(format-alist): Doc fix.
[bpt/emacs.git] / lisp / format.el
CommitLineData
a8581027 1;;; format.el --- read and save files in multiple formats
b578f267 2
7bfd055d
GM
3;; Copyright (C) 1994, 1995, 1997, 1999, 2001, 2002, 2003, 2004, 2005,
4;; 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
029894b9 5
823139fb 6;; Author: Boris Goldowsky <boris@gnu.org>
029894b9
BG
7
8;; This file is part of GNU Emacs.
9
eb3fa2cf 10;; GNU Emacs is free software: you can redistribute it and/or modify
029894b9 11;; it under the terms of the GNU General Public License as published by
eb3fa2cf
GM
12;; the Free Software Foundation, either version 3 of the License, or
13;; (at your option) any later version.
b578f267 14
029894b9
BG
15;; GNU Emacs is distributed in the hope that it will be useful,
16;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18;; GNU General Public License for more details.
b578f267 19
029894b9 20;; You should have received a copy of the GNU General Public License
eb3fa2cf 21;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
029894b9
BG
22
23;;; Commentary:
b578f267
EN
24
25;; This file defines a unified mechanism for saving & loading files stored
26;; in different formats. `format-alist' contains information that directs
029894b9 27;; Emacs to call an encoding or decoding function when reading or writing
823139fb 28;; files that match certain conditions.
029894b9 29;;
b578f267
EN
30;; When a file is visited, its format is determined by matching the
31;; beginning of the file against regular expressions stored in
32;; `format-alist'. If this fails, you can manually translate the buffer
33;; using `format-decode-buffer'. In either case, the formats used are
34;; listed in the variable `buffer-file-format', and become the default
35;; format for saving the buffer. To save a buffer in a different format,
36;; change this variable, or use `format-write-file'.
029894b9
BG
37;;
38;; Auto-save files are normally created in the same format as the visited
4a2255ad 39;; file, but the variable `buffer-auto-save-file-format' can be set to a
b578f267
EN
40;; particularly fast or otherwise preferred format to be used for
41;; auto-saving (or nil to do no encoding on auto-save files, but then you
42;; risk losing any text-properties in the buffer).
029894b9 43;;
b578f267
EN
44;; You can manually translate a buffer into or out of a particular format
45;; with the functions `format-encode-buffer' and `format-decode-buffer'.
46;; To translate just the region use the functions `format-encode-region'
823139fb 47;; and `format-decode-region'.
029894b9 48;;
b578f267
EN
49;; You can define a new format by writing the encoding and decoding
50;; functions, and adding an entry to `format-alist'. See enriched.el for
51;; an example of how to implement a file format. There are various
52;; functions defined in this file that may be useful for writing the
53;; encoding and decoding functions:
54;; * `format-annotate-region' and `format-deannotate-region' allow a
55;; single alist of information to be used for encoding and decoding.
56;; The alist defines a correspondence between strings in the file
57;; ("annotations") and text-properties in the buffer.
029894b9
BG
58;; * `format-replace-strings' is similarly useful for doing simple
59;; string->string translations in a reversible manner.
60
b578f267
EN
61;;; Code:
62
029894b9 63(put 'buffer-file-format 'permanent-local t)
4a2255ad 64(put 'buffer-auto-save-file-format 'permanent-local t)
029894b9 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 (TeX "TeX (encoding)"
75 nil
c84cf2bf 76 iso-tex2iso iso-iso2tex t nil)
823139fb
DL
77 (gtex "German TeX (encoding)"
78 nil
c84cf2bf 79 iso-gtex2iso iso-iso2gtex t nil)
f1d6fe69 80 (html "HTML/SGML \"ISO 8879:1986//ENTITIES Added Latin 1//EN\" (encoding)"
823139fb 81 nil
f1d6fe69 82 iso-sgml2iso iso-iso2sgml t nil)
823139fb
DL
83 (rot13 "rot13"
84 nil
c84cf2bf 85 "tr a-mn-z n-za-m" "tr a-mn-z n-za-m" t nil)
823139fb
DL
86 (duden "Duden Ersatzdarstellung"
87 nil
88 "diac" iso-iso2duden t nil)
89 (de646 "German ASCII (ISO 646)"
90 nil
271af233 91 "recode -f iso646-ge:latin1" "recode -f latin1:iso646-ge" t nil)
823139fb
DL
92 (denet "net German"
93 nil
c84cf2bf 94 iso-german iso-cvt-read-only t nil)
823139fb
DL
95 (esnet "net Spanish"
96 nil
8d0dd9d2 97 iso-spanish iso-cvt-read-only t nil))
029894b9 98 "List of information about understood file formats.
ce44e932
GM
99Elements are of the form
100\(NAME DOC-STR REGEXP FROM-FN TO-FN MODIFY MODE-FN PRESERVE).
d1ae737b 101
029894b9 102NAME is a symbol, which is stored in `buffer-file-format'.
d1ae737b 103
029894b9
BG
104DOC-STR should be a single line providing more information about the
105 format. It is currently unused, but in the future will be shown to
106 the user if they ask for more information.
d1ae737b 107
029894b9 108REGEXP is a regular expression to match against the beginning of the file;
7bfd055d
GM
109 it should match only files in that format. REGEXP may be nil, in
110 which case the format will never be applied automatically to a file.
111 Use this for formats that you only ever want to apply manually.
d1ae737b 112
12008438 113FROM-FN is called to decode files in that format; it takes two args, BEGIN
029894b9
BG
114 and END, and can make any modifications it likes, returning the new
115 end. It must make sure that the beginning of the file no longer
116 matches REGEXP, or else it will get called again.
d1ae737b
RS
117 Alternatively, FROM-FN can be a string, which specifies a shell command
118 (including options) to be used as a filter to perform the conversion.
