Commit | Line | Data |
---|---|---|
f600cf3a | 1 | ;; unidata-gen.el -- Create files containing character property data. |
5df4f04c | 2 | ;; Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2011 |
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 | ||
9ad5de0c | 8 | ;; GNU Emacs is free software: you can redistribute it and/or modify |
f600cf3a | 9 | ;; it under the terms of the GNU General Public License as published by |
9ad5de0c GM |
10 | ;; the Free Software Foundation, either version 3 of the License, or |
11 | ;; (at your option) any later version. | |
f600cf3a KH |
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 | |
9ad5de0c | 19 | ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. |
f600cf3a KH |
20 | |
21 | ;;; Commentary: | |
22 | ||
1275c1dc KH |
23 | ;; SPECIAL NOTICE |
24 | ;; | |
25 | ;; This file must be byte-compilable/loadable by `temacs' and also | |
26 | ;; the entry function `unidata-gen-files' must be runnable by | |
27 | ;; `temacs'. | |
28 | ||
f600cf3a KH |
29 | ;; FILES TO BE GENERATED |
30 | ;; | |
1275c1dc | 31 | ;; The entry function `unidata-gen-files' generates these files in |
f600cf3a KH |
32 | ;; the current directory. |
33 | ;; | |
34 | ;; charprop.el | |
35 | ;; It contains a series of forms of this format: | |
c805dec0 | 36 | ;; (define-char-code-property PROP FILE) |
f600cf3a | 37 | ;; where PROP is a symbol representing a character property |
c805dec0 | 38 | ;; (name, general-category, etc), and FILE is a name of one of |
f600cf3a KH |
39 | ;; the following files. |
40 | ;; | |
dd7d19d7 JB |
41 | ;; uni-name.el, uni-category.el, uni-combining.el, uni-bidi.el, |
42 | ;; uni-decomposition.el, uni-decimal.el, uni-digit.el, uni-numeric.el, | |
43 | ;; uni-mirrored.el, uni-old-name.el, uni-comment.el, uni-uppercase.el, | |
44 | ;; uni-lowercase.el, uni-titlecase.el | |
c805dec0 KH |
45 | ;; They contain one or more forms of this format: |
46 | ;; (define-char-code-property PROP CHAR-TABLE) | |
f600cf3a | 47 | ;; where PROP is the same as above, and CHAR-TABLE is a |
1275c1dc | 48 | ;; char-table containing property values in a compressed format. |
f600cf3a KH |
49 | ;; |
50 | ;; When they are installed in .../lisp/international/, the file | |
51 | ;; "charprop.el" is preloaded in loadup.el. The other files are | |
c805dec0 KH |
52 | ;; automatically loaded when the Lisp functions |
53 | ;; `get-char-code-property' and `put-char-code-property', and C | |
54 | ;; function uniprop_table are called. | |
f600cf3a KH |
55 | ;; |
56 | ;; FORMAT OF A CHAR TABLE | |
57 | ;; | |
58 | ;; We want to make a file size containing a char-table small. We | |
59 | ;; also want to load the file and get a property value fast. We | |
60 | ;; also want to reduce the used memory after loading it. So, | |
61 | ;; instead of naively storing a property value for each character in | |
62 | ;; a char-table (and write it out into a file), we store compressed | |
63 | ;; data in a char-table as below. | |
64 | ;; | |
65 | ;; If succeeding 128*N characters have the same property value, we | |
c805dec0 KH |
66 | ;; store that value (or the encoded one) for them. Otherwise, |
67 | ;; compress values (or the encoded ones) for succeeding 128 | |
68 | ;; characters into a single string and store it for those | |
69 | ;; characters. The way of compression depends on a property. See | |
70 | ;; the section "SIMPLE TABLE", "RUN-LENGTH TABLE", and "WORD-LIST | |
71 | ;; TABLE". | |
72 | ||
73 | ;; The char table has five extra slots: | |
f600cf3a | 74 | ;; 1st: property symbol |
c805dec0 KH |
75 | ;; 2nd: function to call to get a property value, |
76 | ;; or an index number of C function to decode the value, | |
77 | ;; or nil if the value can be directly got from the table. | |
78 | ;; 3nd: function to call to put a property value, | |
79 | ;; or an index number of C function to encode the value, | |
80 | ;; or nil if the value can be directly stored in the table. | |
81 | ;; 4th: function to call to get a description of a property value, or nil | |
f600cf3a | 82 | ;; 5th: data referred by the above functions |
f600cf3a | 83 | |
f600cf3a KH |
84 | ;; List of elements of this form: |
85 | ;; (CHAR-or-RANGE PROP1 PROP2 ... PROPn) | |
86 | ;; CHAR-or-RANGE: a character code or a cons of character codes | |
87 | ;; PROPn: string representing the nth property value | |
88 | ||
3921ea21 KH |
89 | (defvar unidata-list nil) |
90 | ||
c805dec0 KH |
91 | ;; Name of the directory containing files of Unicode Character |
92 | ;; Database. | |
93 | ||
94 | (defvar unidata-dir nil) | |
95 | ||
3921ea21 | 96 | (defun unidata-setup-list (unidata-text-file) |
f600cf3a KH |
97 | (let* ((table (list nil)) |
98 | (tail table) | |
99 | (block-names '(("^<CJK Ideograph" . CJK\ IDEOGRAPH) | |
100 | ("^<Hangul Syllable" . HANGUL\ SYLLABLE) | |
101 | ("^<.*Surrogate" . nil) | |
102 | ("^<.*Private Use" . PRIVATE\ USE))) | |
103 | val char name) | |
c805dec0 | 104 | (setq unidata-text-file (expand-file-name unidata-text-file unidata-dir)) |
f600cf3a KH |
105 | (or (file-readable-p unidata-text-file) |
106 | (error "File not readable: %s" unidata-text-file)) | |
107 | (with-temp-buffer | |
149353a4 KH |
108 | ;; Insert a file of this format: |
109 | ;; (CHAR NAME CATEGORY ...) | |
c0943d3d | 110 | ;; where CHAR is a character code, the following elements are strings |
149353a4 | 111 | ;; representing character properties. |
1275c1dc | 112 | (insert-file-contents unidata-text-file) |
f600cf3a KH |
113 | (goto-char (point-min)) |
114 | (condition-case nil | |
115 | (while t | |
116 | (setq val (read (current-buffer)) | |
117 | char (car val) | |
118 | name (cadr val)) | |
119 | ||
120 | ;; Check this kind of block. | |
121 | ;; 4E00;<CJK Ideograph, First>;Lo;0;L;;;;;N;;;;; | |
149353a4 | 122 | ;; 9FCB;<CJK Ideograph, Last>;Lo;0;L;;;;;N;;;;; |
f600cf3a KH |
123 | (if (and (= (aref name 0) ?<) |
124 | (string-match ", First>$" name)) | |
125 | (let ((first char) | |
126 | (l block-names) | |
127 | block-name) | |
128 | (setq val (read (current-buffer)) | |
129 | char (car val) | |
130 | block-name (cadr val) | |
131 | name nil) | |
132 | (while l | |
133 | (if (string-match (caar l) block-name) | |
134 | (setq name (cdar l) l nil) | |
135 | (setq l (cdr l)))) | |
136 | (if (not name) | |
137 | ;; As this is a surrogate pair range, ignore it. | |
138 | (setq val nil) | |
139 | (setcar val (cons first char)) | |
140 | (setcar (cdr val) name)))) | |
141 | ||
142 | (when val | |
143 | (setcdr tail (list val)) | |
144 | (setq tail (cdr tail)))) | |
145 | (error nil))) | |
3921ea21 | 146 | (setq unidata-list (cdr table)))) |
f600cf3a KH |
147 | |
148 | ;; Alist of this form: | |
