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

   1 ;;; knd-util.el --- Support for composing Kannada characters
   2
   3 ;; Copyright (C) 2003, 2004, 2005, 2006, 2007, 2008
   4 ;;   Free Software Foundation, Inc.
   5
   6 ;; Maintainer:  Maintainer:  CHOWKSEY, Kailash C. <klchxbec@m-net.arbornet.org>
   7 ;; Keywords: multilingual, Kannada
   8
   9 ;; This file is part of GNU Emacs.
  10
  11 ;; GNU Emacs is free software; you can redistribute it and/or modify
  12 ;; it under the terms of the GNU General Public License as published by
  13 ;; the Free Software Foundation; either version 3, or (at your option)
  14 ;; any later version.
  15
  16 ;; GNU Emacs is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;; GNU General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  23 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  24 ;; Boston, MA 02110-1301, USA.
  25
  26 ;; Created: Jul. 14. 2003
  27
  28 ;;; Commentary:
  29
  30 ;; This file provides character(Unicode) to glyph(CDAC) conversion and
  31 ;; composition of Kannada script characters.
  32
  33 ;;; Code:
  34
  35 ;;;###autoload
  36
  37 ;; Kannada Composable Pattern
  38 ;;    C .. Consonants
  39 ;;    V .. Vowel
  40 ;;    H .. Virama
  41 ;;    M .. Matra
  42 ;;    V .. Vowel
  43 ;;    (N .. Zerowidth Non Joiner)
  44 ;;    (J .. Zerowidth Joiner.  )
  45 ;; 1. vowel
  46 ;;  V(A)?
  47 ;; 2. syllable : maximum of 5 consecutive consonants.  (e.g. kartsnya)
  48 ;;  ((CH)?(CH)?(CH)?CH)?C(H|M?)?
  49
  50 (defconst kannada-consonant
  51   "[\e$,1>u\e(B-\e$,1?9\e(B]")
  52
  53 (defconst kannada-consonant-needs-twirl
  54   "[\e$,1>u>w\e(B-\e$,1>{>}\e(B-\e$,1>~? \e(B-\e$,1?"?$\e(B-\e$,1?+?-?0?3\e(B-\e$,1?9\e(B]\\(\e$,1?M\e(B[\e$,1>u\e(B-\e$,1?9\e(B]\\)*[\e$,1?A?B?C?D>b\e(B]?$")
  55
  56 (defconst kannada-composable-pattern
  57   (concat
  58    "\\([\e$,1>b\e(B-\e$,1>t?`>l\e(B]\\)\\|[\e$,1>c\e(B]"
  59    "\\|\\("
  60    "\\(?:\\(?:[\e$,1>u\e(B-\e$,1?9\e(B]\e$,1?M\e(B\\)?\\(?:[\e$,1>u\e(B-\e$,1?9\e(B]\e$,1?M\e(B\\)?\\(?:[\e$,1>u\e(B-\e$,1?9\e(B]\e$,1?M\e(B\\)?[\e$,1>u\e(B-\e$,1?9\e(B]\e$,1?M\e(B\\)?"
  61    "[\e$,1>u\e(B-\e$,1?9\e(B]\\(?:\e$,1?M\e(B\\|[\e$,1?>\e(B-\e$,1?M?U?C\e(B]?\\)?"
  62    "\\)")
  63   "Regexp matching a composable sequence of Kannada characters.")
  64
  65 ;;;###autoload
  66 (defun kannada-compose-region (from to)
  67   (interactive "r")
  68   (save-excursion
  69     (save-restriction
  70       (narrow-to-region from to)
  71       (goto-char (point-min))
  72       (while (re-search-forward kannada-composable-pattern nil t)
  73         (kannada-compose-syllable-region (match-beginning 0)
  74                                             (match-end 0))))))
  75 ;;;###autoload
  76 (defun kannada-compose-string (string)
  77   (with-temp-buffer
  78     (insert (decompose-string string))
  79     (kannada-compose-region (point-min) (point-max))
  80     (buffer-string)))
  81
  82 ;;;###autoload
  83 (defun kannada-post-read-conversion (len)
  84   (save-excursion
  85     (save-restriction
  86       (let ((buffer-modified-p (buffer-modified-p)))
  87         (narrow-to-region (point) (+ (point) len))
  88         (kannada-compose-region (point-min) (point-max))
  89         (set-buffer-modified-p buffer-modified-p)
  90         (- (point-max) (point-min))))))
  91
  92 (defun kannada-range (from to)
  93   "Make the list of the integers of range FROM to TO."
  94   (let (result)
  95     (while (<= from to) (setq result (cons to result) to (1- to))) result))
  96
  97 (defun kannada-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 (defun kannada-regexp-of-hashtbl-vals (hashtbl)
 108   "Return a regular expression that matches all values in hashtable HASHTBL."
 109   (let ((max-specpdl-size 1000))
 110     (regexp-opt
 111      (sort
 112       (let (dummy)
 113         (maphash (function (lambda (key val) (setq dummy (cons val dummy)))) hashtbl)
 114         dummy)
 115       (function (lambda (x y) (> (length x) (length y))))))))
 116
 117 ;;;###autoload
 118 (defun kannada-composition-function (pos &optional string)
 119   "Compose Kannada characters after the position POS.
