lisp/language/devan-util.el

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