Re-indent.
[emacs.git] / lisp / format.el
blobfaae46466c6551e3997e9e1700e355036fe05072
1 ;;; format.el --- read and save files in multiple formats
3 ;; Copyright (C) 1994, 1995, 1997, 1999, 2001, 2002, 2003, 2004,
4 ;; 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
6 ;; Author: Boris Goldowsky <boris@gnu.org>
8 ;; This file is part of GNU Emacs.
10 ;; GNU Emacs is free software; you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published by
12 ;; the Free Software Foundation; either version 3, or (at your option)
13 ;; any later version.
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.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs; see the file COPYING. If not, write to the
22 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
23 ;; Boston, MA 02110-1301, USA.
25 ;;; Commentary:
27 ;; This file defines a unified mechanism for saving & loading files stored
28 ;; in different formats. `format-alist' contains information that directs
29 ;; Emacs to call an encoding or decoding function when reading or writing
30 ;; files that match certain conditions.
32 ;; When a file is visited, its format is determined by matching the
33 ;; beginning of the file against regular expressions stored in
34 ;; `format-alist'. If this fails, you can manually translate the buffer
35 ;; using `format-decode-buffer'. In either case, the formats used are
36 ;; listed in the variable `buffer-file-format', and become the default
37 ;; format for saving the buffer. To save a buffer in a different format,
38 ;; change this variable, or use `format-write-file'.
40 ;; Auto-save files are normally created in the same format as the visited
41 ;; file, but the variable `buffer-auto-save-file-format' can be set to a
42 ;; particularly fast or otherwise preferred format to be used for
43 ;; auto-saving (or nil to do no encoding on auto-save files, but then you
44 ;; risk losing any text-properties in the buffer).
46 ;; You can manually translate a buffer into or out of a particular format
47 ;; with the functions `format-encode-buffer' and `format-decode-buffer'.
48 ;; To translate just the region use the functions `format-encode-region'
49 ;; and `format-decode-region'.
51 ;; You can define a new format by writing the encoding and decoding
52 ;; functions, and adding an entry to `format-alist'. See enriched.el for
53 ;; an example of how to implement a file format. There are various
54 ;; functions defined in this file that may be useful for writing the
55 ;; encoding and decoding functions:
56 ;; * `format-annotate-region' and `format-deannotate-region' allow a
57 ;; single alist of information to be used for encoding and decoding.
58 ;; The alist defines a correspondence between strings in the file
59 ;; ("annotations") and text-properties in the buffer.
60 ;; * `format-replace-strings' is similarly useful for doing simple
61 ;; string->string translations in a reversible manner.
63 ;;; Code:
65 (put 'buffer-file-format 'permanent-local t)
66 (put 'buffer-auto-save-file-format 'permanent-local t)
68 (defvar format-alist
69 '((text/enriched "Extended MIME text/enriched format."
70 "Content-[Tt]ype:[ \t]*text/enriched"
71 enriched-decode enriched-encode t enriched-mode)
72 (plain "ISO 8859-1 standard format, no text properties."
73 ;; Plain only exists so that there is an obvious neutral choice in
74 ;; the completion list.
75 nil nil nil nil nil)
76 (TeX "TeX (encoding)"
77 nil
78 iso-tex2iso iso-iso2tex t nil)
79 (gtex "German TeX (encoding)"
80 nil
81 iso-gtex2iso iso-iso2gtex t nil)
82 (html "HTML/SGML \"ISO 8879:1986//ENTITIES Added Latin 1//EN\" (encoding)"
83 nil
84 iso-sgml2iso iso-iso2sgml t nil)
85 (rot13 "rot13"
86 nil
87 "tr a-mn-z n-za-m" "tr a-mn-z n-za-m" t nil)
88 (duden "Duden Ersatzdarstellung"
89 nil
90 "diac" iso-iso2duden t nil)
91 (de646 "German ASCII (ISO 646)"
92 nil
93 "recode -f iso646-ge:latin1" "recode -f latin1:iso646-ge" t nil)
94 (denet "net German"
95 nil
96 iso-german iso-cvt-read-only t nil)
97 (esnet "net Spanish"
98 nil
99 iso-spanish iso-cvt-read-only t nil))
100 "List of information about understood file formats.
101 Elements are of the form \(NAME DOC-STR REGEXP FROM-FN TO-FN MODIFY MODE-FN).
103 NAME is a symbol, which is stored in `buffer-file-format'.
105 DOC-STR should be a single line providing more information about the
106 format. It is currently unused, but in the future will be shown to
107 the user if they ask for more information.
109 REGEXP is a regular expression to match against the beginning of the file;
110 it should match only files in that format. Use nil to avoid
111 matching at all for formats for which it isn't appropriate to
112 require explicit encoding/decoding.
114 FROM-FN is called to decode files in that format; it takes two args, BEGIN
115 and END, and can make any modifications it likes, returning the new
116 end. It must make sure that the beginning of the file no longer
117 matches REGEXP, or else it will get called again.
118 Alternatively, FROM-FN can be a string, which specifies a shell command
119 (including options) to be used as a filter to perform the conversion.
121 TO-FN is called to encode a region into that format; it takes three
122 arguments: BEGIN, END, and BUFFER. BUFFER is the original buffer that
123 the data being written came from, which the function could use, for
124 example, to find the values of local variables. TO-FN should either
125 return a list of annotations like `write-region-annotate-functions',
126 or modify the region and return the new end.