0902a04e | 149 | ;; (PROP INDEX GENERATOR FILENAME DOCSTRING DESCRIBER DEFAULT VAL-LIST) |
f600cf3a | 150 | ;; PROP: character property |
c805dec0 KH |
151 | ;; INDEX: index to each element of unidata-list for PROP. |
152 | ;; It may be a function that generates an alist of character codes | |
153 | ;; vs. the corresponding property values. | |
f600cf3a KH |
154 | ;; GENERATOR: function to generate a char-table |
155 | ;; FILENAME: filename to store the char-table | |
c805dec0 | 156 | ;; DOCSTRING: docstring for the property |
f600cf3a | 157 | ;; DESCRIBER: function to call to get a description string of property value |
0902a04e KH |
158 | ;; DEFAULT: the default value of the property. It may have the form |
159 | ;; (VAL0 (FROM1 TO1 VAL1) ...) which indicates that the default | |
160 | ;; value is VAL0 except for characters in the ranges specified by | |
161 | ;; FROMn and TOn (incusive). The default value of characters | |
162 | ;; between FROMn and TOn is VALn. | |
c805dec0 | 163 | ;; VAL-LIST: list of specially ordered property values |
f600cf3a KH |
164 | |
165 | (defconst unidata-prop-alist | |
166 | '((name | |
167 | 1 unidata-gen-table-name "uni-name.el" | |
168 | "Unicode character name. | |
0902a04e KH |
169 | Property value is a string." |
170 | nil | |
171 | "") | |
f600cf3a KH |
172 | (general-category |
173 | 2 unidata-gen-table-symbol "uni-category.el" | |
174 | "Unicode general category. | |
175 | Property value is one of the following symbols: | |
176 | Lu, Ll, Lt, Lm, Lo, Mn, Mc, Me, Nd, Nl, No, Pc, Pd, Ps, Pe, Pi, Pf, Po, | |
177 | Sm, Sc, Sk, So, Zs, Zl, Zp, Cc, Cf, Cs, Co, Cn" | |
c805dec0 | 178 | unidata-describe-general-category |
0902a04e | 179 | Cn |
c805dec0 KH |
180 | ;; The order of elements must be in sync with unicode_category_t |
181 | ;; in src/character.h. | |
182 | (Lu Ll Lt Lm Lo Mn Mc Me Nd Nl No Pc Pd Ps Pe Pi Pf Po | |
183 | Sm Sc Sk So Zs Zl Zp Cc Cf Cs Co Cn)) | |
f600cf3a KH |
184 | (canonical-combining-class |
185 | 3 unidata-gen-table-integer "uni-combining.el" | |
186 | "Unicode canonical combining class. | |
187 | Property value is an integer." | |
0902a04e KH |
188 | unidata-describe-canonical-combining-class |
189 | 0) | |
f600cf3a KH |
190 | (bidi-class |
191 | 4 unidata-gen-table-symbol "uni-bidi.el" | |
192 | "Unicode bidi class. | |
193 | Property value is one of the following symbols: | |
194 | L, LRE, LRO, R, AL, RLE, RLO, PDF, EN, ES, ET, | |
195 | AN, CS, NSM, BN, B, S, WS, ON" | |
c805dec0 | 196 | unidata-describe-bidi-class |
0902a04e KH |
197 | (L (#x0600 #x06FF AL) (#x0590 #x05FF R) (#x07C0 #x08FF R) |
198 | (#xFB1D #xFB4F R) (#x10800 #x10FFF R) (#x1E800 #x1EFFF R)) | |
c805dec0 KH |
199 | ;; The order of elements must be in sync with bidi_type_t in |
200 | ;; src/dispextern.h. | |
201 | (L R EN AN BN B AL LRE LRO RLE RLO PDF ES ET CS NSM S WS ON)) | |
f600cf3a KH |
202 | (decomposition |
203 | 5 unidata-gen-table-decomposition "uni-decomposition.el" | |
204 | "Unicode decomposition mapping. | |
205 | Property value is a list of characters. The first element may be | |
1275c1dc | 206 | one of these symbols representing compatibility formatting tag: |
f1478738 KH |
207 | font, noBreak, initial, medial, final, isolated, circle, super, |
208 | sub, vertical, wide, narrow, small, square, fraction, compat" | |
f600cf3a KH |
209 | unidata-describe-decomposition) |
210 | (decimal-digit-value | |
211 | 6 unidata-gen-table-integer "uni-decimal.el" | |
212 | "Unicode numeric value (decimal digit). | |
0902a04e KH |
213 | Property value is an integer 0..9, or nil. |
214 | The value nil stands for NaN \"Numeric_Value\".") | |
f600cf3a KH |
215 | (digit-value |
216 | 7 unidata-gen-table-integer "uni-digit.el" | |
217 | "Unicode numeric value (digit). | |
0902a04e KH |
218 | Property value is an integer 0..9, or nil. |
219 | The value nil stands for NaN \"Numeric_Value\".") | |
f600cf3a | 220 | (numeric-value |
dd787998 | 221 | 8 unidata-gen-table-numeric "uni-numeric.el" |
f600cf3a | 222 | "Unicode numeric value (numeric). |
0902a04e KH |
223 | Property value is an integer, a floating point, or nil. |
224 | The value nil stands for NaN \"Numeric_Value\".") | |
f600cf3a KH |
225 | (mirrored |
226 | 9 unidata-gen-table-symbol "uni-mirrored.el" | |
227 | "Unicode bidi mirrored flag. | |
0902a04e KH |
228 | Property value is a symbol `Y' or `N'. See also the property `mirroring'." |
229 | nil | |
230 | N) | |
f600cf3a KH |
231 | (old-name |
232 | 10 unidata-gen-table-name "uni-old-name.el" | |
233 | "Unicode old names as published in Unicode 1.0. | |
234 | Property value is a string.") | |
235 | (iso-10646-comment | |
236 | 11 unidata-gen-table-name "uni-comment.el" | |
237 | "Unicode ISO 10646 comment. | |
238 | Property value is a string.") | |
239 | (uppercase | |
240 | 12 unidata-gen-table-character "uni-uppercase.el" | |
241 | "Unicode simple uppercase mapping. | |
0902a04e KH |
242 | Property value is a character or nil. |
243 | The value nil means that the actual property value of a character | |
244 | is the character itself." | |
f600cf3a KH |
245 | string) |
246 | (lowercase | |
247 | 13 unidata-gen-table-character "uni-lowercase.el" | |
248 | "Unicode simple lowercase mapping. | |
0902a04e KH |
249 | Property value is a character or nil. |
250 | The value nil means that the actual property value of a character | |
251 | is the character itself." | |
f600cf3a KH |
252 | string) |
253 | (titlecase | |
254 | 14 unidata-gen-table-character "uni-titlecase.el" | |
255 | "Unicode simple titlecase mapping. | |
0902a04e KH |
256 | Property value is a character or nil. |
257 | The value nil means that the actual property value of a character | |
258 | is the character itself." | |
c805dec0 KH |
259 | string) |
260 | (mirroring | |
261 | unidata-gen-mirroring-list unidata-gen-table-character "uni-mirrored.el" | |
262 | "Unicode bidi-mirroring characters. | |
0902a04e KH |
263 | Property value is a character that has the corresponding mirroring image or nil. |
264 | The value nil means that the actual property value of a character | |
265 | is the character itself."))) | |
f600cf3a KH |
266 | |
267 | ;; Functions to access the above data. | |
268 | (defsubst unidata-prop-index (prop) (nth 1 (assq prop unidata-prop-alist))) | |
269 | (defsubst unidata-prop-generator (prop) (nth 2 (assq prop unidata-prop-alist))) | |
270 | (defsubst unidata-prop-file (prop) (nth 3 (assq prop unidata-prop-alist))) | |
271 | (defsubst unidata-prop-docstring (prop) (nth 4 (assq prop unidata-prop-alist))) | |
272 | (defsubst unidata-prop-describer (prop) (nth 5 (assq prop unidata-prop-alist))) | |
c805dec0 KH |
273 | (defsubst unidata-prop-default (prop) (nth 6 (assq prop unidata-prop-alist))) |
