*** empty log message ***
[emacs.git] / lisp / format.el
blobaeadb68d60fbcb8d1e6c789a62c8c0c2753d5e88
1 ;;; format.el --- read and save files in multiple formats
3 ;; Copyright (c) 1994, 1995, 1997, 1999 Free Software Foundation
5 ;; Author: Boris Goldowsky <boris@gnu.org>
7 ;; This file is part of GNU Emacs.
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.
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.
19 ;; You should have received a copy of the GNU General Public License
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.
24 ;;; Commentary:
26 ;; This file defines a unified mechanism for saving & loading files stored
27 ;; in different formats. `format-alist' contains information that directs
28 ;; Emacs to call an encoding or decoding function when reading or writing
29 ;; files that match certain conditions.
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'.
39 ;; Auto-save files are normally created in the same format as the visited
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).
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'
48 ;; and `format-decode-region'.
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.
59 ;; * `format-replace-strings' is similarly useful for doing simple
60 ;; string->string translations in a reversible manner.
62 ;;; Code:
64 (put 'buffer-file-format 'permanent-local t)
66 (defvar format-alist
67 '((text/enriched "Extended MIME text/enriched format."
68 "Content-[Tt]ype:[ \t]*text/enriched"
69 enriched-decode enriched-encode t enriched-mode)
70 (plain "ISO 8859-1 standard format, no text properties."
71 ;; Plain only exists so that there is an obvious neutral choice in
72 ;; the completion list.
73 nil nil nil nil nil)
74 (ibm "IBM Code Page 850 (DOS)"
75 nil ; The original "1\\(^\\)" is obscure.
76 "recode -f ibm-pc:latin1" "recode -f latin1:ibm-pc" t nil)
77 (mac "Apple Macintosh"
78 nil
79 "recode -f mac:latin1" "recode -f latin1:mac" t nil)
80 (hp "HP Roman8"
81 nil
82 "recode -f roman8:latin1" "recode -f latin1:roman8" t nil)
83 (TeX "TeX (encoding)"
84 nil
85 iso-tex2iso iso-iso2tex t nil)
86 (gtex "German TeX (encoding)"
87 nil
88 iso-gtex2iso iso-iso2gtex t nil)
89 (html "HTML/SGML \"ISO 8879:1986//ENTITIES Added Latin 1//EN\" (encoding)"
90 nil
91 iso-sgml2iso iso-iso2sgml t nil)
92 (rot13 "rot13"
93 nil
94 "tr a-mn-z n-za-m" "tr a-mn-z n-za-m" t nil)
95 (duden "Duden Ersatzdarstellung"
96 nil
97 "diac" iso-iso2duden t nil)
98 (de646 "German ASCII (ISO 646)"
99 nil
100 "recode -f iso646-ge:latin1" "recode -f latin1:iso646-ge" t nil)
101 (denet "net German"
103 iso-german iso-cvt-read-only t nil)
104 (esnet "net Spanish"
106 iso-spanish iso-cvt-read-only t nil))
107 "List of information about understood file formats.
108 Elements are of the form \(NAME DOC-STR REGEXP FROM-FN TO-FN MODIFY MODE-FN).
110 NAME is a symbol, which is stored in `buffer-file-format'.
112 DOC-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.
116 REGEXP is a regular expression to match against the beginning of the file;
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.
121 FROM-FN is called to decode files in that format; it gets two args, BEGIN
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.
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.
128 TO-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.
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.
137 MODIFY, 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
139 annotations.
141 MODE-FN, if specified, is called when visiting a file with that format.
142 It is called with a single positive argument, on the assumption
143 that it turns on some Emacs mode.
145 PRESERVE, if non-nil, means that `format-write-file' should not remove
146 this format from `buffer-file-formats'.")
148 ;;; Basic Functions (called from Lisp)
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.
152 If METHOD is a string, it is a shell command;
153 otherwise, it should be a Lisp function.
154 BUFFER should be the buffer that the output originally came from."
155 (if (stringp method)
156 (let ((error-buff (get-buffer-create "*Format Errors*"))
157 (coding-system-for-read 'no-conversion)
158 format-alist)
159 (with-current-buffer error-buff
160 (widen)
161 (erase-buffer))
162 (if (and (zerop (save-window-excursion
163 (shell-command-on-region from to method t t
164 error-buff)))
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)
170 (error "Format encoding failed")))
171 (funcall method from to buffer)))
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.
175 If METHOD is a string, it is a shell command; otherwise, it should be
176 a Lisp function. Decoding is done for the given BUFFER."
177 (if (stringp method)
178 (let ((error-buff (get-buffer-create "*Format Errors*"))
179 (coding-system-for-write 'no-conversion)
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"))
196 (point))
197 (funcall method from to)))
199 (defun format-annotate-function (format from to orig-buf format-count)
200 "Return annotations for writing region as FORMAT.