127 Alternatively, TO-FN can be a string, which specifies a shell command
128 (including options) to be used as a filter to perform the conversion.
130 MODIFY, if non-nil, means the TO-FN wants to modify the region. If nil,
131 TO-FN will not make any changes but will instead return a list of
132 annotations.
134 MODE-FN, if specified, is called when visiting a file with that format.
135 It is called with a single positive argument, on the assumption
136 that this would turn on some minor mode.
138 PRESERVE, if non-nil, means that `format-write-file' should not remove
139 this format from `buffer-file-format'.")
141 ;;; Basic Functions (called from Lisp)
143 (defun format-encode-run-method (method from to &optional buffer)
144 "Translate using METHOD the text from FROM to TO.
145 If METHOD is a string, it is a shell command (including options);
146 otherwise, it should be a Lisp function.
147 BUFFER should be the buffer that the output originally came from."
148 (if (stringp method)
149 (let ((error-buff (get-buffer-create "*Format Errors*"))
150 (coding-system-for-read 'no-conversion)
151 format-alist)
152 (with-current-buffer error-buff
153 (widen)
154 (erase-buffer))
155 (if (and (zerop (save-window-excursion
156 (shell-command-on-region from to method t t
157 error-buff)))
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)
163 (error "Format encoding failed")))
164 (funcall method from to buffer)))
166 (defun format-decode-run-method (method from to &optional buffer)
167 "Decode using METHOD the text from FROM to TO.
168 If METHOD is a string, it is a shell command (including options); otherwise,
169 it should be a Lisp function. Decoding is done for the given BUFFER."
170 (if (stringp method)
171 (let ((error-buff (get-buffer-create "*Format Errors*"))
172 (coding-system-for-write 'no-conversion)
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"))
189 (point))
190 (funcall method from to)))
192 (defun format-annotate-function (format from to orig-buf format-count)
193 "Return annotations for writing region as FORMAT.
194 FORMAT is a symbol naming one of the formats defined in `format-alist'.
195 It must be a single symbol, not a list like `buffer-file-format'.
196 FROM and TO delimit the region to be operated on in the current buffer.
197 ORIG-BUF is the original buffer that the data came from.
199 FORMAT-COUNT is an integer specifying how many times this function has
200 been called in the process of decoding ORIG-BUF.
202 This function works like a function in `write-region-annotate-functions':
203 it either returns a list of annotations, or returns with a different buffer
204 current, which contains the modified text to write. In the latter case,
205 this function's value is nil.
207 For most purposes, consider using `format-encode-region' instead."
208 ;; This function is called by write-region (actually
209 ;; build_annotations) for each element of buffer-file-format.
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.
216 (let ((copy-buf (get-buffer-create (format " *Format Temp %d*"
217 format-count)))
218 (sel-disp selective-display)
219 (multibyte enable-multibyte-characters)
220 (coding-system buffer-file-coding-system))
221 (with-current-buffer copy-buf
222 (setq selective-display sel-disp)
223 (set-buffer-multibyte multibyte)
224 (setq buffer-file-coding-system coding-system))
225 (copy-to-buffer copy-buf from to)
226 (set-buffer copy-buf)
227 (format-insert-annotations write-region-annotations-so-far from)
228 (format-encode-run-method to-fn (point-min) (point-max) orig-buf)
229 nil)
230 ;; Otherwise just call function, it will return annotations.
231 (funcall to-fn from to orig-buf)))))
233 (defun format-decode (format length &optional visit-flag)
234 ;; This function is called by insert-file-contents whenever a file is read.
235 "Decode text from any known FORMAT.
236 FORMAT is a symbol appearing in `format-alist' or a list of such symbols,
237 or nil, in which case this function tries to guess the format of the data by
238 matching against the regular expressions in `format-alist'. After a match is
239 found and the region decoded, the alist is searched again from the beginning
240 for another match.
242 Second arg LENGTH is the number of characters following point to operate on.
243 If optional third arg VISIT-FLAG is true, set `buffer-file-format'
244 to the reverted list of formats used, and call any mode functions defined
245 for those formats.
247 Return the new length of the decoded region.
249 For most purposes, consider using `format-decode-region' instead."
250 (let ((mod (buffer-modified-p))
251 (begin (point))
252 (end (+ (point) length)))
253 (unwind-protect
254 (progn
255 ;; Don't record undo information for the decoding.
257 (if (null format)
258 ;; Figure out which format it is in, remember list in `format'.
259 (let ((try format-alist))
260 (while try
261 (let* ((f (car try))
262 (regexp (nth 2 f))
263 (p (point)))
264 (if (and regexp (looking-at regexp)
265 (< (match-end 0) (+ begin length)))
266 (progn
267 (push (car f) format)
268 ;; Decode it
269 (if (nth 3 f)
270 (setq end (format-decode-run-method (nth 3 f) begin end)))
271 ;; Call visit function if required
272 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
273 ;; Safeguard against either of the functions changing pt.
274 (goto-char p)
275 ;; Rewind list to look for another format
276 (setq try format-alist))
277 (setq try (cdr try))))))
278 ;; Deal with given format(s)
279 (or (listp format) (setq format (list format)))
280 (let ((do format) f)
281 (while do
282 (or (setq f (assq (car do) format-alist))
283 (error "Unknown format %s" (car do)))
284 ;; Decode:
285 (if (nth 3 f)
286 (setq end (format-decode-run-method (nth 3 f) begin end)))
287 ;; Call visit function if required
288 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
289 (setq do (cdr do))))
290 ;; Encode in the opposite order.