 120 If STRING is not nil, it is a string, and POS is an index to the string.
 121 In this case, compose characters after POS of the string."
 122   (if string
 123       ;; Not yet implemented.
 124       nil
 125     (goto-char pos)
 126     (if (looking-at kannada-composable-pattern)
 127         (prog1 (match-end 0)
 128           (kannada-compose-syllable-region pos (match-end 0))))))
 129
 130 ;; Notes on conversion steps.
 131
 132 ;; 1. chars to glyphs
 133 ;;
 134 ;; Rules will not be applied to the virama appeared at the end of the
 135 ;; text.  Also, the preceding/following "r" will be treated as special case.
 136
 137 ;; 2. glyphs reordering.
 138 ;;
 139 ;; The glyphs are split by virama, and each glyph groups are
 140 ;; re-ordered in the following order.
 141 ;;
 142 ;; Note that `consonant-glyph' mentioned here does not contain the
 143 ;; vertical bar (right modifier) attached at the right of the
 144 ;; consonant.
 145 ;;
 146 ;; If the glyph-group contains right modifier,
 147 ;;  (1) consonant-glyphs/vowels
 148 ;;  (2) spacing
 149 ;;  (3) right modifier (may be matra)
 150 ;;  (4) top matra
 151 ;;  (5) preceding "r"
 152 ;;  (7) following "r"
 153 ;;  (8) bottom matra or virama.
 154 ;;
 155 ;; Otherwise,
 156 ;;  (1) consonant-glyph/vowels, with nukta sign
 157 ;;  (3) left matra
 158 ;;  (4) top matra
 159 ;;  (5) preceding "r"
 160 ;;  (7) following "r"
 161 ;;  (8) bottom matra or virama.
 162 ;;  (2) spacing
 163
 164 ;; 3. glyph to glyph
 165 ;;
 166 ;; For better display, some glyph display would be tuned.
 167
 168 ;; 4. Composition.
 169 ;;
 170 ;; left modifiers will be attached at the left.
 171 ;; others will be attached right.
 172
 173 ;; Problem::
 174 ;;  Can we generalize this methods to other Indian scripts?
 175
 176 (defvar knd-char-glyph
 177   '(("\e$,1>e\e(B" . "\e$,43@\e(B")
 178     ("\e$,1>f\e(B" . "\e$,43A\e(B")
 179     ("\e$,1?>\e(B" . "\e$,44{\e(B")
 180     ("\e$,1>g\e(B" . "\e$,43B\e(B")
 181     ("\e$,1??\e(B" . nil)
 182     ("\e$,1>h\e(B" . "\e$,43C\e(B")
 183     ("\e$,1?@\e(B" . nil)
 184     ("\e$,1>i\e(B" . "\e$,43D\e(B")
 185     ("\e$,1?A\e(B" . "\\e$,44\7f\e(B")
 186     ("\e$,1>j\e(B" . "\e$,43E\e(B")
 187     ("\e$,1?B\e(B" . "\\e$,45 \e(B")
 188     ("\e$,1>k\e(B" . "\e$,43F4\7f\e(B")
 189     ("\e$,1?C\e(B" . "\\e$,45$\e(B")
 190     ("\e$,1?`\e(B" . "\e$,43F5 \e(B")
 191     ("\e$,1?D\e(B" . "\\e$,45%\e(B")
 192     ;;("\e$,1>l\e(B" . nil) ; not implemented.
 193     ;;("\e$,1?a\e(B" . nil)
 194     ("\e$,1>n\e(B" . "\e$,43G\e(B")
 195     ("\e$,1>o\e(B" . "\e$,43H\e(B")
 196     ("\e$,1>p\e(B" . "\e$,43I\e(B")
