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) 1997, 1998, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
   4 ;;   Free Software Foundation, Inc.
   5 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   6 ;;   2007, 2008
   7 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   8 ;;   Registration Number H14PRO021
   9
  10 ;; Maintainer:  KAWABATA, Taichi <kawabata@m17n.org>
  11 ;; Keywords: multilingual, Devanagari
  12
  13 ;; This file is part of GNU Emacs.
  14
  15 ;; GNU Emacs is free software; you can redistribute it and/or modify
  16 ;; it under the terms of the GNU General Public License as published by
  17 ;; the Free Software Foundation; either version 3, or (at your option)
  18 ;; any later version.
  19
  20 ;; GNU Emacs is distributed in the hope that it will be useful,
  21 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  22 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  23 ;; GNU General Public License for more details.
  24
  25 ;; You should have received a copy of the GNU General Public License
  26 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  27 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  28 ;; Boston, MA 02110-1301, USA.
  29
  30 ;; Created: Feb. 17. 2001
  31
  32 ;;; Commentary:
  33
  34 ;; This file provides character(Unicode) to glyph(CDAC) conversion and
  35 ;; composition of Devanagari script characters.
  36
  37 ;;; Code:
  38
  39 ;; Devanagari Composable Pattern
  40 ;;    C .. Consonants
  41 ;;    V .. Vowel
  42 ;;    H .. Halant
  43 ;;    M .. Matra
  44 ;;    V .. Vowel
  45 ;;    A .. Anuswar
  46 ;;    D .. Chandrabindu
  47 ;;    (N .. Zerowidth Non Joiner)
  48 ;;    (J .. Zerowidth Joiner.  )
  49 ;; 1. vowel
  50 ;;  V(A/D)?
  51 ;; 2. syllable : maximum of 5 consecutive consonants.  (e.g. kartsnya)
  52 ;;  ((CH)?(CH)?(CH)?CH)?C(H|M?(A|D)?)?
  53
  54 (defconst devanagari-consonant
  55   "[\e$,15U\e(B-\e$,15y68\e(B-\e$,16?\e(B]")
  56
  57     ;;("\e$,16B\e(B" . nil)
  58     ;;("\e$,16A\e(B" . nil)
  59     ;;("\e$,16C\e(B" . nil)
  60
  61
  62 (defconst devanagari-composable-pattern
  63   (concat
  64    "\\([\e$,15E\e(B-\e$,15T6@6A\e(B][\e$,15A5B\e(B]?\\)\\|[\e$,15C6D\e(B]"
  65    "\\|\\("
  66    "\\(?:\\(?:[\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\\)?"
  67    "[\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]?\\)?"
  68    "\\)")
  69   "Regexp matching a composable sequence of Devanagari characters.")
  70
  71 (dolist (range '((#x0903 . #x0903)
  72                  (#x0905 . #x0939)
  73                  (#x0958 . #x0961)))
  74   (set-char-table-range indian-composable-pattern range
  75                         devanagari-composable-pattern))
  76
  77 ;;;###autoload
  78 (defun devanagari-compose-region (from to)
  79   (interactive "r")
  80   (save-excursion
  81     (save-restriction
  82       (narrow-to-region from to)
  83       (goto-char (point-min))
  84       (while (re-search-forward devanagari-composable-pattern nil t)
  85         (devanagari-compose-syllable-region (match-beginning 0)
  86                                             (match-end 0))))))
  87 (defun devanagari-compose-string (string)
  88   (with-temp-buffer
  89     (insert (decompose-string string))
  90     (devanagari-compose-region (point-min) (point-max))
  91     (buffer-string)))
  92
  93 ;;;###autoload
  94 (defun devanagari-post-read-conversion (len)
  95   (save-excursion
  96     (save-restriction
  97       (let ((buffer-modified-p (buffer-modified-p)))
  98         (narrow-to-region (point) (+ (point) len))
  99         (devanagari-compose-region (point-min) (point-max))
 100         (set-buffer-modified-p buffer-modified-p)
 101         (- (point-max) (point-min))))))
 102
 103 (defun devanagari-regexp-of-hashtbl-keys (hashtbl)
 104   "Return a regular expression that matches all keys in hashtable HASHTBL."
 105   (let ((max-specpdl-size 1000))
 106     (regexp-opt
 107      (sort
 108       (let (dummy)
 109         (maphash (function (lambda (key val) (setq dummy (cons key dummy)))) hashtbl)
 110         dummy)
 111       (function (lambda (x y) (> (length x) (length y))))))))
 112
 113 ;;;###autoload
 114 (defun devanagari-composition-function (pos &optional string)
 115   "Compose Devanagari characters after the position POS.