291 (setq format (reverse format)))
292 (if visit-flag
293 (setq buffer-file-format format)))
295 (set-buffer-modified-p mod))
297 ;; Return new length of region
298 (- end begin)))
301 ;;; Interactive functions & entry points
304 (defun format-decode-buffer (&optional format)
305 "Translate the buffer from some FORMAT.
306 If the format is not specified, attempt a regexp-based guess.
307 Set `buffer-file-format' to the format used, and call any
308 format-specific mode functions."
309 (interactive
310 (list (format-read "Translate buffer from format (default guess): ")))
311 (save-excursion
312 (goto-char (point-min))
313 (format-decode format (buffer-size) t)))
315 (defun format-decode-region (from to &optional format)
316 "Decode the region from some format.
317 Arg FORMAT is optional; if omitted the format will be determined by looking
318 for identifying regular expressions at the beginning of the region."
319 (interactive
320 (list (region-beginning) (region-end)
321 (format-read "Translate region from format (default guess): ")))
322 (save-excursion
323 (goto-char from)
324 (format-decode format (- to from) nil)))
326 (defun format-encode-buffer (&optional format)
327 "Translate the buffer into FORMAT.
328 FORMAT defaults to `buffer-file-format'. It is a symbol naming one of the
329 formats defined in `format-alist', or a list of such symbols."
330 (interactive
331 (list (format-read (format "Translate buffer to format (default %s): "
332 buffer-file-format))))
333 (format-encode-region (point-min) (point-max) format))
335 (defun format-encode-region (beg end &optional format)
336 "Translate the region into some FORMAT.
337 FORMAT defaults to `buffer-file-format'. It is a symbol naming
338 one of the formats defined in `format-alist', or a list of such symbols."
339 (interactive
340 (list (region-beginning) (region-end)
341 (format-read (format "Translate region to format (default %s): "
342 buffer-file-format))))
343 (if (null format) (setq format buffer-file-format))
344 (if (symbolp format) (setq format (list format)))
345 (save-excursion
346 (goto-char end)
347 (let ((cur-buf (current-buffer))
348 (end (point-marker)))
349 (while format
350 (let* ((info (assq (car format) format-alist))
351 (to-fn (nth 4 info))
352 (modify (nth 5 info))
353 result)
354 (if to-fn
355 (if modify
356 (setq end (format-encode-run-method to-fn beg end
357 (current-buffer)))
358 (format-insert-annotations
359 (funcall to-fn beg end (current-buffer)))))
360 (setq format (cdr format)))))))
362 (defun format-write-file (filename format &optional confirm)
363 "Write current buffer into file FILENAME using some FORMAT.
364 Make buffer visit that file and set the format as the default for future
365 saves. If the buffer is already visiting a file, you can specify a directory
366 name as FILENAME, to write a file of the same old name in that directory.
368 If optional third arg CONFIRM is non-nil, ask for confirmation before
369 overwriting an existing file. Interactively, confirmation is required
370 unless you supply a prefix argument."
371 (interactive
372 ;; Same interactive spec as write-file, plus format question.
373 (let* ((file (if buffer-file-name
374 (read-file-name "Write file: "
375 nil nil nil nil)
376 (read-file-name "Write file: "
377 (cdr (assq 'default-directory
378 (buffer-local-variables)))
379 nil nil (buffer-name))))
380 (fmt (format-read (format "Write file `%s' in format: "
381 (file-name-nondirectory file)))))
382 (list file fmt (not current-prefix-arg))))
383 (let ((old-formats buffer-file-format)
384 preserve-formats)
385 (dolist (fmt old-formats)
386 (let ((aelt (assq fmt format-alist)))
387 (if (nth 7 aelt)
388 (push fmt preserve-formats))))
389 (setq buffer-file-format format)
390 (dolist (fmt preserve-formats)
391 (unless (memq fmt buffer-file-format)
392 (setq buffer-file-format (append buffer-file-format (list fmt))))))
393 (write-file filename confirm))
395 (defun format-find-file (filename format)
396 "Find the file FILENAME using data format FORMAT.
397 If FORMAT is nil then do not do any format conversion."
398 (interactive
399 ;; Same interactive spec as write-file, plus format question.
400 (let* ((file (read-file-name "Find file: "))
401 (fmt (format-read (format "Read file `%s' in format: "
402 (file-name-nondirectory file)))))
403 (list file fmt)))
404 (let ((format-alist nil))
405 (find-file filename))
406 (if format
407 (format-decode-buffer format)))
409 (defun format-insert-file (filename format &optional beg end)
410 "Insert the contents of file FILENAME using data format FORMAT.
411 If FORMAT is nil then do not do any format conversion.
412 The optional third and fourth arguments BEG and END specify
413 the part (in bytes) of the file to read.
415 The return value is like the value of `insert-file-contents':
416 a list (ABSOLUTE-FILE-NAME SIZE)."
417 (interactive
418 ;; Same interactive spec as write-file, plus format question.