274 | (defsubst unidata-prop-val-list (prop) (nth 7 (assq prop unidata-prop-alist))) | |
f600cf3a KH |
275 | |
276 | \f | |
277 | ;; SIMPLE TABLE | |
278 | ;; | |
279 | ;; If the type of character property value is character, and the | |
280 | ;; values of succeeding character codes are usually different, we use | |
281 | ;; a char-table described here to store such values. | |
282 | ;; | |
c805dec0 KH |
283 | ;; A char-table divides character code space (#x0..#x3FFFFF) into |
284 | ;; #x8000 blocks (each block contains 128 characters). | |
285 | ||
286 | ;; If all characters of a block have no property, a char-table has the | |
287 | ;; symbol nil for that block. Otherwise a char-table has a string of | |
288 | ;; the following format for it. | |
f600cf3a | 289 | ;; |
c805dec0 KH |
290 | ;; The first character of the string is ?\001. |
291 | ;; The second character of the string is FIRST-INDEX. | |
292 | ;; The Nth (N > 1) character of the string is a property value of the | |
293 | ;; character (BLOCK-HEAD + FIRST-INDEX + N - 2), where BLOCK-HEAD is | |
294 | ;; the first character of the block. | |
f600cf3a | 295 | ;; |
c805dec0 KH |
296 | ;; This kind of char-table has these extra slots: |
297 | ;; 1st: the property symbol | |
298 | ;; 2nd: nil | |
299 | ;; 3rd: 0 (corresponding to uniprop_encode_character in chartab.c) | |
300 | ;; 4th to 5th: nil | |
f600cf3a | 301 | |
c805dec0 | 302 | (defun unidata-gen-table-character (prop &rest ignore) |
f600cf3a KH |
303 | (let ((table (make-char-table 'char-code-property-table)) |
304 | (prop-idx (unidata-prop-index prop)) | |
305 | (vec (make-vector 128 0)) | |
306 | (tail unidata-list) | |
307 | elt range val idx slot) | |
c805dec0 KH |
308 | (if (functionp prop-idx) |
309 | (setq tail (funcall prop-idx) | |
310 | prop-idx 1)) | |
f600cf3a KH |
311 | (while tail |
312 | (setq elt (car tail) tail (cdr tail)) | |
313 | (setq range (car elt) | |
314 | val (nth prop-idx elt)) | |
315 | (if (= (length val) 0) | |
316 | (setq val nil) | |
317 | (setq val (string-to-number val 16))) | |
318 | (if (consp range) | |
319 | (if val | |
320 | (set-char-table-range table range val)) | |
321 | (let* ((start (lsh (lsh range -7) 7)) | |
322 | (limit (+ start 127)) | |
323 | first-index last-index) | |
324 | (fillarray vec 0) | |
325 | (if val | |
326 | (aset vec (setq last-index (setq first-index (- range start))) | |
327 | val)) | |
328 | (while (and (setq elt (car tail) range (car elt)) | |
329 | (integerp range) | |
330 | (<= range limit)) | |
331 | (setq val (nth prop-idx elt)) | |
332 | (when (> (length val) 0) | |
333 | (aset vec (setq last-index (- range start)) | |
334 | (string-to-number val 16)) | |
335 | (or first-index | |
336 | (setq first-index last-index))) | |
337 | (setq tail (cdr tail))) | |
338 | (when first-index | |
c805dec0 | 339 | (let ((str (string 1 first-index)) |
f600cf3a KH |
340 | c) |
341 | (while (<= first-index last-index) | |
342 | (setq str (format "%s%c" str (or (aref vec first-index) 0)) | |
343 | first-index (1+ first-index))) | |
344 | (set-char-table-range table (cons start limit) str)))))) | |
345 | ||
346 | (set-char-table-extra-slot table 0 prop) | |
c805dec0 | 347 | (set-char-table-extra-slot table 2 0) |
f600cf3a KH |
348 | table)) |
349 | ||
350 | ||
351 | \f | |
352 | ;; RUN-LENGTH TABLE | |
353 | ;; | |
c805dec0 KH |
354 | ;; If many characters of successive character codes have the same |
355 | ;; property value, we use a char-table described here to store the | |
356 | ;; values. | |
f600cf3a | 357 | ;; |
c805dec0 KH |
358 | ;; At first, instead of a value itself, we store an index number to |
359 | ;; the VAL-TABLE (5th extra slot) in the table. We call that index | |
360 | ;; number as VAL-CODE here after. | |
361 | ;; | |
362 | ;; A char-table divides character code space (#x0..#x3FFFFF) into | |
363 | ;; #x8000 blocks (each block contains 128 characters). | |
364 | ;; | |
365 | ;; If all characters of a block have the same value, a char-table has | |
366 | ;; VAL-CODE for that block. Otherwise a char-table has a string of | |
367 | ;; the following format for that block. | |
368 | ;; | |
369 | ;; The first character of the string is ?\002. | |
370 | ;; The following characters has this form: | |
f600cf3a KH |
371 | ;; ( VAL-CODE RUN-LENGTH ? ) + |
372 | ;; where: | |
c805dec0 | 373 | ;; VAL-CODE (0..127): index into VAL-TABLE. |
f600cf3a KH |
374 | ;; RUN-LENGTH (130..255): |
375 | ;; (RUN-LENGTH - 128) specifies how many characters have the same | |
376 | ;; value. If omitted, it means 1. | |
c805dec0 KH |
377 | ;; |
378 | ;; This kind of char-table has these extra slots: | |
379 | ;; 1st: the property symbol | |
380 | ;; 2nd: 0 (corresponding to uniprop_decode_value in chartab.c) | |
381 | ;; 3rd: 1..3 (corresponding to uniprop_encode_xxx in chartab.c) | |
382 | ;; 4th: function or nil | |
383 | ;; 5th: VAL-TABLE | |
dd787998 | 384 | |
f600cf3a KH |
385 | ;; Encode the character property value VAL into an integer value by |
386 | ;; VAL-LIST. By side effect, VAL-LIST is modified. | |
387 | ;; VAL-LIST has this form: | |
c805dec0 KH |
388 | ;; ((nil . 0) (VAL1 . 1) (VAL2 . 2) ...) |
389 | ;; If VAL is one of VALn, just return n. | |
390 | ;; Otherwise, VAL-LIST is modified to this: | |
391 | ;; ((nil . 0) (VAL1 . 1) (VAL2 . 2) ... (VAL . n+1)) | |
f600cf3a KH |
392 | |
393 | (defun unidata-encode-val (val-list val) | |
dd787998 | 394 | (let ((slot (assoc val val-list)) |
f600cf3a KH |
395 | val-code) |
396 | (if slot | |
397 | (cdr slot) | |
c805dec0 KH |
398 | (setq val-code (length val-list)) |
399 | (nconc val-list (list (cons val val-code))) | |
f600cf3a KH |
400 | val-code))) |
401 | ||
402 | ;; Generate a char-table for the character property PROP. | |
403 | ||
c805dec0 | 404 | (defun unidata-gen-table (prop val-func default-value val-list) |
f600cf3a KH |
405 | (let ((table (make-char-table 'char-code-property-table)) |
406 | (prop-idx (unidata-prop-index prop)) | |
f600cf3a | 407 | (vec (make-vector 128 0)) |
149353a4 KH |
408 | tail elt range val val-code idx slot |
409 | prev-range-data) | |
c805dec0 KH |
410 | (setq val-list (cons nil (copy-sequence val-list))) |
411 | (setq tail val-list val-code 0) | |
412 | ;; Convert (nil A B ...) to ((nil . 0) (A . 1) (B . 2) ...) | |
413 | (while tail | |
414 | (setcar tail (cons (car tail) val-code)) | |
415 | (setq tail (cdr tail) val-code (1+ val-code))) | |
0902a04e KH |
416 | (if (consp default-value) |
417 | (setq default-value (copy-sequence default-value)) | |
418 | (setq default-value (list default-value))) | |
419 | (setcar default-value | |
420 | (unidata-encode-val val-list (car default-value))) | |