201 FORMAT is a symbol naming one of the formats defined in `format-alist',
202 it must be a single symbol, not a list like `buffer-file-format'.
203 FROM and TO delimit the region to be operated on in the current buffer.
204 ORIG-BUF is the original buffer that the data came from.
206 FORMAT-COUNT is an integer specifying how many times this function has
207 been called in the process of decoding ORIG-BUF.
209 This function works like a function on `write-region-annotate-functions':
210 it either returns a list of annotations, or returns with a different buffer
211 current, which contains the modified text to write. In the latter case,
212 this function's value is nil.
214 For most purposes, consider using `format-encode-region' instead."
215 ;; This function is called by write-region (actually
216 ;; build_annotations) for each element of buffer-file-format.
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.
223 (let ((copy-buf (get-buffer-create (format " *Format Temp %d*"
224 format-count)))
225 (sel-disp selective-display)
226 (multibyte enable-multibyte-characters)
227 (coding-system buffer-file-coding-system))
228 (with-current-buffer copy-buf
229 (setq selective-display sel-disp)
230 (set-buffer-multibyte multibyte)
231 (setq buffer-file-coding-system coding-system))
232 (copy-to-buffer copy-buf from to)
233 (set-buffer copy-buf)
234 (format-insert-annotations write-region-annotations-so-far from)
235 (format-encode-run-method to-fn (point-min) (point-max) orig-buf)
236 nil)
237 ;; Otherwise just call function, it will return annotations.
238 (funcall to-fn from to orig-buf)))))
240 (defun format-decode (format length &optional visit-flag)
241 ;; This function is called by insert-file-contents whenever a file is read.
242 "Decode text from any known FORMAT.
243 FORMAT is a symbol appearing in `format-alist' or a list of such symbols,
244 or nil, in which case this function tries to guess the format of the data by
245 matching against the regular expressions in `format-alist'. After a match is
246 found and the region decoded, the alist is searched again from the beginning
247 for another match.
249 Second arg LENGTH is the number of characters following point to operate on.
250 If optional third arg VISIT-FLAG is true, set `buffer-file-format'
251 to the reverted list of formats used, and call any mode functions defined
252 for those formats.
254 Returns the new length of the decoded region.
256 For most purposes, consider using `format-decode-region' instead."
257 (let ((mod (buffer-modified-p))
258 (begin (point))
259 (end (+ (point) length)))
260 (unwind-protect
261 (progn
262 ;; Don't record undo information for the decoding.
264 (if (null format)
265 ;; Figure out which format it is in, remember list in `format'.
266 (let ((try format-alist))
267 (while try
268 (let* ((f (car try))
269 (regexp (nth 2 f))
270 (p (point)))
271 (if (and regexp (looking-at regexp)
272 (< (match-end 0) (+ begin length)))
273 (progn
274 (push (car f) format)
275 ;; Decode it
276 (if (nth 3 f)
277 (setq end (format-decode-run-method (nth 3 f) begin end)))
278 ;; Call visit function if required
279 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
280 ;; Safeguard against either of the functions changing pt.
281 (goto-char p)
282 ;; Rewind list to look for another format
283 (setq try format-alist))
284 (setq try (cdr try))))))
285 ;; Deal with given format(s)
286 (or (listp format) (setq format (list format)))
287 (let ((do format) f)
288 (while do
289 (or (setq f (assq (car do) format-alist))
290 (error "Unknown format" (car do)))
291 ;; Decode:
292 (if (nth 3 f)
293 (setq end (format-decode-run-method (nth 3 f) begin end)))
294 ;; Call visit function if required
295 (if (and visit-flag (nth 6 f)) (funcall (nth 6 f) 1))
296 (setq do (cdr do))))
297 ;; Encode in the opposite order.
298 (setq format (reverse format)))
299 (if visit-flag
300 (setq buffer-file-format format)))
302 (set-buffer-modified-p mod))
304 ;; Return new length of region
305 (- end begin)))
308 ;;; Interactive functions & entry points
311 (defun format-decode-buffer (&optional format)
312 "Translate the buffer from some FORMAT.
313 If the format is not specified, this function attempts to guess.
314 `buffer-file-format' is set to the format used, and any mode-functions
315 for the format are called."
316 (interactive
317 (list (format-read "Translate buffer from format (default: guess): ")))
318 (save-excursion
319 (goto-char (point-min))
320 (format-decode format (buffer-size) t)))
322 (defun format-decode-region (from to &optional format)
323 "Decode the region from some format.
324 Arg FORMAT is optional; if omitted the format will be determined by looking
325 for identifying regular expressions at the beginning of the region."
326 (interactive
327 (list (region-beginning) (region-end)
328 (format-read "Translate region from format (default: guess): ")))
329 (save-excursion
330 (goto-char from)
331 (format-decode format (- to from) nil)))
333 (defun format-encode-buffer (&optional format)
334 "Translate the buffer into FORMAT.