119
12008438 120TO-FN is called to encode a region into that format; it takes three
d1abb42f
RS
121 arguments: BEGIN, END, and BUFFER. BUFFER is the original buffer that
122 the data being written came from, which the function could use, for
123 example, to find the values of local variables. TO-FN should either
124 return a list of annotations like `write-region-annotate-functions',
125 or modify the region and return the new end.
d1ae737b
RS
126 Alternatively, TO-FN can be a string, which specifies a shell command
127 (including options) to be used as a filter to perform the conversion.
128
d1abb42f
RS
129MODIFY, if non-nil, means the TO-FN wants to modify the region. If nil,
130 TO-FN will not make any changes but will instead return a list of
823139fb 131 annotations.
d1ae737b 132
3e224645
EZ
133MODE-FN, if specified, is called when visiting a file with that format.
134 It is called with a single positive argument, on the assumption
12008438 135 that this would turn on some minor mode.
34f30f79
RS
136
137PRESERVE, if non-nil, means that `format-write-file' should not remove
ec8dda2f 138 this format from `buffer-file-format'.")
7bfd055d 139;;;###autoload(put 'format-alist 'risky-local-variable t)
029894b9
BG
140
141;;; Basic Functions (called from Lisp)
142
c84cf2bf 143(defun format-encode-run-method (method from to &optional buffer)
12008438
TTN
144 "Translate using METHOD the text from FROM to TO.
145If METHOD is a string, it is a shell command (including options);
c84cf2bf
RS
146otherwise, it should be a Lisp function.
147BUFFER should be the buffer that the output originally came from."
148 (if (stringp method)
823139fb 149 (let ((error-buff (get-buffer-create "*Format Errors*"))
7cc45a35 150 (coding-system-for-read 'no-conversion)
823139fb
DL
151 format-alist)
152 (with-current-buffer error-buff
153 (widen)
154 (erase-buffer))
10fa0073
GM
155 (if (and (zerop (save-window-excursion
156 (shell-command-on-region from to method t t
157 error-buff)))
823139fb
DL
158 ;; gzip gives zero exit status with bad args, for instance.
159 (zerop (with-current-buffer error-buff
160 (buffer-size))))
161 (bury-buffer error-buff)
162 (switch-to-buffer-other-window error-buff)
10fa0073 163 (error "Format encoding failed")))
c84cf2bf
RS
164 (funcall method from to buffer)))
165
166(defun format-decode-run-method (method from to &optional buffer)
12008438
TTN
167 "Decode using METHOD the text from FROM to TO.
168If METHOD is a string, it is a shell command (including options); otherwise,
169it should be a Lisp function. Decoding is done for the given BUFFER."
c84cf2bf 170 (if (stringp method)
823139fb 171 (let ((error-buff (get-buffer-create "*Format Errors*"))
7cc45a35 172 (coding-system-for-write 'no-conversion)
823139fb
DL
173 format-alist)
174 (with-current-buffer error-buff
175 (widen)
176 (erase-buffer))
177 ;; We should perhaps go via a temporary buffer and copy it
178 ;; back, in case of errors.
179 (if (and (zerop (save-window-excursion
180 (shell-command-on-region (point-min) (point-max)
181 method t t
182 error-buff)))
183 ;; gzip gives zero exit status with bad args, for instance.
184 (zerop (with-current-buffer error-buff
185 (buffer-size))))
186 (bury-buffer error-buff)
187 (switch-to-buffer-other-window error-buff)
188 (error "Format decoding failed"))
c84cf2bf
RS
189 (point))
190 (funcall method from to)))
191
10fa0073 192(defun format-annotate-function (format from to orig-buf format-count)
823139fb 193 "Return annotations for writing region as FORMAT.
12008438
TTN
194FORMAT is a symbol naming one of the formats defined in `format-alist'.
195It must be a single symbol, not a list like `buffer-file-format'.
d1abb42f
RS
196FROM and TO delimit the region to be operated on in the current buffer.
197ORIG-BUF is the original buffer that the data came from.
10fa0073
GM
198
199FORMAT-COUNT is an integer specifying how many times this function has
200been called in the process of decoding ORIG-BUF.
201
12008438 202This function works like a function in `write-region-annotate-functions':
029894b9 203it either returns a list of annotations, or returns with a different buffer
10fa0073
GM
204current, which contains the modified text to write. In the latter case,
205this function's value is nil.
029894b9
BG
206
207For most purposes, consider using `format-encode-region' instead."
10fa0073
GM
208 ;; This function is called by write-region (actually
209 ;; build_annotations) for each element of buffer-file-format.
029894b9
BG
210 (let* ((info (assq format format-alist))
211 (to-fn (nth 4 info))
212 (modify (nth 5 info)))
213 (if to-fn
214 (if modify
215 ;; To-function wants to modify region. Copy to safe place.
10fa0073 216 (let ((copy-buf (get-buffer-create (format " *Format Temp %d*"
8a2f1176
RS
217 format-count)))
218 (sel-disp selective-display)
5258c763
JB
219 (multibyte enable-multibyte-characters)
220 (coding-system buffer-file-coding-system))
8a2f1176 221 (with-current-buffer copy-buf
1d1cf99f
CY
222 (set (make-local-variable
223 'write-region-post-annotation-function)
224 'kill-buffer)
8a2f1176 225 (setq selective-display sel-disp)
5258c763
JB
226 (set-buffer-multibyte multibyte)
227 (setq buffer-file-coding-system coding-system))
029894b9
BG
228 (copy-to-buffer copy-buf from to)
229 (set-buffer copy-buf)
230 (format-insert-annotations write-region-annotations-so-far from)
c84cf2bf 231 (format-encode-run-method to-fn (point-min) (point-max) orig-buf)
029894b9
BG
232 nil)
233 ;; Otherwise just call function, it will return annotations.
d1abb42f 234 (funcall to-fn from to orig-buf)))))
029894b9
BG
235
236(defun format-decode (format length &optional visit-flag)
237 ;; This function is called by insert-file-contents whenever a file is read.