419 (let* ((file (read-file-name "Find file: "))
420 (fmt (format-read (format "Read file `%s' in format: "
421 (file-name-nondirectory file)))))
422 (list file fmt)))
423 (let (value size old-undo)
424 ;; Record only one undo entry for the insertion. Inhibit point-motion and
425 ;; modification hooks as with `insert-file-contents'.
426 (let ((inhibit-point-motion-hooks t)
427 (inhibit-modification-hooks t))
428 ;; Don't bind `buffer-undo-list' to t here to assert that
429 ;; `insert-file-contents' may record whether the buffer was unmodified
430 ;; before.
431 (let ((format-alist nil))
432 (setq value (insert-file-contents filename nil beg end))
433 (setq size (nth 1 value)))
434 (when (consp buffer-undo-list)
435 (let ((head (car buffer-undo-list)))
436 (when (and (consp head)
437 (equal (car head) (point))
438 (equal (cdr head) (+ (point) size)))
439 ;; Remove first entry from `buffer-undo-list', we shall insert
440 ;; another one below.
441 (setq old-undo (cdr buffer-undo-list)))))
442 (when format
443 (let ((buffer-undo-list t))
444 (setq size (format-decode format size)
445 value (list (car value) size)))
446 (unless (eq buffer-undo-list t)
447 (setq buffer-undo-list
448 (cons (cons (point) (+ (point) size)) old-undo)))))
449 (unless inhibit-modification-hooks
450 (run-hook-with-args 'after-change-functions (point) (+ (point) size) 0))
451 value))
453 (defun format-read (&optional prompt)
454 "Read and return the name of a format.
455 Return value is a list, like `buffer-file-format'; it may be nil.
456 Formats are defined in `format-alist'. Optional arg is the PROMPT to use."
457 (let* ((table (mapcar (lambda (x) (list (symbol-name (car x))))
458 format-alist))
459 (ans (completing-read (or prompt "Format: ") table nil t)))
460 (if (not (equal "" ans)) (list (intern ans)))))
464 ;;; Below are some functions that may be useful in writing encoding and
465 ;;; decoding functions for use in format-alist.
468 (defun format-replace-strings (alist &optional reverse beg end)
469 "Do multiple replacements on the buffer.
470 ALIST is a list of (FROM . TO) pairs, which should be proper arguments to
471 `search-forward' and `replace-match', respectively.
472 Optional second arg REVERSE, if non-nil, means the pairs are (TO . FROM),
473 so that you can use the same list in both directions if it contains only
474 literal strings.
475 Optional args BEG and END specify a region of the buffer on which to operate."
476 (save-excursion
477 (save-restriction
478 (or beg (setq beg (point-min)))
479 (if end (narrow-to-region (point-min) end))
480 (while alist
481 (let ((from (if reverse (cdr (car alist)) (car (car alist))))
482 (to (if reverse (car (car alist)) (cdr (car alist)))))
483 (goto-char beg)
484 (while (search-forward from nil t)
485 (goto-char (match-beginning 0))
486 (insert to)
487 (set-text-properties (- (point) (length to)) (point)
488 (text-properties-at (point)))
489 (delete-region (point) (+ (point) (- (match-end 0)
490 (match-beginning 0)))))
491 (setq alist (cdr alist)))))))
493 ;;; Some list-manipulation functions that we need.
495 (defun format-delq-cons (cons list)
496 "Remove the given CONS from LIST by side effect and return the new LIST.
497 Since CONS could be the first element of LIST, write
498 `\(setq foo \(format-delq-cons element foo))' to be sure of changing
499 the value of `foo'."
500 (if (eq cons list)
501 (cdr list)
502 (let ((p list))
503 (while (not (eq (cdr p) cons))
504 (if (null p) (error "format-delq-cons: not an element"))
505 (setq p (cdr p)))
506 ;; Now (cdr p) is the cons to delete
507 (setcdr p (cdr cons))
508 list)))
510 (defun format-make-relatively-unique (a b)
511 "Delete common elements of lists A and B, return as pair.
512 Compare using `equal'."
513 (let* ((acopy (copy-sequence a))
514 (bcopy (copy-sequence b))
515 (tail acopy))
516 (while tail
517 (let ((dup (member (car tail) bcopy))
518 (next (cdr tail)))
519 (if dup (setq acopy (format-delq-cons tail acopy)
520 bcopy (format-delq-cons dup bcopy)))
521 (setq tail next)))
522 (cons acopy bcopy)))
524 (defun format-common-tail (a b)
525 "Given two lists that have a common tail, return it.
526 Compare with `equal', and return the part of A that is equal to the
527 equivalent part of B. If even the last items of the two are not equal,
528 return nil."
529 (let ((la (length a))
530 (lb (length b)))
531 ;; Make sure they are the same length
532 (if (> la lb)
533 (setq a (nthcdr (- la lb) a))
534 (setq b (nthcdr (- lb la) b))))
535 (while (not (equal a b))
536 (setq a (cdr a)
537 b (cdr b)))
540 (defun format-proper-list-p (list)
541 "Return t if LIST is a proper list.
542 A proper list is a list ending with a nil cdr, not with an atom "
543 (when (listp list)
544 (while (consp list)
545 (setq list (cdr list)))
546 (null list)))
548 (defun format-reorder (items order)
549 "Arrange ITEMS to follow partial ORDER.
550 Elements of ITEMS equal to elements of ORDER will be rearranged
551 to follow the ORDER. Unmatched items will go last."
