Commit | Line | Data |
---|---|---|
f600cf3a | 1 | ;; unidata-gen.el -- Create files containing character property data. |
835a3da7 | 2 | ;; Copyright (C) 2005, 2006, 2007, 2008 |
f600cf3a KH |
3 | ;; National Institute of Advanced Industrial Science and Technology (AIST) |
4 | ;; Registration Number H13PRO009 | |
5 | ||
6 | ;; This file is part of GNU Emacs. | |
7 | ||
8 | ;; GNU Emacs is free software; you can redistribute it and/or modify | |
9 | ;; it under the terms of the GNU General Public License as published by | |
835a3da7 | 10 | ;; the Free Software Foundation; either version 3, or (at your option) |
f600cf3a KH |
11 | ;; any later version. |
12 | ||
13 | ;; GNU Emacs is distributed in the hope that it will be useful, | |
14 | ;; but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | ;; GNU General Public License for more details. | |
17 | ||
18 | ;; You should have received a copy of the GNU General Public License | |
19 | ;; along with GNU Emacs; see the file COPYING. If not, write to the | |
835a3da7 GM |
20 | ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
21 | ;; Boston, MA 02110-1301, USA. | |
f600cf3a KH |
22 | |
23 | ;;; Commentary: | |
24 | ||
1275c1dc KH |
25 | ;; SPECIAL NOTICE |
26 | ;; | |
27 | ;; This file must be byte-compilable/loadable by `temacs' and also | |
28 | ;; the entry function `unidata-gen-files' must be runnable by | |
29 | ;; `temacs'. | |
30 | ||
f600cf3a KH |
31 | ;; FILES TO BE GENERATED |
32 | ;; | |
1275c1dc | 33 | ;; The entry function `unidata-gen-files' generates these files in |
f600cf3a KH |
34 | ;; the current directory. |
35 | ;; | |
36 | ;; charprop.el | |
37 | ;; It contains a series of forms of this format: | |
38 | ;; (char-code-property-register PROP FILE) | |
39 | ;; where PROP is a symbol representing a character property | |
1275c1dc | 40 | ;; (name, generic-category, etc), and FILE is a name of one of |
f600cf3a KH |
41 | ;; the following files. |
42 | ;; | |
43 | ;; uni-name.el, uni-cat.el, uni-comb.el, uni-bidi.el | |
44 | ;; It contains a single form of this format: | |
45 | ;; (char-code-property-register PROP CHAR-TABLE) | |
46 | ;; where PROP is the same as above, and CHAR-TABLE is a | |
1275c1dc | 47 | ;; char-table containing property values in a compressed format. |
f600cf3a KH |
48 | ;; |
49 | ;; When they are installed in .../lisp/international/, the file | |
50 | ;; "charprop.el" is preloaded in loadup.el. The other files are | |
51 | ;; automatically loaded when the functions `get-char-code-property' | |
52 | ;; and `put-char-code-property' are called. | |
53 | ;; | |
54 | ;; FORMAT OF A CHAR TABLE | |
55 | ;; | |
56 | ;; We want to make a file size containing a char-table small. We | |
57 | ;; also want to load the file and get a property value fast. We | |
58 | ;; also want to reduce the used memory after loading it. So, | |
59 | ;; instead of naively storing a property value for each character in | |
60 | ;; a char-table (and write it out into a file), we store compressed | |
61 | ;; data in a char-table as below. | |
62 | ;; | |
63 | ;; If succeeding 128*N characters have the same property value, we | |
1275c1dc | 64 | ;; store that value for them. Otherwise, compress values for |
f600cf3a KH |
65 | ;; succeeding 128 characters into a single string and store it as a |
66 | ;; value for those characters. The way of compression depends on a | |
67 | ;; property. See the section "SIMPLE TABLE", "RUN-LENGTH TABLE", | |
68 | ;; and "WORD-LIST TABLE". | |
69 | ||
70 | ;; The char table has four extra slots: | |
71 | ;; 1st: property symbol | |
72 | ;; 2nd: function to call to get a property value | |
73 | ;; 3nd: function to call to put a property value | |
74 | ;; 4th: function to call to get a description of a property value | |
75 | ;; 5th: data referred by the above functions | |
f600cf3a | 76 | |
f600cf3a KH |
77 | ;; List of elements of this form: |
78 | ;; (CHAR-or-RANGE PROP1 PROP2 ... PROPn) | |
79 | ;; CHAR-or-RANGE: a character code or a cons of character codes | |
80 | ;; PROPn: string representing the nth property value | |
81 | ||
3921ea21 KH |
82 | (defvar unidata-list nil) |
83 | ||
84 | (defun unidata-setup-list (unidata-text-file) | |
f600cf3a KH |
85 | (let* ((table (list nil)) |
86 | (tail table) | |
87 | (block-names '(("^<CJK Ideograph" . CJK\ IDEOGRAPH) | |
88 | ("^<Hangul Syllable" . HANGUL\ SYLLABLE) | |
89 | ("^<.*Surrogate" . nil) | |
90 | ("^<.*Private Use" . PRIVATE\ USE))) | |
91 | val char name) | |
92 | (or (file-readable-p unidata-text-file) | |
93 | (error "File not readable: %s" unidata-text-file)) | |
94 | (with-temp-buffer | |
1275c1dc | 95 | (insert-file-contents unidata-text-file) |
f600cf3a KH |
96 | (goto-char (point-min)) |
97 | (condition-case nil | |
98 | (while t | |
99 | (setq val (read (current-buffer)) | |
100 | char (car val) | |
101 | name (cadr val)) | |
102 | ||
103 | ;; Check this kind of block. | |
104 | ;; 4E00;<CJK Ideograph, First>;Lo;0;L;;;;;N;;;;; | |
105 | ;; 9FA5;<CJK Ideograph, Last>;Lo;0;L;;;;;N;;;;; | |
106 | (if (and (= (aref name 0) ?<) | |
107 | (string-match ", First>$" name)) | |
108 | (let ((first char) | |
109 | (l block-names) | |
110 | block-name) | |
111 | (setq val (read (current-buffer)) | |
112 | char (car val) | |
113 | block-name (cadr val) | |
114 | name nil) | |
115 | (while l | |
116 | (if (string-match (caar l) block-name) | |
117 | (setq name (cdar l) l nil) | |
118 | (setq l (cdr l)))) | |
119 | (if (not name) | |
120 | ;; As this is a surrogate pair range, ignore it. | |
121 | (setq val nil) | |
122 | (setcar val (cons first char)) | |
123 | (setcar (cdr val) name)))) | |
124 | ||
125 | (when val | |
126 | (setcdr tail (list val)) | |
127 | (setq tail (cdr tail)))) | |
128 | (error nil))) | |
3921ea21 | 129 | (setq unidata-list (cdr table)))) |
f600cf3a KH |
130 | |
131 | ;; Alist of this form: | |
132 | ;; (PROP INDEX GENERATOR FILENAME) | |
