code.delx.au - gnu-emacs/blob - lisp/language/devan-util.el

   1 ;;; devan-util.el --- Support for composing Devanagari characters -*-coding: iso-2022-7bit;-*-
   2
   3 ;; Copyright (C) 2001 Free Software Foundation, Inc.
   4
   5 ;; Maintainer:  KAWABATA, Taichi <kawabata@m17n.org>
   6 ;; Keywords: multilingual, Devanagari
   7
   8 ;; This file is part of GNU Emacs.
   9
  10 ;; GNU Emacs is free software; you can redistribute it and/or modify
  11 ;; it under the terms of the GNU General Public License as published by
  12 ;; the Free Software Foundation; either version 2, or (at your option)
  13 ;; any later version.
  14
  15 ;; GNU Emacs is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 ;; GNU General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  22 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  23 ;; Boston, MA 02111-1307, USA.
  24
  25 ;; Created: Feb. 17. 2001
  26
  27 ;;; Commentary:
  28
  29 ;; This file provides character(Unicode) to glyph(CDAC) conversion and
  30 ;; composition of Devanagari script characters.
  31
  32 ;;; Code:
  33
  34 ;; Devanagari Composable Pattern
  35 ;;    C .. Consonants
  36 ;;    V .. Vowel
  37 ;;    H .. Halant
  38 ;;    M .. Matra
  39 ;;    V .. Vowel
  40 ;;    A .. Anuswar
  41 ;;    D .. Chandrabindu
  42 ;;    (N .. Zerowidth Non Joiner)
  43 ;;    (J .. Zerowidth Joiner.  )
  44 ;; 1. vowel
  45 ;;  V(A/D)?
  46 ;; 2. syllable : maximum of 5 consecutive consonants.  (e.g. kartsnya)
  47 ;;  ((CH)?(CH)?(CH)?CH)?C(H|M?(A|D)?)?
  48
  49 (defconst devanagari-consonant
  50   "[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]")
  51
  52     ;;("\e$,16B\e(B" . nil)
  53     ;;("\e$,16A\e(B" . nil)
  54     ;;("\e$,16C\e(B" . nil)
  55
  56
  57 (defconst devanagari-composable-pattern
  58   (concat
  59    "\\([\e$,15E\e(B-\e$,15T6@6A\e(B][\e$,15A5B\e(B]?\\)\\|[\e$,15C6D\e(B]"
  60    "\\|\\("
  61    "\\(?:\\(?:[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]\e$,16-\e(B\\)?\\(?:[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]\e$,16-\e(B\\)?\\(?:[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]\e$,16-\e(B\\)?[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]\e$,16-\e(B\\)?"
  62    "[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]\\(?:\e$,16-\e(B\\|[\e$,15~\e(B-\e$,16-6B6C\e(B]?[\e$,15B5A\e(B]?\\)?"
  63    "\\)")
  64   "Regexp matching a composable sequence of Devanagari characters.")
  65
  66 (dolist (range '((#x0903 . #x0903)
  67                  (#x0905 . #x0939)
  68                  (#x0958 . #x0961)))
  69   (set-char-table-range indian-composable-pattern range
  70                         devanagari-composable-pattern))
  71
  72 ;;;###autoload
  73 (defun devanagari-compose-region (from to)
  74   (interactive "r")
  75   (save-excursion
  76     (save-restriction
  77       (narrow-to-region from to)
  78       (goto-char (point-min))
  79       (while (re-search-forward devanagari-composable-pattern nil t)
  80         (devanagari-compose-syllable-region (match-beginning 0)
  81                                             (match-end 0))))))
  82 (defun devanagari-compose-string (string)
  83   (with-temp-buffer
  84     (insert (decompose-string string))
  85     (devanagari-compose-region (point-min) (point-max))
  86     (buffer-string)))
  87
  88 (defun devanagari-post-read-conversion (len)
  89   (save-excursion
  90     (save-restriction
  91       (let ((buffer-modified-p (buffer-modified-p)))
  92         (narrow-to-region (point) (+ (point) len))
  93         (devanagari-compose-region (point-min) (point-max))
  94         (set-buffer-modified-p buffer-modified-p)
  95         (- (point-max) (point-min))))))
  96
  97 (defun devanagari-regexp-of-hashtbl-keys (hashtbl)
  98   "Return a regular expression that matches all keys in hashtable HASHTBL."
  99   (let ((max-specpdl-size 1000))
 100     (regexp-opt
 101      (sort
 102       (let (dummy)
 103         (maphash (function (lambda (key val) (setq dummy (cons key dummy)))) hashtbl)
 104         dummy)
 105       (function (lambda (x y) (> (length x) (length y))))))))
 106
 107 ;;;###autoload
 108 (defun devanagari-composition-function (pos &optional string)
 109   "Compose Devanagari characters after the position POS.