 116 If STRING is not nil, it is a string, and POS is an index to the string.
 117 In this case, compose characters after POS of the string."
 118   (if string
 119       (if auto-compose-current-font
 120           (if (eq (string-match "[\e$,15@\e(B-\e$,16_\e(B]+" pos) pos)
 121               (or (font-shape-text 0 (match-end 0) auto-compose-current-font
 122                                    string)
 123                   pos)))
 124     (goto-char pos)
 125     (if auto-compose-current-font
 126         (if (looking-at "[\e$,15@\e(B-\e$,16_\e(B]+")
 127             (or (font-shape-text pos (match-end 0) auto-compose-current-font)
 128                 pos)
 129           (if (looking-at devanagari-composable-pattern)
 130               (prog1 (match-end 0)
 131                 (devanagari-compose-syllable-region pos (match-end 0))))))))
 132
 133 ;; Notes on conversion steps.
 134
 135 ;; 1. chars to glyphs
 136 ;;
 137 ;; Rules will not be applied to the halant appeared at the end of the
 138 ;; text.  Also, the preceding/following "r" will be treated as special case.
 139
 140 ;; 2. glyphs reordering.
 141 ;;
 142 ;; The glyphs are split by halant, and each glyph groups are
 143 ;; re-ordered in the following order.
 144 ;;
 145 ;; Note that `consonant-glyph' mentioned here does not contain the
 146 ;; vertical bar (right modifier) attached at the right of the
 147 ;; consonant.
 148 ;;
 149 ;; If the glyph-group contains right modifier,
 150 ;;  (1) consonant-glyphs/vowels, with nukta sign
 151 ;;  (2) spacing
 152 ;;  (3) right modifier (may be matra)
 153 ;;  (4) top matra
 154 ;;  (5) preceding "r"
 155 ;;  (6) anuswar
 156 ;;  (7) following "r"
 157 ;;  (8) bottom matra or halant.
 158 ;;
 159 ;; Otherwise,
 160 ;;  (1) consonant-glyph/vowels, with nukta sign
 161 ;;  (3) left matra
 162 ;;  (4) top matra
 163 ;;  (5) preceding "r"
 164 ;;  (6) anuswar
 165 ;;  (7) following "r"
 166 ;;  (8) bottom matra or halant.
 167 ;;  (2) spacing
 168
 169 ;; 3. glyph to glyph
 170 ;;
 171 ;; For better display, some glyph display would be tuned.
 172
 173 ;; 4. Composition.
 174 ;;
 175 ;; left modifiers will be attached at the left.
 176 ;; others will be attached right.
 177
 178 ;; Problem::
 179 ;;  Can we generalize this methods to other Indian scripts?
 180
 181 (defvar dev-char-glyph
 182   '(("\e$,15E\e(B" . "\e$,4 K\e(B")
 183     ("\e$,15F\e(B" . "\e$,4 K")\e(B")
 184     ("\e$,15~\e(B" . "\e$,4")\e(B")
 185     ("\e$,15G\e(B" . "\e$,4 \\e(B")
 186     ("\e$,15\7f\e(B" . "\e$,4"*\e(B")
 187     ("\e$,15\7f5A\e(B" . "\e$,4"*\e(B\\e$,4"&\e(B")
 188     ("\e$,15H\e(B" . "\e$,4 \"'\e(B")
 189     ("\e$,15H5A\e(B" . "\e$,4 \"'"&\e(B")
 190     ("\e$,16 \e(B" . "\e$,4"2\e(B")
 191     ("\e$,16 5A\e(B" . "\e$,4"2"&\e(B")
 192     ("\e$,15I\e(B" . "\e$,4 ]\e(B")
 193     ("\e$,16!\e(B" . "\e$,4"6\e(B")
 194     ("\e$,15J\e(B" . "\e$,4 ^"P\e(B")
 195     ("\e$,16"\e(B" . "\e$,4":\e(B")
 196     ("\e$,15K\e(B" . "\e$,4 `"Q\e(B")
 197     ("\e$,16#\e(B" . "\e$,4">\e(B")
 198     ;;("\e$,15L\e(B" . nil) ; not implemented.
 199     ("\e$,16$\e(B" . "\e$,4"?\e(B")
 200     ("\e$,15M\e(B" . "\e$,4 b"L\e(B")
 201     ("\e$,15M5A\e(B" . "\e$,4 b"$\e(B")
 202     ("\e$,15M5B\e(B" . "\e$,4 b"$\e(B")
 203     ("\e$,16%\e(B" . "\\e$,4"L\e(B")
 204     ("\e$,15N\e(B" . "\e$,4 b"@\e(B")
 205     ("\e$,15N5A\e(B" . "\e$,4 b"@"&\e(B")
 206     ("\e$,16&\e(B" . "\\e$,4"@\e(B")
 207     ("\e$,16&5A\e(B" . "\\e$,4"@\e(B\\e$,4"&\e(B")
 208     ("\e$,15O\e(B" . "\e$,4 b\e(B")
 209     ("\e$,16'\e(B" . "\\e$,4"D\e(B")
 210     ("\e$,16'5A\e(B" . "\\e$,4"D\e(B\\e$,4"&\e(B")
 211     ("\e$,15P\e(B" . "\e$,4 b"D\e(B")
 212     ("\e$,15P5A\e(B" . "\e$,4 b"D"&\e(B")
 213     ("\e$,16(\e(B" . "\\e$,4"H\e(B")
 214     ("\e$,16(5A\e(B" . "\\e$,4"H\e(B\\e$,4"&\e(B")
 215     ("\e$,15Q\e(B" . "\e$,4 K")"L\e(B") ;; special rule for reodering.