238 "Decode text from any known FORMAT.
823139fb 239FORMAT is a symbol appearing in `format-alist' or a list of such symbols,
029894b9
BG
240or nil, in which case this function tries to guess the format of the data by
241matching against the regular expressions in `format-alist'. After a match is
242found and the region decoded, the alist is searched again from the beginning
243for another match.
244
245Second arg LENGTH is the number of characters following point to operate on.
246If optional third arg VISIT-FLAG is true, set `buffer-file-format'
be227f22
SM
247to the reverted list of formats used, and call any mode functions defined
248for those formats.
029894b9 249
12008438 250Return the new length of the decoded region.
029894b9
BG
251
252For most purposes, consider using `format-decode-region' instead."
253 (let ((mod (buffer-modified-p))
10fa0073 254 (begin (point))
029894b9 255 (end (+ (point) length)))
10fa0073
GM
256 (unwind-protect
257 (progn
258 ;; Don't record undo information for the decoding.
71296446 259
10fa0073
GM
260 (if (null format)
261 ;; Figure out which format it is in, remember list in `format'.
262 (let ((try format-alist))
263 (while try
264 (let* ((f (car try))
265 (regexp (nth 2 f))
266 (p (point)))
267 (if (and regexp (looking-at regexp)
268 (< (match-end 0) (+ begin length)))
269 (progn
be227f22 270 (push (car f) format)
10fa0073
GM
271 ;; Decode it
272 (if (nth 3 f)
273 (setq end (format-decode-run-method (nth 3 f) begin end)))
274 ;; Call visit function if required
275 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
276 ;; Safeguard against either of the functions changing pt.
277 (goto-char p)
278 ;; Rewind list to look for another format
279 (setq try format-alist))
280 (setq try (cdr try))))))
281 ;; Deal with given format(s)
282 (or (listp format) (setq format (list format)))
283 (let ((do format) f)
284 (while do
285 (or (setq f (assq (car do) format-alist))
6b61353c 286 (error "Unknown format %s" (car do)))
10fa0073
GM
287 ;; Decode:
288 (if (nth 3 f)
289 (setq end (format-decode-run-method (nth 3 f) begin end)))
290 ;; Call visit function if required
291 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
be227f22
SM
292 (setq do (cdr do))))
293 ;; Encode in the opposite order.
294 (setq format (reverse format)))
10fa0073
GM
295 (if visit-flag
296 (setq buffer-file-format format)))
71296446 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.
12008438
TTN
309If the format is not specified, attempt a regexp-based guess.
310Set `buffer-file-format' to the format used, and call any
311format-specific mode functions."
029894b9 312 (interactive
5b76833f 313 (list (format-read "Translate buffer from format (default guess): ")))
029894b9
BG
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)
5b76833f 324 (format-read "Translate region from format (default guess): ")))
029894b9
BG
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.
12008438 340FORMAT defaults to `buffer-file-format'. It is a symbol naming
029894b9 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 364
6b61353c 365(defun format-write-file (filename format &optional confirm)
84da0f17
GM
366 "Write current buffer into FILENAME, using a format based on FORMAT.
367Constructs the actual format starting from FORMAT, then appending
368any elements from the value of `buffer-file-format' with a non-nil
369`preserve' flag (see the documentation of `format-alist'), if they
370are not already present in FORMAT. It then updates `buffer-file-format'
371with this format, making it the default for future saves.
372
373If the buffer is already visiting a file, you can specify a
374directory name as FILENAME, to write a file of the same old name
08d55d13 375in that directory.
6b61353c 376
84da0f17
GM
377If optional third arg CONFIRM is non-nil, asks for confirmation before
378overwriting an existing file. Interactively, requires confirmation
379unless you supply a prefix argument."
029894b9
BG
380 (interactive
381 ;; Same interactive spec as write-file, plus format question.
382 (let* ((file (if buffer-file-name
383 (read-file-name "Write file: "
384 nil nil nil nil)
385 (read-file-name "Write file: "
386 (cdr (assq 'default-directory
387 (buffer-local-variables)))
388 nil nil (buffer-name))))
823139fb 389 (fmt (format-read (format "Write file `%s' in format: "
029894b9 390 (file-name-nondirectory file)))))
6b61353c 391 (list file fmt (not current-prefix-arg))))
34f30f79
RS
392 (let ((old-formats buffer-file-format)
393 preserve-formats)
394 (dolist (fmt old-formats)
395 (let ((aelt (assq fmt format-alist)))
396 (if (nth 7 aelt)
397 (push fmt preserve-formats))))
398 (setq buffer-file-format format)
399 (dolist (fmt preserve-formats)
400 (unless (memq fmt buffer-file-format)
401 (setq buffer-file-format (append buffer-file-format (list fmt))))))
6b61353c 402 (write-file filename confirm))
029894b9 403
ce6af52b 404(defun format-find-file (filename format)
823139fb 405 "Find the file FILENAME using data format FORMAT.
ce6af52b
KH
406If FORMAT is nil then do not do any format conversion."
407 (interactive
408 ;; Same interactive spec as write-file, plus format question.
409 (let* ((file (read-file-name "Find file: "))
823139fb 410 (fmt (format-read (format "Read file `%s' in format: "
ce6af52b
KH
411 (file-name-nondirectory file)))))
412 (list file fmt)))
413 (let ((format-alist nil))
414 (find-file filename))
415 (if format
416 (format-decode-buffer format)))
417
418(defun format-insert-file (filename format &optional beg end)
823139fb 419 "Insert the contents of file FILENAME using data format FORMAT.
ce6af52b
KH
420If FORMAT is nil then do not do any format conversion.
421The optional third and fourth arguments BEG and END specify
12008438 422the part (in bytes) of the file to read.
ce6af52b
KH
423
424The return value is like the value of `insert-file-contents':
6b61353c 425a list (ABSOLUTE-FILE-NAME SIZE)."
ce6af52b
KH
426 (interactive
427 ;; Same interactive spec as write-file, plus format question.