 110 If STRING is not nil, it is a string, and POS is an index to the string.
 111 In this case, compose characters after POS of the string."
 112   (if string
 113       ;; Not yet implemented.
 114       nil
 115     (goto-char pos)
 116     (if (looking-at devanagari-composable-pattern)
 117         (prog1 (match-end 0)
 118           (devanagari-compose-syllable-region pos (match-end 0))))))
 119
 120 ;; Notes on conversion steps.
 121
 122 ;; 1. chars to glyphs
 123 ;;
 124 ;; Rules will not be applied to the halant appeared at the end of the
 125 ;; text.  Also, the preceding/following "r" will be treated as special case.
 126
 127 ;; 2. glyphs reordering.
 128 ;;
 129 ;; The glyphs are split by halant, and each glyph groups are
 130 ;; re-ordered in the following order.
 131 ;;
 132 ;; Note that `consonant-glyph' mentioned here does not contain the
 133 ;; vertical bar (right modifier) attached at the right of the
 134 ;; consonant.
 135 ;;
 136 ;; If the glyph-group contains right modifier,
 137 ;;  (1) consonant-glyphs/vowels, with nukta sign
 138 ;;  (2) spacing
 139 ;;  (3) right modifier (may be matra)
 140 ;;  (4) top matra
 141 ;;  (5) preceding "r"
 142 ;;  (6) anuswar
 143 ;;  (7) following "r"
 144 ;;  (8) bottom matra or halant.
 145 ;;
 146 ;; Otherwise,
 147 ;;  (1) consonant-glyph/vowels, with nukta sign
 148 ;;  (3) left matra
 149 ;;  (4) top matra
 150 ;;  (5) preceding "r"
 151 ;;  (6) anuswar
 152 ;;  (7) following "r"
 153 ;;  (8) bottom matra or halant.
 154 ;;  (2) spacing
 155
 156 ;; 3. glyph to glyph
 157 ;;
 158 ;; For better display, some glyph display would be tuned.
 159
 160 ;; 4. Composition.
 161 ;;
 162 ;; left modifiers will be attached at the left.
 163 ;; others will be attached right.
 164
 165 ;; Problem::
 166 ;;  Can we generalize this methods to other Indian scripts?
 167
 168 (defvar dev-char-glyph
 169   '(("\e$,15E\e(B" . "\e$,4 K\e(B")
 170     ("\e$,15F\e(B" . "\e$,4 K")\e(B")
 171     ("\e$,15~\e(B" . "\e$,4")\e(B")
 172     ("\e$,15G\e(B" . "\e$,4 \\e(B")
 173     ("\e$,15\7f\e(B" . "\e$,4"*\e(B")
 174     ("\e$,15\7f5A\e(B" . "\e$,4"*\e(B\\e$,4"&\e(B")
 175     ("\e$,15H\e(B" . "\e$,4 \"'\e(B")
 176     ("\e$,15H5A\e(B" . "\e$,4 \"'"&\e(B")
 177     ("\e$,16 \e(B" . "\e$,4"2\e(B")
 178     ("\e$,16 5A\e(B" . "\e$,4"2"&\e(B")
 179     ("\e$,15I\e(B" . "\e$,4 ]\e(B")
 180     ("\e$,16!\e(B" . "\e$,4"6\e(B")
 181     ("\e$,15J\e(B" . "\e$,4 ^"P\e(B")
 182     ("\e$,16"\e(B" . "\e$,4":\e(B")
 183     ("\e$,15K\e(B" . "\e$,4 `"Q\e(B")
 184     ("\e$,16#\e(B" . "\e$,4">\e(B")
 185     ;;("\e$,15L\e(B" . nil) ; not implemented.
 186     ("\e$,16$\e(B" . "\e$,4"?\e(B")
 187     ("\e$,15M\e(B" . "\e$,4 b"L\e(B")
 188     ("\e$,15M5A\e(B" . "\e$,4 b"$\e(B")
 189     ("\e$,15M5B\e(B" . "\e$,4 b"$\e(B")
 190     ("\e$,16%\e(B" . "\\e$,4"L\e(B")
 191     ("\e$,15N\e(B" . "\e$,4 b"@\e(B")
 192     ("\e$,15N5A\e(B" . "\e$,4 b"@"&\e(B")
 193     ("\e$,16&\e(B" . "\\e$,4"@\e(B")
 194     ("\e$,16&5A\e(B" . "\\e$,4"@\e(B\\e$,4"&\e(B")
 195     ("\e$,15O\e(B" . "\e$,4 b\e(B")
 196     ("\e$,16'\e(B" . "\\e$,4"D\e(B")
 197     ("\e$,16'5A\e(B" . "\\e$,4"D\e(B\\e$,4"&\e(B")
 198     ("\e$,15P\e(B" . "\e$,4 b"D\e(B")
 199     ("\e$,15P5A\e(B" . "\e$,4 b"D"&\e(B")
 200     ("\e$,16(\e(B" . "\\e$,4"H\e(B")
 201     ("\e$,16(5A\e(B" . "\\e$,4"H\e(B\\e$,4"&\e(B")
 202     ("\e$,15Q\e(B" . "\e$,4 K")"L\e(B") ;; special rule for reodering.