335 FORMAT defaults to `buffer-file-format'. It is a symbol naming one of the
336 formats defined in `format-alist', or a list of such symbols."
337 (interactive
338 (list (format-read (format "Translate buffer to format (default %s): "
339 buffer-file-format))))
340 (format-encode-region (point-min) (point-max) format))
342 (defun format-encode-region (beg end &optional format)
343 "Translate the region into some FORMAT.
344 FORMAT defaults to `buffer-file-format', it is a symbol naming
345 one of the formats defined in `format-alist', or a list of such symbols."
346 (interactive
347 (list (region-beginning) (region-end)
348 (format-read (format "Translate region to format (default %s): "
349 buffer-file-format))))
350 (if (null format) (setq format buffer-file-format))
351 (if (symbolp format) (setq format (list format)))
352 (save-excursion
353 (goto-char end)
354 (let ((cur-buf (current-buffer))
355 (end (point-marker)))
356 (while format
357 (let* ((info (assq (car format) format-alist))
358 (to-fn (nth 4 info))
359 (modify (nth 5 info))
360 result)
361 (if to-fn
362 (if modify
363 (setq end (format-encode-run-method to-fn beg end
364 (current-buffer)))
365 (format-insert-annotations
366 (funcall to-fn beg end (current-buffer)))))
367 (setq format (cdr format)))))))
369 (defun format-write-file (filename format)
370 "Write current buffer into file FILENAME using some FORMAT.
371 Makes buffer visit that file and sets the format as the default for future
372 saves. If the buffer is already visiting a file, you can specify a directory
373 name as FILENAME, to write a file of the same old name in that directory."
374 (interactive
375 ;; Same interactive spec as write-file, plus format question.
376 (let* ((file (if buffer-file-name
377 (read-file-name "Write file: "
378 nil nil nil nil)
379 (read-file-name "Write file: "
380 (cdr (assq 'default-directory
381 (buffer-local-variables)))
382 nil nil (buffer-name))))
383 (fmt (format-read (format "Write file `%s' in format: "
384 (file-name-nondirectory file)))))
385 (list file fmt)))
386 (let ((old-formats buffer-file-format)
387 preserve-formats)
388 (dolist (fmt old-formats)
389 (let ((aelt (assq fmt format-alist)))
390 (if (nth 7 aelt)
391 (push fmt preserve-formats))))
392 (setq buffer-file-format format)
393 (dolist (fmt preserve-formats)
394 (unless (memq fmt buffer-file-format)
395 (setq buffer-file-format (append buffer-file-format (list fmt))))))
396 (write-file filename))
398 (defun format-find-file (filename format)
399 "Find the file FILENAME using data format FORMAT.
400 If FORMAT is nil then do not do any format conversion."
401 (interactive
402 ;; Same interactive spec as write-file, plus format question.
403 (let* ((file (read-file-name "Find file: "))
404 (fmt (format-read (format "Read file `%s' in format: "
405 (file-name-nondirectory file)))))
406 (list file fmt)))
407 (let ((format-alist nil))
408 (find-file filename))
409 (if format
410 (format-decode-buffer format)))
412 (defun format-insert-file (filename format &optional beg end)
413 "Insert the contents of file FILENAME using data format FORMAT.
414 If FORMAT is nil then do not do any format conversion.
415 The optional third and fourth arguments BEG and END specify
416 the part of the file to read.
418 The return value is like the value of `insert-file-contents':
419 a list (ABSOLUTE-FILE-NAME . SIZE)."
420 (interactive
421 ;; Same interactive spec as write-file, plus format question.
422 (let* ((file (read-file-name "Find file: "))
423 (fmt (format-read (format "Read file `%s' in format: "
424 (file-name-nondirectory file)))))
425 (list file fmt)))
426 (let (value size)
427 (let ((format-alist nil))
428 (setq value (insert-file-contents filename nil beg end))
429 (setq size (nth 1 value)))
430 (if format
431 (setq size (format-decode format size)
432 value (cons (car value) size)))
433 value))
435 (defun format-read (&optional prompt)
436 "Read and return the name of a format.
437 Return value is a list, like `buffer-file-format'; it may be nil.
438 Formats are defined in `format-alist'. Optional arg is the PROMPT to use."
439 (let* ((table (mapcar (lambda (x) (list (symbol-name (car x))))
440 format-alist))
441 (ans (completing-read (or prompt "Format: ") table nil t)))
442 (if (not (equal "" ans)) (list (intern ans)))))
446 ;;; Below are some functions that may be useful in writing encoding and
447 ;;; decoding functions for use in format-alist.
450 (defun format-replace-strings (alist &optional reverse beg end)
451 "Do multiple replacements on the buffer.
452 ALIST is a list of (FROM . TO) pairs, which should be proper arguments to
453 `search-forward' and `replace-match' respectively.