552 (if order
553 (let ((item (member (car order) items)))
554 (if item
555 (cons (car item)
556 (format-reorder (format-delq-cons item items)
557 (cdr order)))
558 (format-reorder items (cdr order))))
559 items))
561 (put 'face 'format-list-valued t) ; These text-properties take values
562 (put 'unknown 'format-list-valued t) ; that are lists, the elements of which
563 ; should be considered separately.
564 ; See format-deannotate-region and
565 ; format-annotate-region.
567 ;; This text property has list values, but they are treated atomically.
569 (put 'display 'format-list-atomic-p t)
572 ;;; Decoding
575 (defun format-deannotate-region (from to translations next-fn)
576 "Translate annotations in the region into text properties.
577 This sets text properties between FROM to TO as directed by the
578 TRANSLATIONS and NEXT-FN arguments.
580 NEXT-FN is a function that searches forward from point for an annotation.
581 It should return a list of 4 elements: \(BEGIN END NAME POSITIVE). BEGIN and
582 END are buffer positions bounding the annotation, NAME is the name searched
583 for in TRANSLATIONS, and POSITIVE should be non-nil if this annotation marks
584 the beginning of a region with some property, or nil if it ends the region.
585 NEXT-FN should return nil if there are no annotations after point.
587 The basic format of the TRANSLATIONS argument is described in the
588 documentation for the `format-annotate-region' function. There are some
589 additional things to keep in mind for decoding, though:
591 When an annotation is found, the TRANSLATIONS list is searched for a
592 text-property name and value that corresponds to that annotation. If the
593 text-property has several annotations associated with it, it will be used only
594 if the other annotations are also in effect at that point. The first match
595 found whose annotations are all present is used.
597 The text property thus determined is set to the value over the region between
598 the opening and closing annotations. However, if the text-property name has a
599 non-nil `format-list-valued' property, then the value will be consed onto the
600 surrounding value of the property, rather than replacing that value.
602 There are some special symbols that can be used in the \"property\" slot of
603 the TRANSLATIONS list: PARAMETER and FUNCTION \(spelled in uppercase).
604 Annotations listed under the pseudo-property PARAMETER are considered to be
605 arguments of the immediately surrounding annotation; the text between the
606 opening and closing parameter annotations is deleted from the buffer but saved
607 as a string.
609 The surrounding annotation should be listed under the pseudo-property
610 FUNCTION. Instead of inserting a text-property for this annotation,
611 the function listed in the VALUE slot is called to make whatever
612 changes are appropriate. It can also return a list of the form
613 \(START LOC PROP VALUE) which specifies a property to put on. The
614 function's first two arguments are the START and END locations, and
615 the rest of the arguments are any PARAMETERs found in that region.
617 Any annotations that are found by NEXT-FN but not defined by TRANSLATIONS
618 are saved as values of the `unknown' text-property \(which is list-valued).
619 The TRANSLATIONS list should usually contain an entry of the form
620 \(unknown \(nil format-annotate-value))
621 to write these unknown annotations back into the file."
622 (save-excursion
623 (save-restriction
624 (narrow-to-region (point-min) to)
625 (goto-char from)
626 (let (next open-ans todo loc unknown-ans)
627 (while (setq next (funcall next-fn))
628 (let* ((loc (nth 0 next))
629 (end (nth 1 next))
630 (name (nth 2 next))
631 (positive (nth 3 next))
632 (found nil))
634 ;; Delete the annotation
635 (delete-region loc end)
636 (cond
637 ;; Positive annotations are stacked, remembering location
638 (positive (push `(,name ((,loc . nil))) open-ans))
639 ;; It is a negative annotation:
640 ;; Close the top annotation & add its text property.
641 ;; If the file's nesting is messed up, the close might not match
642 ;; the top thing on the open-annotations stack.
643 ;; If no matching annotation is open, just ignore the close.
644 ((not (assoc name open-ans))
645 (message "Extra closing annotation (%s) in file" name))
646 ;; If one is open, but not on the top of the stack, close
647 ;; the things in between as well. Set `found' when the real
648 ;; one is closed.
650 (while (not found)
651 (let* ((top (car open-ans)) ; first on stack: should match.
652 (top-name (car top)) ; text property name
653 (top-extents (nth 1 top)) ; property regions
654 (params (cdr (cdr top))) ; parameters
655 (aalist translations)
656 (matched nil))
657 (if (equal name top-name)
658 (setq found t)
659 (message "Improper nesting in file."))
660 ;; Look through property names in TRANSLATIONS
661 (while aalist
662 (let ((prop (car (car aalist)))
663 (alist (cdr (car aalist))))
664 ;; And look through values for each property
665 (while alist
666 (let ((value (car (car alist)))
667 (ans (cdr (car alist))))
668 (if (member top-name ans)
669 ;; This annotation is listed, but still have to
670 ;; check if multiple annotations are satisfied
671 (if (member nil (mapcar (lambda (r)
672 (assoc r open-ans))
673 ans))
674 nil ; multiple ans not satisfied
675 ;; If there are multiple annotations going
676 ;; into one text property, split up the other
677 ;; annotations so they apply individually to
678 ;; the other regions.
