
641 (alist (cdr (car aalist))))
642 ;; And look through values for each property
643 (while alist
644 (let ((value (car (car alist)))
645 (ans (cdr (car alist))))
646 (if (member top-name ans)
647 ;; This annotation is listed, but still have to
648 ;; check if multiple annotations are satisfied
649 (if (member nil (mapcar (lambda (r)
650 (assoc r open-ans))
651 ans))
652 nil ; multiple ans not satisfied
653 ;; If there are multiple annotations going
654 ;; into one text property, split up the other
655 ;; annotations so they apply individually to
656 ;; the other regions.
657 (setcdr (car top-extents) loc)
658 (let ((to-split ans) this-one extents)
659 (while to-split
660 (setq this-one
661 (assoc (car to-split) open-ans)
662 extents (nth 1 this-one))
663 (if (not (eq this-one top))
664 (setcar (cdr this-one)
665 (format-subtract-regions
666 extents top-extents)))
667 (setq to-split (cdr to-split))))
668 ;; Set loop variables to nil so loop
669 ;; will exit.
670 (setq alist nil aalist nil matched t
671 ;; pop annotation off stack.
672 open-ans (cdr open-ans))
673 (let ((extents top-extents)
674 (start (car (car top-extents)))
675 (loc (cdr (car top-extents))))
676 (while extents
677 (cond
678 ;; Check for pseudo-properties
679 ((eq prop 'PARAMETER)
680 ;; A parameter of the top open ann:
681 ;; delete text and use as arg.
682 (if open-ans
683 ;; (If nothing open, discard).
684 (setq open-ans
685 (cons
686 (append (car open-ans)
687 (list
688 (buffer-substring
689 start loc)))
690 (cdr open-ans))))
691 (delete-region start loc))
692 ((eq prop 'FUNCTION)
693 ;; Not a property, but a function.
694 (let ((rtn
695 (apply value start loc params)))
696 (if rtn (setq todo (cons rtn todo)))))
697 (t
698 ;; Normal property/value pair
699 (setq todo
700 (cons (list start loc prop value)
701 todo))))
702 (setq extents (cdr extents)
703 start (car (car extents))
704 loc (cdr (car extents))))))))
705 (setq alist (cdr alist))))
706 (setq aalist (cdr aalist)))
707 (if (not matched)
029894b9
BG
708 ;; Didn't find any match for the annotation:
709 ;; Store as value of text-property `unknown'.
22828206
RS
710 (let ((extents top-extents)
711 (start (car (car top-extents)))
0b140219 712 (loc (or (cdr (car top-extents)) loc)))
22828206
RS
713 (while extents
714 (setq open-ans (cdr open-ans)
715 todo (cons (list start loc 'unknown top-name)
716 todo)
717 unknown-ans (cons name unknown-ans)
718 extents (cdr extents)
719 start (car (car extents))
720 loc (cdr (car extents))))))))))))
029894b9
BG
721
722 ;; Once entire file has been scanned, add the properties.
723 (while todo
724 (let* ((item (car todo))
725 (from (nth 0 item))
726 (to (nth 1 item))
727 (prop (nth 2 item))
728 (val (nth 3 item)))
22828206
RS
729
730 (if (numberp val) ; add to ambient value if numeric
731 (format-property-increment-region from to prop val 0)
732 (put-text-property
029894b9 733 from to prop
22828206 734 (cond ((get prop 'format-list-valued) ; value gets consed onto
029894b9
BG
735 ; list-valued properties
736 (let ((prev (get-text-property from prop)))
737 (cons val (if (listp prev) prev (list prev)))))
22828206 738 (t val))))) ; normally, just set to val.
029894b9 739 (setq todo (cdr todo)))
22828206 740
029894b9
BG
741 (if unknown-ans
742 (message "Unknown annotations: %s" unknown-ans))))))
743
22828206 744(defun format-subtract-regions (minu subtra)
823139fb 745 "Remove from the regions in MINUend the regions in SUBTRAhend.
d0fc7e3b 746A region is a dotted pair (FROM . TO). Both parameters are lists of
823139fb
DL
747regions. Each list must contain nonoverlapping, noncontiguous
748regions, in descending order. The result is also nonoverlapping,
749noncontiguous, and in descending order. The first element of MINUEND
750can have a cdr of nil, indicating that the end of that region is not
751yet known."
22828206
RS
752 (let* ((minuend (copy-alist minu))
753 (subtrahend (copy-alist subtra))
754 (m (car minuend))
755 (s (car subtrahend))
756 results)
757 (while (and minuend subtrahend)
823139fb 758 (cond
22828206
RS
759 ;; The minuend starts after the subtrahend ends; keep it.