454 Optional 2nd arg REVERSE, if non-nil, means the pairs are (TO . FROM), so that
455 you can use the same list in both directions if it contains only literal
456 strings.
457 Optional args BEG and END specify a region of the buffer on which to operate."
458 (save-excursion
459 (save-restriction
460 (or beg (setq beg (point-min)))
461 (if end (narrow-to-region (point-min) end))
462 (while alist
463 (let ((from (if reverse (cdr (car alist)) (car (car alist))))
464 (to (if reverse (car (car alist)) (cdr (car alist)))))
465 (goto-char beg)
466 (while (search-forward from nil t)
467 (goto-char (match-beginning 0))
468 (insert to)
469 (set-text-properties (- (point) (length to)) (point)
470 (text-properties-at (point)))
471 (delete-region (point) (+ (point) (- (match-end 0)
472 (match-beginning 0)))))
473 (setq alist (cdr alist)))))))
475 ;;; Some list-manipulation functions that we need.
477 (defun format-delq-cons (cons list)
478 "Remove the given CONS from LIST by side effect and return the new LIST.
479 Since CONS could be the first element of LIST, write
480 `\(setq foo \(format-delq-cons element foo))' to be sure of changing
481 the value of `foo'."
482 (if (eq cons list)
483 (cdr list)
484 (let ((p list))
485 (while (not (eq (cdr p) cons))
486 (if (null p) (error "format-delq-cons: not an element"))
487 (setq p (cdr p)))
488 ;; Now (cdr p) is the cons to delete
489 (setcdr p (cdr cons))
490 list)))
492 (defun format-make-relatively-unique (a b)
493 "Delete common elements of lists A and B, return as pair.
494 Compares using `equal'."
495 (let* ((acopy (copy-sequence a))
496 (bcopy (copy-sequence b))
497 (tail acopy))
498 (while tail
499 (let ((dup (member (car tail) bcopy))
500 (next (cdr tail)))
501 (if dup (setq acopy (format-delq-cons tail acopy)
502 bcopy (format-delq-cons dup bcopy)))
503 (setq tail next)))
504 (cons acopy bcopy)))
506 (defun format-common-tail (a b)
507 "Given two lists that have a common tail, return it.
508 Compares with `equal', and returns the part of A that is equal to the
509 equivalent part of B. If even the last items of the two are not equal,
510 returns nil."
511 (let ((la (length a))
512 (lb (length b)))
513 ;; Make sure they are the same length
514 (if (> la lb)
515 (setq a (nthcdr (- la lb) a))
516 (setq b (nthcdr (- lb la) b))))
517 (while (not (equal a b))
518 (setq a (cdr a)
519 b (cdr b)))
522 (defun format-proper-list-p (list)
523 "Return t if LIST is a proper list.
524 A proper list is a list ending with a nil cdr, not with an atom "
525 (when (listp list)
526 (while (consp list)
527 (setq list (cdr list)))
528 (null list)))
530 (defun format-reorder (items order)
531 "Arrange ITEMS to following partial ORDER.
532 Elements of ITEMS equal to elements of ORDER will be rearranged to follow the
533 ORDER. Unmatched items will go last."
534 (if order
535 (let ((item (member (car order) items)))
536 (if item
537 (cons (car item)
538 (format-reorder (format-delq-cons item items)
539 (cdr order)))
540 (format-reorder items (cdr order))))
541 items))
543 (put 'face 'format-list-valued t) ; These text-properties take values
544 (put 'unknown 'format-list-valued t) ; that are lists, the elements of which
545 ; should be considered separately.
546 ; See format-deannotate-region and
547 ; format-annotate-region.
549 ;; This text property has list values, but they are treated atomically.
551 (put 'display 'format-list-atomic-p t)
554 ;;; Decoding
557 (defun format-deannotate-region (from to translations next-fn)
558 "Translate annotations in the region into text properties.
559 This sets text properties between FROM to TO as directed by the
560 TRANSLATIONS and NEXT-FN arguments.
562 NEXT-FN is a function that searches forward from point for an annotation.
563 It should return a list of 4 elements: \(BEGIN END NAME POSITIVE). BEGIN and
564 END are buffer positions bounding the annotation, NAME is the name searched
565 for in TRANSLATIONS, and POSITIVE should be non-nil if this annotation marks
566 the beginning of a region with some property, or nil if it ends the region.
567 NEXT-FN should return nil if there are no annotations after point.
569 The basic format of the TRANSLATIONS argument is described in the
570 documentation for the `format-annotate-region' function. There are some
571 additional things to keep in mind for decoding, though:
573 When an annotation is found, the TRANSLATIONS list is searched for a
574 text-property name and value that corresponds to that annotation. If the
575 text-property has several annotations associated with it, it will be used only
576 if the other annotations are also in effect at that point. The first match
577 found whose annotations are all present is used.