 203     ("\e$,15Q5A\e(B" . "\e$,4 K")"$\e(B")
 204     ("\e$,15Q5B\e(B" . "\e$,4 K")"$\e(B")
 205     ("\e$,16)\e(B" . "\\e$,4")"L\e(B")
 206     ("\e$,16)5A\e(B" . "\\e$,4")"$\e(B")
 207     ("\e$,16)5B\e(B" . "\\e$,4")"$\e(B")
 208     ("\e$,15R\e(B" . "\e$,4 K")"@\e(B")
 209     ("\e$,15R5A\e(B" . "\e$,4 K")"@"&\e(B")
 210     ("\e$,16*\e(B" . "\\e$,4")"@\e(B")
 211     ("\e$,16*5A\e(B" . "\\e$,4")"@"&\e(B")
 212     ("\e$,15S\e(B" . "\e$,4 K")"D\e(B")
 213     ("\e$,15S5A\e(B" . "\e$,4 K")"D"&\e(B")
 214     ("\e$,16+\e(B" . "\\e$,4")"D\e(B")
 215     ("\e$,16+5A\e(B" . "\\e$,4")"D"&\e(B")
 216     ("\e$,15T\e(B" . "\e$,4 K")"H\e(B")
 217     ("\e$,15T5A\e(B" . "\e$,4 K")"H"&\e(B")
 218     ("\e$,16,\e(B" . "\\e$,4")"H\e(B")
 219     ("\e$,16,5A\e(B" . "\\e$,4")"H"&\e(B")
 220     ("\e$,16@\e(B" . "\e$,4 a"Q\e(B")
 221     ;;("\e$,16B\e(B" . nil)
 222     ;;("\e$,16A\e(B" . nil)
 223     ;;("\e$,16C\e(B" . nil)
 224
 225     ;; GRUTTALS
 226     ("\e$,15U\e(B" . "\e$,4 e"R\e(B")
 227     ("\e$,15U6-\e(B" . "\e$,4 c\e(B")
 228     ("\e$,15U6-5p\e(B" . "\e$,4 g"R\e(B")
 229     ("\e$,15U6-5d\e(B" . "\e$,4 h"R\e(B")
 230     ("\e$,15U6-5w\e(B" . "\e$,4 i")\e(B")
 231     ("\e$,15U6-5w6-\e(B" . "\e$,4 i\e(B")
 232
 233     ("\e$,15V\e(B" . "\e$,4 j")\e(B")
 234     ("\e$,15V6-\e(B" . "\e$,4 j\e(B")
 235     ("\e$,15V6-5p\e(B" . "\e$,4 l")\e(B")
 236     ("\e$,15V6-5p6-\e(B" . "\e$,4 l\e(B")
 237
 238     ("\e$,15W\e(B" . "\e$,4 m")\e(B")
 239     ("\e$,15W6-\e(B" . "\e$,4 m\e(B")
 240     ("\e$,15W6-5p\e(B" . "\e$,4 o")\e(B")
 241     ("\e$,15W6-5p6-\e(B" . "\e$,4 o\e(B")
 242
 243     ("\e$,15X\e(B" . "\e$,4 p")\e(B")
 244     ("\e$,15X6-\e(B" . "\e$,4 p\e(B")
 245     ("\e$,15X6-5p\e(B" . "\e$,4 q")\e(B")
 246     ("\e$,15X6-5p6-\e(B" . "\e$,4 q\e(B")
 247
 248     ("\e$,15Y\e(B" . "\e$,4 r"S\e(B")
 249     ;; PALATALS
 250     ("\e$,15Z\e(B" . "\e$,4 s")\e(B")
 251     ("\e$,15Z6-\e(B" . "\e$,4 s\e(B")
 252     ("\e$,15Z6-5p\e(B" . "\e$,4 t")\e(B")
 253     ("\e$,15Z6-5p6-\e(B" . "\e$,4 t\e(B")
 254
 255     ("\e$,15[\e(B" . "\e$,4 u"T\e(B")
 256
 257     ("\e$,15\\e(B" . "\e$,4 v")\e(B")
 258     ("\e$,15\6-\e(B" . "\e$,4 v\e(B")
 259     ("\e$,15\6-5p\e(B" . "\e$,4 x")\e(B")
 260     ("\e$,15\6-5p6-\e(B" . "\e$,4 x\e(B")
 261     ("\e$,15\6-5^\e(B" . "\e$,4 y")\e(B")
 262     ("\e$,15\6-5^6-\e(B" . "\e$,4 y\e(B")
 263
 264     ("\e$,15]\e(B" . "\e$,4 z")\e(B")
 265     ("\e$,15]6-\e(B" . "\e$,4 z\e(B")
 266     ("\e$,15]6-5p\e(B" . "\e$,4 {")\e(B")
 267     ("\e$,15]6-5p6-\e(B" . "\e$,4 {\e(B")
 268
 269     ("\e$,15^\e(B" . "\e$,4 |")\e(B")
 270     ("\e$,15^6-\e(B" . "\e$,4 |\e(B")