760 ((> (car m) (cdr s))
761 (setq results (cons m results)
762 minuend (cdr minuend)
763 m (car minuend)))
764 ;; The minuend extends beyond the end of the subtrahend. Chop it off.
765 ((or (null (cdr m)) (> (cdr m) (cdr s)))
766 (setq results (cons (cons (1+ (cdr s)) (cdr m)) results))
767 (setcdr m (cdr s)))
768 ;; The subtrahend starts after the minuend ends; throw it away.
769 ((< (cdr m) (car s))
770 (setq subtrahend (cdr subtrahend) s (car subtrahend)))
771 ;; The subtrahend extends beyond the end of the minuend. Chop it off.
772 (t ;(<= (cdr m) (cdr s)))
773 (if (>= (car m) (car s))
774 (setq minuend (cdr minuend) m (car minuend))
775 (setcdr m (1- (car s)))
776 (setq subtrahend (cdr subtrahend) s (car subtrahend))))))
777 (nconc (nreverse results) minuend)))
778
779;; This should probably go somewhere other than format.el. Then again,
780;; indent.el has alter-text-property. NOTE: We can also use
781;; next-single-property-change instead of text-property-not-all, but then
782;; we have to see if we passed TO.
783(defun format-property-increment-region (from to prop delta default)
823139fb
DL
784 "Over the region between FROM and TO increment property PROP by amount DELTA.
785DELTA may be negative. If property PROP is nil anywhere
22828206
RS
786in the region, it is treated as though it were DEFAULT."
787 (let ((cur from) val newval next)
788 (while cur
789 (setq val (get-text-property cur prop)
790 newval (+ (or val default) delta)
791 next (text-property-not-all cur to prop val))
792 (put-text-property cur (or next to) prop newval)
793 (setq cur next))))
794
029894b9
BG
795;;;
796;;; Encoding
797;;;
798
799(defun format-insert-annotations (list &optional offset)
800 "Apply list of annotations to buffer as `write-region' would.
801Inserts each element of the given LIST of buffer annotations at its
802appropriate place. Use second arg OFFSET if the annotations' locations are
803not relative to the beginning of the buffer: annotations will be inserted
804at their location-OFFSET+1 \(ie, the offset is treated as the character number
805of the first character in the buffer)."
823139fb 806 (if (not offset)
029894b9
BG
807 (setq offset 0)
808 (setq offset (1- offset)))
809 (let ((l (reverse list)))
810 (while l
811 (goto-char (- (car (car l)) offset))
812 (insert (cdr (car l)))
813 (setq l (cdr l)))))
814
815(defun format-annotate-value (old new)
0b1ce6ba 816 "Return OLD and NEW as a \(CLOSE . OPEN) annotation pair.
029894b9
BG
817Useful as a default function for TRANSLATIONS alist when the value of the text
818property is the name of the annotation that you want to use, as it is for the
819`unknown' text property."
820 (cons (if old (list old))
821 (if new (list new))))
822
823139fb 823(defun format-annotate-region (from to translations format-fn ignore)
029894b9
BG
824 "Generate annotations for text properties in the region.
825Searches for changes between FROM and TO, and describes them with a list of
826annotations as defined by alist TRANSLATIONS and FORMAT-FN. IGNORE lists text
827properties not to consider; any text properties that are neither ignored nor
828listed in TRANSLATIONS are warned about.
829If you actually want to modify the region, give the return value of this
830function to `format-insert-annotations'.
831
832Format of the TRANSLATIONS argument:
833
834Each element is a list whose car is a PROPERTY, and the following
d0fc7e3b
RS
835elements have the form (VALUE ANNOTATIONS...).
836Whenever the property takes on the value VALUE, the annotations
029894b9
BG
837\(as formatted by FORMAT-FN) are inserted into the file.
838When the property stops having that value, the matching negated annotation
839will be inserted \(it may actually be closed earlier and reopened, if
823139fb 840necessary, to keep proper nesting).
029894b9 841
d0fc7e3b 842If VALUE is a list, then each element of the list is dealt with
029894b9
BG
843separately.
844
845If a VALUE is numeric, then it is assumed that there is a single annotation
846and each occurrence of it increments the value of the property by that number.
847Thus, given the entry \(left-margin \(4 \"indent\")), if the left margin
848changes from 4 to 12, two <indent> annotations will be generated.