 197     ("\e$,1?F\e(B" . "\\e$,45&\e(B")
 198     ("\e$,1?G\e(B" . "\\e$,45&4~\e(B")
 199     ("\e$,1?H\e(B" . "\\e$,45&5'\e(B")
 200     ("\e$,1>r\e(B" . "\e$,43J\e(B")
 201     ("\e$,1?J\e(B" . "\e$,45&5 \e(B")
 202     ("\e$,1>s\e(B" . "\e$,43K\e(B")
 203     ("\e$,1?K\e(B" . "\\e$,45&5 4~\e(B")
 204     ("\e$,1>t\e(B" . "\e$,43L\e(B")
 205     ("\e$,1?L\e(B" . "\\e$,45(\e(B")
 206     ("\e$,1>b\e(B" . "\e$,43M\e(B")
 207     ("\e$,1>c\e(B" . "\e$,43N\e(B")
 208     ("\e$,1>u?M\e(B" . "\e$,43O5)\e(B") ("\e$,1>u\e(B" . "\e$,43O\e(B") ("\e$,1>u??\e(B" . "\e$,43P\e(B") ("\e$,1>u?@\e(B" . "\e$,43P4~\e(B")
 209     ("\e$,1>v?M\e(B" . "\e$,43S5)\e(B") ("\e$,1>v\e(B" .  "\e$,43S\e(B") ("\e$,1>v??\e(B" . "\e$,43T\e(B") ("\e$,1>v?@\e(B" . "\e$,43T4~\e(B") ("\e$,1>v?F\e(B" . "\e$,43S5&\e(B") ("\e$,1>v?G\e(B" . "\e$,43S5&4~\e(B") ("\e$,1>v?H\e(B" . "\e$,43S5&5'\e(B") ("\e$,1>v?J\e(B" .  "\e$,43S5&5&5 \e(B") ("\e$,1>v?K\e(B" . "\e$,43S5&5&5 4~\e(B") ("\e$,1>v?L\e(B" . "\e$,43S5(\e(B")
 210     ("\e$,1>w?M\e(B" . "\e$,43V5)\e(B") ("\e$,1>w\e(B" . "\e$,43V\e(B") ("\e$,1>w??\e(B" . "\e$,43W\e(B") ("\e$,1>w?@\e(B" . "\e$,43W4~\e(B")
 211     ("\e$,1>x?M\e(B" . "\e$,43Y5)\e(B") ("\e$,1>x\e(B" . "\e$,43Y\e(B") ("\e$,1>x??\e(B" . "\e$,43Z\e(B") ("\e$,1>x?@\e(B" . "\e$,43Z4~\e(B")
 212     ("\e$,1>y?M\e(B" . "\e$,43\5)\e(B") ("\e$,1>y\e(B" . "\e$,43\\e(B")
 213     ("\e$,1>z?M\e(B" . "\e$,43^5)\e(B") ("\e$,1>z\e(B" . "\e$,43^\e(B") ("\e$,1>z??\e(B" . "\e$,43_\e(B") ("\e$,1>z?@\e(B" . "\e$,43_4~\e(B")
 214     ("\e$,1>{?M\e(B" . "\e$,43a5)\e(B") ("\e$,1>{\e(B" . "\e$,43a\e(B") ("\e$,1>{??\e(B" . "\e$,43b\e(B") ("\e$,1>{?@\e(B" . "\e$,43b4~\e(B")
 215     ("\e$,1>|?M\e(B" . "\e$,43d5)\e(B") ("\e$,1>|\e(B" . "\e$,43d\e(B") ("\e$,1>|??\e(B" . "\e$,43f\e(B") ("\e$,1>|?@\e(B" . "\e$,43f4~\e(B") ("\e$,1>|?F\e(B" . "\e$,43e5&\e(B") ("\e$,1>|?G\e(B" . "\e$,43e5&4~\e(B") ("\e$,1>|?H\e(B" . "\e$,43e5&5'\e(B") ("\e$,1>|?J\e(B" .  "\e$,43e5&5&5 \e(B") ("\e$,1>|?K\e(B" . "\e$,43e5&5&5 4~\e(B") ("\e$,1>|?L\e(B" . "\e$,43e5(\e(B")
 216     ("\e$,1>}?M\e(B" . "\e$,44a4z3h4\7f5)\e(B") ("\e$,1>}\e(B" . "\e$,44a4z3h4\7f\e(B") ("\e$,1>}??\e(B" . "\e$,44b3h4\7f\e(B") ("\e$,1>}?@\e(B" . "\e$,44b3h4\7f4~\e(B") ("\e$,1>}?B\e(B". "\e$,44a4z3h5 \e(B") ("\e$,1>}?J\e(B". "\e$,44a5&3h5 \e(B") ("\e$,1>}?K\e(B". "\e$,44a5&3h5 4~\e(B")
 217     ("\e$,1>~?M\e(B" . "\e$,43j5)\e(B") ("\e$,1>~\e(B" . "\e$,43j\e(B")
 218     ("\e$,1>\7f?M\e(B" . "\e$,43m5)\e(B") ("\e$,1>\7f\e(B" . "\e$,43l\e(B") ("\e$,1?#?>\e(B" . "\e$,43m4{\e(B") ("\e$,1>\7f??\e(B" . "\e$,43n\e(B") ("\e$,1>\7f?@\e(B" . "\e$,43n4~\e(B") ("\e$,1>\7f?F\e(B" . "\e$,43m5&\e(B") ("\e$,1>\7f?G\e(B" . "\e$,43m5&4~\e(B") ("\e$,1>\7f?H\e(B" . "\e$,43m5&5'\e(B") ("\e$,1>\7f?J\e(B" .  "\e$,43m5&5&5 \e(B") ("\e$,1>\7f?K\e(B" . "\e$,43m5&5&5 4~\e(B") ("\e$,1>\7f?L\e(B" . "\e$,43m5(\e(B")