 271     ;; CEREBRALS
 272     ("\e$,15_\e(B" . "\e$,4 }"U\e(B")
 273     ("\e$,15_6-5_\e(B" . "\e$,4 ~"U\e(B")
 274     ("\e$,15_6-5`\e(B" . "\e$,4 \7f"U\e(B")
 275
 276     ("\e$,15`\e(B" . "\e$,4! "V\e(B")
 277     ("\e$,15`6-5`\e(B" . "\e$,4!!"V\e(B")
 278
 279     ("\e$,15a\e(B" . "\e$,4!""W\e(B")
 280     ("\e$,15a6-5a\e(B" . "\e$,4!$"W\e(B")
 281     ("\e$,15a6-5b\e(B" . "\e$,4!%"W\e(B")
 282
 283     ("\e$,15b\e(B" . "\e$,4!&"X\e(B")
 284
 285     ("\e$,15c\e(B" . "\e$,4!(")\e(B")
 286     ("\e$,15c6-\e(B" . "\e$,4!(\e(B")
 287     ;; DENTALS
 288     ("\e$,15d\e(B" . "\e$,4!)")\e(B")
 289     ("\e$,15d6-\e(B" . "\e$,4!)\e(B")
 290     ("\e$,15d6-5p\e(B" . "\e$,4!*")\e(B")
 291     ("\e$,15d6-5p6-\e(B" . "\e$,4!*\e(B")
 292     ("\e$,15d6-5d\e(B" . "\e$,4!+")\e(B")
 293     ("\e$,15d6-5d6-\e(B" . "\e$,4!+\e(B")
 294
 295     ("\e$,15e\e(B" . "\e$,4!,")\e(B")
 296     ("\e$,15e6-\e(B" . "\e$,4!,\e(B")
 297     ("\e$,15e6-5p\e(B" . "\e$,4!-")\e(B")
 298     ("\e$,15e6-5p6-\e(B" . "\e$,4!-\e(B")
 299
 300     ("\e$,15f\e(B" . "\e$,4!."Y\e(B")
 301     ("\e$,15f6#\e(B" . "\e$,4!/"Y\e(B")
 302     ("\e$,15f6-5p\e(B" . "\e$,4!0"Y\e(B")
 303     ("\e$,15f6-5f\e(B" . "\e$,4!1"Y\e(B")
 304     ("\e$,15f6-5g\e(B" . "\e$,4!2"Y\e(B")
 305     ("\e$,15f6-5n\e(B" . "\e$,4!3\e(B")
 306     ("\e$,15f6-5o\e(B" . "\e$,4!4\e(B")
 307     ("\e$,15f6-5u\e(B" . "\e$,4!5"Y\e(B")
 308
 309     ("\e$,15g\e(B" . "\e$,4!6")\e(B")
 310     ("\e$,15g6-\e(B" . "\e$,4!6\e(B")
 311     ("\e$,15g6-5p\e(B" . "\e$,4!7")\e(B")
 312     ("\e$,15g6-5p6-\e(B" . "\e$,4!7\e(B")
 313
 314     ("\e$,15h\e(B" . "\e$,4!8")\e(B")
 315     ("\e$,15h6-\e(B" . "\e$,4!8\e(B")
 316     ("\e$,15h6-5p\e(B" . "\e$,4!9")\e(B")
 317     ("\e$,15h6-5p6-\e(B" . "\e$,4!9")\e(B")
 318     ("\e$,15h6-5h\e(B" . "\e$,4!:")\e(B")
 319     ("\e$,15h6-5h6-\e(B" . "\e$,4!:\e(B")
 320
 321     ("\e$,15i\e(B" . "\e$,4!8"#")\e(B")
 322     ;; LABIALS
 323     ("\e$,15j\e(B" . "\e$,4!;")\e(B")
 324     ("\e$,15j6-\e(B" . "\e$,4!;\e(B")
 325     ("\e$,15j6-5p\e(B" . "\e$,4!<")\e(B")
 326     ("\e$,15j6-5p6-\e(B" . "\e$,4!<\e(B")
 327
 328     ("\e$,15k\e(B" . "\e$,4!a"[\e(B")
 329     ("\e$,15k6-\e(B" . "\e$,4!=\e(B")
 330     ("\e$,15k6-5p\e(B" . "\e$,4!c"[\e(B")
 331
 332     ("\e$,15l\e(B" . "\e$,4!d")\e(B")
 333     ("\e$,15l6-\e(B" . "\e$,4!d\e(B")
 334     ("\e$,15l6-5p\e(B" . "\e$,4!e")\e(B")
 335     ("\e$,15l6-5p6-\e(B" . "\e$,4!e\e(B")
 336
 337     ("\e$,15m\e(B" . "\e$,4!f")\e(B")
 338     ("\e$,15m6-\e(B" . "\e$,4!f\e(B")
 339     ("\e$,15m6-5p\e(B" . "\e$,4!g")\e(B")
 340     ("\e$,15m6-5p6-\e(B" . "\e$,4!g\e(B")
 341
 342     ("\e$,15n\e(B" . "\e$,4!h")\e(B")