579 The text property thus determined is set to the value over the region between
580 the opening and closing annotations. However, if the text-property name has a
581 non-nil `format-list-valued' property, then the value will be consed onto the
582 surrounding value of the property, rather than replacing that value.
584 There are some special symbols that can be used in the \"property\" slot of
585 the TRANSLATIONS list: PARAMETER and FUNCTION \(spelled in uppercase).
586 Annotations listed under the pseudo-property PARAMETER are considered to be
587 arguments of the immediately surrounding annotation; the text between the
588 opening and closing parameter annotations is deleted from the buffer but saved
589 as a string.
591 The surrounding annotation should be listed under the pseudo-property
592 FUNCTION. Instead of inserting a text-property for this annotation,
593 the function listed in the VALUE slot is called to make whatever
594 changes are appropriate. It can also return a list of the form
595 \(START LOC PROP VALUE) which specifies a property to put on. The
596 function's first two arguments are the START and END locations, and
597 the rest of the arguments are any PARAMETERs found in that region.
599 Any annotations that are found by NEXT-FN but not defined by TRANSLATIONS
600 are saved as values of the `unknown' text-property \(which is list-valued).
601 The TRANSLATIONS list should usually contain an entry of the form
602 \(unknown \(nil format-annotate-value))
603 to write these unknown annotations back into the file."
604 (save-excursion
605 (save-restriction
606 (narrow-to-region (point-min) to)
607 (goto-char from)
608 (let (next open-ans todo loc unknown-ans)
609 (while (setq next (funcall next-fn))
610 (let* ((loc (nth 0 next))
611 (end (nth 1 next))
612 (name (nth 2 next))
613 (positive (nth 3 next))
614 (found nil))
616 ;; Delete the annotation
617 (delete-region loc end)
618 (cond
619 ;; Positive annotations are stacked, remembering location
620 (positive (push `(,name ((,loc . nil))) open-ans))
621 ;; It is a negative annotation:
622 ;; Close the top annotation & add its text property.
623 ;; If the file's nesting is messed up, the close might not match
624 ;; the top thing on the open-annotations stack.
625 ;; If no matching annotation is open, just ignore the close.
626 ((not (assoc name open-ans))
627 (message "Extra closing annotation (%s) in file" name))
628 ;; If one is open, but not on the top of the stack, close
629 ;; the things in between as well. Set `found' when the real
630 ;; one is closed.
632 (while (not found)
633 (let* ((top (car open-ans)) ; first on stack: should match.
634 (top-name (car top)) ; text property name
635 (top-extents (nth 1 top)) ; property regions
636 (params (cdr (cdr top))) ; parameters
637 (aalist translations)
638 (matched nil))
639 (if (equal name top-name)
640 (setq found t)
641 (message "Improper nesting in file."))
642 ;; Look through property names in TRANSLATIONS
643 (while aalist
644 (let ((prop (car (car aalist)))
645 (alist (cdr (car aalist))))
646 ;; And look through values for each property
647 (while alist
648 (let ((value (car (car alist)))
649 (ans (cdr (car alist))))
650 (if (member top-name ans)
651 ;; This annotation is listed, but still have to
652 ;; check if multiple annotations are satisfied
653 (if (member nil (mapcar (lambda (r)
654 (assoc r open-ans))
655 ans))
656 nil ; multiple ans not satisfied
657 ;; If there are multiple annotations going
658 ;; into one text property, split up the other
659 ;; annotations so they apply individually to
660 ;; the other regions.
661 (setcdr (car top-extents) loc)
662 (let ((to-split ans) this-one extents)
663 (while to-split
664 (setq this-one
665 (assoc (car to-split) open-ans)
666 extents (nth 1 this-one))
667 (if (not (eq this-one top))
668 (setcar (cdr this-one)
669 (format-subtract-regions
670 extents top-extents)))
671 (setq to-split (cdr to-split))))
672 ;; Set loop variables to nil so loop
673 ;; will exit.
674 (setq alist nil aalist nil matched t
675 ;; pop annotation off stack.
676 open-ans (cdr open-ans))
677 (let ((extents top-extents)
678 (start (car (car top-extents)))
679 (loc (cdr (car top-extents))))
680 (while extents
681 (cond
682 ;; Check for pseudo-properties
683 ((eq prop 'PARAMETER)
684 ;; A parameter of the top open ann:
685 ;; delete text and use as arg.
686 (if open-ans
687 ;; (If nothing open, discard).
688 (setq open-ans
689 (cons
690 (append (car open-ans)
691 (list
692 (buffer-substring
693 start loc)))
694 (cdr open-ans))))
695 (delete-region start loc))
696 ((eq prop 'FUNCTION)
697 ;; Not a property, but a function.
698 (let ((rtn
699 (apply value start loc params)))
700 (if rtn (push rtn todo))))
702 ;; Normal property/value pair
703 (setq todo
704 (cons (list start loc prop value)
705 todo))))
706 (setq extents (cdr extents)
707 start (car (car extents))
708 loc (cdr (car extents))))))))
709 (setq alist (cdr alist))))
710 (setq aalist (cdr aalist)))
711 (if (not matched)
712 ;; Didn't find any match for the annotation:
713 ;; Store as value of text-property `unknown'.