428 (let* ((file (read-file-name "Find file: "))
823139fb 429 (fmt (format-read (format "Read file `%s' in format: "
ce6af52b
KH
430 (file-name-nondirectory file)))))
431 (list file fmt)))
f70aa678
MR
432 (let (value size old-undo)
433 ;; Record only one undo entry for the insertion. Inhibit point-motion and
434 ;; modification hooks as with `insert-file-contents'.
435 (let ((inhibit-point-motion-hooks t)
436 (inhibit-modification-hooks t))
437 ;; Don't bind `buffer-undo-list' to t here to assert that
438 ;; `insert-file-contents' may record whether the buffer was unmodified
439 ;; before.
440 (let ((format-alist nil))
441 (setq value (insert-file-contents filename nil beg end))
442 (setq size (nth 1 value)))
443 (when (consp buffer-undo-list)
444 (let ((head (car buffer-undo-list)))
445 (when (and (consp head)
446 (equal (car head) (point))
447 (equal (cdr head) (+ (point) size)))
448 ;; Remove first entry from `buffer-undo-list', we shall insert
449 ;; another one below.
450 (setq old-undo (cdr buffer-undo-list)))))
451 (when format
452 (let ((buffer-undo-list t))
453 (setq size (format-decode format size)
454 value (list (car value) size)))
455 (unless (eq buffer-undo-list t)
456 (setq buffer-undo-list
457 (cons (cons (point) (+ (point) size)) old-undo)))))
458 (unless inhibit-modification-hooks
459 (run-hook-with-args 'after-change-functions (point) (+ (point) size) 0))
ce6af52b
KH
460 value))
461
029894b9
BG
462(defun format-read (&optional prompt)
463 "Read and return the name of a format.
464Return value is a list, like `buffer-file-format'; it may be nil.
465Formats are defined in `format-alist'. Optional arg is the PROMPT to use."
466 (let* ((table (mapcar (lambda (x) (list (symbol-name (car x))))
467 format-alist))
468 (ans (completing-read (or prompt "Format: ") table nil t)))
469 (if (not (equal "" ans)) (list (intern ans)))))
470
471
472;;;
473;;; Below are some functions that may be useful in writing encoding and
474;;; decoding functions for use in format-alist.
475;;;
476
477(defun format-replace-strings (alist &optional reverse beg end)
478 "Do multiple replacements on the buffer.
d0fc7e3b 479ALIST is a list of (FROM . TO) pairs, which should be proper arguments to
12008438
TTN
480`search-forward' and `replace-match', respectively.
481Optional second arg REVERSE, if non-nil, means the pairs are (TO . FROM),
482so that you can use the same list in both directions if it contains only
483literal strings.
823139fb 484Optional args BEG and END specify a region of the buffer on which to operate."
029894b9
BG
485 (save-excursion
486 (save-restriction
487 (or beg (setq beg (point-min)))
488 (if end (narrow-to-region (point-min) end))
489 (while alist
490 (let ((from (if reverse (cdr (car alist)) (car (car alist))))
2c3ac81d 491 (to (if reverse (car (car alist)) (cdr (car alist)))))
029894b9
BG
492 (goto-char beg)
493 (while (search-forward from nil t)
494 (goto-char (match-beginning 0))
495 (insert to)
496 (set-text-properties (- (point) (length to)) (point)
497 (text-properties-at (point)))
498 (delete-region (point) (+ (point) (- (match-end 0)
499 (match-beginning 0)))))
500 (setq alist (cdr alist)))))))
501
502;;; Some list-manipulation functions that we need.
503
504(defun format-delq-cons (cons list)
823139fb
DL
505 "Remove the given CONS from LIST by side effect and return the new LIST.
506Since CONS could be the first element of LIST, write
507`\(setq foo \(format-delq-cons element foo))' to be sure of changing
508the value of `foo'."
029894b9
BG
509 (if (eq cons list)
510 (cdr list)
511 (let ((p list))
512 (while (not (eq (cdr p) cons))
e8af40ee 513 (if (null p) (error "format-delq-cons: not an element"))
029894b9
BG
514 (setq p (cdr p)))
515 ;; Now (cdr p) is the cons to delete
516 (setcdr p (cdr cons))
517 list)))
71296446 518
029894b9
BG
519(defun format-make-relatively-unique (a b)
520 "Delete common elements of lists A and B, return as pair.
12008438 521Compare using `equal'."
029894b9
BG
522 (let* ((acopy (copy-sequence a))
523 (bcopy (copy-sequence b))
524 (tail acopy))
525 (while tail
526 (let ((dup (member (car tail) bcopy))
527 (next (cdr tail)))
528 (if dup (setq acopy (format-delq-cons tail acopy)
529 bcopy (format-delq-cons dup bcopy)))
530 (setq tail next)))
531 (cons acopy bcopy)))
532
533(defun format-common-tail (a b)
534 "Given two lists that have a common tail, return it.
12008438 535Compare with `equal', and return the part of A that is equal to the
029894b9 536equivalent part of B. If even the last items of the two are not equal,
12008438 537return nil."
029894b9
BG
538 (let ((la (length a))
539 (lb (length b)))
540 ;; Make sure they are the same length
823139fb 541 (if (> la lb)
029894b9
BG
542 (setq a (nthcdr (- la lb) a))
543 (setq b (nthcdr (- lb la) b))))
544 (while (not (equal a b))
545 (setq a (cdr a)
546 b (cdr b)))
547 a)
548
f0a6c717
GM
549(defun format-proper-list-p (list)
550 "Return t if LIST is a proper list.
551A proper list is a list ending with a nil cdr, not with an atom "
552 (when (listp list)
553 (while (consp list)
554 (setq list (cdr list)))
555 (null list)))
556
029894b9 557(defun format-reorder (items order)
12008438
TTN
558 "Arrange ITEMS to follow partial ORDER.