 219     ("\e$,1? ?M\e(B" . "\e$,43p5)\e(B") ("\e$,1? \e(B" . "\e$,43p\e(B") ("\e$,1? ??\e(B" . "\e$,43q\e(B") ("\e$,1? ?@\e(B" . "\e$,43q4~\e(B")
 220     ("\e$,1?!?M\e(B" . "\e$,43s5)\e(B") ("\e$,1?!\e(B" . "\e$,43s\e(B") ("\e$,1?!??\e(B" . "\e$,43t\e(B") ("\e$,1?!?@\e(B" . "\e$,43t4~\e(B")
 221     ("\e$,1?"?M\e(B" . "\e$,43v5)\e(B") ("\e$,1?"\e(B" . "\e$,43v\e(B") ("\e$,1?"??\e(B" . "\e$,43w\e(B") ("\e$,1?"?@\e(B" . "\e$,43w4~\e(B")
 222     ("\e$,1?#?M\e(B" . "\e$,43z5)\e(B") ("\e$,1?#\e(B" . "\e$,43y\e(B") ("\e$,1?#?>\e(B" . "\e$,43z4{\e(B") ("\e$,1?#??\e(B" . "\e$,43{\e(B") ("\e$,1?#?@\e(B" . "\e$,43{4~\e(B") ("\e$,1?#?F\e(B" . "\e$,43z5&\e(B") ("\e$,1?#?G\e(B" . "\e$,43z5&4~\e(B") ("\e$,1?#?H\e(B" . "\e$,43z5&5'\e(B") ("\e$,1?#?J\e(B" .  "\e$,43z5&5&5 \e(B") ("\e$,1?#?K\e(B" . "\e$,43z5&5&5 4~\e(B") ("\e$,1?#?L\e(B" . "\e$,43z5(\e(B")
 223     ("\e$,1?$?M\e(B" . "\e$,43}5)\e(B") ("\e$,1?$\e(B" . "\e$,43}\e(B") ("\e$,1?$??\e(B" . "\e$,43~\e(B") ("\e$,1?$?@\e(B" . "\e$,43~4~\e(B")
 224     ("\e$,1?%?M\e(B" . "\e$,44B5)\e(B") ("\e$,1?%\e(B" . "\e$,44B\e(B") ("\e$,1?%??\e(B" . "\e$,44C\e(B") ("\e$,1?%?@\e(B" . "\e$,44C4~\e(B")
 225     ("\e$,1?&?M\e(B" . "\e$,44E5)\e(B") ("\e$,1?&\e(B" . "\e$,44E\e(B") ("\e$,1?&??\e(B" . "\e$,44F\e(B") ("\e$,1?&?@\e(B" . "\e$,44F4~\e(B")
 226     ("\e$,1?'?M\e(B" . "\e$,44H5)\e(B") ("\e$,1?'\e(B" . "\e$,44H\e(B") ("\e$,1?'??\e(B" . "\e$,44I\e(B") ("\e$,1?'?@\e(B" . "\e$,44I4~\e(B")
 227     ("\e$,1?(?M\e(B" . "\e$,44K5)\e(B") ("\e$,1?(\e(B" . "\e$,44K\e(B") ("\e$,1?(??\e(B" . "\e$,44L\e(B") ("\e$,1?(?@\e(B" . "\e$,44L4~\e(B")
 228     ("\e$,1?*?M\e(B" . "\e$,44N5)\e(B") ("\e$,1?*\e(B" . "\e$,44N\e(B") ("\e$,1?*??\e(B" . "\e$,44O\e(B") ("\e$,1?*?@\e(B" . "\e$,44O4~\e(B") ("\e$,1?*?A\e(B" . "\e$,44N5"\e(B") ("\e$,1?*?B\e(B" . "\e$,44N5#\e(B") ("\e$,1?*?J\e(B" . "\e$,44N5&5#\e(B") ("\e$,1?*?K\e(B" . "\e$,44N5&5#4~\e(B")
 229     ("\e$,1?+?M\e(B" . "\e$,44Q5)\e(B") ("\e$,1?+\e(B" . "\e$,44Q\e(B") ("\e$,1?+??\e(B" . "\e$,44R\e(B") ("\e$,1?+?@\e(B" . "\e$,44R4~\e(B") ("\e$,1?+?A\e(B" . "\e$,44Q5"\e(B") ("\e$,1?+?B\e(B" . "\e$,44Q5#\e(B") ("\e$,1?+?J\e(B" . "\e$,44Q5&5#\e(B") ("\e$,1?+?K\e(B" . "\e$,44Q5&5#4~\e(B")
 230     ("\e$,1?,?M\e(B" . "\e$,44W5)\e(B") ("\e$,1?,\e(B" . "\e$,44V\e(B") ("\e$,1?,?>\e(B". "\e$,44W4{\e(B") ("\e$,1?,??\e(B" . "\e$,44X\e(B") ("\e$,1?,?@\e(B" . "\e$,44X4~\e(B") ("\e$,1?,?F\e(B" . "\e$,44W5&\e(B") ("\e$,1?,?G\e(B" . "\e$,44W5&4~\e(B") ("\e$,1?,?H\e(B" . "\e$,44W5&5'\e(B") ("\e$,1?,?J\e(B" .  "\e$,44W5&5&5 \e(B") ("\e$,1?,?K\e(B" . "\e$,44W5&5&5 4~\e(B") ("\e$,1?,?L\e(B" . "\e$,44W5(\e(B")