 216     ("\e$,15Q5A\e(B" . "\e$,4 K")"$\e(B")
 217     ("\e$,15Q5B\e(B" . "\e$,4 K")"$\e(B")
 218     ("\e$,16)\e(B" . "\\e$,4")"L\e(B")
 219     ("\e$,16)5A\e(B" . "\\e$,4")"$\e(B")
 220     ("\e$,16)5B\e(B" . "\\e$,4")"$\e(B")
 221     ("\e$,15R\e(B" . "\e$,4 K")"@\e(B")
 222     ("\e$,15R5A\e(B" . "\e$,4 K")"@"&\e(B")
 223     ("\e$,16*\e(B" . "\\e$,4")"@\e(B")
 224     ("\e$,16*5A\e(B" . "\\e$,4")"@"&\e(B")
 225     ("\e$,15S\e(B" . "\e$,4 K")"D\e(B")
 226     ("\e$,15S5A\e(B" . "\e$,4 K")"D"&\e(B")
 227     ("\e$,16+\e(B" . "\\e$,4")"D\e(B")
 228     ("\e$,16+5A\e(B" . "\\e$,4")"D"&\e(B")
 229     ("\e$,15T\e(B" . "\e$,4 K")"H\e(B")
 230     ("\e$,15T5A\e(B" . "\e$,4 K")"H"&\e(B")
 231     ("\e$,16,\e(B" . "\\e$,4")"H\e(B")
 232     ("\e$,16,5A\e(B" . "\\e$,4")"H"&\e(B")
 233     ("\e$,16@\e(B" . "\e$,4 a"Q\e(B")
 234     ;;("\e$,16B\e(B" . nil)
 235     ;;("\e$,16A\e(B" . nil)
 236     ;;("\e$,16C\e(B" . nil)
 237
 238     ;; GRUTTALS
 239     ("\e$,15U\e(B" . "\e$,4 e"R\e(B")
 240     ("\e$,15U6-\e(B" . "\e$,4 c\e(B")
 241     ("\e$,15U6-5p\e(B" . "\e$,4 g"R\e(B")
 242     ("\e$,15U6-5d\e(B" . "\e$,4 h"R\e(B")
 243     ("\e$,15U6-5w\e(B" . "\e$,4 i")\e(B")
 244     ("\e$,15U6-5w6-\e(B" . "\e$,4 i\e(B")
 245
 246     ("\e$,15V\e(B" . "\e$,4 j")\e(B")
 247     ("\e$,15V6-\e(B" . "\e$,4 j\e(B")
 248     ("\e$,15V6-5p\e(B" . "\e$,4 l")\e(B")
 249     ("\e$,15V6-5p6-\e(B" . "\e$,4 l\e(B")
 250
 251     ("\e$,15W\e(B" . "\e$,4 m")\e(B")
 252     ("\e$,15W6-\e(B" . "\e$,4 m\e(B")
 253     ("\e$,15W6-5p\e(B" . "\e$,4 o")\e(B")
 254     ("\e$,15W6-5p6-\e(B" . "\e$,4 o\e(B")
 255
 256     ("\e$,15X\e(B" . "\e$,4 p")\e(B")
 257     ("\e$,15X6-\e(B" . "\e$,4 p\e(B")
 258     ("\e$,15X6-5p\e(B" . "\e$,4 q")\e(B")
 259     ("\e$,15X6-5p6-\e(B" . "\e$,4 q\e(B")
 260
 261     ("\e$,15Y\e(B" . "\e$,4 r"S\e(B")
 262     ;; PALATALS
 263     ("\e$,15Z\e(B" . "\e$,4 s")\e(B")
 264     ("\e$,15Z6-\e(B" . "\e$,4 s\e(B")
 265     ("\e$,15Z6-5p\e(B" . "\e$,4 t")\e(B")
 266     ("\e$,15Z6-5p6-\e(B" . "\e$,4 t\e(B")
 267
 268     ("\e$,15[\e(B" . "\e$,4 u"T\e(B")
 269
 270     ("\e$,15\\e(B" . "\e$,4 v")\e(B")
 271     ("\e$,15\6-\e(B" . "\e$,4 v\e(B")
 272     ("\e$,15\6-5p\e(B" . "\e$,4 x")\e(B")
 273     ("\e$,15\6-5p6-\e(B" . "\e$,4 x\e(B")
 274     ("\e$,15\6-5^\e(B" . "\e$,4 y")\e(B")
 275     ("\e$,15\6-5^6-\e(B" . "\e$,4 y\e(B")
 276
 277     ("\e$,15]\e(B" . "\e$,4 z")\e(B")
 278     ("\e$,15]6-\e(B" . "\e$,4 z\e(B")
 279     ("\e$,15]6-5p\e(B" . "\e$,4 {")\e(B")
 280     ("\e$,15]6-5p6-\e(B" . "\e$,4 {\e(B")
 281
 282     ("\e$,15^\e(B" . "\e$,4 |")\e(B")
 283     ("\e$,15^6-\e(B" . "\e$,4 |\e(B")