679 (setcdr (car top-extents) loc)
680 (let ((to-split ans) this-one extents)
681 (while to-split
682 (setq this-one
683 (assoc (car to-split) open-ans)
684 extents (nth 1 this-one))
685 (if (not (eq this-one top))
686 (setcar (cdr this-one)
687 (format-subtract-regions
688 extents top-extents)))
689 (setq to-split (cdr to-split))))
690 ;; Set loop variables to nil so loop
691 ;; will exit.
692 (setq alist nil aalist nil matched t
693 ;; pop annotation off stack.
694 open-ans (cdr open-ans))
695 (let ((extents top-extents)
696 (start (car (car top-extents)))
697 (loc (cdr (car top-extents))))
698 (while extents
699 (cond
700 ;; Check for pseudo-properties
701 ((eq prop 'PARAMETER)
702 ;; A parameter of the top open ann:
703 ;; delete text and use as arg.
704 (if open-ans
705 ;; (If nothing open, discard).
706 (setq open-ans
707 (cons
708 (append (car open-ans)
709 (list
710 (buffer-substring
711 start loc)))
712 (cdr open-ans))))
713 (delete-region start loc))
714 ((eq prop 'FUNCTION)
715 ;; Not a property, but a function.
716 (let ((rtn
717 (apply value start loc params)))
718 (if rtn (push rtn todo))))
720 ;; Normal property/value pair
721 (setq todo
722 (cons (list start loc prop value)
723 todo))))
724 (setq extents (cdr extents)
725 start (car (car extents))
726 loc (cdr (car extents))))))))
727 (setq alist (cdr alist))))
728 (setq aalist (cdr aalist)))
729 (if (not matched)
730 ;; Didn't find any match for the annotation:
731 ;; Store as value of text-property `unknown'.
732 (let ((extents top-extents)
733 (start (car (car top-extents)))
734 (loc (or (cdr (car top-extents)) loc)))
735 (while extents
736 (setq open-ans (cdr open-ans)
737 todo (cons (list start loc 'unknown top-name)
738 todo)
739 unknown-ans (cons name unknown-ans)
740 extents (cdr extents)
741 start (car (car extents))
742 loc (cdr (car extents))))))))))))
744 ;; Once entire file has been scanned, add the properties.
745 (while todo
746 (let* ((item (car todo))
747 (from (nth 0 item))
748 (to (nth 1 item))
749 (prop (nth 2 item))
750 (val (nth 3 item)))
752 (if (numberp val) ; add to ambient value if numeric
753 (format-property-increment-region from to prop val 0)
754 (put-text-property
755 from to prop
756 (cond ((get prop 'format-list-valued) ; value gets consed onto
757 ; list-valued properties
758 (let ((prev (get-text-property from prop)))
759 (cons val (if (listp prev) prev (list prev)))))
760 (t val))))) ; normally, just set to val.
761 (setq todo (cdr todo)))
763 (if unknown-ans
764 (message "Unknown annotations: %s" unknown-ans))))))
766 (defun format-subtract-regions (minu subtra)
767 "Remove from the regions in MINUEND the regions in SUBTRAHEND.
768 A region is a dotted pair (FROM . TO). Both parameters are lists of
769 regions. Each list must contain nonoverlapping, noncontiguous
770 regions, in descending order. The result is also nonoverlapping,
771 noncontiguous, and in descending order. The first element of MINUEND
772 can have a cdr of nil, indicating that the end of that region is not
773 yet known.
775 \(fn MINUEND SUBTRAHEND)"
776 (let* ((minuend (copy-alist minu))
777 (subtrahend (copy-alist subtra))
778 (m (car minuend))
779 (s (car subtrahend))
780 results)
781 (while (and minuend subtrahend)
782 (cond
783 ;; The minuend starts after the subtrahend ends; keep it.
784 ((> (car m) (cdr s))
785 (push m results)
786 (setq minuend (cdr minuend)
787 m (car minuend)))
788 ;; The minuend extends beyond the end of the subtrahend. Chop it off.
789 ((or (null (cdr m)) (> (cdr m) (cdr s)))
790 (push (cons (1+ (cdr s)) (cdr m)) results)
791 (setcdr m (cdr s)))
792 ;; The subtrahend starts after the minuend ends; throw it away.
793 ((< (cdr m) (car s))
794 (setq subtrahend (cdr subtrahend) s (car subtrahend)))
795 ;; The subtrahend extends beyond the end of the minuend. Chop it off.
796 (t ;(<= (cdr m) (cdr s)))
797 (if (>= (car m) (car s))
798 (setq minuend (cdr minuend) m (car minuend))
799 (setcdr m (1- (car s)))
800 (setq subtrahend (cdr subtrahend) s (car subtrahend))))))
801 (nconc (nreverse results) minuend)))
803 ;; This should probably go somewhere other than format.el. Then again,
804 ;; indent.el has alter-text-property. NOTE: We can also use
805 ;; next-single-property-change instead of text-property-not-all, but then
806 ;; we have to see if we passed TO.
807 (defun format-property-increment-region (from to prop delta default)
808 "In the region from FROM to TO increment property PROP by amount DELTA.
809 DELTA may be negative. If property PROP is nil anywhere
810 in the region, it is treated as though it were DEFAULT."
811 (let ((cur from) val newval next)
812 (while cur
813 (setq val (get-text-property cur prop)
814 newval (+ (or val default) delta)
815 next (text-property-not-all cur to prop val))
816 (put-text-property cur (or next to) prop newval)
817 (setq cur next))))
820 ;;; Encoding
823 (defun format-insert-annotations (list &optional offset)
824 "Apply list of annotations to buffer as `write-region' would.