559Elements of ITEMS equal to elements of ORDER will be rearranged
560to follow the ORDER. Unmatched items will go last."
029894b9
BG
561 (if order
562 (let ((item (member (car order) items)))
563 (if item
823139fb 564 (cons (car item)
029894b9
BG
565 (format-reorder (format-delq-cons item items)
566 (cdr order)))
567 (format-reorder items (cdr order))))
568 items))
569
570(put 'face 'format-list-valued t) ; These text-properties take values
571(put 'unknown 'format-list-valued t) ; that are lists, the elements of which
572 ; should be considered separately.
573 ; See format-deannotate-region and
574 ; format-annotate-region.
575
f3bbef87
GM
576;; This text property has list values, but they are treated atomically.
577
578(put 'display 'format-list-atomic-p t)
579
029894b9
BG
580;;;
581;;; Decoding
582;;;
583
584(defun format-deannotate-region (from to translations next-fn)
585 "Translate annotations in the region into text properties.
823139fb 586This sets text properties between FROM to TO as directed by the
029894b9
BG
587TRANSLATIONS and NEXT-FN arguments.
588
589NEXT-FN is a function that searches forward from point for an annotation.
590It should return a list of 4 elements: \(BEGIN END NAME POSITIVE). BEGIN and
591END are buffer positions bounding the annotation, NAME is the name searched
592for in TRANSLATIONS, and POSITIVE should be non-nil if this annotation marks
593the beginning of a region with some property, or nil if it ends the region.
594NEXT-FN should return nil if there are no annotations after point.
595
596The basic format of the TRANSLATIONS argument is described in the
597documentation for the `format-annotate-region' function. There are some
598additional things to keep in mind for decoding, though:
599
600When an annotation is found, the TRANSLATIONS list is searched for a
601text-property name and value that corresponds to that annotation. If the
602text-property has several annotations associated with it, it will be used only
603if the other annotations are also in effect at that point. The first match
604found whose annotations are all present is used.
605
606The text property thus determined is set to the value over the region between
607the opening and closing annotations. However, if the text-property name has a
608non-nil `format-list-valued' property, then the value will be consed onto the
609surrounding value of the property, rather than replacing that value.
610
611There are some special symbols that can be used in the \"property\" slot of
612the TRANSLATIONS list: PARAMETER and FUNCTION \(spelled in uppercase).
613Annotations listed under the pseudo-property PARAMETER are considered to be
614arguments of the immediately surrounding annotation; the text between the
615opening and closing parameter annotations is deleted from the buffer but saved
92308d3a
RS
616as a string.
617
618The surrounding annotation should be listed under the pseudo-property
619FUNCTION. Instead of inserting a text-property for this annotation,
620the function listed in the VALUE slot is called to make whatever
621changes are appropriate. It can also return a list of the form
622\(START LOC PROP VALUE) which specifies a property to put on. The
623function's first two arguments are the START and END locations, and
624the rest of the arguments are any PARAMETERs found in that region.
029894b9
BG
625
626Any annotations that are found by NEXT-FN but not defined by TRANSLATIONS
627are saved as values of the `unknown' text-property \(which is list-valued).
628The TRANSLATIONS list should usually contain an entry of the form
629 \(unknown \(nil format-annotate-value))
630to write these unknown annotations back into the file."
631 (save-excursion
632 (save-restriction
633 (narrow-to-region (point-min) to)
634 (goto-char from)
635 (let (next open-ans todo loc unknown-ans)
636 (while (setq next (funcall next-fn))
637 (let* ((loc (nth 0 next))
638 (end (nth 1 next))
639 (name (nth 2 next))
640 (positive (nth 3 next))
641 (found nil))
642
643 ;; Delete the annotation
644 (delete-region loc end)
22828206
RS
645 (cond
646 ;; Positive annotations are stacked, remembering location
be227f22 647 (positive (push `(,name ((,loc . nil))) open-ans))
22828206
RS
648 ;; It is a negative annotation:
649 ;; Close the top annotation & add its text property.
650 ;; If the file's nesting is messed up, the close might not match
651 ;; the top thing on the open-annotations stack.
652 ;; If no matching annotation is open, just ignore the close.
653 ((not (assoc name open-ans))
654 (message "Extra closing annotation (%s) in file" name))
655 ;; If one is open, but not on the top of the stack, close
656 ;; the things in between as well. Set `found' when the real
657 ;; one is closed.
658 (t
659 (while (not found)
660 (let* ((top (car open-ans)) ; first on stack: should match.
661 (top-name (car top)) ; text property name
662 (top-extents (nth 1 top)) ; property regions
663 (params (cdr (cdr top))) ; parameters
664 (aalist translations)
665 (matched nil))
666 (if (equal name top-name)
667 (setq found t)
668 (message "Improper nesting in file."))
669 ;; Look through property names in TRANSLATIONS
670 (while aalist
671 (let ((prop (car (car aalist)))
672 (alist (cdr (car aalist))))
673 ;; And look through values for each property
674 (while alist
675 (let ((value (car (car alist)))
676 (ans (cdr (car alist))))
677 (if (member top-name ans)
678 ;; This annotation is listed, but still have to
679 ;; check if multiple annotations are satisfied
680 (if (member nil (mapcar (lambda (r)
681 (assoc r open-ans))
682 ans))
683 nil ; multiple ans not satisfied
684 ;; If there are multiple annotations going
685 ;; into one text property, split up the other
686 ;; annotations so they apply individually to
687 ;; the other regions.
688 (setcdr (car top-extents) loc)
689 (let ((to-split ans) this-one extents)
690 (while to-split
691 (setq this-one
692 (assoc (car to-split) open-ans)
693 extents (nth 1 this-one))
694 (if (not (eq this-one top))
695 (setcar (cdr this-one)
696 (format-subtract-regions
697 extents top-extents)))
698 (setq to-split (cdr to-split))))
699 ;; Set loop variables to nil so loop
700 ;; will exit.