 231     ("\e$,1?-?M\e(B" . "\e$,44Z5)\e(B") ("\e$,1?-\e(B" . "\e$,44Z\e(B") ("\e$,1?-??\e(B" . "\e$,44[\e(B") ("\e$,1?-?@\e(B" . "\e$,44[4~\e(B")
 232     ("\e$,1?.?M\e(B" . "\e$,44h5!5)\e(B") ("\e$,1?.\e(B" . "\e$,44h4z4\7f\e(B") ("\e$,1?.?>\e(B" . "\e$,44h4z5!4{\e(B") ("\e$,1?.??\e(B" . "\e$,44i4\7f\e(B") ("\e$,1?.?@\e(B" . "\e$,44i4\7f4~\e(B") ("\e$,1?.?J\e(B". "\e$,44h5&5 \e(B") ("\e$,1?.?K\e(B". "\e$,44h5&5 4~\e(B")
 233     ("\e$,1?/?M\e(B" . "\e$,44^4z5!5)\e(B") ("\e$,1?/\e(B" . "\e$,44^4z4\7f\e(B") ("\e$,1?/?>\e(B" . "\e$,44^4z5!4{\e(B")("\e$,1?/??\e(B" . "\e$,44_4\7f\e(B") ("\e$,1?/?@\e(B" . "\e$,44_4\7f4~\e(B") ("\e$,1?/?J\e(B" . "\e$,44^5&5 \e(B") ("\e$,1?/?K\e(B" . "\e$,44^5&5 4~\e(B")
 234     ("\e$,1?0?M\e(B" . "\e$,44a5)\e(B") ("\e$,1?0\e(B" . "\e$,44a\e(B") ("\e$,1?0??\e(B" . "\e$,44b\e(B") ("\e$,1?0?@\e(B" . "\e$,44b4~\e(B")
 235     ("\e$,1?0?M\e(B" . "\e$,44a5)\e(B") ("\e$,1?0\e(B" . "\e$,44a\e(B") ("\e$,1?0??\e(B" . "\e$,44b\e(B") ("\e$,1?0?@\e(B" . "\e$,44b4~\e(B")
 236     ("\e$,1?2?M\e(B" . "\e$,44e5)\e(B") ("\e$,1?2\e(B" . "\e$,44d\e(B") ("\e$,1?2?>\e(B" . "\e$,44e4{\e(B") ("\e$,1?2??\e(B" . "\e$,44f\e(B") ("\e$,1?2?@\e(B" . "\e$,44f4~\e(B") ("\e$,1?2?F\e(B" . "\e$,44e5&\e(B") ("\e$,1?2?G\e(B" . "\e$,44e5&4~\e(B") ("\e$,1?2?H\e(B" . "\e$,44e5&5'\e(B") ("\e$,1?2?J\e(B" .  "\e$,44e5&5&5 \e(B") ("\e$,1?2?K\e(B" . "\e$,44e5&5&5 4~\e(B") ("\e$,1?2?L\e(B" . "\e$,44e5(\e(B")
 237     ("\e$,1?5?M\e(B" . "\e$,44h5)\e(B") ("\e$,1?5\e(B" . "\e$,44h\e(B") ("\e$,1?5??\e(B" . "\e$,44i\e(B") ("\e$,1?5?@\e(B" . "\e$,44i4~\e(B") ("\e$,1?5?A\e(B" . "\e$,44h5"\e(B") ("\e$,1?5?B\e(B" . "\e$,44h5#\e(B") ("\e$,1?5?J\e(B" . "\e$,44h5&5#\e(B") ("\e$,1?5?K\e(B" . "\e$,44h5&5#4~\e(B")
 238     ("\e$,1?6?M\e(B" . "\e$,44k5)\e(B") ("\e$,1?6\e(B" . "\e$,44k\e(B") ("\e$,1?6??\e(B" . "\e$,44l\e(B") ("\e$,1?6?@\e(B" . "\e$,44l4~\e(B")
 239     ("\e$,1?7?M\e(B" . "\e$,44n5)\e(B") ("\e$,1?7\e(B" . "\e$,44n\e(B") ("\e$,1?7??\e(B" . "\e$,44o\e(B") ("\e$,1?7?@\e(B" . "\e$,44o4~\e(B")
 240     ("\e$,1?8?M\e(B" . "\e$,44q5)\e(B") ("\e$,1?8\e(B" . "\e$,44q\e(B") ("\e$,1?8??\e(B" . "\e$,44r\e(B") ("\e$,1?8?@\e(B" . "\e$,44r4~\e(B")
 241     ("\e$,1?9?M\e(B" . "\e$,44t5)\e(B") ("\e$,1?9\e(B" . "\e$,44t\e(B") ("\e$,1?9??\e(B" . "\e$,44u\e(B") ("\e$,1?9?@\e(B" . "\e$,44u4~\e(B")
 242     ("\e$,1?3?M\e(B" . "\e$,44w5)\e(B") ("\e$,1?3\e(B" . "\e$,44w\e(B") ("\e$,1?3??\e(B" . "\e$,44x\e(B") ("\e$,1?3?@\e(B" . "\e$,44x4~\e(B"))
 243 "Kannada characters to glyphs conversion table.