421 | (set-char-table-range table t (car default-value)) | |
422 | (set-char-table-range table nil (car default-value)) | |
423 | (dolist (elm (cdr default-value)) | |
424 | (setcar (nthcdr 2 elm) | |
425 | (unidata-encode-val val-list (nth 2 elm))) | |
426 | (set-char-table-range table (cons (car elm) (nth 1 elm)) (nth 2 elm))) | |
427 | ||
f600cf3a KH |
428 | (setq tail unidata-list) |
429 | (while tail | |
430 | (setq elt (car tail) tail (cdr tail)) | |
431 | (setq range (car elt) | |
432 | val (funcall val-func (nth prop-idx elt))) | |
433 | (setq val-code (if val (unidata-encode-val val-list val))) | |
434 | (if (consp range) | |
149353a4 | 435 | (when val-code |
c805dec0 | 436 | (set-char-table-range table range val-code) |
149353a4 KH |
437 | (let ((from (car range)) (to (cdr range))) |
438 | ;; If RANGE doesn't end at the char-table boundary (each | |
439 | ;; 128 characters), we may have to carry over the data | |
440 | ;; for the last several characters (at most 127 chars) | |
441 | ;; to the next loop. In that case, set PREV-RANGE-DATA | |
442 | ;; to ((FROM . TO) . VAL-CODE) where (FROM . TO) | |
443 | ;; specifies the range of characters handled in the next | |
444 | ;; loop. | |
445 | (when (< (logand to #x7F) #x7F) | |
446 | (if (< from (logand to #x1FFF80)) | |
447 | (setq from (logand to #x1FFF80))) | |
448 | (setq prev-range-data (cons (cons from to) val-code))))) | |
f600cf3a KH |
449 | (let* ((start (lsh (lsh range -7) 7)) |
450 | (limit (+ start 127)) | |
0902a04e KH |
451 | str count new-val from to vcode) |
452 | (fillarray vec (car default-value)) | |
453 | (dolist (elm (cdr default-value)) | |
454 | (setq from (car elm) to (nth 1 elm)) | |
455 | (when (and (<= from limit) | |
456 | (or (>= from start) (>= to start))) | |
457 | (setq from (max from start) | |
458 | to (min to limit) | |
459 | vcode (nth 2 elm)) | |
149353a4 KH |
460 | (while (<= from to) |
461 | (aset vec (- from start) vcode) | |
462 | (setq from (1+ from))))) | |
0902a04e KH |
463 | ;; See the comment above. |
464 | (when (and prev-range-data | |
465 | (>= (cdr (car prev-range-data)) start)) | |
466 | (setq from (car (car prev-range-data)) | |
467 | to (cdr (car prev-range-data)) | |
468 | vcode (cdr prev-range-data)) | |
469 | (while (<= from to) | |
470 | (aset vec (- from start) vcode) | |
471 | (setq from (1+ from)))) | |
149353a4 | 472 | (setq prev-range-data nil) |
f600cf3a KH |
473 | (if val-code |
474 | (aset vec (- range start) val-code)) | |
475 | (while (and (setq elt (car tail) range (car elt)) | |
476 | (integerp range) | |
477 | (<= range limit)) | |
478 | (setq new-val (funcall val-func (nth prop-idx elt))) | |
479 | (if (not (eq val new-val)) | |
480 | (setq val new-val | |
481 | val-code (if val (unidata-encode-val val-list val)))) | |
482 | (if val-code | |
483 | (aset vec (- range start) val-code)) | |
484 | (setq tail (cdr tail))) | |
c805dec0 | 485 | (setq str "\002" val-code -1 count 0) |
f600cf3a KH |
486 | (mapc #'(lambda (x) |
487 | (if (= val-code x) | |
488 | (setq count (1+ count)) | |
489 | (if (> count 2) | |
490 | (setq str (concat str (string val-code | |
491 | (+ count 128)))) | |
492 | (if (= count 2) | |
493 | (setq str (concat str (string val-code val-code))) | |
494 | (if (= count 1) | |
495 | (setq str (concat str (string val-code)))))) | |
496 | (setq val-code x count 1))) | |
497 | vec) | |
498 | (if (= count 128) | |
499 | (if val | |
c805dec0 | 500 | (set-char-table-range table (cons start limit) val-code)) |
f600cf3a KH |
501 | (if (= val-code 0) |
502 | (set-char-table-range table (cons start limit) str) | |
503 | (if (> count 2) | |
504 | (setq str (concat str (string val-code (+ count 128)))) | |
505 | (if (= count 2) | |
506 | (setq str (concat str (string val-code val-code))) | |
507 | (setq str (concat str (string val-code))))) | |
508 | (set-char-table-range table (cons start limit) str)))))) | |
509 | ||
dd7d19d7 | 510 | (set-char-table-extra-slot table 0 prop) |
f600cf3a KH |
511 | (set-char-table-extra-slot table 4 (vconcat (mapcar 'car val-list))) |
512 | table)) | |
513 | ||
c805dec0 | 514 | (defun unidata-gen-table-symbol (prop default-value val-list) |
f600cf3a KH |
515 | (let ((table (unidata-gen-table prop |
516 | #'(lambda (x) (and (> (length x) 0) | |
517 | (intern x))) | |
c805dec0 KH |
518 | default-value val-list))) |
519 | (set-char-table-extra-slot table 1 0) | |
520 | (set-char-table-extra-slot table 2 1) | |
f600cf3a KH |
521 | table)) |
522 | ||
c805dec0 | 523 | (defun unidata-gen-table-integer (prop default-value val-list) |
f600cf3a KH |
524 | (let ((table (unidata-gen-table prop |
525 | #'(lambda (x) (and (> (length x) 0) | |
526 | (string-to-number x))) | |
c805dec0 KH |
527 | default-value val-list))) |
528 | (set-char-table-extra-slot table 1 0) | |
529 | (set-char-table-extra-slot table 2 1) | |
f600cf3a KH |
530 | table)) |
531 | ||
c805dec0 | 532 | (defun unidata-gen-table-numeric (prop default-value val-list) |
dd787998 KH |
533 | (let ((table (unidata-gen-table prop |
534 | #'(lambda (x) | |
535 | (if (string-match "/" x) | |
536 | (/ (float (string-to-number x)) | |
537 | (string-to-number | |
538 | (substring x (match-end 0)))) | |
539 | (if (> (length x) 0) | |
540 | (string-to-number x)))) | |
c805dec0 KH |
541 | default-value val-list))) |
542 | (set-char-table-extra-slot table 1 0) | |
543 | (set-char-table-extra-slot table 2 2) | |
dd787998 KH |
544 | table)) |
545 | ||
f600cf3a KH |
546 | \f |
547 | ;; WORD-LIST TABLE | |
548 | ||
549 | ;; If the table is for `name' property, each character in the string | |
550 | ;; is one of these: | |
551 | ;; DIFF-HEAD-CODE (0, 1, or 2): | |
552 | ;; specifies how to decode the following characters. | |
553 | ;; WORD-CODE (3..#x7FF excluding '-', '0'..'9', 'A'..'Z'): | |
554 | ;; specifies an index number into WORD-TABLE (see below) | |
555 | ;; Otherwise (' ', '-', '0'..'9', 'A'..'Z'): | |
556 | ;; specifies a literal word. | |
557 | ;; | |
558 | ;; The 4th slots is a vector: | |
559 | ;; [ WORD-TABLE BLOCK-NAME HANGUL-JAMO-TABLE ] | |
560 | ;; WORD-TABLE is a vector of word symbols. | |
561 | ;; BLOCK-NAME is a vector of name symbols for a block of characters. | |
562 | ;; HANGUL-JAMO-TABLE is `unidata-name-jamo-name-table'. | |
563 | ||
564 | ;; Return the difference of symbol list L1 and L2 in this form: | |
565 | ;; (DIFF-HEAD SYM1 SYM2 ...) | |
566 | ;; DIFF-HEAD is ((SAME-HEAD-LENGTH * 16) + SAME-TAIL-LENGTH). | |
567 | ;; Ex: If L1 is (a b c d e f) and L2 is (a g h e f), this function | |
568 | ;; returns ((+ (* 1 16) 2) g h). | |
569 | ;; It means that we can get L2 from L1 by prepending the first element | |