 284     ;; CEREBRALS
 285     ("\e$,15_\e(B" . "\e$,4 }"U\e(B")
 286     ("\e$,15_6-5_\e(B" . "\e$,4 ~"U\e(B")
 287     ("\e$,15_6-5`\e(B" . "\e$,4 \7f"U\e(B")
 288
 289     ("\e$,15`\e(B" . "\e$,4! "V\e(B")
 290     ("\e$,15`6-5`\e(B" . "\e$,4!!"V\e(B")
 291
 292     ("\e$,15a\e(B" . "\e$,4!""W\e(B")
 293     ("\e$,15a6-5a\e(B" . "\e$,4!$"W\e(B")
 294     ("\e$,15a6-5b\e(B" . "\e$,4!%"W\e(B")
 295
 296     ("\e$,15b\e(B" . "\e$,4!&"X\e(B")
 297
 298     ("\e$,15c\e(B" . "\e$,4!(")\e(B")
 299     ("\e$,15c6-\e(B" . "\e$,4!(\e(B")
 300     ;; DENTALS
 301     ("\e$,15d\e(B" . "\e$,4!)")\e(B")
 302     ("\e$,15d6-\e(B" . "\e$,4!)\e(B")
 303     ("\e$,15d6-5p\e(B" . "\e$,4!*")\e(B")
 304     ("\e$,15d6-5p6-\e(B" . "\e$,4!*\e(B")
 305     ("\e$,15d6-5d\e(B" . "\e$,4!+")\e(B")
 306     ("\e$,15d6-5d6-\e(B" . "\e$,4!+\e(B")
 307
 308     ("\e$,15e\e(B" . "\e$,4!,")\e(B")
 309     ("\e$,15e6-\e(B" . "\e$,4!,\e(B")
 310     ("\e$,15e6-5p\e(B" . "\e$,4!-")\e(B")
 311     ("\e$,15e6-5p6-\e(B" . "\e$,4!-\e(B")
 312
 313     ("\e$,15f\e(B" . "\e$,4!."Y\e(B")
 314     ("\e$,15f6#\e(B" . "\e$,4!/"Y\e(B")
 315     ("\e$,15f6-5p\e(B" . "\e$,4!0"Y\e(B")
 316     ("\e$,15f6-5f\e(B" . "\e$,4!1"Y\e(B")
 317     ("\e$,15f6-5g\e(B" . "\e$,4!2"Y\e(B")
 318     ("\e$,15f6-5n\e(B" . "\e$,4!3\e(B")
 319     ("\e$,15f6-5o\e(B" . "\e$,4!4\e(B")
 320     ("\e$,15f6-5u\e(B" . "\e$,4!5"Y\e(B")
 321
 322     ("\e$,15g\e(B" . "\e$,4!6")\e(B")
 323     ("\e$,15g6-\e(B" . "\e$,4!6\e(B")
 324     ("\e$,15g6-5p\e(B" . "\e$,4!7")\e(B")
 325     ("\e$,15g6-5p6-\e(B" . "\e$,4!7\e(B")
 326
 327     ("\e$,15h\e(B" . "\e$,4!8")\e(B")
 328     ("\e$,15h6-\e(B" . "\e$,4!8\e(B")
 329     ("\e$,15h6-5p\e(B" . "\e$,4!9")\e(B")
 330     ("\e$,15h6-5p6-\e(B" . "\e$,4!9")\e(B")
 331     ("\e$,15h6-5h\e(B" . "\e$,4!:")\e(B")
 332     ("\e$,15h6-5h6-\e(B" . "\e$,4!:\e(B")
 333
 334     ("\e$,15i\e(B" . "\e$,4!8"#")\e(B")
 335     ;; LABIALS
 336     ("\e$,15j\e(B" . "\e$,4!;")\e(B")
 337     ("\e$,15j6-\e(B" . "\e$,4!;\e(B")
 338     ("\e$,15j6-5p\e(B" . "\e$,4!<")\e(B")
 339     ("\e$,15j6-5p6-\e(B" . "\e$,4!<\e(B")
 340
 341     ("\e$,15k\e(B" . "\e$,4!a"[\e(B")
 342     ("\e$,15k6-\e(B" . "\e$,4!=\e(B")
 343     ("\e$,15k6-5p\e(B" . "\e$,4!c"[\e(B")
 344
 345     ("\e$,15l\e(B" . "\e$,4!d")\e(B")
 346     ("\e$,15l6-\e(B" . "\e$,4!d\e(B")
 347     ("\e$,15l6-5p\e(B" . "\e$,4!e")\e(B")
 348     ("\e$,15l6-5p6-\e(B" . "\e$,4!e\e(B")
 349
 350     ("\e$,15m\e(B" . "\e$,4!f")\e(B")
 351     ("\e$,15m6-\e(B" . "\e$,4!f\e(B")
 352     ("\e$,15m6-5p\e(B" . "\e$,4!g")\e(B")
 353     ("\e$,15m6-5p6-\e(B" . "\e$,4!g\e(B")
 354
 355     ("\e$,15n\e(B" . "\e$,4!h")\e(B")