 244 Default value contains only the basic rules.")
 245
 246 (defvar knd-char-glyph-hash
 247   (let* ((hash (make-hash-table :test 'equal)))
 248     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 249           knd-char-glyph)
 250     hash))
 251
 252 (defvar knd-char-glyph-regexp
 253   (kannada-regexp-of-hashtbl-keys knd-char-glyph-hash))
 254
 255 (defvar knd-conjunct-glyph
 256   '(("\e$,1>u\e(B" . "\e$,43Q\e(B") ("\e$,1>v\e(B" . "\e$,43U\e(B") ("\e$,1>w\e(B" . "\e$,43X\e(B") ("\e$,1>x\e(B" . "\e$,43[\e(B") ("\e$,1>y\e(B" . "\e$,43]\e(B")
 257     ("\e$,1>z\e(B" . "\e$,43`\e(B") ("\e$,1>{\e(B" . "\e$,43c\e(B") ("\e$,1>|\e(B" . "\e$,43g\e(B") ("\e$,1>}\e(B" . "\e$,43i\e(B") ("\e$,1>~\e(B" . "\e$,43k\e(B")
 258     ("\e$,1>\7f\e(B" . "\e$,43o\e(B") ("\e$,1? \e(B" . "\e$,43r\e(B") ("\e$,1?!\e(B" . "\e$,43u\e(B") ("\e$,1?"\e(B" . "\e$,43x\e(B") ("\e$,1?#\e(B" . "\e$,43|\e(B")
 259     ("\e$,1?$\e(B" . "\e$,44A\e(B") ("\e$,1?%\e(B" . "\e$,44D\e(B") ("\e$,1?&\e(B" . "\e$,44G\e(B") ("\e$,1?'\e(B" . "\e$,44J\e(B") ("\e$,1?(\e(B" . "\e$,44M\e(B")
 260     ("\e$,1?*\e(B" . "\e$,44P\e(B") ("\e$,1?+\e(B" . "\e$,44U\e(B") ("\e$,1?,\e(B" . "\e$,44Y\e(B") ("\e$,1?-\e(B" . "\e$,44\\e(B") ("\e$,1?.\e(B" . "\e$,44]\e(B")
 261     ("\e$,1?/\e(B" . "\e$,44`\e(B") ("\e$,1?0\e(B" . "\e$,44c\e(B") ("\e$,1?2\e(B" . "\e$,44g\e(B") ("\e$,1?3\e(B" . "\e$,44y\e(B") ("\e$,1?5\e(B" . "\e$,44j\e(B")
 262     ("\e$,1?6\e(B" . "\e$,44m\e(B") ("\e$,1?7\e(B" . "\e$,44p\e(B") ("\e$,1?8\e(B" . "\e$,44s\e(B") ("\e$,1?9\e(B" . "\e$,44v\e(B"))
 263 "Kannada characters to conjunct glyphs conversion table.")
 264
 265 (defvar knd-conjunct-glyph-hash
 266   (let* ((hash (make-hash-table :test 'equal)))
 267     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 268           knd-conjunct-glyph)
 269     hash))
 270
 271 (defvar knd-conjunct-glyph-regexp
 272   (kannada-regexp-of-hashtbl-vals knd-conjunct-glyph-hash))
 273
 274 (mapc
 275   (function (lambda (x)
 276     (put-char-code-property (aref (cdr x) 0) 'reference-point '(5 . 3))))
 277   knd-conjunct-glyph)
 278
 279 ;; glyph-to-glyph conversion table.
 280 ;; it is supposed that glyphs are ordered in
 281 ;;   [consonant/nukta] - [matra/virama] - [preceding-r] - [anuswar].