570 | ;; of L1 and appending the last 2 elements of L1 to the list (g h). | |
571 | ;; If L1 and L2 don't have common elements at the head and tail, | |
572 | ;; set DIFF-HEAD to -1 and SYM1 ... to the elements of L2. | |
573 | ||
574 | (defun unidata-word-list-diff (l1 l2) | |
575 | (let ((beg 0) | |
576 | (end 0) | |
577 | (len1 (length l1)) | |
578 | (len2 (length l2)) | |
579 | result) | |
580 | (when (< len1 16) | |
581 | (while (and l1 (eq (car l1) (car l2))) | |
582 | (setq beg (1+ beg) | |
583 | l1 (cdr l1) len1 (1- len1) l2 (cdr l2) len2 (1- len2))) | |
dd7d19d7 | 584 | (while (and (< end len1) (< end len2) |
f600cf3a KH |
585 | (eq (nth (- len1 end 1) l1) (nth (- len2 end 1) l2))) |
586 | (setq end (1+ end)))) | |
587 | (if (= (+ beg end) 0) | |
588 | (setq result (list -1)) | |
589 | (setq result (list (+ (* beg 16) (+ beg (- len1 end)))))) | |
590 | (while (< end len2) | |
591 | (setcdr result (cons (nth (- len2 end 1) l2) (cdr result))) | |
592 | (setq end (1+ end))) | |
593 | result)) | |
594 | ||
595 | ;; Return a compressed form of the vector VEC. Each element of VEC is | |
596 | ;; a list of symbols of which names can be concatenated to form a | |
597 | ;; character name. This function changes those elements into | |
598 | ;; compressed forms by utilizing the fact that diff of consecutive | |
599 | ;; elements is usually small. | |
600 | ||
601 | (defun unidata-word-list-compress (vec) | |
602 | (let (last-elt last-idx diff-head tail elt val) | |
603 | (dotimes (i 128) | |
604 | (setq elt (aref vec i)) | |
605 | (when elt | |
606 | (if (null last-elt) | |
607 | (setq diff-head -1 | |
608 | val (cons 0 elt)) | |
609 | (setq val (unidata-word-list-diff last-elt elt)) | |
610 | (if (= (car val) -1) | |
611 | (setq diff-head -1 | |
612 | val (cons 0 (cdr val))) | |
613 | (if (eq diff-head (car val)) | |
614 | (setq val (cons 2 (cdr val))) | |
615 | (setq diff-head (car val)) | |
616 | (if (>= diff-head 0) | |
617 | (setq val (cons 1 val)))))) | |
618 | (aset vec i val) | |
619 | (setq last-idx i last-elt elt))) | |
620 | (if (not last-idx) | |
621 | (setq vec nil) | |
622 | (if (< last-idx 127) | |
623 | (let ((shorter (make-vector (1+ last-idx) nil))) | |
624 | (dotimes (i (1+ last-idx)) | |
625 | (aset shorter i (aref vec i))) | |
626 | (setq vec shorter)))) | |
627 | vec)) | |
628 | ||
629 | ;; Encode the word index IDX into a characters code that can be | |
630 | ;; embedded in a string. | |
631 | ||
632 | (defsubst unidata-encode-word (idx) | |
633 | ;; Exclude 0, 1, 2. | |
634 | (+ idx 3)) | |
635 | ||
636 | ;; Decode the character code CODE (that is embedded in a string) into | |
637 | ;; the corresponding word name by looking up WORD-TABLE. | |
638 | ||
639 | (defsubst unidata-decode-word (code word-table) | |
640 | (setq code (- code 3)) | |
641 | (if (< code (length word-table)) | |
642 | (aref word-table code))) | |
643 | ||
644 | ;; Table of short transliterated name symbols of Hangul Jamo divided | |
645 | ;; into Choseong, Jungseong, and Jongseong. | |
646 | ||
647 | (defconst unidata-name-jamo-name-table | |
648 | [[G GG N D DD R M B BB S SS nil J JJ C K T P H] | |
649 | [A AE YA YAE EO E YEO YE O WA WAE OE YO U WEO WE WI YU EU YI I] | |
650 | [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]]) | |
651 | ||
652 | ;; Return a name of CHAR. VAL is the current value of (aref TABLE | |
653 | ;; CHAR). | |
654 | ||
655 | (defun unidata-get-name (char val table) | |
dd7d19d7 | 656 | (cond |
f600cf3a KH |
657 | ((stringp val) |
658 | (if (> (aref val 0) 0) | |
659 | val | |
660 | (let* ((first-char (lsh (lsh char -7) 7)) | |
661 | (word-table (aref (char-table-extra-slot table 4) 0)) | |
662 | (i 1) | |
663 | (len (length val)) | |
664 | (vec (make-vector 128 nil)) | |
665 | (idx 0) | |
666 | (case-fold-search nil) | |
667 | c word-list tail-list last-list word diff-head) | |
668 | (while (< i len) | |
669 | (setq c (aref val i)) | |
670 | (if (< c 3) | |
671 | (progn | |
672 | (if (or word-list tail-list) | |
673 | (aset vec idx | |
674 | (setq last-list (nconc word-list tail-list)))) | |
675 | (setq i (1+ i) idx (1+ idx) | |
676 | word-list nil tail-list nil) | |
677 | (if (> c 0) | |
678 | (let ((l last-list)) | |
679 | (if (= c 1) | |
680 | (setq diff-head | |
681 | (prog1 (aref val i) (setq i (1+ i))))) | |
682 | (setq tail-list (nthcdr (% diff-head 16) last-list)) | |
683 | (dotimes (i (/ diff-head 16)) | |
684 | (setq word-list (nconc word-list (list (car l))) | |
685 | l (cdr l)))))) | |
686 | (setq word-list | |
dd7d19d7 | 687 | (nconc word-list |
f600cf3a KH |
688 | (list (symbol-name |
689 | (unidata-decode-word c word-table)))) | |
690 | i (1+ i)))) | |
691 | (if (or word-list tail-list) | |
692 | (aset vec idx (nconc word-list tail-list))) | |
693 | (setq val nil) | |
694 | (dotimes (i 128) | |
695 | (setq c (+ first-char i)) | |
696 | (let ((name (aref vec i))) | |
697 | (if name | |
698 | (let ((tail (cdr (setq name (copy-sequence name)))) | |
699 | elt) | |
700 | (while tail | |
701 | (setq elt (car tail)) | |
702 | (or (string= elt "-") | |
703 | (progn | |
704 | (setcdr tail (cons elt (cdr tail))) | |
705 | (setcar tail " "))) | |
706 | (setq tail (cddr tail))) | |
707 | (setq name (apply 'concat name)))) | |
708 | (aset table c name) | |
709 | (if (= c char) | |
710 | (setq val name)))) | |
0902a04e | 711 | (or val "")))) |
f600cf3a KH |
712 | |
713 | ((and (integerp val) (> val 0)) | |
714 | (let* ((symbol-table (aref (char-table-extra-slot table 4) 1)) | |
715 | (sym (aref symbol-table (1- val)))) | |
716 | (cond ((eq sym 'HANGUL\ SYLLABLE) | |
717 | (let ((jamo-name-table (aref (char-table-extra-slot table 4) 2))) | |
718 | ;; SIndex = S - SBase | |
719 | (setq char (- char #xAC00)) | |
720 | (let ( ;; LIndex = SIndex / NCount | |
721 | (L (/ char 588)) | |
722 | ;; VIndex = (SIndex % NCount) * TCount | |
723 | (V (/ (% char 588) 28)) | |
724 | ;; TIndex = SIndex % TCount | |
725 | (T (% char 28))) | |
dd7d19d7 | 726 | (format "HANGUL SYLLABLE %s%s%s" |
752fc44b KH |
727 | ;; U+110B is nil in this table. |
728 | (or (aref (aref jamo-name-table 0) L) "") | |
f600cf3a KH |
729 | (aref (aref jamo-name-table 1) V) |
730 | (if (= T 0) "" | |
731 | (aref (aref jamo-name-table 2) (1- T))))))) | |
e60b06d7 KH |
732 | ((eq sym 'CJK\ IDEOGRAPH) |
733 | (format "%s-%04X" sym char)) | |
f600cf3a KH |
734 | ((eq sym 'CJK\ COMPATIBILITY\ IDEOGRAPH) |
735 | (format "%s-%04X" sym char)) | |
736 | ((eq sym 'VARIATION\ SELECTOR) | |
0902a04e KH |
737 | (format "%s-%d" sym (+ (- char #xe0100) 17)))))) |
738 | ||
739 | (t ""))) | |
f600cf3a KH |
740 | |
741 | ;; Store VAL as the name of CHAR in TABLE. | |
742 | ||