133 | ;; PROP: character property | |
134 | ;; INDEX: index to each element of unidata-list for PROP | |
135 | ;; GENERATOR: function to generate a char-table | |
136 | ;; FILENAME: filename to store the char-table | |
137 | ;; DESCRIBER: function to call to get a description string of property value | |
138 | ||
139 | (defconst unidata-prop-alist | |
140 | '((name | |
141 | 1 unidata-gen-table-name "uni-name.el" | |
142 | "Unicode character name. | |
143 | Property value is a string.") | |
144 | (general-category | |
145 | 2 unidata-gen-table-symbol "uni-category.el" | |
146 | "Unicode general category. | |
147 | Property value is one of the following symbols: | |
148 | Lu, Ll, Lt, Lm, Lo, Mn, Mc, Me, Nd, Nl, No, Pc, Pd, Ps, Pe, Pi, Pf, Po, | |
149 | Sm, Sc, Sk, So, Zs, Zl, Zp, Cc, Cf, Cs, Co, Cn" | |
150 | unidata-describe-general-category) | |
151 | (canonical-combining-class | |
152 | 3 unidata-gen-table-integer "uni-combining.el" | |
153 | "Unicode canonical combining class. | |
154 | Property value is an integer." | |
155 | unidata-describe-canonical-combining-class) | |
156 | (bidi-class | |
157 | 4 unidata-gen-table-symbol "uni-bidi.el" | |
158 | "Unicode bidi class. | |
159 | Property value is one of the following symbols: | |
160 | L, LRE, LRO, R, AL, RLE, RLO, PDF, EN, ES, ET, | |
161 | AN, CS, NSM, BN, B, S, WS, ON" | |
162 | unidata-describe-bidi-class) | |
163 | (decomposition | |
164 | 5 unidata-gen-table-decomposition "uni-decomposition.el" | |
165 | "Unicode decomposition mapping. | |
166 | Property value is a list of characters. The first element may be | |
1275c1dc | 167 | one of these symbols representing compatibility formatting tag: |
f600cf3a KH |
168 | <font>, <noBreak>, <initial>, <medial>, <final>, <isolated>, <circle>, |
169 | <super>, <sub>, <vertical>, <wide>, <narrow>, <small>, <square>, <fraction>, | |
170 | <compat>" | |
171 | unidata-describe-decomposition) | |
172 | (decimal-digit-value | |
173 | 6 unidata-gen-table-integer "uni-decimal.el" | |
174 | "Unicode numeric value (decimal digit). | |
175 | Property value is an integer.") | |
176 | (digit-value | |
177 | 7 unidata-gen-table-integer "uni-digit.el" | |
178 | "Unicode numeric value (digit). | |
179 | Property value is an integer.") | |
180 | (numeric-value | |
181 | 8 unidata-gen-table-symbol "uni-numeric.el" | |
182 | "Unicode numeric value (numeric). | |
183 | Property value is an symbol.") | |
184 | (mirrored | |
185 | 9 unidata-gen-table-symbol "uni-mirrored.el" | |
186 | "Unicode bidi mirrored flag. | |
187 | Property value is a symbol `Y' or `N'.") | |
188 | (old-name | |
189 | 10 unidata-gen-table-name "uni-old-name.el" | |
190 | "Unicode old names as published in Unicode 1.0. | |
191 | Property value is a string.") | |
192 | (iso-10646-comment | |
193 | 11 unidata-gen-table-name "uni-comment.el" | |
194 | "Unicode ISO 10646 comment. | |
195 | Property value is a string.") | |
196 | (uppercase | |
197 | 12 unidata-gen-table-character "uni-uppercase.el" | |
198 | "Unicode simple uppercase mapping. | |
199 | Property value is a character." | |
200 | string) | |
201 | (lowercase | |
202 | 13 unidata-gen-table-character "uni-lowercase.el" | |
203 | "Unicode simple lowercase mapping. | |
204 | Property value is a character." | |
205 | string) | |
206 | (titlecase | |
207 | 14 unidata-gen-table-character "uni-titlecase.el" | |
208 | "Unicode simple titlecase mapping. | |
209 | Property value is a character." | |
210 | string))) | |
211 | ||
212 | ;; Functions to access the above data. | |
213 | (defsubst unidata-prop-index (prop) (nth 1 (assq prop unidata-prop-alist))) | |
214 | (defsubst unidata-prop-generator (prop) (nth 2 (assq prop unidata-prop-alist))) | |
215 | (defsubst unidata-prop-file (prop) (nth 3 (assq prop unidata-prop-alist))) | |
216 | (defsubst unidata-prop-docstring (prop) (nth 4 (assq prop unidata-prop-alist))) | |
217 | (defsubst unidata-prop-describer (prop) (nth 5 (assq prop unidata-prop-alist))) | |
218 | ||
219 | \f | |
220 | ;; SIMPLE TABLE | |
221 | ;; | |
222 | ;; If the type of character property value is character, and the | |
223 | ;; values of succeeding character codes are usually different, we use | |
224 | ;; a char-table described here to store such values. | |
225 | ;; | |
226 | ;; If succeeding 128 characters has no property, a char-table has the | |
227 | ;; symbol t is for them. Otherwise a char-table has a string of the | |
228 | ;; following format for them. | |
229 | ;; | |
230 | ;; The first character of the string is FIRST-INDEX. | |
231 | ;; The Nth (N > 0) character of the string is a property value of the | |
1275c1dc | 232 | ;; character (BLOCK-HEAD + FIRST-INDEX + N - 1), where BLOCK-HEAD is |
f600cf3a KH |
233 | ;; the first of the characters in the block. |
234 | ;; | |
235 | ;; The 4th extra slot of a char-table is nil. | |
236 | ||
237 | (defun unidata-get-character (char val table) | |
238 | (cond | |
239 | ((characterp val) | |
240 | val) | |
241 | ||
242 | ((stringp val) | |
243 | (let* ((len (length val)) | |
244 | (block-head (lsh (lsh char -7) 7)) | |
245 | (vec (make-vector 128 nil)) | |
246 | (first-index (aref val 0))) | |
247 | (dotimes (i (1- len)) | |
248 | (let ((elt (aref val (1+ i)))) | |
249 | (if (> elt 0) | |
250 | (aset vec (+ first-index i) elt)))) | |
251 | (dotimes (i 128) | |
252 | (aset table (+ block-head i) (aref vec i))) | |
253 | (aref vec (- char block-head)))))) | |
254 | ||
255 | (defun unidata-put-character (char val table) | |
256 | (or (characterp val) | |
257 | (not val) | |
258 | (error "Not an character nor nil: %S" val)) | |
259 | (let ((current-val (aref table char))) | |
260 | (unless (eq current-val val) | |
261 | (if (stringp current-val) | |
262 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
263 | (aset table char val)))) | |
264 | ||
265 | (defun unidata-gen-table-character (prop) | |
266 | (let ((table (make-char-table 'char-code-property-table)) | |