849
850If the VALUE is nil, then instead of annotations, a function should be
851specified. This function is used as a default: it is called for all
852transitions not explicitly listed in the table. The function is called with
853two arguments, the OLD and NEW values of the property. It should return
d0fc7e3b 854a cons cell (CLOSE . OPEN) as `format-annotate-single-property-change' does.
029894b9 855
d0fc7e3b 856The same TRANSLATIONS structure can be used in reverse for reading files."
029894b9
BG
857 (let ((all-ans nil) ; All annotations - becomes return value
858 (open-ans nil) ; Annotations not yet closed
859 (loc nil) ; Current location
860 (not-found nil)) ; Properties that couldn't be saved
861 (while (or (null loc)
862 (and (setq loc (next-property-change loc nil to))
863 (< loc to)))
864 (or loc (setq loc from))
823139fb 865 (let* ((ans (format-annotate-location loc (= loc from) ignore translations))
029894b9
BG
866 (neg-ans (format-reorder (aref ans 0) open-ans))
867 (pos-ans (aref ans 1))
868 (ignored (aref ans 2)))
869 (setq not-found (append ignored not-found)
870 ignore (append ignored ignore))
871 ;; First do the negative (closing) annotations
872 (while neg-ans
873 ;; Check if it's missing. This can happen (eg, a numeric property
874 ;; going negative can generate closing annotations before there are
875 ;; any open). Warn user & ignore.
876 (if (not (member (car neg-ans) open-ans))
877 (message "Can't close %s: not open." (car neg-ans))
878 (while (not (equal (car neg-ans) (car open-ans)))
879 ;; To close anno. N, need to first close ans 1 to N-1,
880 ;; remembering to re-open them later.
881 (setq pos-ans (cons (car open-ans) pos-ans))
823139fb 882 (setq all-ans
029894b9
BG
883 (cons (cons loc (funcall format-fn (car open-ans) nil))
884 all-ans))
885 (setq open-ans (cdr open-ans)))
886 ;; Now remove the one we're really interested in from open list.
887 (setq open-ans (cdr open-ans))
888 ;; And put the closing annotation here.
823139fb 889 (setq all-ans
029894b9
BG
890 (cons (cons loc (funcall format-fn (car neg-ans) nil))
891 all-ans)))
892 (setq neg-ans (cdr neg-ans)))
893 ;; Now deal with positive (opening) annotations
894 (let ((p pos-ans))
895 (while pos-ans
896 (setq open-ans (cons (car pos-ans) open-ans))
823139fb 897 (setq all-ans
029894b9
BG
898 (cons (cons loc (funcall format-fn (car pos-ans) t))
899 all-ans))
900 (setq pos-ans (cdr pos-ans))))))
901
902 ;; Close any annotations still open
903 (while open-ans
823139fb 904 (setq all-ans
029894b9
BG
905 (cons (cons to (funcall format-fn (car open-ans) nil))
906 all-ans))
907 (setq open-ans (cdr open-ans)))
908 (if not-found
909 (message "These text properties could not be saved:\n %s"
910 not-found))
911 (nreverse all-ans)))
912
913;;; Internal functions for format-annotate-region.
914
823139fb
DL
915(defun format-annotate-location (loc all ignore translations)
916 "Return annotation(s) needed at location LOC.
029894b9
BG
917This includes any properties that change between LOC-1 and LOC.
918If ALL is true, don't look at previous location, but generate annotations for
919all non-nil properties.
920Third argument IGNORE is a list of text-properties not to consider.
d0fc7e3b 921Use the TRANSLATIONS alist (see `format-annotate-region' for doc).
029894b9
BG
922
923Return value is a vector of 3 elements:
d0fc7e3b
RS
9241. List of annotations to close
9252. List of annotations to open.
9263. List of properties that were ignored or couldn't be annotated.
927
928The annotations in lists 1 and 2 need not be strings.
929They can be whatever the FORMAT-FN in `format-annotate-region'
930can handle. If that is `enriched-make-annotation', they can be
931either strings, or lists of the form (PARAMETER VALUE)."
029894b9
BG
932 (let* ((prev-loc (1- loc))
933 (before-plist (if all nil (text-properties-at prev-loc)))
934 (after-plist (text-properties-at loc))
935 p negatives positives prop props not-found)
936 ;; make list of all property names involved
937 (setq p before-plist)
938 (while p
939 (if (not (memq (car p) props))
940 (setq props (cons (car p) props)))
941 (setq p (cdr (cdr p))))
942 (setq p after-plist)
943 (while p
944 (if (not (memq (car p) props))
945 (setq props (cons (car p) props)))
946 (setq p (cdr (cdr p))))
947
948 (while props
949 (setq prop (car props)
950 props (cdr props))
951 (if (memq prop ignore)
952 nil ; If it's been ignored before, ignore it now.