 343     ("\e$,15n6-\e(B" . "\e$,4!h\e(B")
 344     ("\e$,15n6-5p\e(B" . "\e$,4!i")\e(B")
 345     ("\e$,15n6-5p6-\e(B" . "\e$,4!i\e(B")
 346     ;; SEMIVOWELS
 347     ("\e$,15o\e(B" . "\e$,4!j")\e(B")
 348     ("\e$,15o6-\e(B" . "\e$,4!j\e(B")
 349     ("\e$,15o6-5p\e(B" . "\e$,4!k")\e(B")
 350     ("\e$,15o6-5p6-\e(B" . "\e$,4!k\e(B")
 351     ("\e$,16-5o\e(B" . "\e$,4!l\e(B") ;; when every ohter lig. fails.
 352
 353     ("\e$,15p\e(B" . "\e$,4!n"W\e(B")
 354     ;; ("\e$,15p6-\e(B" . "\\e$,4"'\e(B") ;; special case.  only the topmost pos.
 355     ("\e$,15q\e(B" . "\e$,4!n"#"W\e(B")
 356     ("\e$,15q6-\e(B" . "\e$,4!m\e(B") ;; IS 13194 speical rule.
 357     ("\e$,15p6!\e(B" . "\e$,4!o"[\e(B")
 358     ("\e$,15p6"\e(B" . "\e$,4!p"\\e(B")
 359
 360     ("\e$,15r\e(B" . "\e$,4!q")\e(B")
 361     ("\e$,15r6-\e(B" . "\e$,4!q\e(B")
 362     ("\e$,15s\e(B" . "\e$,4!s\e(B")
 363     ("\e$,15s6-\e(B" . "\e$,4!r\e(B")
 364     ("\e$,15t\e(B" . "\e$,4!s"#\e(B")
 365     ("\e$,15t6-\e(B" . "\e$,4!r"#\e(B")
 366
 367     ("\e$,15u\e(B" . "\e$,4!t")\e(B")
 368     ("\e$,15u6-\e(B" . "\e$,4!t\e(B")
 369     ("\e$,15u6-5p\e(B" . "\e$,4!u")\e(B")
 370     ("\e$,15u6-5p6-\e(B" . "\e$,4!u\e(B")
 371     ;; SIBILANTS
 372     ("\e$,15v\e(B" . "\e$,4!v")\e(B")
 373     ("\e$,15v6-\e(B" . "\e$,4!v\e(B")
 374     ("\e$,15v6-5u\e(B" . "\e$,4!w")\e(B")
 375     ("\e$,15v6-5u6-\e(B" . "\e$,4!w\e(B")
 376     ("\e$,15v6-5p\e(B" . "\e$,4!x")\e(B")
 377     ("\e$,15v6-5p6-\e(B" . "\e$,4!x\e(B")
 378
 379     ("\e$,15w\e(B" . "\e$,4!y")\e(B")
 380     ("\e$,15w6-\e(B" . "\e$,4!y\e(B")
 381     ("\e$,15x\e(B" . "\e$,4!z")\e(B")
 382     ("\e$,15x6-\e(B" . "\e$,4!z\e(B")
 383     ("\e$,15x6-5p\e(B" . "\e$,4!{")\e(B")
 384     ("\e$,15x6-5p6-\e(B" . "\e$,4!{\e(B")
 385
 386     ("\e$,15y\e(B" . "\e$,4!}\e(B")
 387     ("\e$,15y6-\e(B" . "\e$,4!|\e(B")
 388     ("\e$,15y6#\e(B" . "\e$,4!~\e(B")
 389     ("\e$,15y6-5p\e(B" . "\e$,4!\7f\e(B")
 390     ("\e$,15y6-5n\e(B" . "\e$,4" \e(B")
 391     ("\e$,15y6-5o\e(B" . "\e$,4"!\e(B")
 392     ;; NUKTAS
 393     ("\e$,168\e(B" . "\e$,4 f"R"S\e(B")
 394     ("\e$,1686-\e(B" . "\e$,4 d\e(B")
 395     ("\e$,169\e(B" . "\e$,4 k")\e(B")
 396     ("\e$,1696-\e(B" . "\e$,4 k\e(B")
 397     ("\e$,16:\e(B" . "\e$,4 n")\e(B")
 398     ("\e$,16:6-\e(B" . "\e$,4 n\e(B")
 399     ("\e$,16;\e(B" . "\e$,4 w")\e(B")
 400     ("\e$,16;6-\e(B" . "\e$,4 w\e(B")
 401     ("\e$,16<\e(B" . "\e$,4!#"W\e(B")
 402     ("\e$,16=\e(B" . "\e$,4!'"X\e(B")
 403     ("\e$,16>\e(B" . "\e$,4!b"[\e(B")
 404     ("\e$,16>6-\e(B" . "\e$,4!>\e(B")
 405     ("\e$,16?\e(B" . "\e$,4!j"#")\e(B")
 406     ;; misc modifiers.