267 | (prop-idx (unidata-prop-index prop)) | |
268 | (vec (make-vector 128 0)) | |
269 | (tail unidata-list) | |
270 | elt range val idx slot) | |
271 | (set-char-table-range table (cons 0 (max-char)) t) | |
272 | (while tail | |
273 | (setq elt (car tail) tail (cdr tail)) | |
274 | (setq range (car elt) | |
275 | val (nth prop-idx elt)) | |
276 | (if (= (length val) 0) | |
277 | (setq val nil) | |
278 | (setq val (string-to-number val 16))) | |
279 | (if (consp range) | |
280 | (if val | |
281 | (set-char-table-range table range val)) | |
282 | (let* ((start (lsh (lsh range -7) 7)) | |
283 | (limit (+ start 127)) | |
284 | first-index last-index) | |
285 | (fillarray vec 0) | |
286 | (if val | |
287 | (aset vec (setq last-index (setq first-index (- range start))) | |
288 | val)) | |
289 | (while (and (setq elt (car tail) range (car elt)) | |
290 | (integerp range) | |
291 | (<= range limit)) | |
292 | (setq val (nth prop-idx elt)) | |
293 | (when (> (length val) 0) | |
294 | (aset vec (setq last-index (- range start)) | |
295 | (string-to-number val 16)) | |
296 | (or first-index | |
297 | (setq first-index last-index))) | |
298 | (setq tail (cdr tail))) | |
299 | (when first-index | |
300 | (let ((str (string first-index)) | |
301 | c) | |
302 | (while (<= first-index last-index) | |
303 | (setq str (format "%s%c" str (or (aref vec first-index) 0)) | |
304 | first-index (1+ first-index))) | |
305 | (set-char-table-range table (cons start limit) str)))))) | |
306 | ||
307 | (set-char-table-extra-slot table 0 prop) | |
308 | (byte-compile 'unidata-get-character) | |
309 | (byte-compile 'unidata-put-character) | |
310 | (set-char-table-extra-slot table 1 (symbol-function 'unidata-get-character)) | |
311 | (set-char-table-extra-slot table 2 (symbol-function 'unidata-put-character)) | |
312 | ||
313 | table)) | |
314 | ||
315 | ||
316 | \f | |
317 | ;; RUN-LENGTH TABLE | |
318 | ;; | |
319 | ;; If the type of character property value is symbol, integer, | |
320 | ;; boolean, or character, we use a char-table described here to store | |
321 | ;; the values. | |
322 | ;; | |
323 | ;; The 4th extra slot is a vector of property values (VAL-TABLE), and | |
324 | ;; values for succeeding 128 characters are encoded into this | |
325 | ;; character sequence: | |
326 | ;; ( VAL-CODE RUN-LENGTH ? ) + | |
327 | ;; where: | |
328 | ;; VAL-CODE (0..127): | |
329 | ;; (VAL-CODE - 1) is an index into VAL-TABLE. | |
330 | ;; The value 0 means no-value. | |
331 | ;; RUN-LENGTH (130..255): | |
332 | ;; (RUN-LENGTH - 128) specifies how many characters have the same | |
333 | ;; value. If omitted, it means 1. | |
334 | ||
335 | ||
336 | ;; Return a symbol-type character property value of CHAR. VAL is the | |
337 | ;; current value of (aref TABLE CHAR). | |
338 | ||
339 | (defun unidata-get-symbol (char val table) | |
340 | (let ((val-table (char-table-extra-slot table 4))) | |
341 | (cond ((symbolp val) | |
342 | val) | |
343 | ((stringp val) | |
344 | (let ((first-char (lsh (lsh char -7) 7)) | |
345 | (str val) | |
346 | (len (length val)) | |
347 | (idx 0) | |
348 | this-val count) | |
349 | (set-char-table-range table (cons first-char (+ first-char 127)) | |
350 | nil) | |
351 | (while (< idx len) | |
352 | (setq val (aref str idx) idx (1+ idx) | |
353 | count (if (< idx len) (aref str idx) 1)) | |
354 | (setq val (and (> val 0) (aref val-table (1- val))) | |
355 | count (if (< count 128) | |
356 | 1 | |
357 | (prog1 (- count 128) (setq idx (1+ idx))))) | |
358 | (dotimes (i count) | |
359 | (if val | |
360 | (aset table first-char val)) | |
361 | (if (= first-char char) | |
362 | (setq this-val val)) | |
363 | (setq first-char (1+ first-char)))) | |
364 | this-val)) | |
365 | ((> val 0) | |
366 | (aref val-table (1- val)))))) | |
367 | ||
368 | ;; Return a integer-type character property value of CHAR. VAL is the | |
369 | ;; current value of (aref TABLE CHAR). | |
370 | ||
371 | (defun unidata-get-integer (char val table) | |
372 | (let ((val-table (char-table-extra-slot table 4))) | |
373 | (cond ((integerp val) | |
374 | val) | |
375 | ((stringp val) | |
376 | (let ((first-char (lsh (lsh char -7) 7)) | |
377 | (str val) | |
378 | (len (length val)) | |
379 | (idx 0) | |
380 | this-val count) | |
381 | (while (< idx len) | |
382 | (setq val (aref str idx) idx (1+ idx) | |
383 | count (if (< idx len) (aref str idx) 1)) | |
384 | (setq val (and (> val 0) (aref val-table (1- val))) | |
385 | count (if (< count 128) | |
386 | 1 | |
387 | (prog1 (- count 128) (setq idx (1+ idx))))) | |
388 | (dotimes (i count) | |
389 | (aset table first-char val) | |
390 | (if (= first-char char) | |
391 | (setq this-val val)) | |
392 | (setq first-char (1+ first-char)))) | |
393 | this-val))))) | |
394 | ||
395 | ;; Store VAL (symbol) as a character property value of CHAR in TABLE. | |
396 | ||
397 | (defun unidata-put-symbol (char val table) | |
398 | (or (symbolp val) | |
399 | (error "Not a symbol: %S" val)) | |
400 | (let ((current-val (aref table char))) | |
401 | (unless (eq current-val val) | |
402 | (if (stringp current-val) | |
403 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
404 | (aset table char val)))) | |
405 | ||
406 | ;; Store VAL (integer) as a character property value of CHAR in TABLE. | |
407 | ||
408 | (defun unidata-put-integer (char val table) | |
409 | (or (integerp val) | |
410 | (not val) | |
411 | (error "Not an integer nor nil: %S" val)) | |
412 | (let ((current-val (aref table char))) | |
413 | (unless (eq current-val val) | |
414 | (if (stringp current-val) | |
415 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
416 | (aset table char val)))) | |
417 | ||
418 | ;; Encode the character property value VAL into an integer value by | |
419 | ;; VAL-LIST. By side effect, VAL-LIST is modified. | |
420 | ;; VAL-LIST has this form: | |
421 | ;; (t (VAL1 . VAL-CODE1) (VAL2 . VAL-CODE2) ...) | |