 282
 283 (defvar knd-glyph-glyph
 284   '(("\e$,45$4A\e(B" . "\e$,45*\e(B")
 285     ("\e$,45'4A\e(B" . "\e$,45+\e(B")
 286     ("\e$,44A3g\e(B" . "\e$,45,\e(B")
 287     ("\e$,45$3Q\e(B" . "\e$,45-\e(B")))
 288
 289 (defvar knd-glyph-glyph-hash
 290   (let* ((hash (make-hash-table :test 'equal)))
 291     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 292           knd-glyph-glyph)
 293     hash))
 294 (defvar knd-glyph-glyph-regexp
 295   (kannada-regexp-of-hashtbl-keys knd-glyph-glyph-hash))
 296
 297 (defun knd-charseq (from &optional to)
 298   (if (null to) (setq to from))
 299   (number-sequence (decode-char 'kannada-cdac from)
 300                    (decode-char 'kannada-cdac to)))
 301
 302 (defvar knd-glyph-cv
 303   (append
 304    (knd-charseq #x40 #x50)
 305    (knd-charseq #x52 #x54)
 306    (knd-charseq #x56 #x57)
 307    (knd-charseq #x59 #x5a)
 308    (knd-charseq #x5c)
 309    (knd-charseq #x5e #x5f)
 310    (knd-charseq #x61 #x62)
 311    (knd-charseq #x64 #x66)
 312    (knd-charseq #x6a)
 313    (knd-charseq #x6c #x6e)
 314    (knd-charseq #x70 #x71)
 315    (knd-charseq #x73 #x74)
 316    (knd-charseq #x76 #x77)
 317    (knd-charseq #x79 #x7b)
 318    (knd-charseq #x7d #x7e)
 319    (knd-charseq #xa2 #xa3)
 320    (knd-charseq #xa5 #xa6)
 321    (knd-charseq #xa8 #xa9)
 322    (knd-charseq #xab #xac)
 323    (knd-charseq #xae #xaf)
 324    (knd-charseq #xb1 #xb2)
 325    (knd-charseq #xb6 #xb8)
 326    (knd-charseq #xb6 #xb8)
 327    (knd-charseq #xba #xbb)
 328    (knd-charseq #xbe #xbf)
 329    (knd-charseq #xc1 #xc2)
 330    (knd-charseq #xc4 #xc6)
 331    (knd-charseq #xc8 #xc9)
 332    (knd-charseq #xcb #xcc)
 333    (knd-charseq #xce #xcf)
 334    (knd-charseq #xd1 #xd2)
 335    (knd-charseq #xd4 #xd5)
 336    (knd-charseq #xd7 #xd8)
 337    (knd-charseq #xc3))
 338   "Kannada Consonants/Vowels/Nukta Glyphs")
 339
 340 (defvar knd-glyph-space
 341   (knd-charseq #xb3 #xb4)
 342   "Kannada Spacing Glyphs")
 343
 344 (defvar knd-glyph-right-modifier
 345   (append
 346    (knd-charseq #xdb #xdd)
 347    (knd-charseq #xdf)
 348    (knd-charseq #xe0 #xe3)
 349    (knd-charseq #xe9))
 350   "Kannada Modifiers attached at the right side.")
 351
 352 (defvar knd-glyph-right-modifier-regexp
 353   (concat "[" knd-glyph-right-modifier "]"))
 354
 355 (defvar knd-glyph-jha-tail
 356    (knd-charseq #x68)
 357   "Kannada tail for jha.")
 358
 359 (defvar knd-glyph-top-matra
 360   (append
 361    (knd-charseq #xda)
 362    (knd-charseq #xdd)
 363    (knd-charseq #xe6)
 364    (knd-charseq #xe8))
 365   "Kannada Matras attached at the top side.")
 366
 367 (defvar knd-glyph-bottom-matra
 368   (append
 369    (knd-charseq #xe4 #xe5)
 370    (knd-charseq #xe7))
 371   "Kannada Matras attached at the bottom.")
 372
 373 (defvar knd-glyph-end-marks
 374   (append
 375    (knd-charseq #x25)
 376    (knd-charseq #x4d #x4e)
 377    (knd-charseq #xde))
 378   "Kannada end marks: arkavattu, virama, au and diirghaa.")
 379
 380 (defvar knd-glyph-bottom-modifier
 381   (append
 382    (knd-charseq #x51)
 383    (knd-charseq #x55)
 384    (knd-charseq #x58)
 385    (knd-charseq #x5b)
 386    (knd-charseq #x5d)
 387    (knd-charseq #x60)
 388    (knd-charseq #x63)
 389    (knd-charseq #x67)
 390    (knd-charseq #x69)
 391    (knd-charseq #x6b)
 392    (knd-charseq #x6f)
 393    (knd-charseq #x72)
 394    (knd-charseq #x75)
 395    (knd-charseq #x78)
 396    (knd-charseq #x7c)
 397    (knd-charseq #xa1)
 398    (knd-charseq #xa4)
 399    (knd-charseq #xa7)
 400    (knd-charseq #xaa)
 401    (knd-charseq #xad)
 402    (knd-charseq #xb0)
 403    (knd-charseq #xb5)
 404    (knd-charseq #xb9)
 405    (knd-charseq #xbc #xbd)
 406    (knd-charseq #xc0)
 407    (knd-charseq #xc3)
 408    (knd-charseq #xc7)
 409    (knd-charseq #xca)
 410    (knd-charseq #xcd)
 411    (knd-charseq #xd0)
 412    (knd-charseq #xd3)
 413    (knd-charseq #xd6)
 414    (knd-charseq #xd9)
 415    (knd-charseq #xea #xef))
 416   "Kannada Modifiers attached at the bottom.")
 417
 418 (defvar knd-glyph-order
 419   `((,knd-glyph-cv . 1)
 420     (,knd-glyph-top-matra . 2)
 421     (,knd-glyph-jha-tail . 3)
 422     (,knd-glyph-right-modifier . 4)
 423     (,knd-glyph-space . 5)
 424     (,knd-glyph-bottom-modifier . 5)
 425     (,knd-glyph-bottom-matra . 6)
 426     (,knd-glyph-end-marks . 7)
 427     ))
 428
 429 (mapc
 430  (function (lambda (x)
 431    (mapc
 432      (function (lambda (y)
 433        (put-char-code-property y 'composition-order (cdr x))))
 434      (car x))))
 435   knd-glyph-order)
 436
 437 (defun kannada-compose-syllable-string (string)
 438   (with-temp-buffer
 439     (insert (decompose-string string))
 440     (kannada-compose-syllable-region (point-min) (point-max))
 441     (buffer-string)))
 442
 443 ;; kch
 444 (defun kannada-compose-syllable-region (from to)
 445   "Compose kannada syllable in region FROM to TO."