953 (let ((before (if all nil (car (cdr (memq prop before-plist)))))
954 (after (car (cdr (memq prop after-plist)))))
955 (if (equal before after)
956 nil ; no change; ignore
957 (let ((result (format-annotate-single-property-change
823139fb 958 prop before after translations)))
029894b9
BG
959 (if (not result)
960 (setq not-found (cons prop not-found))
961 (setq negatives (nconc negatives (car result))
962 positives (nconc positives (cdr result)))))))))
963 (vector negatives positives not-found)))
964
d0fc7e3b 965(defun format-annotate-single-property-change (prop old new translations)
823139fb 966 "Return annotations for property PROP changing from OLD to NEW.
d0fc7e3b
RS
967These are searched for in the translations alist TRANSLATIONS
968 (see `format-annotate-region' for the format).
029894b9
BG
969If NEW does not appear in the list, but there is a default function, then that
970function is called.
d0fc7e3b
RS
971Returns a cons of the form (CLOSE . OPEN)
972where CLOSE is a list of annotations to close
973and OPEN is a list of annotations to open.
974
975The annotations in CLOSE and OPEN need not be strings.
976They can be whatever the FORMAT-FN in `format-annotate-region'
977can handle. If that is `enriched-make-annotation', they can be
978either strings, or lists of the form (PARAMETER VALUE)."
979
980 (let ((prop-alist (cdr (assoc prop translations)))
029894b9
BG
981 default)
982 (if (not prop-alist)
983 nil
029894b9 984 ;; If either old or new is a list, have to treat both that way.
f0a6c717 985 (if (and (or (listp old) (listp new))
f3bbef87 986 (not (get prop 'format-list-atomic-p)))
f0a6c717
GM
987 (if (or (not (format-proper-list-p old))
988 (not (format-proper-list-p new)))
989 (format-annotate-atomic-property-change prop-alist old new)
990 (let* ((old (if (listp old) old (list old)))
991 (new (if (listp new) new (list new)))
992 (tail (format-common-tail old new))
993 close open)
994 (while old
995 (setq close
996 (append (car (format-annotate-atomic-property-change
997 prop-alist (car old) nil))
998 close)
999 old (cdr old)))
1000 (while new
1001 (setq open
1002 (append (cdr (format-annotate-atomic-property-change
1003 prop-alist nil (car new)))
1004 open)
1005 new (cdr new)))
1006 (format-make-relatively-unique close open)))
029894b9
BG
1007 (format-annotate-atomic-property-change prop-alist old new)))))
1008
1009(defun format-annotate-atomic-property-change (prop-alist old new)
1010 "Internal function annotate a single property change.
d0fc7e3b 1011PROP-ALIST is the relevant element of a TRANSLATIONS list.
029894b9 1012OLD and NEW are the values."
e5a60108
RS
1013 (let (num-ann)
1014 ;; If old and new values are numbers,
1015 ;; look for a number in PROP-ALIST.
5f525595
RS
1016 (if (and (or (null old) (numberp old))
1017 (or (null new) (numberp new)))
e5a60108
RS
1018 (progn
1019 (setq num-ann prop-alist)
1020 (while (and num-ann (not (numberp (car (car num-ann)))))
1021 (setq num-ann (cdr num-ann)))))
1022 (if num-ann
5f525595 1023 ;; Numerical annotation - use difference
34fbecde
RS
1024 (progn
1025 ;; If property is numeric, nil means 0
1026 (cond ((and (numberp old) (null new))
1027 (setq new 0))
1028 ((and (numberp new) (null old))
1029 (setq old 0)))
1030
1031 (let* ((entry (car num-ann))
1032 (increment (car entry))
1033 (n (ceiling (/ (float (- new old)) (float increment))))
1034 (anno (car (cdr entry))))
1035 (if (> n 0)
1036 (cons nil (make-list n anno))
1037 (cons (make-list (- n) anno) nil))))
e5a60108
RS
1038
1039 ;; Standard annotation
1040 (let ((close (and old (cdr (assoc old prop-alist))))
029894b9
BG
1041 (open (and new (cdr (assoc new prop-alist)))))
1042 (if (or close open)
1043 (format-make-relatively-unique close open)
1044 ;; Call "Default" function, if any
1045 (let ((default (assq nil prop-alist)))
1046 (if default
1047 (funcall (car (cdr default)) old new))))))))
1048
c84cf2bf 1049(provide 'format)
823139fb
DL
1050
1051;;; format.el ends here