 407     ("\e$,15A\e(B" . "\\e$,4"$\e(B")
 408     ("\e$,15B\e(B" . "\\e$,4"&\e(B")
 409     ("\e$,15C\e(B" . "\e$,4 F\e(B")
 410     ("\e$,15|\e(B" . "\e$,4"#\e(B")
 411     ("\e$,15}\e(B" . "\e$,4 E\e(B")
 412     ("\e$,16-\e(B" . "\e$,4""\e(B")
 413     ("\e$,16-5p\e(B" . "\e$,4"%\e(B") ;; following "r"
 414     ;; ("\e$,160\e(B" . "\e$,4 D\e(B")
 415     ("\e$,16D\e(B" . "\e$,4 J\e(B")
 416     ;; ("\e$,16F\e(B" . "")
 417     ;; ("\e$,16G\e(B" . "")
 418     ;; ("\e$,16H\e(B" . "")
 419     ;; ("\e$,16I\e(B" . "")
 420     ;; ("\e$,16J\e(B" . "")
 421     ;; ("\e$,16K\e(B" . "")
 422     ;; ("\e$,16L\e(B" . "")
 423     ;; ("\e$,16M\e(B" . "")
 424     ;; ("\e$,16N\e(B" . "")
 425     ;; ("\e$,16O\e(B" . "")
 426     )
 427   "Devanagari characters to glyphs conversion table.
 428 Default value contains only the basic rules.  You may add your own
 429 preferred rule from the sanskrit fonts."  )
 430
 431 (defvar dev-char-glyph-hash
 432   (let* ((hash (make-hash-table :test 'equal)))
 433     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 434           dev-char-glyph)
 435     hash))
 436
 437 (defvar dev-char-glyph-regexp
 438   (devanagari-regexp-of-hashtbl-keys dev-char-glyph-hash))
 439
 440 ;; glyph-to-glyph conversion table.
 441 ;; it is supposed that glyphs are ordered in
 442 ;;   [consonant/nukta] - [matra/halant] - [preceding-r] - [anuswar].
 443
 444 (defvar dev-glyph-glyph
 445   '(("\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"(\e(B")
 446     ("\\e$,4"'\e(B\\e$,4"$\e(B" . "\\e$,4"(\e(B")
 447     ("\e$,4"*\e(B\\e$,4"&\e(B" . "\e$,4"+\e(B")
 448     ("\e$,4"*\e(B\\e$,4"'\e(B" . "\e$,4",\e(B")
 449     ("\e$,4"*\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\e$,4"-\e(B")
 450     ("\e$,4"2\e(B\\e$,4"&\e(B" . "\e$,4"3\e(B")
 451     ("\e$,4"2\e(B\\e$,4"'\e(B" . "\e$,4"4\e(B")
 452     ("\e$,4"2\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\e$,4"5\e(B")
 453     ("\e$,4"#\e(B\\e$,4"6\e(B" . "\e$,4"7\e(B")
 454     ("\e$,4"%\e(B\\e$,4"6\e(B" . "\e$,4"8\e(B")
 455     ;;("\e$,4"6\e(B" . "\e$,4"9\e(B")
 456     ("\e$,4"#\e(B\\e$,4":\e(B" . "\e$,4";\e(B")
 457     ("\e$,4"%\e(B\\e$,4":\e(B" . "\e$,4"<\e(B")
 458     ;;("\e$,4":\e(B" . "\e$,4"=\e(B")
 459     ("\\e$,4"@\e(B\\e$,4"&\e(B" . "\\e$,4"A\e(B")
 460     ("\\e$,4"@\e(B\\e$,4"'\e(B" . "\\e$,4"B\e(B")
 461     ("\\e$,4"@\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"C\e(B")
 462     ("\\e$,4"D\e(B\\e$,4"&\e(B" . "\\e$,4"E\e(B")
 463     ("\\e$,4"D\e(B\\e$,4"'\e(B" . "\\e$,4"F\e(B")
 464     ("\\e$,4"D\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"G\e(B")
 465     ("\\e$,4"H\e(B\\e$,4"&\e(B" . "\\e$,4"I\e(B")
 466     ("\\e$,4"H\e(B\\e$,4"'\e(B" . "\\e$,4"J\e(B")
 467     ("\\e$,4"H\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"K\e(B")
 468     ("\\e$,4"L\e(B\\e$,4"&\e(B" . "\\e$,4"M\e(B")
 469     ("\\e$,4"L\e(B\\e$,4"'\e(B" . "\\e$,4"N\e(B")
 470     ("\\e$,4"L\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"O\e(B")
 471     ))
 472 (defvar dev-glyph-glyph-hash
 473   (let* ((hash (make-hash-table :test 'equal)))