714 (let ((extents top-extents)
715 (start (car (car top-extents)))
716 (loc (or (cdr (car top-extents)) loc)))
717 (while extents
718 (setq open-ans (cdr open-ans)
719 todo (cons (list start loc 'unknown top-name)
720 todo)
721 unknown-ans (cons name unknown-ans)
722 extents (cdr extents)
723 start (car (car extents))
724 loc (cdr (car extents))))))))))))
726 ;; Once entire file has been scanned, add the properties.
727 (while todo
728 (let* ((item (car todo))
729 (from (nth 0 item))
730 (to (nth 1 item))
731 (prop (nth 2 item))
732 (val (nth 3 item)))
734 (if (numberp val) ; add to ambient value if numeric
735 (format-property-increment-region from to prop val 0)
736 (put-text-property
737 from to prop
738 (cond ((get prop 'format-list-valued) ; value gets consed onto
739 ; list-valued properties
740 (let ((prev (get-text-property from prop)))
741 (cons val (if (listp prev) prev (list prev)))))
742 (t val))))) ; normally, just set to val.
743 (setq todo (cdr todo)))
745 (if unknown-ans
746 (message "Unknown annotations: %s" unknown-ans))))))
748 (defun format-subtract-regions (minu subtra)
749 "Remove from the regions in MINUend the regions in SUBTRAhend.
750 A region is a dotted pair (FROM . TO). Both parameters are lists of
751 regions. Each list must contain nonoverlapping, noncontiguous
752 regions, in descending order. The result is also nonoverlapping,
753 noncontiguous, and in descending order. The first element of MINUEND
754 can have a cdr of nil, indicating that the end of that region is not
755 yet known."
756 (let* ((minuend (copy-alist minu))
757 (subtrahend (copy-alist subtra))
758 (m (car minuend))
759 (s (car subtrahend))
760 results)
761 (while (and minuend subtrahend)
762 (cond
763 ;; The minuend starts after the subtrahend ends; keep it.
764 ((> (car m) (cdr s))
765 (push m results)
766 (setq minuend (cdr minuend)
767 m (car minuend)))
768 ;; The minuend extends beyond the end of the subtrahend. Chop it off.
769 ((or (null (cdr m)) (> (cdr m) (cdr s)))
770 (push (cons (1+ (cdr s)) (cdr m)) results)
771 (setcdr m (cdr s)))
772 ;; The subtrahend starts after the minuend ends; throw it away.
773 ((< (cdr m) (car s))
774 (setq subtrahend (cdr subtrahend) s (car subtrahend)))
775 ;; The subtrahend extends beyond the end of the minuend. Chop it off.
776 (t ;(<= (cdr m) (cdr s)))
777 (if (>= (car m) (car s))
778 (setq minuend (cdr minuend) m (car minuend))
779 (setcdr m (1- (car s)))
780 (setq subtrahend (cdr subtrahend) s (car subtrahend))))))
781 (nconc (nreverse results) minuend)))
783 ;; This should probably go somewhere other than format.el. Then again,
784 ;; indent.el has alter-text-property. NOTE: We can also use
785 ;; next-single-property-change instead of text-property-not-all, but then
786 ;; we have to see if we passed TO.
787 (defun format-property-increment-region (from to prop delta default)
788 "Over the region between FROM and TO increment property PROP by amount DELTA.
789 DELTA may be negative. If property PROP is nil anywhere
790 in the region, it is treated as though it were DEFAULT."
791 (let ((cur from) val newval next)
792 (while cur
793 (setq val (get-text-property cur prop)
794 newval (+ (or val default) delta)
795 next (text-property-not-all cur to prop val))
796 (put-text-property cur (or next to) prop newval)
797 (setq cur next))))
800 ;;; Encoding
803 (defun format-insert-annotations (list &optional offset)
804 "Apply list of annotations to buffer as `write-region' would.
805 Inserts each element of the given LIST of buffer annotations at its
806 appropriate place. Use second arg OFFSET if the annotations' locations are
807 not relative to the beginning of the buffer: annotations will be inserted
808 at their location-OFFSET+1 \(ie, the offset is treated as the character number
809 of the first character in the buffer)."
810 (if (not offset)
811 (setq offset 0)
812 (setq offset (1- offset)))
813 (let ((l (reverse list)))
814 (while l
815 (goto-char (- (car (car l)) offset))
816 (insert (cdr (car l)))
817 (setq l (cdr l)))))
819 (defun format-annotate-value (old new)
820 "Return OLD and NEW as a \(CLOSE . OPEN) annotation pair.
