lisp/language/knd-util.el

   1 ;;; knd-util.el --- Support for composing Kannada characters
   2
   3 ;; Copyright (C) 2003 Free Software Foundation, Inc.
   4
   5 ;; Maintainer:  Maintainer:  CHOWKSEY, Kailash C. <klchxbec@m-net.arbornet.org>
   6 ;; Keywords: multilingual, Kannada
   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., 51 Franklin Street, Fifth Floor,
  23 ;; Boston, MA 02110-1301, USA.
  24
  25 ;; Created: Jul. 14. 2003
  26
  27 ;;; Commentary:
  28
  29 ;; This file provides character(Unicode) to glyph(CDAC) conversion and
  30 ;; composition of Kannada script characters.
  31
  32 ;;; Code:
  33
  34 ;;;###autoload
  35
  36 ;; Kannada Composable Pattern
  37 ;;    C .. Consonants
  38 ;;    V .. Vowel
  39 ;;    H .. Virama
  40 ;;    M .. Matra
  41 ;;    V .. Vowel
  42 ;;    (N .. Zerowidth Non Joiner)
  43 ;;    (J .. Zerowidth Joiner.  )
  44 ;; 1. vowel
  45 ;;  V(A)?
  46 ;; 2. syllable : maximum of 5 consecutive consonants.  (e.g. kartsnya)
  47 ;;  ((CH)?(CH)?(CH)?CH)?C(H|M?)?
  48
  49 (defconst kannada-consonant
  50   "[\e$,1>u\e(B-\e$,1?9\e(B]")
  51
  52 (defconst kannada-consonant-needs-twirl
  53   "[\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$,1?A?B?C?D>b\e(B]?$")
  54
  55 (defconst kannada-composable-pattern
  56   (concat
  57    "\\([\e$,1>b\e(B-\e$,1>t?`>l\e(B]\\)\\|[\e$,1>c\e(B]"
  58    "\\|\\("
  59    "\\(?:\\(?:[\e$,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$,1?M\e(B\\)?"
  60    "[\e$,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]?\\)?"
  61    "\\)")
  62   "Regexp matching a composable sequence of Kannada characters.")
  63
  64 ;;;###autoload
  65 (defun kannada-compose-region (from to)
  66   (interactive "r")
  67   (save-excursion
  68     (save-restriction
  69       (narrow-to-region from to)
  70       (goto-char (point-min))
  71       (while (re-search-forward kannada-composable-pattern nil t)
  72         (kannada-compose-syllable-region (match-beginning 0)
  73                                             (match-end 0))))))
  74 ;;;###autoload
  75 (defun kannada-compose-string (string)
  76   (with-temp-buffer
  77     (insert (decompose-string string))
  78     (kannada-compose-region (point-min) (point-max))
  79     (buffer-string)))
  80
  81 ;;;###autoload
  82 (defun kannada-post-read-conversion (len)
  83   (save-excursion
  84     (save-restriction
  85       (let ((buffer-modified-p (buffer-modified-p)))
  86         (narrow-to-region (point) (+ (point) len))
  87         (kannada-compose-region (point-min) (point-max))
  88         (set-buffer-modified-p buffer-modified-p)
  89         (- (point-max) (point-min))))))
  90
  91 (defun kannada-range (from to)
  92   "Make the list of the integers of range FROM to TO."
  93   (let (result)
  94     (while (<= from to) (setq result (cons to result) to (1- to))) result))
  95
  96 (defun kannada-regexp-of-hashtbl-keys (hashtbl)
  97   "Return a regular expression that matches all keys in hashtable HASHTBL."
  98   (let ((max-specpdl-size 1000))
  99     (regexp-opt
 100      (sort
 101       (let (dummy)
 102         (maphash (function (lambda (key val) (setq dummy (cons key dummy)))) hashtbl)
 103         dummy)
 104       (function (lambda (x y) (> (length x) (length y))))))))
 105
 106 (defun kannada-regexp-of-hashtbl-vals (hashtbl)
 107   "Return a regular expression that matches all values in hashtable HASHTBL."