422 | ;; If VAL is one of VALn, just return VAL-CODEn. Otherwise, | |
423 | ;; VAL-LIST is modified to this: | |
424 | ;; (t (VAL . (1+ VAL-CODE1)) (VAL1 . VAL-CODE1) (VAL2 . VAL-CODE2) ...) | |
425 | ||
426 | (defun unidata-encode-val (val-list val) | |
427 | (let ((slot (assq val val-list)) | |
428 | val-code) | |
429 | (if slot | |
430 | (cdr slot) | |
431 | (setq val-code (if (cdr val-list) (1+ (cdr (nth 1 val-list))) 1)) | |
432 | (setcdr val-list (cons (cons val val-code) (cdr val-list))) | |
433 | val-code))) | |
434 | ||
435 | ;; Generate a char-table for the character property PROP. | |
436 | ||
437 | (defun unidata-gen-table (prop val-func default-value) | |
438 | (let ((table (make-char-table 'char-code-property-table)) | |
439 | (prop-idx (unidata-prop-index prop)) | |
440 | (val-list (list t)) | |
441 | (vec (make-vector 128 0)) | |
442 | tail elt range val val-code idx slot) | |
443 | (set-char-table-range table (cons 0 (max-char)) default-value) | |
444 | (setq tail unidata-list) | |
445 | (while tail | |
446 | (setq elt (car tail) tail (cdr tail)) | |
447 | (setq range (car elt) | |
448 | val (funcall val-func (nth prop-idx elt))) | |
449 | (setq val-code (if val (unidata-encode-val val-list val))) | |
450 | (if (consp range) | |
451 | (if val-code | |
452 | (set-char-table-range table range val)) | |
453 | (let* ((start (lsh (lsh range -7) 7)) | |
454 | (limit (+ start 127)) | |
455 | str count new-val) | |
456 | (fillarray vec 0) | |
457 | (if val-code | |
458 | (aset vec (- range start) val-code)) | |
459 | (while (and (setq elt (car tail) range (car elt)) | |
460 | (integerp range) | |
461 | (<= range limit)) | |
462 | (setq new-val (funcall val-func (nth prop-idx elt))) | |
463 | (if (not (eq val new-val)) | |
464 | (setq val new-val | |
465 | val-code (if val (unidata-encode-val val-list val)))) | |
466 | (if val-code | |
467 | (aset vec (- range start) val-code)) | |
468 | (setq tail (cdr tail))) | |
469 | (setq str "" val-code -1 count 0) | |
470 | (mapc #'(lambda (x) | |
471 | (if (= val-code x) | |
472 | (setq count (1+ count)) | |
473 | (if (> count 2) | |
474 | (setq str (concat str (string val-code | |
475 | (+ count 128)))) | |
476 | (if (= count 2) | |
477 | (setq str (concat str (string val-code val-code))) | |
478 | (if (= count 1) | |
479 | (setq str (concat str (string val-code)))))) | |
480 | (setq val-code x count 1))) | |
481 | vec) | |
482 | (if (= count 128) | |
483 | (if val | |
484 | (set-char-table-range table (cons start limit) val)) | |
485 | (if (= val-code 0) | |
486 | (set-char-table-range table (cons start limit) str) | |
487 | (if (> count 2) | |
488 | (setq str (concat str (string val-code (+ count 128)))) | |
489 | (if (= count 2) | |
490 | (setq str (concat str (string val-code val-code))) | |
491 | (setq str (concat str (string val-code))))) | |
492 | (set-char-table-range table (cons start limit) str)))))) | |
493 | ||
494 | (setq val-list (nreverse (cdr val-list))) | |
495 | (set-char-table-extra-slot table 0 prop) | |
496 | (set-char-table-extra-slot table 4 (vconcat (mapcar 'car val-list))) | |
497 | table)) | |
498 | ||
499 | (defun unidata-gen-table-symbol (prop) | |
500 | (let ((table (unidata-gen-table prop | |
501 | #'(lambda (x) (and (> (length x) 0) | |
502 | (intern x))) | |
503 | 0))) | |
504 | (byte-compile 'unidata-get-symbol) | |
505 | (byte-compile 'unidata-put-symbol) | |
506 | (set-char-table-extra-slot table 1 (symbol-function 'unidata-get-symbol)) | |
507 | (set-char-table-extra-slot table 2 (symbol-function 'unidata-put-symbol)) | |
508 | table)) | |
509 | ||
510 | (defun unidata-gen-table-integer (prop) | |
511 | (let ((table (unidata-gen-table prop | |
512 | #'(lambda (x) (and (> (length x) 0) | |
513 | (string-to-number x))) | |
514 | t))) | |
515 | (byte-compile 'unidata-get-integer) | |
516 | (byte-compile 'unidata-put-integer) | |
517 | (set-char-table-extra-slot table 1 (symbol-function 'unidata-get-integer)) | |
518 | (set-char-table-extra-slot table 2 (symbol-function 'unidata-put-integer)) | |
519 | table)) | |
520 | ||
521 | \f | |
522 | ;; WORD-LIST TABLE | |
523 | ||
524 | ;; If the table is for `name' property, each character in the string | |
525 | ;; is one of these: | |
526 | ;; DIFF-HEAD-CODE (0, 1, or 2): | |
527 | ;; specifies how to decode the following characters. | |
528 | ;; WORD-CODE (3..#x7FF excluding '-', '0'..'9', 'A'..'Z'): | |
529 | ;; specifies an index number into WORD-TABLE (see below) | |
530 | ;; Otherwise (' ', '-', '0'..'9', 'A'..'Z'): | |
531 | ;; specifies a literal word. | |
532 | ;; | |
533 | ;; The 4th slots is a vector: | |
534 | ;; [ WORD-TABLE BLOCK-NAME HANGUL-JAMO-TABLE ] | |
535 | ;; WORD-TABLE is a vector of word symbols. | |
536 | ;; BLOCK-NAME is a vector of name symbols for a block of characters. | |
537 | ;; HANGUL-JAMO-TABLE is `unidata-name-jamo-name-table'. | |
538 | ||
539 | ;; Return the difference of symbol list L1 and L2 in this form: | |
540 | ;; (DIFF-HEAD SYM1 SYM2 ...) | |
541 | ;; DIFF-HEAD is ((SAME-HEAD-LENGTH * 16) + SAME-TAIL-LENGTH). | |
542 | ;; Ex: If L1 is (a b c d e f) and L2 is (a g h e f), this function | |
543 | ;; returns ((+ (* 1 16) 2) g h). | |
544 | ;; It means that we can get L2 from L1 by prepending the first element | |
545 | ;; of L1 and appending the last 2 elements of L1 to the list (g h). | |
546 | ;; If L1 and L2 don't have common elements at the head and tail, | |
547 | ;; set DIFF-HEAD to -1 and SYM1 ... to the elements of L2. | |
548 | ||
549 | (defun unidata-word-list-diff (l1 l2) | |
550 | (let ((beg 0) | |
551 | (end 0) | |
552 | (len1 (length l1)) | |
553 | (len2 (length l2)) | |
554 | result) | |
555 | (when (< len1 16) | |
556 | (while (and l1 (eq (car l1) (car l2))) | |
557 | (setq beg (1+ beg) | |
558 | l1 (cdr l1) len1 (1- len1) l2 (cdr l2) len2 (1- len2))) | |
559 | (while (and (< end len1) (< end len2) | |
560 | (eq (nth (- len1 end 1) l1) (nth (- len2 end 1) l2))) | |