825 Insert each element of the given LIST of buffer annotations at its
826 appropriate place. Use second arg OFFSET if the annotations' locations are
827 not relative to the beginning of the buffer: annotations will be inserted
828 at their location-OFFSET+1 \(ie, the offset is treated as the position of
829 the first character in the buffer)."
830 (if (not offset)
831 (setq offset 0)
832 (setq offset (1- offset)))
833 (let ((l (reverse list)))
834 (while l
835 (goto-char (- (car (car l)) offset))
836 (insert (cdr (car l)))
837 (setq l (cdr l)))))
839 (defun format-annotate-value (old new)
840 "Return OLD and NEW as a \(CLOSE . OPEN) annotation pair.
841 Useful as a default function for TRANSLATIONS alist when the value of the text
842 property is the name of the annotation that you want to use, as it is for the
843 `unknown' text property."
844 (cons (if old (list old))
845 (if new (list new))))
847 (defun format-annotate-region (from to translations format-fn ignore)
848 "Generate annotations for text properties in the region.
849 Search for changes between FROM and TO, and describe them with a list of
850 annotations as defined by alist TRANSLATIONS and FORMAT-FN. IGNORE lists text
851 properties not to consider; any text properties that are neither ignored nor
852 listed in TRANSLATIONS are warned about.
853 If you actually want to modify the region, give the return value of this
854 function to `format-insert-annotations'.
856 Format of the TRANSLATIONS argument:
858 Each element is a list whose car is a PROPERTY, and the following
859 elements have the form (VALUE ANNOTATIONS...).
860 Whenever the property takes on the value VALUE, the annotations
861 \(as formatted by FORMAT-FN) are inserted into the file.
862 When the property stops having that value, the matching negated annotation
863 will be inserted \(it may actually be closed earlier and reopened, if
864 necessary, to keep proper nesting).
866 If VALUE is a list, then each element of the list is dealt with
867 separately.
869 If a VALUE is numeric, then it is assumed that there is a single annotation
870 and each occurrence of it increments the value of the property by that number.
871 Thus, given the entry \(left-margin \(4 \"indent\")), if the left margin
872 changes from 4 to 12, two <indent> annotations will be generated.
874 If the VALUE is nil, then instead of annotations, a function should be
875 specified. This function is used as a default: it is called for all
876 transitions not explicitly listed in the table. The function is called with
877 two arguments, the OLD and NEW values of the property. It should return
878 a cons cell (CLOSE . OPEN) as `format-annotate-single-property-change' does.
880 The same TRANSLATIONS structure can be used in reverse for reading files."
881 (let ((all-ans nil) ; All annotations - becomes return value
882 (open-ans nil) ; Annotations not yet closed
883 (loc nil) ; Current location
884 (not-found nil)) ; Properties that couldn't be saved
885 (while (or (null loc)
886 (and (setq loc (next-property-change loc nil to))
887 (< loc to)))
888 (or loc (setq loc from))
889 (let* ((ans (format-annotate-location loc (= loc from) ignore translations))
890 (neg-ans (format-reorder (aref ans 0) open-ans))
891 (pos-ans (aref ans 1))
892 (ignored (aref ans 2)))
893 (setq not-found (append ignored not-found)
894 ignore (append ignored ignore))
895 ;; First do the negative (closing) annotations
896 (while neg-ans
897 ;; Check if it's missing. This can happen (eg, a numeric property
898 ;; going negative can generate closing annotations before there are
899 ;; any open). Warn user & ignore.
900 (if (not (member (car neg-ans) open-ans))
901 (message "Can't close %s: not open." (car neg-ans))
902 (while (not (equal (car neg-ans) (car open-ans)))
903 ;; To close anno. N, need to first close ans 1 to N-1,
904 ;; remembering to re-open them later.
905 (push (car open-ans) pos-ans)
906 (setq all-ans
907 (cons (cons loc (funcall format-fn (car open-ans) nil))
908 all-ans))
909 (setq open-ans (cdr open-ans)))
910 ;; Now remove the one we're really interested in from open list.
911 (setq open-ans (cdr open-ans))
912 ;; And put the closing annotation here.
913 (push (cons loc (funcall format-fn (car neg-ans) nil))
914 all-ans))
915 (setq neg-ans (cdr neg-ans)))
916 ;; Now deal with positive (opening) annotations
917 (let ((p pos-ans))
918 (while pos-ans
919 (push (car pos-ans) open-ans)
920 (push (cons loc (funcall format-fn (car pos-ans) t))
921 all-ans)
922 (setq pos-ans (cdr pos-ans))))))
924 ;; Close any annotations still open
925 (while open-ans
926 (setq all-ans
927 (cons (cons to (funcall format-fn (car open-ans) nil))
928 all-ans))
929 (setq open-ans (cdr open-ans)))
930 (if not-found
931 (message "These text properties could not be saved:\n %s"
932 not-found))
933 (nreverse all-ans)))
935 ;;; Internal functions for format-annotate-region.
937 (defun format-annotate-location (loc all ignore translations)
938 "Return annotation(s) needed at location LOC.
939 This includes any properties that change between LOC - 1 and LOC.
940 If ALL is true, don't look at previous location, but generate annotations for
941 all non-nil properties.