701 (setq alist nil aalist nil matched t
702 ;; pop annotation off stack.
703 open-ans (cdr open-ans))
704 (let ((extents top-extents)
705 (start (car (car top-extents)))
706 (loc (cdr (car top-extents))))
707 (while extents
708 (cond
709 ;; Check for pseudo-properties
710 ((eq prop 'PARAMETER)
711 ;; A parameter of the top open ann:
712 ;; delete text and use as arg.
713 (if open-ans
714 ;; (If nothing open, discard).
715 (setq open-ans
716 (cons
717 (append (car open-ans)
718 (list
719 (buffer-substring
720 start loc)))
721 (cdr open-ans))))
722 (delete-region start loc))
723 ((eq prop 'FUNCTION)
724 ;; Not a property, but a function.
725 (let ((rtn
726 (apply value start loc params)))
be227f22 727 (if rtn (push rtn todo))))
22828206
RS
728 (t
729 ;; Normal property/value pair
730 (setq todo
731 (cons (list start loc prop value)
732 todo))))
733 (setq extents (cdr extents)
734 start (car (car extents))
735 loc (cdr (car extents))))))))
736 (setq alist (cdr alist))))
737 (setq aalist (cdr aalist)))
738 (if (not matched)
029894b9
BG
739 ;; Didn't find any match for the annotation:
740 ;; Store as value of text-property `unknown'.
22828206
RS
741 (let ((extents top-extents)
742 (start (car (car top-extents)))
0b140219 743 (loc (or (cdr (car top-extents)) loc)))
22828206
RS
744 (while extents
745 (setq open-ans (cdr open-ans)
746 todo (cons (list start loc 'unknown top-name)
747 todo)
748 unknown-ans (cons name unknown-ans)
749 extents (cdr extents)
750 start (car (car extents))
751 loc (cdr (car extents))))))))))))
029894b9
BG
752
753 ;; Once entire file has been scanned, add the properties.
754 (while todo
755 (let* ((item (car todo))
756 (from (nth 0 item))
757 (to (nth 1 item))
758 (prop (nth 2 item))
759 (val (nth 3 item)))
22828206
RS
760
761 (if (numberp val) ; add to ambient value if numeric
762 (format-property-increment-region from to prop val 0)
763 (put-text-property
029894b9 764 from to prop
22828206 765 (cond ((get prop 'format-list-valued) ; value gets consed onto
029894b9
BG
766 ; list-valued properties
767 (let ((prev (get-text-property from prop)))
768 (cons val (if (listp prev) prev (list prev)))))
22828206 769 (t val))))) ; normally, just set to val.
029894b9 770 (setq todo (cdr todo)))
22828206 771
029894b9
BG
772 (if unknown-ans
773 (message "Unknown annotations: %s" unknown-ans))))))
774
22828206 775(defun format-subtract-regions (minu subtra)
9c3054da 776 "Remove from the regions in MINUEND the regions in SUBTRAHEND.
d0fc7e3b 777A region is a dotted pair (FROM . TO). Both parameters are lists of
823139fb
DL
778regions. Each list must contain nonoverlapping, noncontiguous
779regions, in descending order. The result is also nonoverlapping,
780noncontiguous, and in descending order. The first element of MINUEND
781can have a cdr of nil, indicating that the end of that region is not
9c3054da
JB
782yet known.
783
784\(fn MINUEND SUBTRAHEND)"
22828206
RS
785 (let* ((minuend (copy-alist minu))
786 (subtrahend (copy-alist subtra))
787 (m (car minuend))
788 (s (car subtrahend))
789 results)
790 (while (and minuend subtrahend)
823139fb 791 (cond
22828206
RS
792 ;; The minuend starts after the subtrahend ends; keep it.
793 ((> (car m) (cdr s))
be227f22
SM
794 (push m results)
795 (setq minuend (cdr minuend)
22828206
RS
796 m (car minuend)))
797 ;; The minuend extends beyond the end of the subtrahend. Chop it off.
798 ((or (null (cdr m)) (> (cdr m) (cdr s)))
be227f22 799 (push (cons (1+ (cdr s)) (cdr m)) results)
22828206
RS
800 (setcdr m (cdr s)))
801 ;; The subtrahend starts after the minuend ends; throw it away.
802 ((< (cdr m) (car s))
803 (setq subtrahend (cdr subtrahend) s (car subtrahend)))
804 ;; The subtrahend extends beyond the end of the minuend. Chop it off.
805 (t ;(<= (cdr m) (cdr s)))
806 (if (>= (car m) (car s))
807 (setq minuend (cdr minuend) m (car minuend))
808 (setcdr m (1- (car s)))
809 (setq subtrahend (cdr subtrahend) s (car subtrahend))))))
810 (nconc (nreverse results) minuend)))
811
812;; This should probably go somewhere other than format.el. Then again,
813;; indent.el has alter-text-property. NOTE: We can also use
814;; next-single-property-change instead of text-property-not-all, but then
815;; we have to see if we passed TO.
816(defun format-property-increment-region (from to prop delta default)
12008438 817 "In the region from FROM to TO increment property PROP by amount DELTA.
823139fb 818DELTA may be negative. If property PROP is nil anywhere
22828206
RS
819in the region, it is treated as though it were DEFAULT."
820 (let ((cur from) val newval next)
821 (while cur
822 (setq val (get-text-property cur prop)
823 newval (+ (or val default) delta)
824 next (text-property-not-all cur to prop val))
825 (put-text-property cur (or next to) prop newval)
826 (setq cur next))))
827
029894b9
BG
828;;;
829;;; Encoding
830;;;
831
832(defun format-insert-annotations (list &optional offset)
833 "Apply list of annotations to buffer as `write-region' would.