 356     ("\e$,15n6-\e(B" . "\e$,4!h\e(B")
 357     ("\e$,15n6-5p\e(B" . "\e$,4!i")\e(B")
 358     ("\e$,15n6-5p6-\e(B" . "\e$,4!i\e(B")
 359     ;; SEMIVOWELS
 360     ("\e$,15o\e(B" . "\e$,4!j")\e(B")
 361     ("\e$,15o6-\e(B" . "\e$,4!j\e(B")
 362     ("\e$,15o6-5p\e(B" . "\e$,4!k")\e(B")
 363     ("\e$,15o6-5p6-\e(B" . "\e$,4!k\e(B")
 364     ("\e$,16-5o\e(B" . "\e$,4!l\e(B") ;; when every ohter lig. fails.
 365
 366     ("\e$,15p\e(B" . "\e$,4!n"W\e(B")
 367     ;; ("\e$,15p6-\e(B" . "\\e$,4"'\e(B") ;; special case.  only the topmost pos.
 368     ("\e$,15q\e(B" . "\e$,4!n"#"W\e(B")
 369     ("\e$,15q6-\e(B" . "\e$,4!m\e(B") ;; IS 13194 speical rule.
 370     ("\e$,15p6!\e(B" . "\e$,4!o"[\e(B")
 371     ("\e$,15p6"\e(B" . "\e$,4!p"\\e(B")
 372
 373     ("\e$,15r\e(B" . "\e$,4!q")\e(B")
 374     ("\e$,15r6-\e(B" . "\e$,4!q\e(B")
 375     ("\e$,15s\e(B" . "\e$,4!s\e(B")
 376     ("\e$,15s6-\e(B" . "\e$,4!r\e(B")
 377     ("\e$,15t\e(B" . "\e$,4!s"#\e(B")
 378     ("\e$,15t6-\e(B" . "\e$,4!r"#\e(B")
 379
 380     ("\e$,15u\e(B" . "\e$,4!t")\e(B")
 381     ("\e$,15u6-\e(B" . "\e$,4!t\e(B")
 382     ("\e$,15u6-5p\e(B" . "\e$,4!u")\e(B")
 383     ("\e$,15u6-5p6-\e(B" . "\e$,4!u\e(B")
 384     ;; SIBILANTS
 385     ("\e$,15v\e(B" . "\e$,4!v")\e(B")
 386     ("\e$,15v6-\e(B" . "\e$,4!v\e(B")
 387     ("\e$,15v6-5u\e(B" . "\e$,4!w")\e(B")
 388     ("\e$,15v6-5u6-\e(B" . "\e$,4!w\e(B")
 389     ("\e$,15v6-5p\e(B" . "\e$,4!x")\e(B")
 390     ("\e$,15v6-5p6-\e(B" . "\e$,4!x\e(B")
 391
 392     ("\e$,15w\e(B" . "\e$,4!y")\e(B")
 393     ("\e$,15w6-\e(B" . "\e$,4!y\e(B")
 394     ("\e$,15x\e(B" . "\e$,4!z")\e(B")
 395     ("\e$,15x6-\e(B" . "\e$,4!z\e(B")
 396     ("\e$,15x6-5p\e(B" . "\e$,4!{")\e(B")
 397     ("\e$,15x6-5p6-\e(B" . "\e$,4!{\e(B")
 398
 399     ("\e$,15y\e(B" . "\e$,4!}\e(B")
 400     ("\e$,15y6-\e(B" . "\e$,4!|\e(B")
 401     ("\e$,15y6#\e(B" . "\e$,4!~\e(B")
 402     ("\e$,15y6-5p\e(B" . "\e$,4!\7f\e(B")
 403     ("\e$,15y6-5n\e(B" . "\e$,4" \e(B")
 404     ("\e$,15y6-5o\e(B" . "\e$,4"!\e(B")
 405     ;; NUKTAS
 406     ("\e$,168\e(B" . "\e$,4 f"R"S\e(B")
 407     ("\e$,1686-\e(B" . "\e$,4 d\e(B")
 408     ("\e$,169\e(B" . "\e$,4 k")\e(B")
 409     ("\e$,1696-\e(B" . "\e$,4 k\e(B")
 410     ("\e$,16:\e(B" . "\e$,4 n")\e(B")
 411     ("\e$,16:6-\e(B" . "\e$,4 n\e(B")
 412     ("\e$,16;\e(B" . "\e$,4 w")\e(B")
 413     ("\e$,16;6-\e(B" . "\e$,4 w\e(B")
 414     ("\e$,16<\e(B" . "\e$,4!#"W\e(B")
 415     ("\e$,16=\e(B" . "\e$,4!'"X\e(B")
 416     ("\e$,16>\e(B" . "\e$,4!b"[\e(B")
 417     ("\e$,16>6-\e(B" . "\e$,4!>\e(B")
 418     ("\e$,16?\e(B" . "\e$,4!j"#")\e(B")
 419     ;; misc modifiers.
