Mercurial > emacs

comparison lisp/language/indian.el @ 107925:d95939738dba

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Fix previous change of *-composable-pattern.
author Kenichi Handa <handa@etlken>
date Wed, 07 Apr 2010 15:09:33 +0900
parents 64dd8cad5b43
children 6a2e327273fe
comparison
equal deleted inserted replaced
107920:6ab7fa5331f8 107925:d95939738dba
239 (indian-compose-regexp 239 (indian-compose-regexp
240 (concat 240 (concat
241 ;; syllables with an independent vowel, or 241 ;; syllables with an independent vowel, or
242 "\\(?:RH\\)?Vn?\\(?:J?HR\\)?v*n?a?A?\\|" 242 "\\(?:RH\\)?Vn?\\(?:J?HR\\)?v*n?a?A?\\|"
243 ;; consonant-based syllables, or 243 ;; consonant-based syllables, or
244 "Cn?\\(?:J?HJ?Cn?\\)*\\(?:H[NJ]?|v*n?a?A?\\)\\|" 244 "Cn?\\(?:J?HJ?Cn?\\)*\\(?:H[NJ]?\\|v*n?a?A?\\)\\|"
245 ;; special consonant form, or 245 ;; special consonant form, or
246 "JHR\\|" 246 "JHR\\|"
247 ;; any other singleton characters 247 ;; any other singleton characters
248 "X") 248 "X")
249 table)) 249 table))
254 '(("a" . "\u0B01") ; SIGN CANDRABINDU 254 '(("a" . "\u0B01") ; SIGN CANDRABINDU
255 ("A" . "[\u0B02-\u0B03]") ; SIGN ANUSVARA .. VISARGA 255 ("A" . "[\u0B02-\u0B03]") ; SIGN ANUSVARA .. VISARGA
256 ("V" . "[\u0B05-\u0B14\u0B60-\u0B61]") ; independent vowel 256 ("V" . "[\u0B05-\u0B14\u0B60-\u0B61]") ; independent vowel
257 ("C" . "[\u0B15-\u0B39\u0B5C-\u0B5D\u0B71]") ; consonant 257 ("C" . "[\u0B15-\u0B39\u0B5C-\u0B5D\u0B71]") ; consonant
258 ("B" . "[\u0B15-\u0B17\u0B1B-\u0B1D\u0B1F-\u0B21\u0B23-\u0B24\u0B27-\u0B30\u0B32-\u0B35\u0B38-\u0B39]") ; consonant with below form 258 ("B" . "[\u0B15-\u0B17\u0B1B-\u0B1D\u0B1F-\u0B21\u0B23-\u0B24\u0B27-\u0B30\u0B32-\u0B35\u0B38-\u0B39]") ; consonant with below form
259 ("R" . "\u0B30") ; RA
259 ("n" . "\u0B3C") ; NUKTA 260 ("n" . "\u0B3C") ; NUKTA
260 ("v" . "[\u0B3E-\u0B44\u0B56-\u0B57\u0B62-\u0B63]") ; vowel sign 261 ("v" . "[\u0B3E-\u0B44\u0B47\u0B56-\u0B57\u0B62-\u0B63]") ; vowel sign
261 ("H" . "\u0B4D") ; VIRAMA 262 ("H" . "\u0B4D") ; VIRAMA
262 ("N" . "\u200C") ; ZWNJ 263 ("N" . "\u200C") ; ZWNJ
263 ("J" . "\u200D") ; ZWJ 264 ("J" . "\u200D") ; ZWJ
264 ("X" . "[\u0B00-\u0B7F]")))) ; all coverage 265 ("X" . "[\u0B00-\u0B7F]")))) ; all coverage
265 (indian-compose-regexp 266 (indian-compose-regexp
266 (concat 267 (concat
267 ;; syllables with an independent vowel, or 268 ;; syllables with an independent vowel, or
268 "\\(?:RH\\)?Vn?\\(?:J?HB\\)?v*n?a?A?\\|" 269 "\\(?:RH\\)?Vn?\\(?:J?HB\\)?v*n?a?A?\\|"
269 ;; consonant-based syllables, or 270 ;; consonant-based syllables, or
270 "Cn?\\(?:J?HJ?Cn?\\)*\\(?:H[NJ]?|v*n?a?A?\\)\\|" 271 "Cn?\\(?:J?HJ?Cn?\\)*\\(?:H[NJ]?\\|v*n?a?A?\\)\\|"