743 | (defun unidata-put-name (char val table) | |
744 | (let ((current-val (aref table char))) | |
745 | (if (and (stringp current-val) (= (aref current-val 0) 0)) | |
746 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
747 | (aset table char val))) | |
748 | ||
749 | (defun unidata-get-decomposition (char val table) | |
750 | (cond | |
0902a04e KH |
751 | ((not val) |
752 | (list char)) | |
753 | ||
f600cf3a KH |
754 | ((consp val) |
755 | val) | |
756 | ||
757 | ((stringp val) | |
758 | (if (> (aref val 0) 0) | |
759 | val | |
760 | (let* ((first-char (lsh (lsh char -7) 7)) | |
761 | (word-table (char-table-extra-slot table 4)) | |
762 | (i 1) | |
763 | (len (length val)) | |
764 | (vec (make-vector 128 nil)) | |
765 | (idx 0) | |
766 | (case-fold-search nil) | |
767 | c word-list tail-list last-list word diff-head) | |
768 | (while (< i len) | |
769 | (setq c (aref val i)) | |
770 | (if (< c 3) | |
771 | (progn | |
772 | (if (or word-list tail-list) | |
773 | (aset vec idx | |
774 | (setq last-list (nconc word-list tail-list)))) | |
775 | (setq i (1+ i) idx (1+ idx) | |
776 | word-list nil tail-list nil) | |
777 | (if (> c 0) | |
778 | (let ((l last-list)) | |
779 | (if (= c 1) | |
780 | (setq diff-head | |
781 | (prog1 (aref val i) (setq i (1+ i))))) | |
782 | (setq tail-list (nthcdr (% diff-head 16) last-list)) | |
783 | (dotimes (i (/ diff-head 16)) | |
784 | (setq word-list (nconc word-list (list (car l))) | |
785 | l (cdr l)))))) | |
786 | (setq word-list | |
dd7d19d7 | 787 | (nconc word-list |
f600cf3a KH |
788 | (list (or (unidata-decode-word c word-table) c))) |
789 | i (1+ i)))) | |
790 | (if (or word-list tail-list) | |
791 | (aset vec idx (nconc word-list tail-list))) | |
792 | (dotimes (i 128) | |
793 | (aset table (+ first-char i) (aref vec i))) | |
0902a04e KH |
794 | (setq val (aref vec (- char first-char))) |
795 | (or val (list char))))) | |
f600cf3a | 796 | |
1275c1dc KH |
797 | ;; Hangul syllable |
798 | ((and (eq val 0) (>= char #xAC00) (<= char #xD7A3)) | |
799 | ;; SIndex = S (char) - SBase (#xAC00) | |
800 | (setq char (- char #xAC00)) | |
801 | (let (;; L = LBase + SIndex / NCount | |
802 | (L (+ #x1100 (/ char 588))) | |
803 | ;; V = VBase + (SIndex % NCount) * TCount | |
804 | (V (+ #x1161 (/ (% char 588) 28))) | |
0b4ff5df KH |
805 | ;; LV = SBase + (SIndex / TCount) * TCount |
806 | (LV (+ #xAC00 (* (/ char 28) 28))) | |
1275c1dc KH |
807 | ;; T = TBase + SIndex % TCount |
808 | (T (+ #x11A7 (% char 28)))) | |
d636247a KH |
809 | (if (= T #x11A7) |
810 | (list L V) | |
88758986 | 811 | (list LV T)))) |
1275c1dc KH |
812 | |
813 | )) | |
814 | ||
815 | ;; Store VAL as the decomposition information of CHAR in TABLE. | |
f600cf3a KH |
816 | |
817 | (defun unidata-put-decomposition (char val table) | |
818 | (let ((current-val (aref table char))) | |
819 | (if (and (stringp current-val) (= (aref current-val 0) 0)) | |
820 | (funcall (char-table-extra-slot table 1) char current-val table)) | |
821 | (aset table char val))) | |
822 | ||
823 | ;; UnicodeData.txt contains these lines: | |
824 | ;; 0000;<control>;Cc;0;BN;;;;;N;NULL;;;; | |
825 | ;; ... | |
826 | ;; 0020;SPACE;Zs;0;WS;;;;;N;;;;; | |
827 | ;; ... | |
828 | ;; The following command yields a file of about 96K bytes. | |
829 | ;; % gawk -F ';' '{print $1,$2;}' < UnicodeData.txt | gzip > temp.gz | |
830 | ;; With the following function, we can get a file of almost the same | |
831 | ;; the size. | |
832 | ||
833 | ;; Generate a char-table for character names. | |
834 | ||
835 | (defun unidata-gen-table-word-list (prop val-func) | |
836 | (let ((table (make-char-table 'char-code-property-table)) | |
837 | (prop-idx (unidata-prop-index prop)) | |
838 | (word-list (list nil)) | |
839 | word-table | |
840 | block-list block-word-table block-end | |
841 | tail elt range val idx slot) | |
f600cf3a KH |
842 | (setq tail unidata-list) |
843 | (setq block-end -1) | |
844 | (while tail | |
845 | (setq elt (car tail) tail (cdr tail)) | |
846 | (setq range (car elt) | |
847 | val (funcall val-func (nth prop-idx elt))) | |
848 | ;; Treat the sequence of "CJK COMPATIBILITY IDEOGRAPH-XXXX" and | |
849 | ;; "VARIATION SELECTOR-XXX" as a block. | |
850 | (if (and (consp val) (eq prop 'name) | |
851 | (or (and (eq (car val) 'CJK) | |
852 | (eq (nth 1 val) 'COMPATIBILITY)) | |
853 | (and (>= range #xe0100) | |
854 | (eq (car val) 'VARIATION) | |
855 | (eq (nth 1 val) 'SELECTOR)))) | |
856 | (let ((first (car val)) | |
857 | (second (nth 1 val)) | |
858 | (start range)) | |
859 | (while (and (setq elt (car tail) range (car elt) | |
860 | val (funcall val-func (nth prop-idx elt))) | |
861 | (consp val) | |
862 | (eq first (car val)) | |
863 | (eq second (nth 1 val))) | |
864 | (setq block-end range | |
865 | tail (cdr tail))) | |
866 | (setq range (cons start block-end) | |
867 | val (if (eq first 'CJK) 'CJK\ COMPATIBILITY\ IDEOGRAPH | |
868 | 'VARIATION\ SELECTOR)))) | |
869 | ||
870 | (if (consp range) | |
871 | (if val | |
872 | (let ((slot (assq val block-list))) | |
873 | (setq range (cons (car range) (cdr range))) | |
874 | (setq block-end (cdr range)) | |
875 | (if slot | |
876 | (nconc slot (list range)) | |
877 | (push (list val range) block-list)))) | |
878 | (let* ((start (lsh (lsh range -7) 7)) | |
879 | (limit (+ start 127)) | |
880 | (first tail) | |
881 | (vec (make-vector 128 nil)) | |
882 | c name len) | |
883 | (if (<= start block-end) | |
884 | ;; START overlap with the previous block. | |
885 | (aset table range (nth prop-idx elt)) | |
886 | (if val | |
887 | (aset vec (- range start) val)) | |
888 | (while (and (setq elt (car tail) range (car elt)) | |
889 | (integerp range) | |
890 | (<= range limit)) | |
891 | (setq val (funcall val-func (nth prop-idx elt))) | |
892 | (if val | |
893 | (aset vec (- range start) val)) | |
894 | (setq tail (cdr tail))) | |
895 | (setq vec (unidata-word-list-compress vec)) | |
896 | (when vec | |
897 | (dotimes (i (length vec)) | |
898 | (dolist (elt (aref vec i)) | |
899 | (if (symbolp elt) | |
900 | (let ((slot (assq elt word-list))) | |
901 | (if slot | |
902 | (setcdr slot (1+ (cdr slot))) | |
903 | (setcdr word-list | |
904 | (cons (cons elt 1) (cdr word-list)))))))) | |
905 | (set-char-table-range table (cons start limit) vec)))))) | |
906 | (setq word-list (sort (cdr word-list) | |
907 | #'(lambda (x y) (> (cdr x) (cdr y))))) | |
908 | (setq tail word-list idx 0) | |
909 | (while tail | |
910 | (setcdr (car tail) (unidata-encode-word idx)) | |
911 | (setq idx (1+ idx) tail (cdr tail))) | |
912 | (setq word-table (make-vector (length word-list) nil)) | |
913 | (setq idx 0) | |
914 | (dolist (elt word-list) | |
915 | (aset word-table idx (car elt)) | |