 446   (let ((glyph-str nil) (cons-num 0) (glyph-str-list nil)
 447         (last-virama nil) (preceding-r nil) (last-modifier nil)
 448         (last-char (char-before to)) match-str pos
 449         glyph-block split-pos (conj nil) (rest nil))
 450     (save-excursion
 451       (save-restriction
 452           ;;; *** char-to-glyph conversion ***
 453         ;; Special rule 1. -- Last virama must be preserved.
 454         (if (eq last-char ?\e$,1?M\e(B)
 455             (progn
 456               (setq last-virama t)
 457               (narrow-to-region from (1- to)))
 458           (narrow-to-region from to))
 459         (goto-char (point-min))
 460         ;; Special rule 2. -- preceding "r virama" must be modifier.
 461         (when (looking-at "\e$,1?0?M\e(B.")
 462           (setq preceding-r t)
 463           (goto-char (+ 2 (point))))
 464         ;; remove conjunct consonants
 465         (while (re-search-forward knd-char-glyph-regexp nil t)
 466           (setq match-str (match-string 0))
 467           (if (and (string-match kannada-consonant match-str)
 468                    (> cons-num 0))
 469               (progn
 470                 (setq conj (concat conj (gethash (match-string 0 match-str)
 471                                                  knd-conjunct-glyph-hash)))
 472                 (setq match-str (replace-match "" t nil match-str))
 473                 (if (string-match "\e$,1?M\e(B" rest)
 474                     (setq rest (replace-match "" t nil rest)))))
 475           (setq rest (concat rest match-str))
 476           ;; count the number of consonant-glyhs.
 477           (if (string-match kannada-consonant match-str)
 478               (setq cons-num (1+ cons-num))))
 479         ;; translate the rest characters into glyphs
 480         (setq pos 0)
 481         (while (string-match knd-char-glyph-regexp rest pos)
 482           (setq match-str (match-string 0 rest))
 483           (setq pos (match-end 0))
 484           (setq glyph-str
 485                 (concat glyph-str (gethash match-str knd-char-glyph-hash))))
 486
 487         (if conj (setq glyph-str (concat glyph-str conj)))
 488         (if last-virama (setq glyph-str (concat glyph-str "\e$,45)\e(B"))
 489           (goto-char (point-min))
 490           (if (re-search-forward kannada-consonant-needs-twirl nil t)
 491               (progn
 492                 (setq match-str (match-string 0))
 493                 (setq glyph-str (concat glyph-str "\e$,44z\e(B")))))
 494         ;; preceding-r must be attached
 495         (if preceding-r
 496               (setq glyph-str (concat glyph-str "\e$,43%\e(B")))
 497           ;;; *** glyph-to-glyph conversion ***
 498         (when (string-match knd-glyph-glyph-regexp glyph-str)
 499           (setq glyph-str
 500                 (replace-match (gethash (match-string 0 glyph-str)
 501                                         knd-glyph-glyph-hash)
 502                                nil t glyph-str)))
 503           ;;; *** glyph reordering ***
 504         (while (setq split-pos (string-match "\e$,45)\e(B\\|.$" glyph-str))
 505           (setq glyph-block (substring glyph-str 0 (1+ split-pos)))
 506           (setq glyph-str (substring glyph-str (1+ split-pos)))
 507           (setq
 508            glyph-block
 509                (sort (string-to-list glyph-block)
 510                      (function (lambda (x y)
 511                         (< (get-char-code-property x 'composition-order)
 512                            (get-char-code-property y 'composition-order))))))
 513           (setq glyph-str-list (nconc glyph-str-list glyph-block)))
 514           ;;; *** insert space glyphs for kerning ***
 515         (if (> cons-num 0)
 516             (let ((curr glyph-str-list) (prev nil) (last-bott nil) bott co)
 517               (while curr
 518                 (setq co (get-char-code-property
 519                                 (car curr) 'composition-order)
 520                       bott (or (eq co 5) (eq co 6)))
 521                 (if (and bott last-bott)
 522                     (setcdr prev (cons ?\e$,44T\e(B curr)))
 523                 (setq last-bott bott prev curr curr (cdr curr)))))
 524           ;; concatenate and attach reference-points.
 525         (setq glyph-str
 526               (cdr
 527                (apply
 528                 'nconc
 529                 (mapcar
 530                  (function (lambda (x)
 531                    (list
 532                     (or (get-char-code-property x 'reference-point)
 533                     '(5 . 3) ;; default reference point.
 534                      )
 535                     x)))
 536                  glyph-str-list))))))
 537       (compose-region from to glyph-str)))
 538
 539 (provide 'knd-util)
 540
 541 ;;; arch-tag: 78d32230-a960-46a5-b622-61ed6ffcf8fc
 542 ;;; knd-util.el ends here