561 | (setq end (1+ end)))) | |
562 | (if (= (+ beg end) 0) | |
563 | (setq result (list -1)) | |
564 | (setq result (list (+ (* beg 16) (+ beg (- len1 end)))))) | |
565 | (while (< end len2) | |
566 | (setcdr result (cons (nth (- len2 end 1) l2) (cdr result))) | |
567 | (setq end (1+ end))) | |
568 | result)) | |
569 | ||
570 | ;; Return a compressed form of the vector VEC. Each element of VEC is | |
571 | ;; a list of symbols of which names can be concatenated to form a | |
572 | ;; character name. This function changes those elements into | |
573 | ;; compressed forms by utilizing the fact that diff of consecutive | |
574 | ;; elements is usually small. | |
575 | ||
576 | (defun unidata-word-list-compress (vec) | |
577 | (let (last-elt last-idx diff-head tail elt val) | |
578 | (dotimes (i 128) | |
579 | (setq elt (aref vec i)) | |
580 | (when elt | |
581 | (if (null last-elt) | |
582 | (setq diff-head -1 | |
583 | val (cons 0 elt)) | |
584 | (setq val (unidata-word-list-diff last-elt elt)) | |
585 | (if (= (car val) -1) | |
586 | (setq diff-head -1 | |
587 | val (cons 0 (cdr val))) | |
588 | (if (eq diff-head (car val)) | |
589 | (setq val (cons 2 (cdr val))) | |
590 | (setq diff-head (car val)) | |
591 | (if (>= diff-head 0) | |
592 | (setq val (cons 1 val)))))) | |
593 | (aset vec i val) | |
594 | (setq last-idx i last-elt elt))) | |
595 | (if (not last-idx) | |
596 | (setq vec nil) | |
597 | (if (< last-idx 127) | |
598 | (let ((shorter (make-vector (1+ last-idx) nil))) | |
599 | (dotimes (i (1+ last-idx)) | |
600 | (aset shorter i (aref vec i))) | |
601 | (setq vec shorter)))) | |
602 | vec)) | |
603 | ||
604 | ;; Encode the word index IDX into a characters code that can be | |
605 | ;; embedded in a string. | |
606 | ||
607 | (defsubst unidata-encode-word (idx) | |
608 | ;; Exclude 0, 1, 2. | |
609 | (+ idx 3)) | |
610 | ||
611 | ;; Decode the character code CODE (that is embedded in a string) into | |
612 | ;; the corresponding word name by looking up WORD-TABLE. | |
613 | ||
614 | (defsubst unidata-decode-word (code word-table) | |
615 | (setq code (- code 3)) | |
616 | (if (< code (length word-table)) | |
617 | (aref word-table code))) | |
618 | ||
619 | ;; Table of short transliterated name symbols of Hangul Jamo divided | |
620 | ;; into Choseong, Jungseong, and Jongseong. | |
621 | ||
622 | (defconst unidata-name-jamo-name-table | |
623 | [[G GG N D DD R M B BB S SS nil J JJ C K T P H] | |
624 | [A AE YA YAE EO E YEO YE O WA WAE OE YO U WEO WE WI YU EU YI I] | |
625 | [G GG GS N NJ NH D L LG LM LB LS LT LP LH M B BS S SS NG J C K T P H]]) | |
626 | ||
627 | ;; Return a name of CHAR. VAL is the current value of (aref TABLE | |
628 | ;; CHAR). | |
629 | ||
630 | (defun unidata-get-name (char val table) | |
631 | (cond | |
632 | ((stringp val) | |
633 | (if (> (aref val 0) 0) | |
634 | val | |
635 | (let* ((first-char (lsh (lsh char -7) 7)) | |
636 | (word-table (aref (char-table-extra-slot table 4) 0)) | |
637 | (i 1) | |
638 | (len (length val)) | |
639 | (vec (make-vector 128 nil)) | |
640 | (idx 0) | |
641 | (case-fold-search nil) | |
642 | c word-list tail-list last-list word diff-head) | |
643 | (while (< i len) | |
644 | (setq c (aref val i)) | |
645 | (if (< c 3) | |
646 | (progn | |
647 | (if (or word-list tail-list) | |
648 | (aset vec idx | |
649 | (setq last-list (nconc word-list tail-list)))) | |
650 | (setq i (1+ i) idx (1+ idx) | |
651 | word-list nil tail-list nil) | |
652 | (if (> c 0) | |
653 | (let ((l last-list)) | |
654 | (if (= c 1) | |
655 | (setq diff-head | |
656 | (prog1 (aref val i) (setq i (1+ i))))) | |
657 | (setq tail-list (nthcdr (% diff-head 16) last-list)) | |
658 | (dotimes (i (/ diff-head 16)) | |
659 | (setq word-list (nconc word-list (list (car l))) | |
660 | l (cdr l)))))) | |
661 | (setq word-list | |
662 | (nconc word-list | |
663 | (list (symbol-name | |
664 | (unidata-decode-word c word-table)))) | |
665 | i (1+ i)))) | |
666 | (if (or word-list tail-list) | |
667 | (aset vec idx (nconc word-list tail-list))) | |
668 | (setq val nil) | |
669 | (dotimes (i 128) | |
670 | (setq c (+ first-char i)) | |
671 | (let ((name (aref vec i))) | |
672 | (if name | |
673 | (let ((tail (cdr (setq name (copy-sequence name)))) | |
674 | elt) | |
675 | (while tail | |
676 | (setq elt (car tail)) | |
677 | (or (string= elt "-") | |
678 | (progn | |
679 | (setcdr tail (cons elt (cdr tail))) | |
680 | (setcar tail " "))) | |
681 | (setq tail (cddr tail))) | |
682 | (setq name (apply 'concat name)))) | |
683 | (aset table c name) | |
684 | (if (= c char) | |
685 | (setq val name)))) | |
686 | val))) | |
687 | ||
688 | ((and (integerp val) (> val 0)) | |
689 | (let* ((symbol-table (aref (char-table-extra-slot table 4) 1)) | |
690 | (sym (aref symbol-table (1- val)))) | |
691 | (cond ((eq sym 'HANGUL\ SYLLABLE) | |
692 | (let ((jamo-name-table (aref (char-table-extra-slot table 4) 2))) | |
693 | ;; SIndex = S - SBase | |
694 | (setq char (- char #xAC00)) | |
695 | (let ( ;; LIndex = SIndex / NCount | |
696 | (L (/ char 588)) | |
697 | ;; VIndex = (SIndex % NCount) * TCount | |
698 | (V (/ (% char 588) 28)) | |
699 | ;; TIndex = SIndex % TCount | |
700 | (T (% char 28))) | |
701 | (format "HANGUL SYLLABLE %s%s%s" | |
752fc44b KH |
702 | ;; U+110B is nil in this table. |
703 | (or (aref (aref jamo-name-table 0) L) "") | |
f600cf3a KH |
704 | (aref (aref jamo-name-table 1) V) |
705 | (if (= T 0) "" | |
706 | (aref (aref jamo-name-table 2) (1- T))))))) | |
e60b06d7 KH |
707 | ((eq sym 'CJK\ IDEOGRAPH) |
708 | (format "%s-%04X" sym char)) | |
f600cf3a KH |
709 | ((eq sym 'CJK\ COMPATIBILITY\ IDEOGRAPH) |
710 | (format "%s-%04X" sym char)) | |
711 | ((eq sym 'VARIATION\ SELECTOR) | |
712 | (format "%s-%d" sym (+ (- char #xe0100) 17)))))))) | |
713 | ||
714 | ;; Store VAL as the name of CHAR in TABLE. | |
715 | ||
716 | (defun unidata-put-name (char val table) | |