916 | (setq idx (1+ idx))) | |
917 | ||
918 | (if (and (eq prop 'decomposition) | |
919 | (> idx 32)) | |
1275c1dc | 920 | (error "Too many symbols in decomposition data")) |
f600cf3a KH |
921 | |
922 | (dotimes (i (/ #x110000 128)) | |
923 | (let* ((idx (* i 128)) | |
924 | (vec (aref table idx))) | |
925 | (when (vectorp vec) | |
926 | (dotimes (i (length vec)) | |
927 | (let ((tail (aref vec i)) | |
928 | elt code) | |
929 | (if (not tail) | |
930 | (aset vec i "\0") | |
931 | (while tail | |
932 | (setq elt (car tail) | |
933 | code (if (integerp elt) elt | |
934 | (cdr (assq elt word-list)))) | |
935 | (setcar tail (string code)) | |
936 | (setq tail (cdr tail))) | |
937 | (aset vec i (mapconcat 'identity (aref vec i) ""))))) | |
938 | (set-char-table-range | |
939 | table (cons idx (+ idx 127)) | |
940 | (mapconcat 'identity vec ""))))) | |
941 | ||
942 | (setq block-word-table (make-vector (length block-list) nil)) | |
943 | (setq idx 0) | |
944 | (dolist (elt block-list) | |
945 | (dolist (e (cdr elt)) | |
946 | (set-char-table-range table e (1+ idx))) | |
947 | (aset block-word-table idx (car elt)) | |
948 | (setq idx (1+ idx))) | |
949 | ||
950 | (set-char-table-extra-slot table 0 prop) | |
951 | (set-char-table-extra-slot table 4 (cons word-table block-word-table)) | |
952 | table)) | |
953 | ||
954 | (defun unidata-split-name (str) | |
955 | (if (symbolp str) | |
956 | str | |
957 | (let ((len (length str)) | |
958 | (l nil) | |
959 | (idx 0) | |
960 | c) | |
961 | (if (= len 0) | |
962 | nil | |
963 | (dotimes (i len) | |
964 | (setq c (aref str i)) | |
965 | (if (= c 32) | |
966 | (setq l (cons (intern (substring str idx i)) l) | |
967 | idx (1+ i)) | |
dd7d19d7 | 968 | (if (and (= c ?-) (< idx i) |
f600cf3a KH |
969 | (< (1+ i) len) (/= (aref str (1+ i)) 32)) |
970 | (setq l (cons '- (cons (intern (substring str idx i)) l)) | |
971 | idx (1+ i))))) | |
972 | (nreverse (cons (intern (substring str idx)) l)))))) | |
973 | ||
c805dec0 | 974 | (defun unidata-gen-table-name (prop &rest ignore) |
f600cf3a KH |
975 | (let* ((table (unidata-gen-table-word-list prop 'unidata-split-name)) |
976 | (word-tables (char-table-extra-slot table 4))) | |
977 | (byte-compile 'unidata-get-name) | |
978 | (byte-compile 'unidata-put-name) | |
979 | (set-char-table-extra-slot table 1 (symbol-function 'unidata-get-name)) | |
980 | (set-char-table-extra-slot table 2 (symbol-function 'unidata-put-name)) | |
981 | ||
982 | (if (eq prop 'name) | |
983 | (set-char-table-extra-slot table 4 | |
984 | (vector (car word-tables) | |
985 | (cdr word-tables) | |
986 | unidata-name-jamo-name-table)) | |
987 | (set-char-table-extra-slot table 4 | |
988 | (vector (car word-tables)))) | |
989 | table)) | |
990 | ||
991 | (defun unidata-split-decomposition (str) | |
992 | (if (symbolp str) | |
993 | str | |
994 | (let ((len (length str)) | |
995 | (l nil) | |
996 | (idx 0) | |
997 | c) | |
998 | (if (= len 0) | |
999 | nil | |
1000 | (dotimes (i len) | |
1001 | (setq c (aref str i)) | |
1002 | (if (= c 32) | |
1003 | (setq l (if (= (aref str idx) ?<) | |
f1478738 | 1004 | (cons (intern (substring str (1+ idx) (1- i))) l) |
1275c1dc | 1005 | (cons (string-to-number (substring str idx i) 16) l)) |
f600cf3a KH |
1006 | idx (1+ i)))) |
1007 | (if (= (aref str idx) ?<) | |
f1478738 | 1008 | (setq l (cons (intern (substring str (1+ idx) (1- len))) l)) |
1275c1dc | 1009 | (setq l (cons (string-to-number (substring str idx len) 16) l))) |
f600cf3a KH |
1010 | (nreverse l))))) |
1011 | ||
1012 | ||
c805dec0 | 1013 | (defun unidata-gen-table-decomposition (prop &rest ignore) |
f600cf3a KH |
1014 | (let* ((table (unidata-gen-table-word-list prop 'unidata-split-decomposition)) |
1015 | (word-tables (char-table-extra-slot table 4))) | |
1016 | (byte-compile 'unidata-get-decomposition) | |
1017 | (byte-compile 'unidata-put-decomposition) | |
1018 | (set-char-table-extra-slot table 1 | |
1019 | (symbol-function 'unidata-get-decomposition)) | |
dd7d19d7 | 1020 | (set-char-table-extra-slot table 2 |
f600cf3a KH |
1021 | (symbol-function 'unidata-put-decomposition)) |
1022 | (set-char-table-extra-slot table 4 (car word-tables)) | |
1023 | table)) | |
1024 | ||
1025 | ||
1026 | \f | |
1027 | (defun unidata-describe-general-category (val) | |
1028 | (cdr (assq val | |
c805dec0 KH |
1029 | '((nil . "Uknown") |
1030 | (Lu . "Letter, Uppercase") | |
f600cf3a KH |
1031 | (Ll . "Letter, Lowercase") |
1032 | (Lt . "Letter, Titlecase") | |
1033 | (Lm . "Letter, Modifier") | |
1034 | (Lo . "Letter, Other") | |
1035 | (Mn . "Mark, Nonspacing") | |
1036 | (Mc . "Mark, Spacing Combining") | |
1037 | (Me . "Mark, Enclosing") | |
1038 | (Nd . "Number, Decimal Digit") | |
1039 | (Nl . "Number, Letter") | |
1040 | (No . "Number, Other") | |
1041 | (Pc . "Punctuation, Connector") | |
1042 | (Pd . "Punctuation, Dash") | |
1043 | (Ps . "Punctuation, Open") | |
1044 | (Pe . "Punctuation, Close") | |
1045 | (Pi . "Punctuation, Initial quote") | |
1046 | (Pf . "Punctuation, Final quote") | |
1047 | (Po . "Punctuation, Other") | |
1048 | (Sm . "Symbol, Math") | |
1049 | (Sc . "Symbol, Currency") | |
1050 | (Sk . "Symbol, Modifier") | |
1051 | (So . "Symbol, Other") | |
1052 | (Zs . "Separator, Space") | |
1053 | (Zl . "Separator, Line") | |
1054 | (Zp . "Separator, Paragraph") | |
1055 | (Cc . "Other, Control") | |
1056 | (Cf . "Other, Format") | |
1057 | (Cs . "Other, Surrogate") | |
1058 | (Co . "Other, Private Use") | |
1059 | (Cn . "Other, Not Assigned"))))) | |
1060 | ||
1061 | (defun unidata-describe-canonical-combining-class (val) | |
1062 | (cdr (assq val | |
1063 | '((0 . "Spacing, split, enclosing, reordrant, and Tibetan subjoined") | |
1064 | (1 . "Overlays and interior") | |
1065 | (7 . "Nuktas") | |
1066 | (8 . "Hiragana/Katakana voicing marks") | |
1067 | (9 . "Viramas") | |
1068 | (10 . "Start of fixed position classes") | |
1069 | (199 . "End of fixed position classes") | |
1070 | (200 . "Below left attached") | |
1071 | (202 . "Below attached") | |
1072 | (204 . "Below right attached") | |
1073 | (208 . "Left attached (reordrant around single base character)") | |
1074 | (210 . "Right attached") | |
1075 | (212 . "Above left attached") | |
1076 | (214 . "Above attached") | |
1077 | (216 . "Above right attached") | |
1078 | (218 . "Below left") | |
1079 | (220 . "Below") | |
1080 | (222 . "Below right") | |
1081 | (224 . "Left (reordrant around single base character)") | |
1082 | (226 . "Right") | |
1083 | (228 . "Above left") | |
1084 | (230 . "Above") | |
1085 | (232 . "Above right") | |
1086 | (233 . "Double below") | |
1087 | (234 . "Double above") | |
1088 | (240 . "Below (iota subscript)"))))) | |
1089 | ||