12008438 834Insert each element of the given LIST of buffer annotations at its
029894b9
BG
835appropriate place. Use second arg OFFSET if the annotations' locations are
836not relative to the beginning of the buffer: annotations will be inserted
9c3054da
JB
837at their location-OFFSET+1 \(ie, the offset is treated as the position of
838the first character in the buffer)."
823139fb 839 (if (not offset)
029894b9
BG
840 (setq offset 0)
841 (setq offset (1- offset)))
842 (let ((l (reverse list)))
843 (while l
844 (goto-char (- (car (car l)) offset))
845 (insert (cdr (car l)))
846 (setq l (cdr l)))))
847
848(defun format-annotate-value (old new)
0b1ce6ba 849 "Return OLD and NEW as a \(CLOSE . OPEN) annotation pair.
029894b9
BG
850Useful as a default function for TRANSLATIONS alist when the value of the text
851property is the name of the annotation that you want to use, as it is for the
852`unknown' text property."
853 (cons (if old (list old))
854 (if new (list new))))
855
823139fb 856(defun format-annotate-region (from to translations format-fn ignore)
029894b9 857 "Generate annotations for text properties in the region.
12008438 858Search for changes between FROM and TO, and describe them with a list of
029894b9
BG
859annotations as defined by alist TRANSLATIONS and FORMAT-FN. IGNORE lists text
860properties not to consider; any text properties that are neither ignored nor
861listed in TRANSLATIONS are warned about.
862If you actually want to modify the region, give the return value of this
863function to `format-insert-annotations'.
864
865Format of the TRANSLATIONS argument:
866
867Each element is a list whose car is a PROPERTY, and the following
d0fc7e3b
RS
868elements have the form (VALUE ANNOTATIONS...).
869Whenever the property takes on the value VALUE, the annotations
029894b9
BG
870\(as formatted by FORMAT-FN) are inserted into the file.
871When the property stops having that value, the matching negated annotation
872will be inserted \(it may actually be closed earlier and reopened, if
823139fb 873necessary, to keep proper nesting).
029894b9 874
d0fc7e3b 875If VALUE is a list, then each element of the list is dealt with
029894b9
BG
876separately.
877
878If a VALUE is numeric, then it is assumed that there is a single annotation
879and each occurrence of it increments the value of the property by that number.
880Thus, given the entry \(left-margin \(4 \"indent\")), if the left margin
881changes from 4 to 12, two <indent> annotations will be generated.
882
883If the VALUE is nil, then instead of annotations, a function should be
884specified. This function is used as a default: it is called for all
885transitions not explicitly listed in the table. The function is called with
886two arguments, the OLD and NEW values of the property. It should return
d0fc7e3b 887a cons cell (CLOSE . OPEN) as `format-annotate-single-property-change' does.
029894b9 888
d0fc7e3b 889The same TRANSLATIONS structure can be used in reverse for reading files."
029894b9
BG
890 (let ((all-ans nil) ; All annotations - becomes return value
891 (open-ans nil) ; Annotations not yet closed
892 (loc nil) ; Current location
893 (not-found nil)) ; Properties that couldn't be saved
894 (while (or (null loc)
895 (and (setq loc (next-property-change loc nil to))
896 (< loc to)))
897 (or loc (setq loc from))
823139fb 898 (let* ((ans (format-annotate-location loc (= loc from) ignore translations))
029894b9
BG
899 (neg-ans (format-reorder (aref ans 0) open-ans))
900 (pos-ans (aref ans 1))
901 (ignored (aref ans 2)))
902 (setq not-found (append ignored not-found)
903 ignore (append ignored ignore))
904 ;; First do the negative (closing) annotations
905 (while neg-ans
906 ;; Check if it's missing. This can happen (eg, a numeric property
907 ;; going negative can generate closing annotations before there are
908 ;; any open). Warn user & ignore.
909 (if (not (member (car neg-ans) open-ans))
910 (message "Can't close %s: not open." (car neg-ans))
911 (while (not (equal (car neg-ans) (car open-ans)))
912 ;; To close anno. N, need to first close ans 1 to N-1,
913 ;; remembering to re-open them later.
be227f22 914 (push (car open-ans) pos-ans)
823139fb 915 (setq all-ans
029894b9
BG
916 (cons (cons loc (funcall format-fn (car open-ans) nil))
917 all-ans))
918 (setq open-ans (cdr open-ans)))
919 ;; Now remove the one we're really interested in from open list.
920 (setq open-ans (cdr open-ans))
921 ;; And put the closing annotation here.
be227f22
SM
922 (push (cons loc (funcall format-fn (car neg-ans) nil))
923 all-ans))
029894b9
BG
924 (setq neg-ans (cdr neg-ans)))
925 ;; Now deal with positive (opening) annotations
926 (let ((p pos-ans))
927 (while pos-ans
be227f22
SM
928 (push (car pos-ans) open-ans)
929 (push (cons loc (funcall format-fn (car pos-ans) t))
930 all-ans)
029894b9
BG
931 (setq pos-ans (cdr pos-ans))))))
932
933 ;; Close any annotations still open
934 (while open-ans
823139fb 935 (setq all-ans
029894b9
BG
936 (cons (cons to (funcall format-fn (car open-ans) nil))
937 all-ans))
938 (setq open-ans (cdr open-ans)))
939 (if not-found
940 (message "These text properties could not be saved:\n %s"
941 not-found))
942 (nreverse all-ans)))
943
944;;; Internal functions for format-annotate-region.
945
823139fb
DL
946(defun format-annotate-location (loc all ignore translations)
947 "Return annotation(s) needed at location LOC.
9c3054da 948This includes any properties that change between LOC - 1 and LOC.
029894b9
BG
949If ALL is true, don't look at previous location, but generate annotations for
950all non-nil properties.
951Third argument IGNORE is a list of text-properties not to consider.
d0fc7e3b 952Use the TRANSLATIONS alist (see `format-annotate-region' for doc).