717 | (let ((current-val (aref table char))) | |
718 | (if (and (stringp current-val) (= (aref current-val 0) 0)) | |
719 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
720 | (aset table char val))) | |
721 | ||
722 | (defun unidata-get-decomposition (char val table) | |
723 | (cond | |
724 | ((consp val) | |
725 | val) | |
726 | ||
727 | ((stringp val) | |
728 | (if (> (aref val 0) 0) | |
729 | val | |
730 | (let* ((first-char (lsh (lsh char -7) 7)) | |
731 | (word-table (char-table-extra-slot table 4)) | |
732 | (i 1) | |
733 | (len (length val)) | |
734 | (vec (make-vector 128 nil)) | |
735 | (idx 0) | |
736 | (case-fold-search nil) | |
737 | c word-list tail-list last-list word diff-head) | |
738 | (while (< i len) | |
739 | (setq c (aref val i)) | |
740 | (if (< c 3) | |
741 | (progn | |
742 | (if (or word-list tail-list) | |
743 | (aset vec idx | |
744 | (setq last-list (nconc word-list tail-list)))) | |
745 | (setq i (1+ i) idx (1+ idx) | |
746 | word-list nil tail-list nil) | |
747 | (if (> c 0) | |
748 | (let ((l last-list)) | |
749 | (if (= c 1) | |
750 | (setq diff-head | |
751 | (prog1 (aref val i) (setq i (1+ i))))) | |
752 | (setq tail-list (nthcdr (% diff-head 16) last-list)) | |
753 | (dotimes (i (/ diff-head 16)) | |
754 | (setq word-list (nconc word-list (list (car l))) | |
755 | l (cdr l)))))) | |
756 | (setq word-list | |
757 | (nconc word-list | |
758 | (list (or (unidata-decode-word c word-table) c))) | |
759 | i (1+ i)))) | |
760 | (if (or word-list tail-list) | |
761 | (aset vec idx (nconc word-list tail-list))) | |
762 | (dotimes (i 128) | |
763 | (aset table (+ first-char i) (aref vec i))) | |
1275c1dc | 764 | (aref vec (- char first-char))))) |
f600cf3a | 765 | |
1275c1dc KH |
766 | ;; Hangul syllable |
767 | ((and (eq val 0) (>= char #xAC00) (<= char #xD7A3)) | |
768 | ;; SIndex = S (char) - SBase (#xAC00) | |
769 | (setq char (- char #xAC00)) | |
770 | (let (;; L = LBase + SIndex / NCount | |
771 | (L (+ #x1100 (/ char 588))) | |
772 | ;; V = VBase + (SIndex % NCount) * TCount | |
773 | (V (+ #x1161 (/ (% char 588) 28))) | |
774 | ;; T = TBase + SIndex % TCount | |
775 | (T (+ #x11A7 (% char 28)))) | |
d636247a KH |
776 | (if (= T #x11A7) |
777 | (list L V) | |
778 | (list L V T)))) | |
1275c1dc KH |
779 | |
780 | )) | |
781 | ||
782 | ;; Store VAL as the decomposition information of CHAR in TABLE. | |
f600cf3a KH |
783 | |
784 | (defun unidata-put-decomposition (char val table) | |
785 | (let ((current-val (aref table char))) | |
786 | (if (and (stringp current-val) (= (aref current-val 0) 0)) | |
787 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
788 | (aset table char val))) | |
789 | ||
790 | ;; UnicodeData.txt contains these lines: | |
791 | ;; 0000;<control>;Cc;0;BN;;;;;N;NULL;;;; | |
792 | ;; ... | |
793 | ;; 0020;SPACE;Zs;0;WS;;;;;N;;;;; | |
794 | ;; ... | |
795 | ;; The following command yields a file of about 96K bytes. | |
796 | ;; % gawk -F ';' '{print $1,$2;}' < UnicodeData.txt | gzip > temp.gz | |
797 | ;; With the following function, we can get a file of almost the same | |
798 | ;; the size. | |
799 | ||
800 | ;; Generate a char-table for character names. | |
801 | ||
802 | (defun unidata-gen-table-word-list (prop val-func) | |
803 | (let ((table (make-char-table 'char-code-property-table)) | |
804 | (prop-idx (unidata-prop-index prop)) | |
805 | (word-list (list nil)) | |
806 | word-table | |
807 | block-list block-word-table block-end | |
808 | tail elt range val idx slot) | |
809 | (set-char-table-range table (cons 0 (max-char)) 0) | |
810 | (setq tail unidata-list) | |
811 | (setq block-end -1) | |
812 | (while tail | |
813 | (setq elt (car tail) tail (cdr tail)) | |
814 | (setq range (car elt) | |
815 | val (funcall val-func (nth prop-idx elt))) | |
816 | ;; Treat the sequence of "CJK COMPATIBILITY IDEOGRAPH-XXXX" and | |
817 | ;; "VARIATION SELECTOR-XXX" as a block. | |
818 | (if (and (consp val) (eq prop 'name) | |
819 | (or (and (eq (car val) 'CJK) | |
820 | (eq (nth 1 val) 'COMPATIBILITY)) | |
821 | (and (>= range #xe0100) | |
822 | (eq (car val) 'VARIATION) | |
823 | (eq (nth 1 val) 'SELECTOR)))) | |
824 | (let ((first (car val)) | |
825 | (second (nth 1 val)) | |
826 | (start range)) | |
827 | (while (and (setq elt (car tail) range (car elt) | |
828 | val (funcall val-func (nth prop-idx elt))) | |
829 | (consp val) | |
830 | (eq first (car val)) | |
831 | (eq second (nth 1 val))) | |
832 | (setq block-end range | |
833 | tail (cdr tail))) | |
834 | (setq range (cons start block-end) | |
835 | val (if (eq first 'CJK) 'CJK\ COMPATIBILITY\ IDEOGRAPH | |
836 | 'VARIATION\ SELECTOR)))) | |
837 | ||
838 | (if (consp range) | |
839 | (if val | |
840 | (let ((slot (assq val block-list))) | |
841 | (setq range (cons (car range) (cdr range))) | |
842 | (setq block-end (cdr range)) | |
843 | (if slot | |
844 | (nconc slot (list range)) | |
845 | (push (list val range) block-list)))) | |
846 | (let* ((start (lsh (lsh range -7) 7)) | |
847 | (limit (+ start 127)) | |
848 | (first tail) | |
849 | (vec (make-vector 128 nil)) | |
850 | c name len) | |
851 | (if (<= start block-end) | |
852 | ;; START overlap with the previous block. | |
853 | (aset table range (nth prop-idx elt)) | |
854 | (if val | |
855 | (aset vec (- range start) val)) | |
856 | (while (and (setq elt (car tail) range (car elt)) | |
857 | (integerp range) | |
858 | (<= range limit)) | |
859 | (setq val (funcall val-func (nth prop-idx elt))) | |
860 | (if val | |
861 | (aset vec (- range start) val)) | |
862 | (setq tail (cdr tail))) | |
863 | (setq vec (unidata-word-list-compress vec)) | |
864 | (when vec | |
865 | (dotimes (i (length vec)) | |
866 | (dolist (elt (aref vec i)) | |
867 | (if (symbolp elt) | |
868 | (let ((slot (assq elt word-list))) | |
869 | (if slot | |
870 | (setcdr slot (1+ (cdr slot))) | |
871 | (setcdr word-list | |
872 | (cons (cons elt 1) (cdr word-list)))))))) | |