271 ;; special consonant form, or 272 ;; special consonant form, or
272 "JHB\\|" 273 "JHB\\|"
273 ;; any other singleton characters 274 ;; any other singleton characters
274 "X") 275 "X")
275 table)) 276 table))
286 ("J" . "\u200D") ; ZWJ 287 ("J" . "\u200D") ; ZWJ
287 ("X" . "[\u0B80-\u0BFF]")))) ; all coverage 288 ("X" . "[\u0B80-\u0BFF]")))) ; all coverage
288 (indian-compose-regexp 289 (indian-compose-regexp
289 (concat 290 (concat
290 ;; consonant-based syllables, or 291 ;; consonant-based syllables, or
291 "C\\(?:J?HJ?C\\)*\\(?:H[NJ]?|v*a?\\)\\|" 292 "C\\(?:J?HJ?C\\)*\\(?:H[NJ]?\\|v*a?\\)\\|"
292 ;; syllables with an independent vowel, or 293 ;; syllables with an independent vowel, or
293 "Vv*a?\\|" 294 "Vv*a?\\|"
294 ;; any other singleton characters 295 ;; any other singleton characters
295 "X") 296 "X")
296 table)) 297 table))
300 (let ((table 301 (let ((table
301 '(("a" . "[\u0C01-\u0C03]") ; SIGN CANDRABINDU .. VISARGA 302 '(("a" . "[\u0C01-\u0C03]") ; SIGN CANDRABINDU .. VISARGA
302 ("V" . "[\u0C05-\u0C14\u0C60-\u0C61]") ; independent vowel 303 ("V" . "[\u0C05-\u0C14\u0C60-\u0C61]") ; independent vowel
303 ("C" . "[\u0C15-\u0C39\u0C58-\u0C59]") ; consonant 304 ("C" . "[\u0C15-\u0C39\u0C58-\u0C59]") ; consonant
304 ("v" . "[\u0C3E-\u0C4C\u0C55-\u0C56\u0C62-\u0C63]") ; vowel sign 305 ("v" . "[\u0C3E-\u0C4C\u0C55-\u0C56\u0C62-\u0C63]") ; vowel sign
305 ("H" . "\u0BCD") ; VIRAMA 306 ("H" . "\u0C4D") ; VIRAMA
306 ("N" . "\u200C") ; ZWNJ 307 ("N" . "\u200C") ; ZWNJ
307 ("J" . "\u200D") ; ZWJ 308 ("J" . "\u200D") ; ZWJ
308 ("X" . "[\u0C00-\u0C7F]")))) ; all coverage 309 ("X" . "[\u0C00-\u0C7F]")))) ; all coverage
309 (indian-compose-regexp 310 (indian-compose-regexp
310 (concat 311 (concat
311 ;; consonant-based syllables, or 312 ;; consonant-based syllables, or
312 "C\\(?:J?HJ?C\\)*\\(?:H[NJ]?|v*a?\\)\\|" 313 "C\\(?:J?HJ?C\\)*\\(?:H[NJ]?\\|v*a?\\)\\|"
313 ;; syllables with an independent vowel, or 314 ;; syllables with an independent vowel, or
314 "V\\(?:J?HC\\)?v*a?\\|" 315 "V\\(?:J?HC\\)?v*a?\\|"
315 ;; special consonant form, or 316 ;; special consonant form, or
316 "JHC\\|" 317 "JHC\\|"
317 ;; any other singleton characters 318 ;; any other singleton characters
322 (defconst kannada-composable-pattern 323 (defconst kannada-composable-pattern
323 (let ((table 324 (let ((table
324 '(("A" . "[\u0C82-\u0C83]") ; SIGN ANUSVARA .. VISARGA 325 '(("A" . "[\u0C82-\u0C83]") ; SIGN ANUSVARA .. VISARGA
325 ("V" . "[\u0C85-\u0C94\u0CE0-\u0CE1]") ; independent vowel 326 ("V" . "[\u0C85-\u0C94\u0CE0-\u0CE1]") ; independent vowel