821 Useful as a default function for TRANSLATIONS alist when the value of the text
822 property is the name of the annotation that you want to use, as it is for the
823 `unknown' text property."
824 (cons (if old (list old))
825 (if new (list new))))
827 (defun format-annotate-region (from to translations format-fn ignore)
828 "Generate annotations for text properties in the region.
829 Searches for changes between FROM and TO, and describes them with a list of
830 annotations as defined by alist TRANSLATIONS and FORMAT-FN. IGNORE lists text
831 properties not to consider; any text properties that are neither ignored nor
832 listed in TRANSLATIONS are warned about.
833 If you actually want to modify the region, give the return value of this
834 function to `format-insert-annotations'.
836 Format of the TRANSLATIONS argument:
838 Each element is a list whose car is a PROPERTY, and the following
839 elements have the form (VALUE ANNOTATIONS...).
840 Whenever the property takes on the value VALUE, the annotations
841 \(as formatted by FORMAT-FN) are inserted into the file.
842 When the property stops having that value, the matching negated annotation
843 will be inserted \(it may actually be closed earlier and reopened, if
844 necessary, to keep proper nesting).
846 If VALUE is a list, then each element of the list is dealt with
847 separately.
849 If a VALUE is numeric, then it is assumed that there is a single annotation
850 and each occurrence of it increments the value of the property by that number.
851 Thus, given the entry \(left-margin \(4 \"indent\")), if the left margin
852 changes from 4 to 12, two <indent> annotations will be generated.
854 If the VALUE is nil, then instead of annotations, a function should be
855 specified. This function is used as a default: it is called for all
856 transitions not explicitly listed in the table. The function is called with
857 two arguments, the OLD and NEW values of the property. It should return
858 a cons cell (CLOSE . OPEN) as `format-annotate-single-property-change' does.
860 The same TRANSLATIONS structure can be used in reverse for reading files."
861 (let ((all-ans nil) ; All annotations - becomes return value
862 (open-ans nil) ; Annotations not yet closed
863 (loc nil) ; Current location
864 (not-found nil)) ; Properties that couldn't be saved
865 (while (or (null loc)
866 (and (setq loc (next-property-change loc nil to))
867 (< loc to)))
868 (or loc (setq loc from))
869 (let* ((ans (format-annotate-location loc (= loc from) ignore translations))
870 (neg-ans (format-reorder (aref ans 0) open-ans))
871 (pos-ans (aref ans 1))
872 (ignored (aref ans 2)))
873 (setq not-found (append ignored not-found)
874 ignore (append ignored ignore))
875 ;; First do the negative (closing) annotations
876 (while neg-ans
877 ;; Check if it's missing. This can happen (eg, a numeric property
878 ;; going negative can generate closing annotations before there are
879 ;; any open). Warn user & ignore.
880 (if (not (member (car neg-ans) open-ans))
881 (message "Can't close %s: not open." (car neg-ans))
882 (while (not (equal (car neg-ans) (car open-ans)))
883 ;; To close anno. N, need to first close ans 1 to N-1,
884 ;; remembering to re-open them later.
885 (push (car open-ans) pos-ans)
886 (setq all-ans
887 (cons (cons loc (funcall format-fn (car open-ans) nil))
888 all-ans))
889 (setq open-ans (cdr open-ans)))
890 ;; Now remove the one we're really interested in from open list.
891 (setq open-ans (cdr open-ans))
892 ;; And put the closing annotation here.
893 (push (cons loc (funcall format-fn (car neg-ans) nil))
894 all-ans))
895 (setq neg-ans (cdr neg-ans)))
896 ;; Now deal with positive (opening) annotations
897 (let ((p pos-ans))
898 (while pos-ans
899 (push (car pos-ans) open-ans)
900 (push (cons loc (funcall format-fn (car pos-ans) t))
901 all-ans)
902 (setq pos-ans (cdr pos-ans))))))
904 ;; Close any annotations still open
905 (while open-ans
906 (setq all-ans
907 (cons (cons to (funcall format-fn (car open-ans) nil))
908 all-ans))
909 (setq open-ans (cdr open-ans)))
910 (if not-found
911 (message "These text properties could not be saved:\n %s"
912 not-found))
913 (nreverse all-ans)))
915 ;;; Internal functions for format-annotate-region.
917 (defun format-annotate-location (loc all ignore translations)
918 "Return annotation(s) needed at location LOC.
919 This includes any properties that change between LOC-1 and LOC.
920 If ALL is true, don't look at previous location, but generate annotations for
921 all non-nil properties.
922 Third argument IGNORE is a list of text-properties not to consider.
923 Use the TRANSLATIONS alist (see `format-annotate-region' for doc).
925 Return value is a vector of 3 elements:
926 1. List of annotations to close
927 2. List of annotations to open.
928 3. List of properties that were ignored or couldn't be annotated.