1090 | (defun unidata-describe-bidi-class (val) | |
1091 | (cdr (assq val | |
1092 | '((L . "Left-to-Right") | |
1093 | (LRE . "Left-to-Right Embedding") | |
1094 | (LRO . "Left-to-Right Override") | |
1095 | (R . "Right-to-Left") | |
1096 | (AL . "Right-to-Left Arabic") | |
1097 | (RLE . "Right-to-Left Embedding") | |
1098 | (RLO . "Right-to-Left Override") | |
1099 | (PDF . "Pop Directional Format") | |
1100 | (EN . "European Number") | |
1101 | (ES . "European Number Separator") | |
1102 | (ET . "European Number Terminator") | |
1103 | (AN . "Arabic Number") | |
1104 | (CS . "Common Number Separator") | |
1105 | (NSM . "Non-Spacing Mark") | |
1106 | (BN . "Boundary Neutral") | |
1107 | (B . "Paragraph Separator") | |
1108 | (S . "Segment Separator") | |
1109 | (WS . "Whitespace") | |
1110 | (ON . "Other Neutrals"))))) | |
1111 | ||
1112 | (defun unidata-describe-decomposition (val) | |
0af3fedf KH |
1113 | (mapconcat |
1114 | #'(lambda (x) | |
1115 | (if (symbolp x) (symbol-name x) | |
1116 | (concat (string ?') | |
1117 | (compose-string (string x) 0 1 (string ?\t x ?\t)) | |
1118 | (string ?')))) | |
1119 | val " ")) | |
f600cf3a | 1120 | |
c805dec0 KH |
1121 | (defun unidata-gen-mirroring-list () |
1122 | (let ((head (list nil)) | |
1123 | tail) | |
1124 | (with-temp-buffer | |
1125 | (insert-file-contents (expand-file-name "BidiMirroring.txt" unidata-dir)) | |
1126 | (goto-char (point-min)) | |
1127 | (setq tail head) | |
1128 | (while (re-search-forward "^\\([0-9A-F]+\\);\\s +\\([0-9A-F]+\\)" nil t) | |
1129 | (let ((char (string-to-number (match-string 1) 16)) | |
1130 | (mirror (match-string 2))) | |
1131 | (setq tail (setcdr tail (list (list char mirror))))))) | |
1132 | (cdr head))) | |
1133 | ||
f600cf3a KH |
1134 | ;; Verify if we can retrieve correct values from the generated |
1135 | ;; char-tables. | |
1136 | ||
1137 | (defun unidata-check () | |
1138 | (dolist (elt unidata-prop-alist) | |
1139 | (let* ((prop (car elt)) | |
1140 | (index (unidata-prop-index prop)) | |
1141 | (generator (unidata-prop-generator prop)) | |
dd7d19d7 | 1142 | (table (progn |
f600cf3a KH |
1143 | (message "Generating %S table..." prop) |
1144 | (funcall generator prop))) | |
1145 | (decoder (char-table-extra-slot table 1)) | |
1146 | (check #x400)) | |
1147 | (dolist (e unidata-list) | |
1148 | (let ((char (car e)) | |
1149 | (val1 (nth index e)) | |
1150 | val2) | |
1151 | (if (and (stringp val1) (= (length val1) 0)) | |
1152 | (setq val1 nil)) | |
1153 | (unless (consp char) | |
1154 | (setq val2 (funcall decoder char (aref table char) table)) | |
1155 | (if val1 | |
1156 | (cond ((eq generator 'unidata-gen-table-symbol) | |
1157 | (setq val1 (intern val1))) | |
1158 | ((eq generator 'unidata-gen-table-integer) | |
1275c1dc | 1159 | (setq val1 (string-to-number val1))) |
f600cf3a | 1160 | ((eq generator 'unidata-gen-table-character) |
1275c1dc | 1161 | (setq val1 (string-to-number val1 16))) |
f600cf3a KH |
1162 | ((eq generator 'unidata-gen-table-decomposition) |
1163 | (setq val1 (unidata-split-decomposition val1))))) | |
1164 | (when (>= char check) | |
1165 | (message "%S %04X" prop check) | |
1166 | (setq check (+ check #x400))) | |
1167 | (or (equal val1 val2) | |
dd7d19d7 | 1168 | (insert (format "> %04X %S\n< %04X %S\n" |
f600cf3a KH |
1169 | char val1 char val2))) |
1170 | (sit-for 0))))))) | |
1171 | ||
1172 | ;; The entry function. It generates files described in the header | |
1173 | ;; comment of this file. | |
1174 | ||
c805dec0 KH |
1175 | (defun unidata-gen-files (&optional data-dir unidata-text-file) |
1176 | (or data-dir | |
1177 | (setq data-dir (car command-line-args-left) | |
1178 | command-line-args-left (cdr command-line-args-left) | |
1179 | unidata-text-file (car command-line-args-left) | |
3921ea21 | 1180 | command-line-args-left (cdr command-line-args-left))) |
db516472 | 1181 | (let ((coding-system-for-write 'utf-8-unix) |
c805dec0 KH |
1182 | (charprop-file "charprop.el") |
1183 | (unidata-dir data-dir)) | |
1184 | (dolist (elt unidata-prop-alist) | |
1185 | (let* ((prop (car elt)) | |
1186 | (file (unidata-prop-file prop))) | |
1187 | (if (file-exists-p file) | |
1188 | (delete-file file)))) | |
1189 | (unidata-setup-list unidata-text-file) | |
f600cf3a KH |
1190 | (with-temp-file charprop-file |
1191 | (insert ";; Automatically generated by unidata-gen.el.\n") | |
1192 | (dolist (elt unidata-prop-alist) | |
1193 | (let* ((prop (car elt)) | |
1194 | (generator (unidata-prop-generator prop)) | |
1195 | (file (unidata-prop-file prop)) | |
1196 | (docstring (unidata-prop-docstring prop)) | |
1197 | (describer (unidata-prop-describer prop)) | |
c805dec0 KH |
1198 | (default-value (unidata-prop-default prop)) |
1199 | (val-list (unidata-prop-val-list prop)) | |
f600cf3a KH |
1200 | table) |
1201 | ;; Filename in this comment line is extracted by sed in | |
1202 | ;; Makefile. | |
1203 | (insert (format ";; FILE: %s\n" file)) | |
1204 | (insert (format "(define-char-code-property '%S %S\n %S)\n" | |
1205 | prop file docstring)) | |
c805dec0 | 1206 | (with-temp-buffer |
f600cf3a | 1207 | (message "Generating %s..." file) |
c805dec0 KH |
1208 | (when (file-exists-p file) |
1209 | (insert-file-contents file) | |
1210 | (goto-char (point-max)) | |
1211 | (search-backward ";; Local Variables:")) | |
1212 | (setq table (funcall generator prop default-value val-list)) | |
f600cf3a KH |
1213 | (when describer |
1214 | (unless (subrp (symbol-function describer)) | |
1215 | (byte-compile describer) | |
1216 | (setq describer (symbol-function describer))) | |
1217 | (set-char-table-extra-slot table 3 describer)) | |
c805dec0 KH |
1218 | (if (bobp) |
1219 | (insert ";; Copyright (C) 1991-2009 Unicode, Inc. | |
1220 | ;; This file was generated from the Unicode data files at | |
1221 | ;; http://www.unicode.org/Public/UNIDATA/. | |
1222 | ;; See lisp/international/README for the copyright and permission notice.\n")) | |
1223 | (insert (format "(define-char-code-property '%S %S %S)\n" | |
1224 | prop table docstring)) | |
1225 | (if (eobp) | |
1226 | (insert ";; Local Variables:\n" | |
1227 | ";; coding: utf-8\n" | |
1228 | ";; no-byte-compile: t\n" | |
1229 | ";; End:\n\n" | |
1230 | (format ";; %s ends here\n" file))) | |
1231 | (write-file file) | |
1232 | (message "Generating %s...done" file)))) | |
f600cf3a | 1233 | (message "Writing %s..." charprop-file) |
1275c1dc | 1234 | (insert ";; Local Variables:\n" |
f600cf3a KH |
1235 | ";; coding: utf-8\n" |
1236 | ";; no-byte-compile: t\n" | |
1237 | ";; End:\n\n" | |
1238 | (format ";; %s ends here\n" charprop-file))))) | |
1239 | ||
1275c1dc KH |
1240 | \f |
1241 | ||
f600cf3a | 1242 | ;;; unidata-gen.el ends here |