873 | (set-char-table-range table (cons start limit) vec)))))) | |
874 | (setq word-list (sort (cdr word-list) | |
875 | #'(lambda (x y) (> (cdr x) (cdr y))))) | |
876 | (setq tail word-list idx 0) | |
877 | (while tail | |
878 | (setcdr (car tail) (unidata-encode-word idx)) | |
879 | (setq idx (1+ idx) tail (cdr tail))) | |
880 | (setq word-table (make-vector (length word-list) nil)) | |
881 | (setq idx 0) | |
882 | (dolist (elt word-list) | |
883 | (aset word-table idx (car elt)) | |
884 | (setq idx (1+ idx))) | |
885 | ||
886 | (if (and (eq prop 'decomposition) | |
887 | (> idx 32)) | |
1275c1dc | 888 | (error "Too many symbols in decomposition data")) |
f600cf3a KH |
889 | |
890 | (dotimes (i (/ #x110000 128)) | |
891 | (let* ((idx (* i 128)) | |
892 | (vec (aref table idx))) | |
893 | (when (vectorp vec) | |
894 | (dotimes (i (length vec)) | |
895 | (let ((tail (aref vec i)) | |
896 | elt code) | |
897 | (if (not tail) | |
898 | (aset vec i "\0") | |
899 | (while tail | |
900 | (setq elt (car tail) | |
901 | code (if (integerp elt) elt | |
902 | (cdr (assq elt word-list)))) | |
903 | (setcar tail (string code)) | |
904 | (setq tail (cdr tail))) | |
905 | (aset vec i (mapconcat 'identity (aref vec i) ""))))) | |
906 | (set-char-table-range | |
907 | table (cons idx (+ idx 127)) | |
908 | (mapconcat 'identity vec ""))))) | |
909 | ||
910 | (setq block-word-table (make-vector (length block-list) nil)) | |
911 | (setq idx 0) | |
912 | (dolist (elt block-list) | |
913 | (dolist (e (cdr elt)) | |
914 | (set-char-table-range table e (1+ idx))) | |
915 | (aset block-word-table idx (car elt)) | |
916 | (setq idx (1+ idx))) | |
917 | ||
918 | (set-char-table-extra-slot table 0 prop) | |
919 | (set-char-table-extra-slot table 4 (cons word-table block-word-table)) | |
920 | table)) | |
921 | ||
922 | (defun unidata-split-name (str) | |
923 | (if (symbolp str) | |
924 | str | |
925 | (let ((len (length str)) | |
926 | (l nil) | |
927 | (idx 0) | |
928 | c) | |
929 | (if (= len 0) | |
930 | nil | |
931 | (dotimes (i len) | |
932 | (setq c (aref str i)) | |
933 | (if (= c 32) | |
934 | (setq l (cons (intern (substring str idx i)) l) | |
935 | idx (1+ i)) | |
936 | (if (and (= c ?-) (< idx i) | |
937 | (< (1+ i) len) (/= (aref str (1+ i)) 32)) | |
938 | (setq l (cons '- (cons (intern (substring str idx i)) l)) | |
939 | idx (1+ i))))) | |
940 | (nreverse (cons (intern (substring str idx)) l)))))) | |
941 | ||
942 | (defun unidata-gen-table-name (prop) | |
943 | (let* ((table (unidata-gen-table-word-list prop 'unidata-split-name)) | |
944 | (word-tables (char-table-extra-slot table 4))) | |
945 | (byte-compile 'unidata-get-name) | |
946 | (byte-compile 'unidata-put-name) | |
947 | (set-char-table-extra-slot table 1 (symbol-function 'unidata-get-name)) | |
948 | (set-char-table-extra-slot table 2 (symbol-function 'unidata-put-name)) | |
949 | ||
950 | (if (eq prop 'name) | |
951 | (set-char-table-extra-slot table 4 | |
952 | (vector (car word-tables) | |
953 | (cdr word-tables) | |
954 | unidata-name-jamo-name-table)) | |
955 | (set-char-table-extra-slot table 4 | |
956 | (vector (car word-tables)))) | |
957 | table)) | |
958 | ||
959 | (defun unidata-split-decomposition (str) | |
960 | (if (symbolp str) | |
961 | str | |
962 | (let ((len (length str)) | |
963 | (l nil) | |
964 | (idx 0) | |
965 | c) | |
966 | (if (= len 0) | |
967 | nil | |
968 | (dotimes (i len) | |
969 | (setq c (aref str i)) | |
970 | (if (= c 32) | |
971 | (setq l (if (= (aref str idx) ?<) | |
972 | (cons (intern (substring str idx i)) l) | |
1275c1dc | 973 | (cons (string-to-number (substring str idx i) 16) l)) |
f600cf3a KH |
974 | idx (1+ i)))) |
975 | (if (= (aref str idx) ?<) | |
976 | (setq l (cons (intern (substring str idx len)) l)) | |
1275c1dc | 977 | (setq l (cons (string-to-number (substring str idx len) 16) l))) |
f600cf3a KH |
978 | (nreverse l))))) |
979 | ||
980 | ||
981 | (defun unidata-gen-table-decomposition (prop) | |
982 | (let* ((table (unidata-gen-table-word-list prop 'unidata-split-decomposition)) | |
983 | (word-tables (char-table-extra-slot table 4))) | |
984 | (byte-compile 'unidata-get-decomposition) | |
985 | (byte-compile 'unidata-put-decomposition) | |
986 | (set-char-table-extra-slot table 1 | |
987 | (symbol-function 'unidata-get-decomposition)) | |
988 | (set-char-table-extra-slot table 2 | |
989 | (symbol-function 'unidata-put-decomposition)) | |
990 | (set-char-table-extra-slot table 4 (car word-tables)) | |
991 | table)) | |
992 | ||
993 | ||
994 | \f | |
995 | (defun unidata-describe-general-category (val) | |
996 | (cdr (assq val | |
997 | '((Lu . "Letter, Uppercase") | |
998 | (Ll . "Letter, Lowercase") | |
999 | (Lt . "Letter, Titlecase") | |
1000 | (Lm . "Letter, Modifier") | |
1001 | (Lo . "Letter, Other") | |
1002 | (Mn . "Mark, Nonspacing") | |
1003 | (Mc . "Mark, Spacing Combining") | |
1004 | (Me . "Mark, Enclosing") | |
1005 | (Nd . "Number, Decimal Digit") | |
1006 | (Nl . "Number, Letter") | |
1007 | (No . "Number, Other") | |
1008 | (Pc . "Punctuation, Connector") | |
1009 | (Pd . "Punctuation, Dash") | |
1010 | (Ps . "Punctuation, Open") | |
1011 | (Pe . "Punctuation, Close") | |
1012 | (Pi . "Punctuation, Initial quote") | |
1013 | (Pf . "Punctuation, Final quote") | |
1014 | (Po . "Punctuation, Other") | |
1015 | (Sm . "Symbol, Math") | |
1016 | (Sc . "Symbol, Currency") | |
1017 | (Sk . "Symbol, Modifier") | |
1018 | (So . "Symbol, Other") | |
1019 | (Zs . "Separator, Space") | |
1020 | (Zl . "Separator, Line") | |
1021 | (Zp . "Separator, Paragraph") | |
1022 | (Cc . "Other, Control") | |
1023 | (Cf . "Other, Format") | |
1024 | (Cs . "Other, Surrogate") | |
1025 | (Co . "Other, Private Use") | |
1026 | (Cn . "Other, Not Assigned"))))) | |
1027 | ||
1028 | (defun unidata-describe-canonical-combining-class (val) | |
1029 | (cdr (assq val | |
1030 | '((0 . "Spacing, split, enclosing, reordrant, and Tibetan subjoined") | |