 420     ("\e$,15A\e(B" . "\\e$,4"$\e(B")
 421     ("\e$,15B\e(B" . "\\e$,4"&\e(B")
 422     ("\e$,15C\e(B" . "\e$,4 F\e(B")
 423     ("\e$,15|\e(B" . "\e$,4"#\e(B")
 424     ("\e$,15}\e(B" . "\e$,4 E\e(B")
 425     ("\e$,16-\e(B" . "\e$,4""\e(B")
 426     ("\e$,16-5p\e(B" . "\e$,4"%\e(B") ;; following "r"
 427     ;; ("\e$,160\e(B" . "\e$,4 D\e(B")
 428     ("\e$,16D\e(B" . "\e$,4 J\e(B")
 429     ;; ("\e$,16F\e(B" . "")
 430     ;; ("\e$,16G\e(B" . "")
 431     ;; ("\e$,16H\e(B" . "")
 432     ;; ("\e$,16I\e(B" . "")
 433     ;; ("\e$,16J\e(B" . "")
 434     ;; ("\e$,16K\e(B" . "")
 435     ;; ("\e$,16L\e(B" . "")
 436     ;; ("\e$,16M\e(B" . "")
 437     ;; ("\e$,16N\e(B" . "")
 438     ;; ("\e$,16O\e(B" . "")
 439     )
 440   "Devanagari characters to glyphs conversion table.
 441 Default value contains only the basic rules.  You may add your own
 442 preferred rule from the sanskrit fonts."  )
 443
 444 (defvar dev-char-glyph-hash
 445   (let* ((hash (make-hash-table :test 'equal)))
 446     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 447           dev-char-glyph)
 448     hash))
 449
 450 (defvar dev-char-glyph-regexp
 451   (devanagari-regexp-of-hashtbl-keys dev-char-glyph-hash))
 452
 453 ;; glyph-to-glyph conversion table.
 454 ;; it is supposed that glyphs are ordered in
 455 ;;   [consonant/nukta] - [matra/halant] - [preceding-r] - [anuswar].