 474     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 475           dev-glyph-glyph)
 476     hash))
 477 (defvar dev-glyph-glyph-regexp
 478   (devanagari-regexp-of-hashtbl-keys dev-glyph-glyph-hash))
 479
 480
 481 ;; yet another glyph-to-glyph conversions.
 482 (defvar dev-glyph-glyph-2
 483   '(("\e$,4"*\e(B" . "\e$,4".\e(B")
 484     ("\e$,4"+\e(B" . "\e$,4"/\e(B")
 485     ("\e$,4",\e(B" . "\e$,4"0\e(B")
 486     ("\e$,4"-\e(B" . "\e$,4"1\e(B")))
 487 (defvar dev-glyph-glyph-2-hash
 488   (let* ((hash (make-hash-table :test 'equal)))
 489     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 490           dev-glyph-glyph-2)
 491     hash))
 492 (defvar dev-glyph-glyph-2-regexp
 493   (devanagari-regexp-of-hashtbl-keys dev-glyph-glyph-2-hash))
 494
 495 (defun dev-charseq (from &optional to)
 496   (if (null to) (setq to from))
 497   (number-sequence (decode-char 'devanagari-cdac from)
 498                    (decode-char 'devanagari-cdac to)))
 499
 500 (defvar dev-glyph-cvn
 501   (append
 502    (dev-charseq #x2b)
 503    (dev-charseq #x3c #xc1)
 504    (dev-charseq #xc3))
 505   "Devanagari Consonants/Vowels/Nukta Glyphs")
 506
 507 (defvar dev-glyph-space
 508   (dev-charseq #xf0 #xfe)
 509   "Devanagari Spacing Glyphs")
 510
 511 (defvar dev-glyph-right-modifier
 512   (append
 513    (dev-charseq #xc9)
 514    (dev-charseq #xd2 #xd5))
 515   "Devanagari Modifiers attached at the right side.")
 516
 517 (defvar dev-glyph-right-modifier-regexp
 518   (concat "[" dev-glyph-right-modifier "]"))
 519
 520 (defvar dev-glyph-left-matra
 521   (dev-charseq #xca #xd1)
 522   "Devanagari Matras attached at the left side.")
 523
 524 (defvar dev-glyph-top-matra
 525   (dev-charseq #xe0 #xef)
 526   "Devanagari Matras attached at the top side.")
 527
 528 (defvar dev-glyph-bottom-modifier
 529   (append
 530    (dev-charseq #xd6 #xdf)
 531    (dev-charseq #xc2))
 532   "Devanagari Modifiers attached at the bottom.")
 533
 534 (defvar dev-glyph-order
 535   `((,dev-glyph-cvn . 1)
 536     (,dev-glyph-space . 2)
 537     (,dev-glyph-right-modifier . 3)
 538     (,dev-glyph-left-matra . 3) ;; processed by reference point.
 539     (,dev-glyph-top-matra . 4)
 540     (,(dev-charseq #xc7 #xc8) . 5)
 541     (,(dev-charseq #xc4) . 6)
 542     (,(dev-charseq #xc6) . 6)
 543     (,(dev-charseq #xc5) . 7)
 544     (,dev-glyph-bottom-modifier . 8)))
 545
 546 (mapc
 547  (function (lambda (x)
 548    (mapc
 549      (function (lambda (y)
 550        (put-char-code-property y 'composition-order (cdr x))))
 551      (car x))))
 552   dev-glyph-order)
 553
 554 (mapc
 555   (function (lambda (x)
 556     (put-char-code-property x 'reference-point '(3 . 5))))
 557  dev-glyph-left-matra)
 558
 559 (defun devanagari-compose-syllable-string (string)
 560   (with-temp-buffer
 561     (insert (decompose-string string))
 562     (devanagari-compose-syllable-region (point-min) (point-max))
 563     (buffer-string)))
 564
 565 (defun devanagari-compose-syllable-region (from to)
 566   "Compose devanagari syllable in region FROM to TO."