942 Third argument IGNORE is a list of text-properties not to consider.
943 Use the TRANSLATIONS alist (see `format-annotate-region' for doc).
945 Return value is a vector of 3 elements:
946 1. List of annotations to close
947 2. List of annotations to open.
948 3. List of properties that were ignored or couldn't be annotated.
950 The annotations in lists 1 and 2 need not be strings.
951 They can be whatever the FORMAT-FN in `format-annotate-region'
952 can handle. If that is `enriched-make-annotation', they can be
953 either strings, or lists of the form (PARAMETER VALUE)."
954 (let* ((prev-loc (1- loc))
955 (before-plist (if all nil (text-properties-at prev-loc)))
956 (after-plist (text-properties-at loc))
957 p negatives positives prop props not-found)
958 ;; make list of all property names involved
959 (setq p before-plist)
960 (while p
961 (if (not (memq (car p) props))
962 (push (car p) props))
963 (setq p (cdr (cdr p))))
964 (setq p after-plist)
965 (while p
966 (if (not (memq (car p) props))
967 (push (car p) props))
968 (setq p (cdr (cdr p))))
970 (while props
971 (setq prop (pop props))
972 (if (memq prop ignore)
973 nil ; If it's been ignored before, ignore it now.
974 (let ((before (if all nil (car (cdr (memq prop before-plist)))))
975 (after (car (cdr (memq prop after-plist)))))
976 (if (equal before after)
977 nil ; no change; ignore
978 (let ((result (format-annotate-single-property-change
979 prop before after translations)))
980 (if (not result)
981 (push prop not-found)
982 (setq negatives (nconc negatives (car result))
983 positives (nconc positives (cdr result)))))))))
984 (vector negatives positives not-found)))
986 (defun format-annotate-single-property-change (prop old new translations)
987 "Return annotations for property PROP changing from OLD to NEW.
988 These are searched for in the translations alist TRANSLATIONS
989 (see `format-annotate-region' for the format).
990 If NEW does not appear in the list, but there is a default function,
991 then call that function.
992 Return a cons of the form (CLOSE . OPEN)
993 where CLOSE is a list of annotations to close
994 and OPEN is a list of annotations to open.
996 The annotations in CLOSE and OPEN need not be strings.
997 They can be whatever the FORMAT-FN in `format-annotate-region'
998 can handle. If that is `enriched-make-annotation', they can be
999 either strings, or lists of the form (PARAMETER VALUE)."
1001 (let ((prop-alist (cdr (assoc prop translations)))
1002 default)
1003 (if (not prop-alist)
1005 ;; If either old or new is a list, have to treat both that way.
1006 (if (and (or (listp old) (listp new))
1007 (not (get prop 'format-list-atomic-p)))
1008 (if (or (not (format-proper-list-p old))
1009 (not (format-proper-list-p new)))
1010 (format-annotate-atomic-property-change prop-alist old new)
1011 (let* ((old (if (listp old) old (list old)))
1012 (new (if (listp new) new (list new)))
1013 (tail (format-common-tail old new))
1014 close open)
1015 (while old
1016 (setq close
1017 (append (car (format-annotate-atomic-property-change
1018 prop-alist (car old) nil))
1019 close)
1020 old (cdr old)))
1021 (while new
1022 (setq open
1023 (append (cdr (format-annotate-atomic-property-change
1024 prop-alist nil (car new)))
1025 open)
1026 new (cdr new)))
1027 (format-make-relatively-unique close open)))
1028 (format-annotate-atomic-property-change prop-alist old new)))))
1030 (defun format-annotate-atomic-property-change (prop-alist old new)
1031 "Internal function to annotate a single property change.
1032 PROP-ALIST is the relevant element of a TRANSLATIONS list.
1033 OLD and NEW are the values."
1034 (let (num-ann)
1035 ;; If old and new values are numbers,
1036 ;; look for a number in PROP-ALIST.
1037 (if (and (or (null old) (numberp old))
1038 (or (null new) (numberp new)))
1039 (progn
1040 (setq num-ann prop-alist)
1041 (while (and num-ann (not (numberp (car (car num-ann)))))
1042 (setq num-ann (cdr num-ann)))))
1043 (if num-ann
1044 ;; Numerical annotation - use difference
1045 (progn
1046 ;; If property is numeric, nil means 0
1047 (cond ((and (numberp old) (null new))
1048 (setq new 0))
1049 ((and (numberp new) (null old))
1050 (setq old 0)))
1052 (let* ((entry (car num-ann))
1053 (increment (car entry))
1054 (n (ceiling (/ (float (- new old)) (float increment))))
1055 (anno (car (cdr entry))))
1056 (if (> n 0)
1057 (cons nil (make-list n anno))
1058 (cons (make-list (- n) anno) nil))))
1060 ;; Standard annotation
1061 (let ((close (and old (cdr (assoc old prop-alist))))
1062 (open (and new (cdr (assoc new prop-alist)))))
1063 (if (or close open)
1064 (format-make-relatively-unique close open)
1065 ;; Call "Default" function, if any
1066 (let ((default (assq nil prop-alist)))
1067 (if default
1068 (funcall (car (cdr default)) old new))))))))
1070 (provide 'format)
1072 ;;; arch-tag: c387e9c7-a93d-47bf-89bc-8ca67e96755a
1073 ;;; format.el ends here