1031 | (1 . "Overlays and interior") | |
1032 | (7 . "Nuktas") | |
1033 | (8 . "Hiragana/Katakana voicing marks") | |
1034 | (9 . "Viramas") | |
1035 | (10 . "Start of fixed position classes") | |
1036 | (199 . "End of fixed position classes") | |
1037 | (200 . "Below left attached") | |
1038 | (202 . "Below attached") | |
1039 | (204 . "Below right attached") | |
1040 | (208 . "Left attached (reordrant around single base character)") | |
1041 | (210 . "Right attached") | |
1042 | (212 . "Above left attached") | |
1043 | (214 . "Above attached") | |
1044 | (216 . "Above right attached") | |
1045 | (218 . "Below left") | |
1046 | (220 . "Below") | |
1047 | (222 . "Below right") | |
1048 | (224 . "Left (reordrant around single base character)") | |
1049 | (226 . "Right") | |
1050 | (228 . "Above left") | |
1051 | (230 . "Above") | |
1052 | (232 . "Above right") | |
1053 | (233 . "Double below") | |
1054 | (234 . "Double above") | |
1055 | (240 . "Below (iota subscript)"))))) | |
1056 | ||
1057 | (defun unidata-describe-bidi-class (val) | |
1058 | (cdr (assq val | |
1059 | '((L . "Left-to-Right") | |
1060 | (LRE . "Left-to-Right Embedding") | |
1061 | (LRO . "Left-to-Right Override") | |
1062 | (R . "Right-to-Left") | |
1063 | (AL . "Right-to-Left Arabic") | |
1064 | (RLE . "Right-to-Left Embedding") | |
1065 | (RLO . "Right-to-Left Override") | |
1066 | (PDF . "Pop Directional Format") | |
1067 | (EN . "European Number") | |
1068 | (ES . "European Number Separator") | |
1069 | (ET . "European Number Terminator") | |
1070 | (AN . "Arabic Number") | |
1071 | (CS . "Common Number Separator") | |
1072 | (NSM . "Non-Spacing Mark") | |
1073 | (BN . "Boundary Neutral") | |
1074 | (B . "Paragraph Separator") | |
1075 | (S . "Segment Separator") | |
1076 | (WS . "Whitespace") | |
1077 | (ON . "Other Neutrals"))))) | |
1078 | ||
1079 | (defun unidata-describe-decomposition (val) | |
1080 | (mapconcat #'(lambda (x) (if (symbolp x) (symbol-name x) (string ?' x ?'))) | |
1081 | val " ")) | |
1082 | ||
1083 | ;; Verify if we can retrieve correct values from the generated | |
1084 | ;; char-tables. | |
1085 | ||
1086 | (defun unidata-check () | |
1087 | (dolist (elt unidata-prop-alist) | |
1088 | (let* ((prop (car elt)) | |
1089 | (index (unidata-prop-index prop)) | |
1090 | (generator (unidata-prop-generator prop)) | |
1091 | (table (progn | |
1092 | (message "Generating %S table..." prop) | |
1093 | (funcall generator prop))) | |
1094 | (decoder (char-table-extra-slot table 1)) | |
1095 | (check #x400)) | |
1096 | (dolist (e unidata-list) | |
1097 | (let ((char (car e)) | |
1098 | (val1 (nth index e)) | |
1099 | val2) | |
1100 | (if (and (stringp val1) (= (length val1) 0)) | |
1101 | (setq val1 nil)) | |
1102 | (unless (consp char) | |
1103 | (setq val2 (funcall decoder char (aref table char) table)) | |
1104 | (if val1 | |
1105 | (cond ((eq generator 'unidata-gen-table-symbol) | |
1106 | (setq val1 (intern val1))) | |
1107 | ((eq generator 'unidata-gen-table-integer) | |
1275c1dc | 1108 | (setq val1 (string-to-number val1))) |
f600cf3a | 1109 | ((eq generator 'unidata-gen-table-character) |
1275c1dc | 1110 | (setq val1 (string-to-number val1 16))) |
f600cf3a KH |
1111 | ((eq generator 'unidata-gen-table-decomposition) |
1112 | (setq val1 (unidata-split-decomposition val1))))) | |
1113 | (when (>= char check) | |
1114 | (message "%S %04X" prop check) | |
1115 | (setq check (+ check #x400))) | |
1116 | (or (equal val1 val2) | |
1117 | (insert (format "> %04X %S\n< %04X %S\n" | |
1118 | char val1 char val2))) | |
1119 | (sit-for 0))))))) | |
1120 | ||
1121 | ;; The entry function. It generates files described in the header | |
1122 | ;; comment of this file. | |
1123 | ||
3921ea21 KH |
1124 | (defun unidata-gen-files (&optional unidata-text-file) |
1125 | (or unidata-text-file | |
1126 | (setq unidata-text-file (car command-line-args-left) | |
1127 | command-line-args-left (cdr command-line-args-left))) | |
1128 | (unidata-setup-list unidata-text-file) | |
db516472 | 1129 | (let ((coding-system-for-write 'utf-8-unix) |
f600cf3a KH |
1130 | (charprop-file "charprop.el")) |
1131 | (with-temp-file charprop-file | |
1132 | (insert ";; Automatically generated by unidata-gen.el.\n") | |
1133 | (dolist (elt unidata-prop-alist) | |
1134 | (let* ((prop (car elt)) | |
1135 | (generator (unidata-prop-generator prop)) | |
1136 | (file (unidata-prop-file prop)) | |
1137 | (docstring (unidata-prop-docstring prop)) | |
1138 | (describer (unidata-prop-describer prop)) | |
1139 | table) | |
1140 | ;; Filename in this comment line is extracted by sed in | |
1141 | ;; Makefile. | |
1142 | (insert (format ";; FILE: %s\n" file)) | |
1143 | (insert (format "(define-char-code-property '%S %S\n %S)\n" | |
1144 | prop file docstring)) | |
1145 | (with-temp-file file | |
1146 | (message "Generating %s..." file) | |
1147 | (setq table (funcall generator prop)) | |
1148 | (when describer | |
1149 | (unless (subrp (symbol-function describer)) | |
1150 | (byte-compile describer) | |
1151 | (setq describer (symbol-function describer))) | |
1152 | (set-char-table-extra-slot table 3 describer)) | |
1153 | (insert ";; Automatically generated from UnicodeData.txt.\n" | |
1154 | (format "(define-char-code-property '%S %S %S)\n" | |
1155 | prop table docstring) | |
1275c1dc | 1156 | ";; Local Variables:\n" |
f600cf3a KH |
1157 | ";; coding: utf-8\n" |
1158 | ";; no-byte-compile: t\n" | |
1159 | ";; End:\n\n" | |
1160 | (format ";; %s ends here\n" file))))) | |
1161 | (message "Writing %s..." charprop-file) | |
1275c1dc | 1162 | (insert ";; Local Variables:\n" |
f600cf3a KH |
1163 | ";; coding: utf-8\n" |
1164 | ";; no-byte-compile: t\n" | |
1165 | ";; End:\n\n" | |
1166 | (format ";; %s ends here\n" charprop-file))))) | |
1167 | ||
1275c1dc KH |
1168 | \f |
1169 | ||
e9a8c04f | 1170 | ;; arch-tag: 961c862e-b821-447e-9b8a-bfbab9c2d525 |
f600cf3a | 1171 | ;;; unidata-gen.el ends here |