029894b9
BG
953
954Return value is a vector of 3 elements:
d0fc7e3b
RS
9551. List of annotations to close
9562. List of annotations to open.
9573. List of properties that were ignored or couldn't be annotated.
958
959The annotations in lists 1 and 2 need not be strings.
960They can be whatever the FORMAT-FN in `format-annotate-region'
961can handle. If that is `enriched-make-annotation', they can be
962either strings, or lists of the form (PARAMETER VALUE)."
029894b9
BG
963 (let* ((prev-loc (1- loc))
964 (before-plist (if all nil (text-properties-at prev-loc)))
965 (after-plist (text-properties-at loc))
966 p negatives positives prop props not-found)
967 ;; make list of all property names involved
968 (setq p before-plist)
969 (while p
970 (if (not (memq (car p) props))
be227f22 971 (push (car p) props))
029894b9
BG
972 (setq p (cdr (cdr p))))
973 (setq p after-plist)
974 (while p
975 (if (not (memq (car p) props))
be227f22 976 (push (car p) props))
029894b9
BG
977 (setq p (cdr (cdr p))))
978
979 (while props
be227f22 980 (setq prop (pop props))
029894b9
BG
981 (if (memq prop ignore)
982 nil ; If it's been ignored before, ignore it now.
983 (let ((before (if all nil (car (cdr (memq prop before-plist)))))
984 (after (car (cdr (memq prop after-plist)))))
985 (if (equal before after)
986 nil ; no change; ignore
987 (let ((result (format-annotate-single-property-change
823139fb 988 prop before after translations)))
029894b9 989 (if (not result)
be227f22 990 (push prop not-found)
029894b9
BG
991 (setq negatives (nconc negatives (car result))
992 positives (nconc positives (cdr result)))))))))
993 (vector negatives positives not-found)))
994
d0fc7e3b 995(defun format-annotate-single-property-change (prop old new translations)
823139fb 996 "Return annotations for property PROP changing from OLD to NEW.
d0fc7e3b
RS
997These are searched for in the translations alist TRANSLATIONS
998 (see `format-annotate-region' for the format).
12008438
TTN
999If NEW does not appear in the list, but there is a default function,
1000then call that function.
1001Return a cons of the form (CLOSE . OPEN)
d0fc7e3b
RS
1002where CLOSE is a list of annotations to close
1003and OPEN is a list of annotations to open.
1004
1005The annotations in CLOSE and OPEN need not be strings.
1006They can be whatever the FORMAT-FN in `format-annotate-region'
1007can handle. If that is `enriched-make-annotation', they can be
1008either strings, or lists of the form (PARAMETER VALUE)."
1009
1010 (let ((prop-alist (cdr (assoc prop translations)))
029894b9
BG
1011 default)
1012 (if (not prop-alist)
1013 nil
029894b9 1014 ;; If either old or new is a list, have to treat both that way.
f0a6c717 1015 (if (and (or (listp old) (listp new))
f3bbef87 1016 (not (get prop 'format-list-atomic-p)))
f0a6c717
GM
1017 (if (or (not (format-proper-list-p old))
1018 (not (format-proper-list-p new)))
1019 (format-annotate-atomic-property-change prop-alist old new)
1020 (let* ((old (if (listp old) old (list old)))
1021 (new (if (listp new) new (list new)))
1022 (tail (format-common-tail old new))
1023 close open)
1024 (while old
1025 (setq close
1026 (append (car (format-annotate-atomic-property-change
1027 prop-alist (car old) nil))
1028 close)
1029 old (cdr old)))
1030 (while new
1031 (setq open
1032 (append (cdr (format-annotate-atomic-property-change
1033 prop-alist nil (car new)))
1034 open)
1035 new (cdr new)))
1036 (format-make-relatively-unique close open)))
029894b9
BG
1037 (format-annotate-atomic-property-change prop-alist old new)))))
1038
1039(defun format-annotate-atomic-property-change (prop-alist old new)
12008438 1040 "Internal function to annotate a single property change.
d0fc7e3b 1041PROP-ALIST is the relevant element of a TRANSLATIONS list.
029894b9 1042OLD and NEW are the values."
e5a60108
RS
1043 (let (num-ann)
1044 ;; If old and new values are numbers,
1045 ;; look for a number in PROP-ALIST.
5f525595
RS
1046 (if (and (or (null old) (numberp old))
1047 (or (null new) (numberp new)))
e5a60108
RS
1048 (progn
1049 (setq num-ann prop-alist)
1050 (while (and num-ann (not (numberp (car (car num-ann)))))
1051 (setq num-ann (cdr num-ann)))))
1052 (if num-ann
5f525595 1053 ;; Numerical annotation - use difference
34fbecde
RS
1054 (progn
1055 ;; If property is numeric, nil means 0
1056 (cond ((and (numberp old) (null new))
1057 (setq new 0))
1058 ((and (numberp new) (null old))
1059 (setq old 0)))
1060
1061 (let* ((entry (car num-ann))
1062 (increment (car entry))
1063 (n (ceiling (/ (float (- new old)) (float increment))))
1064 (anno (car (cdr entry))))
1065 (if (> n 0)
1066 (cons nil (make-list n anno))
1067 (cons (make-list (- n) anno) nil))))
e5a60108
RS
1068
1069 ;; Standard annotation
1070 (let ((close (and old (cdr (assoc old prop-alist))))
029894b9
BG
1071 (open (and new (cdr (assoc new prop-alist)))))
1072 (if (or close open)
1073 (format-make-relatively-unique close open)
1074 ;; Call "Default" function, if any
1075 (let ((default (assq nil prop-alist)))
1076 (if default
1077 (funcall (car (cdr default)) old new))))))))
1078
c84cf2bf 1079(provide 'format)
823139fb 1080
cbee283d 1081;; arch-tag: c387e9c7-a93d-47bf-89bc-8ca67e96755a
823139fb 1082;;; format.el ends here