 456
 457 (defvar dev-glyph-glyph
 458   '(("\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"(\e(B")
 459     ("\\e$,4"'\e(B\\e$,4"$\e(B" . "\\e$,4"(\e(B")
 460     ("\e$,4"*\e(B\\e$,4"&\e(B" . "\e$,4"+\e(B")
 461     ("\e$,4"*\e(B\\e$,4"'\e(B" . "\e$,4",\e(B")
 462     ("\e$,4"*\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\e$,4"-\e(B")
 463     ("\e$,4"2\e(B\\e$,4"&\e(B" . "\e$,4"3\e(B")
 464     ("\e$,4"2\e(B\\e$,4"'\e(B" . "\e$,4"4\e(B")
 465     ("\e$,4"2\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\e$,4"5\e(B")
 466     ("\e$,4"#\e(B\\e$,4"6\e(B" . "\e$,4"7\e(B")
 467     ("\e$,4"%\e(B\\e$,4"6\e(B" . "\e$,4"8\e(B")
 468     ;;("\e$,4"6\e(B" . "\e$,4"9\e(B")
 469     ("\e$,4"#\e(B\\e$,4":\e(B" . "\e$,4";\e(B")
 470     ("\e$,4"%\e(B\\e$,4":\e(B" . "\e$,4"<\e(B")
 471     ;;("\e$,4":\e(B" . "\e$,4"=\e(B")
 472     ("\\e$,4"@\e(B\\e$,4"&\e(B" . "\\e$,4"A\e(B")
 473     ("\\e$,4"@\e(B\\e$,4"'\e(B" . "\\e$,4"B\e(B")
 474     ("\\e$,4"@\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"C\e(B")
 475     ("\\e$,4"D\e(B\\e$,4"&\e(B" . "\\e$,4"E\e(B")
 476     ("\\e$,4"D\e(B\\e$,4"'\e(B" . "\\e$,4"F\e(B")
 477     ("\\e$,4"D\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"G\e(B")
 478     ("\\e$,4"H\e(B\\e$,4"&\e(B" . "\\e$,4"I\e(B")
 479     ("\\e$,4"H\e(B\\e$,4"'\e(B" . "\\e$,4"J\e(B")
 480     ("\\e$,4"H\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"K\e(B")
 481     ("\\e$,4"L\e(B\\e$,4"&\e(B" . "\\e$,4"M\e(B")
 482     ("\\e$,4"L\e(B\\e$,4"'\e(B" . "\\e$,4"N\e(B")
 483     ("\\e$,4"L\e(B\\e$,4"'\e(B\\e$,4"&\e(B" . "\\e$,4"O\e(B")
 484     ))
 485 (defvar dev-glyph-glyph-hash
 486   (let* ((hash (make-hash-table :test 'equal)))
 487     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 488           dev-glyph-glyph)
 489     hash))
 490 (defvar dev-glyph-glyph-regexp
 491   (devanagari-regexp-of-hashtbl-keys dev-glyph-glyph-hash))
 492
 493
 494 ;; yet another glyph-to-glyph conversions.
 495 (defvar dev-glyph-glyph-2
 496   '(("\e$,4"*\e(B" . "\e$,4".\e(B")
 497     ("\e$,4"+\e(B" . "\e$,4"/\e(B")
 498     ("\e$,4",\e(B" . "\e$,4"0\e(B")
 499     ("\e$,4"-\e(B" . "\e$,4"1\e(B")))
 500 (defvar dev-glyph-glyph-2-hash
 501   (let* ((hash (make-hash-table :test 'equal)))
 502     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 503           dev-glyph-glyph-2)
 504     hash))
 505 (defvar dev-glyph-glyph-2-regexp
 506   (devanagari-regexp-of-hashtbl-keys dev-glyph-glyph-2-hash))
 507
 508 (defun dev-charseq (from &optional to)
 509   (if (null to) (setq to from))
 510   (number-sequence (decode-char 'devanagari-cdac from)
 511                    (decode-char 'devanagari-cdac to)))
 512
 513 (defvar dev-glyph-cvn
 514   (append
 515    (dev-charseq #x2b)
 516    (dev-charseq #x3c #xc1)
 517    (dev-charseq #xc3))
 518   "Devanagari Consonants/Vowels/Nukta Glyphs")
 519
 520 (defvar dev-glyph-space
 521   (dev-charseq #xf0 #xfe)
 522   "Devanagari Spacing Glyphs")
 523
 524 (defvar dev-glyph-right-modifier
 525   (append
 526    (dev-charseq #xc9)
 527    (dev-charseq #xd2 #xd5))
 528   "Devanagari Modifiers attached at the right side.")
 529
 530 (defvar dev-glyph-right-modifier-regexp
 531   (concat "[" dev-glyph-right-modifier "]"))
 532
 533 (defvar dev-glyph-left-matra
 534   (dev-charseq #xca #xd1)
 535   "Devanagari Matras attached at the left side.")
 536
 537 (defvar dev-glyph-top-matra
 538   (dev-charseq #xe0 #xef)
 539   "Devanagari Matras attached at the top side.")
 540
 541 (defvar dev-glyph-bottom-modifier
 542   (append
 543    (dev-charseq #xd6 #xdf)
 544    (dev-charseq #xc2))
 545   "Devanagari Modifiers attached at the bottom.")
 546
 547 (defvar dev-glyph-order
 548   `((,dev-glyph-cvn . 1)
 549     (,dev-glyph-space . 2)
 550     (,dev-glyph-right-modifier . 3)
 551     (,dev-glyph-left-matra . 3) ;; processed by reference point.
 552     (,dev-glyph-top-matra . 4)
 553     (,(dev-charseq #xc7 #xc8) . 5)
 554     (,(dev-charseq #xc4) . 6)
 555     (,(dev-charseq #xc6) . 6)
 556     (,(dev-charseq #xc5) . 7)
 557     (,dev-glyph-bottom-modifier . 8)))
 558
 559 (mapc
 560  (function (lambda (x)
 561    (mapc
 562      (function (lambda (y)
 563        (put-char-code-property y 'composition-order (cdr x))))
 564      (car x))))
 565   dev-glyph-order)
 566
 567 (mapc
 568   (function (lambda (x)
 569     (put-char-code-property x 'reference-point '(3 . 5))))
 570  dev-glyph-left-matra)
 571
 572 (defun devanagari-compose-syllable-string (string)
 573   (with-temp-buffer
 574     (insert (decompose-string string))
 575     (devanagari-compose-syllable-region (point-min) (point-max))
 576     (buffer-string)))
 577
 578 (defun devanagari-compose-syllable-region (from to)
 579   "Compose devanagari syllable in region FROM to TO."