930 The annotations in lists 1 and 2 need not be strings.
931 They can be whatever the FORMAT-FN in `format-annotate-region'
932 can handle. If that is `enriched-make-annotation', they can be
933 either strings, or lists of the form (PARAMETER VALUE)."
934 (let* ((prev-loc (1- loc))
935 (before-plist (if all nil (text-properties-at prev-loc)))
936 (after-plist (text-properties-at loc))
937 p negatives positives prop props not-found)
938 ;; make list of all property names involved
939 (setq p before-plist)
940 (while p
941 (if (not (memq (car p) props))
942 (push (car p) props))
943 (setq p (cdr (cdr p))))
944 (setq p after-plist)
945 (while p
946 (if (not (memq (car p) props))
947 (push (car p) props))
948 (setq p (cdr (cdr p))))
950 (while props
951 (setq prop (pop props))
952 (if (memq prop ignore)
953 nil ; If it's been ignored before, ignore it now.
954 (let ((before (if all nil (car (cdr (memq prop before-plist)))))
955 (after (car (cdr (memq prop after-plist)))))
956 (if (equal before after)
957 nil ; no change; ignore
958 (let ((result (format-annotate-single-property-change
959 prop before after translations)))
960 (if (not result)
961 (push prop not-found)
962 (setq negatives (nconc negatives (car result))
963 positives (nconc positives (cdr result)))))))))
964 (vector negatives positives not-found)))
966 (defun format-annotate-single-property-change (prop old new translations)
967 "Return annotations for property PROP changing from OLD to NEW.
968 These are searched for in the translations alist TRANSLATIONS
969 (see `format-annotate-region' for the format).
970 If NEW does not appear in the list, but there is a default function, then that
971 function is called.
972 Returns a cons of the form (CLOSE . OPEN)
973 where CLOSE is a list of annotations to close
974 and OPEN is a list of annotations to open.
976 The annotations in CLOSE and OPEN need not be strings.
977 They can be whatever the FORMAT-FN in `format-annotate-region'
978 can handle. If that is `enriched-make-annotation', they can be
979 either strings, or lists of the form (PARAMETER VALUE)."
981 (let ((prop-alist (cdr (assoc prop translations)))
982 default)
983 (if (not prop-alist)
985 ;; If either old or new is a list, have to treat both that way.
986 (if (and (or (listp old) (listp new))
987 (not (get prop 'format-list-atomic-p)))
988 (if (or (not (format-proper-list-p old))
989 (not (format-proper-list-p new)))
990 (format-annotate-atomic-property-change prop-alist old new)
991 (let* ((old (if (listp old) old (list old)))
992 (new (if (listp new) new (list new)))
993 (tail (format-common-tail old new))
994 close open)
995 (while old
996 (setq close
997 (append (car (format-annotate-atomic-property-change
998 prop-alist (car old) nil))
999 close)
1000 old (cdr old)))
1001 (while new
1002 (setq open
1003 (append (cdr (format-annotate-atomic-property-change
1004 prop-alist nil (car new)))
1005 open)
1006 new (cdr new)))
1007 (format-make-relatively-unique close open)))
1008 (format-annotate-atomic-property-change prop-alist old new)))))
1010 (defun format-annotate-atomic-property-change (prop-alist old new)
1011 "Internal function annotate a single property change.
1012 PROP-ALIST is the relevant element of a TRANSLATIONS list.
1013 OLD and NEW are the values."
1014 (let (num-ann)
1015 ;; If old and new values are numbers,
1016 ;; look for a number in PROP-ALIST.
1017 (if (and (or (null old) (numberp old))
1018 (or (null new) (numberp new)))
1019 (progn
1020 (setq num-ann prop-alist)
1021 (while (and num-ann (not (numberp (car (car num-ann)))))
1022 (setq num-ann (cdr num-ann)))))
1023 (if num-ann
1024 ;; Numerical annotation - use difference
1025 (progn
1026 ;; If property is numeric, nil means 0
1027 (cond ((and (numberp old) (null new))
1028 (setq new 0))
1029 ((and (numberp new) (null old))
1030 (setq old 0)))
1032 (let* ((entry (car num-ann))
1033 (increment (car entry))
1034 (n (ceiling (/ (float (- new old)) (float increment))))
1035 (anno (car (cdr entry))))
1036 (if (> n 0)
1037 (cons nil (make-list n anno))
1038 (cons (make-list (- n) anno) nil))))
1040 ;; Standard annotation
1041 (let ((close (and old (cdr (assoc old prop-alist))))
1042 (open (and new (cdr (assoc new prop-alist)))))
1043 (if (or close open)
1044 (format-make-relatively-unique close open)
1045 ;; Call "Default" function, if any
1046 (let ((default (assq nil prop-alist)))
1047 (if default
1048 (funcall (car (cdr default)) old new))))))))
1050 (provide 'format)
1052 ;;; arch-tag: c387e9c7-a93d-47bf-89bc-8ca67e96755a
1053 ;;; format.el ends here