 567   (let ((glyph-str nil) (cons-num 0) glyph-str-list
 568         (last-halant nil) (preceding-r nil) (last-modifier nil)
 569         (last-char (char-before to)) match-str
 570         glyph-block split-pos)
 571     (save-excursion
 572       (save-restriction
 573           ;;; *** char-to-glyph conversion ***
 574         ;; Special rule 1. -- Last halant must be preserved.
 575         (if (eq last-char ?\e$,16-\e(B)
 576             (progn
 577               (setq last-halant t)
 578               (narrow-to-region from (1- to)))
 579           (narrow-to-region from to)
 580           ;; note if the last char is modifier.
 581           (if (or (eq last-char ?\e$,15A\e(B) (eq last-char ?\e$,15B\e(B))
 582               (setq last-modifier t)))
 583         (goto-char (point-min))
 584         ;; Special rule 2. -- preceding "r halant" must be modifier.
 585         (when (looking-at "\e$,15p6-\e(B.")
 586           (setq preceding-r t)
 587           (goto-char (+ 2 (point))))
 588         ;; translate the rest characters into glyphs
 589         (while (not (eobp))
 590           (if (looking-at dev-char-glyph-regexp)
 591               (let ((end (match-end 0)))
 592                 (setq match-str (match-string 0)
 593                       glyph-str
 594                       (concat glyph-str
 595                               (gethash match-str dev-char-glyph-hash)))
 596                 ;; count the number of consonant-glyhs.
 597                 (if (string-match devanagari-consonant match-str)
 598                     (setq cons-num (1+ cons-num)))
 599                 (goto-char end))
 600             (setq glyph-str (concat glyph-str (string (following-char))))
 601             (forward-char 1)))
 602         ;; preceding-r must be attached before the anuswar if exists.
 603         (if preceding-r
 604             (if last-modifier
 605                 (setq glyph-str (concat (substring glyph-str 0 -1)
 606                                         "\e$,4"'\e(B" (substring glyph-str -1)))
 607               (setq glyph-str (concat glyph-str "\e$,4"'\e(B"))))
 608         (if last-halant (setq glyph-str (concat glyph-str "\e$,4""\e(B")))
 609           ;;; *** glyph-to-glyph conversion ***
 610         (when (string-match dev-glyph-glyph-regexp glyph-str)
 611           (setq glyph-str
 612                 (replace-match (gethash (match-string 0 glyph-str)
 613                                         dev-glyph-glyph-hash)
 614                                nil t glyph-str))
 615           (if (and (> cons-num 1)
 616                    (string-match dev-glyph-glyph-2-regexp glyph-str))
 617               (setq glyph-str
 618                     (replace-match (gethash (match-string 0 glyph-str)
 619                                             dev-glyph-glyph-2-hash)
 620                                    nil t glyph-str))))
 621           ;;; *** glyph reordering ***
 622         (while (setq split-pos (string-match "\e$,4""\e(B\\|.$" glyph-str))
 623           (setq glyph-block (substring glyph-str 0 (1+ split-pos)))
 624           (setq glyph-str (substring glyph-str (1+ split-pos)))
 625           (setq
 626            glyph-block
 627            (if (string-match dev-glyph-right-modifier-regexp glyph-block)
 628                (sort (string-to-list glyph-block)
 629                      (function (lambda (x y)
 630                         (< (get-char-code-property x 'composition-order)
 631                            (get-char-code-property y 'composition-order)))))
 632              (sort (string-to-list glyph-block)
 633                    (function (lambda (x y)
 634                       (let ((xo (get-char-code-property x 'composition-order))
 635                             (yo (get-char-code-property y 'composition-order)))
 636                         (if (= xo 2) nil (if (= yo 2) t (< xo yo)))))))))
 637           (setq glyph-str-list (nconc glyph-str-list glyph-block)))
 638           ;; concatenate and attach reference-points.
 639         (setq glyph-str
 640               (cdr
 641                (apply
 642                 'nconc
 643                 (mapcar
 644                  (function (lambda (x)
 645                    (list
 646                     (or (get-char-code-property x 'reference-point)
 647                     '(5 . 3) ;; default reference point.
 648                      )
 649                     x)))
 650                  glyph-str-list))))))
 651       (compose-region from to glyph-str)))
 652
 653 (provide 'devan-util)
 654
 655 ;;; devan-util.el ends here