 580   (let ((glyph-str nil) (cons-num 0) glyph-str-list
 581         (last-halant nil) (preceding-r nil) (last-modifier nil)
 582         (last-char (char-before to)) match-str
 583         glyph-block split-pos)
 584     (save-excursion
 585       (save-restriction
 586           ;;; *** char-to-glyph conversion ***
 587         ;; Special rule 1. -- Last halant must be preserved.
 588         (if (eq last-char ?\e$,16-\e(B)
 589             (progn
 590               (setq last-halant t)
 591               (narrow-to-region from (1- to)))
 592           (narrow-to-region from to)
 593           ;; note if the last char is modifier.
 594           (if (or (eq last-char ?\e$,15A\e(B) (eq last-char ?\e$,15B\e(B))
 595               (setq last-modifier t)))
 596         (goto-char (point-min))
 597         ;; Special rule 2. -- preceding "r halant" must be modifier.
 598         (when (looking-at "\e$,15p6-\e(B.")
 599           (setq preceding-r t)
 600           (goto-char (+ 2 (point))))
 601         ;; translate the rest characters into glyphs
 602         (while (re-search-forward dev-char-glyph-regexp nil t)
 603           (setq match-str (match-string 0))
 604           (setq glyph-str
 605                 (concat glyph-str
 606                         (gethash match-str dev-char-glyph-hash)))
 607           ;; count the number of consonant-glyhs.
 608           (if (string-match devanagari-consonant match-str)
 609               (setq cons-num (1+ cons-num))))
 610         ;; preceding-r must be attached before the anuswar if exists.
 611         (if preceding-r
 612             (if last-modifier
 613                 (setq glyph-str (concat (substring glyph-str 0 -1)
 614                                         "\e$,4"'\e(B" (substring glyph-str -1)))
 615               (setq glyph-str (concat glyph-str "\e$,4"'\e(B"))))
 616         (if last-halant (setq glyph-str (concat glyph-str "\e$,4""\e(B")))
 617           ;;; *** glyph-to-glyph conversion ***
 618         (when (string-match dev-glyph-glyph-regexp glyph-str)
 619           (setq glyph-str
 620                 (replace-match (gethash (match-string 0 glyph-str)
 621                                         dev-glyph-glyph-hash)
 622                                nil t glyph-str))
 623           (if (and (> cons-num 1)
 624                    (string-match dev-glyph-glyph-2-regexp glyph-str))
 625               (setq glyph-str
 626                     (replace-match (gethash (match-string 0 glyph-str)
 627                                             dev-glyph-glyph-2-hash)
 628                                    nil t glyph-str))))
 629           ;;; *** glyph reordering ***
 630         (while (setq split-pos (string-match "\e$,4""\e(B\\|.$" glyph-str))
 631           (setq glyph-block (substring glyph-str 0 (1+ split-pos)))
 632           (setq glyph-str (substring glyph-str (1+ split-pos)))
 633           (setq
 634            glyph-block
 635            (if (string-match dev-glyph-right-modifier-regexp glyph-block)
 636                (sort (string-to-list glyph-block)
 637                      (function (lambda (x y)
 638                         (< (get-char-code-property x 'composition-order)
 639                            (get-char-code-property y 'composition-order)))))
 640              (sort (string-to-list glyph-block)
 641                    (function (lambda (x y)
 642                       (let ((xo (get-char-code-property x 'composition-order))
 643                             (yo (get-char-code-property y 'composition-order)))
 644                         (if (= xo 2) nil (if (= yo 2) t (< xo yo)))))))))
 645           (setq glyph-str-list (nconc glyph-str-list glyph-block)))
 646           ;; concatenate and attach reference-points.
 647         (setq glyph-str
 648               (cdr
 649                (apply
 650                 'nconc
 651                 (mapcar
 652                  (function (lambda (x)
 653                    (list
 654                     (or (get-char-code-property x 'reference-point)
 655                     '(5 . 3) ;; default reference point.
 656                      )
 657                     x)))
 658                  glyph-str-list))))))
 659       (compose-region from to glyph-str)))
 660
 661 (provide 'devan-util)
 662
 663 ;;; arch-tag: 9bc4d6e3-f2b9-4110-886e-ff9b66b7eebc
 664 ;;; devan-util.el ends here