326 ("C" . "[\u0C95-\u0CB9\u0CDE]") ; consonant 327 ("C" . "[\u0C95-\u0CB9\u0CDE]") ; consonant
327 ("B" . "\u0CB0") ; RA 328 ("R" . "\u0CB0") ; RA
328 ("n" . "\u0CBC") ; NUKTA 329 ("n" . "\u0CBC") ; NUKTA
329 ("v" . "[\u0CBE-\u0CCC\u0CD5-\u0CD6\u0CE2-\u0CE3]") ; vowel sign 330 ("v" . "[\u0CBE-\u0CCC\u0CD5-\u0CD6\u0CE2-\u0CE3]") ; vowel sign
330 ("H" . "\u0CCD") ; VIRAMA 331 ("H" . "\u0CCD") ; VIRAMA
331 ("N" . "\u200C") ; ZWNJ 332 ("N" . "\u200C") ; ZWNJ
332 ("J" . "\u200D") ; ZWJ 333 ("J" . "\u200D") ; ZWJ
334 (indian-compose-regexp 335 (indian-compose-regexp
335 (concat 336 (concat
336 ;; syllables with an independent vowel, or 337 ;; syllables with an independent vowel, or
337 "\\(?:RH\\)?Vn?\\(?:J?HC\\)?v?A?\\|" 338 "\\(?:RH\\)?Vn?\\(?:J?HC\\)?v?A?\\|"
338 ;; consonant-based syllables, or 339 ;; consonant-based syllables, or
339 "Cn?\\(?:J?HJ?Cn?\\)*\\(?:H[NJ]?|v*n?A?\\)\\|" 340 "Cn?\\(?:J?HJ?Cn?\\)*\\(?:H[NJ]?\\|v*n?A?\\)\\|"
340 ;; special consonant form, or 341 ;; special consonant form, or
341 "JHB\\|" 342 "JHC\\|"
342 ;; any other singleton characters 343 ;; any other singleton characters
343 "X") 344 "X")
344 table)) 345 table))
345 "Regexp matching a composable sequence of Kannada characters.") 346 "Regexp matching a composable sequence of Kannada characters.")
346 347
349 '(("A" . "[\u0D02-\u0D03]") ; SIGN ANUSVARA .. VISARGA 350 '(("A" . "[\u0D02-\u0D03]") ; SIGN ANUSVARA .. VISARGA
350 ("V" . "[\u0D05-\u0D14\u0D60-\u0D61]") ; independent vowel 351 ("V" . "[\u0D05-\u0D14\u0D60-\u0D61]") ; independent vowel
351 ("C" . "[\u0D15-\u0D39]") ; consonant 352 ("C" . "[\u0D15-\u0D39]") ; consonant
352 ("Y" . "[\u0D2F-\u0D30\u0D32\u0D35]") ; YA, RA, LA, VA 353 ("Y" . "[\u0D2F-\u0D30\u0D32\u0D35]") ; YA, RA, LA, VA
353 ("v" . "[\u0D3E-\u0D48\u0D57\u0D62-\u0D63]") ; postbase matra 354 ("v" . "[\u0D3E-\u0D48\u0D57\u0D62-\u0D63]") ; postbase matra
355 ("H" . "\u0D4D") ; SIGN VIRAMA
354 ("N" . "\u200C") ; ZWNJ 356 ("N" . "\u200C") ; ZWNJ
355 ("J" . "\u200D") ; ZWJ 357 ("J" . "\u200D") ; ZWJ
356 ("X" . "[\u0D00-\u0D7F]")))) ; all coverage 358 ("X" . "[\u0D00-\u0D7F]")))) ; all coverage
357 (indian-compose-regexp 359 (indian-compose-regexp
358 (concat 360 (concat
359 ;; consonant-based syllables, or 361 ;; consonant-based syllables, or
360 "\\(?:CJ?HJ?C\\)*\\(?:H[NJ]?\\|v?A?\\)\\|" 362 "C\\(?:J?HJ?C\\)*\\(?:H[NJ]?\\|v?A?\\)\\|"
361 ;; syllables with an independent vowel, or 363 ;; syllables with an independent vowel, or
362 "V\\(?:J?HY\\)?v*?A?\\|" 364 "V\\(?:J?HY\\)?v*?A?\\|"
363 ;; special consonant form, or 365 ;; special consonant form, or
364 "JHY\\|" 366 "JHY\\|"
365 ;; any other singleton characters 367 ;; any other singleton characters