 108   (let ((max-specpdl-size 1000))
 109     (regexp-opt
 110      (sort
 111       (let (dummy)
 112         (maphash (function (lambda (key val) (setq dummy (cons val dummy)))) hashtbl)
 113         dummy)
 114       (function (lambda (x y) (> (length x) (length y))))))))
 115
 116 (defun kannada-composition-function (from to pattern &optional string)
 117   "Compose Kannada characters in REGION, or STRING if specified.
 118 Assume that the REGION or STRING must fully match the composable
 119 PATTERN regexp."
 120   (if string (kannada-compose-syllable-string string)
 121     (kannada-compose-syllable-region from to))
 122   (- to from))
 123
 124 ;; Register a function to compose Kannada characters.
 125 (mapc
 126  (function (lambda (ucs)
 127    (aset composition-function-table (decode-char 'ucs ucs)
 128          (list (cons kannada-composable-pattern
 129                      'kannada-composition-function)))))
 130  (kannada-range #x0c80 #x0cff))
 131
 132 ;; Notes on conversion steps.
 133
 134 ;; 1. chars to glyphs
 135 ;;
 136 ;; Rules will not be applied to the virama appeared at the end of the
 137 ;; text.  Also, the preceding/following "r" will be treated as special case.
 138
 139 ;; 2. glyphs reordering.
 140 ;;
 141 ;; The glyphs are split by virama, and each glyph groups are
 142 ;; re-ordered in the following order.
 143 ;;
 144 ;; Note that `consonant-glyph' mentioned here does not contain the
 145 ;; vertical bar (right modifier) attached at the right of the
 146 ;; consonant.
 147 ;;
 148 ;; If the glyph-group contains right modifier,
 149 ;;  (1) consonant-glyphs/vowels
 150 ;;  (2) spacing
 151 ;;  (3) right modifier (may be matra)
 152 ;;  (4) top matra
 153 ;;  (5) preceding "r"
 154 ;;  (7) following "r"
 155 ;;  (8) bottom matra or virama.
 156 ;;
 157 ;; Otherwise,
 158 ;;  (1) consonant-glyph/vowels, with nukta sign
 159 ;;  (3) left matra
 160 ;;  (4) top matra
 161 ;;  (5) preceding "r"
 162 ;;  (7) following "r"
 163 ;;  (8) bottom matra or virama.
 164 ;;  (2) spacing
 165
 166 ;; 3. glyph to glyph
 167 ;;
 168 ;; For better display, some glyph display would be tuned.
 169
 170 ;; 4. Composition.
 171 ;;
 172 ;; left modifiers will be attached at the left.
 173 ;; others will be attached right.
 174
 175 ;; Problem::
 176 ;;  Can we generalize this methods to other Indian scripts?
 177
 178 (defvar knd-char-glyph
 179   '(("\e$,1>e\e(B" . "\e$,43@\e(B")
 180     ("\e$,1>f\e(B" . "\e$,43A\e(B")
 181     ("\e$,1?>\e(B" . "\e$,44{\e(B")
 182     ("\e$,1>g\e(B" . "\e$,43B\e(B")
 183     ("\e$,1??\e(B" . nil)
 184     ("\e$,1>h\e(B" . "\e$,43C\e(B")
 185     ("\e$,1?@\e(B" . nil)
 186     ("\e$,1>i\e(B" . "\e$,43D\e(B")
 187     ("\e$,1?A\e(B" . "\\e$,44\x7f\e(B")
 188     ("\e$,1>j\e(B" . "\e$,43E\e(B")
 189     ("\e$,1?B\e(B" . "\\e$,45 \e(B")
 190     ("\e$,1>k\e(B" . "\e$,43F4\x7f\e(B")
 191     ("\e$,1?C\e(B" . "\\e$,45$\e(B")
 192     ("\e$,1?`\e(B" . "\e$,43F5 \e(B")
 193     ("\e$,1?D\e(B" . "\\e$,45%\e(B")
 194     ;;("\e$,1>l\e(B" . nil) ; not implemented.
 195     ;;("\e$,1?a\e(B" . nil)
 196     ("\e$,1>n\e(B" . "\e$,43G\e(B")
 197     ("\e$,1>o\e(B" . "\e$,43H\e(B")
 198     ("\e$,1>p\e(B" . "\e$,43I\e(B")
 199     ("\e$,1?F\e(B" . "\\e$,45&\e(B")
 200     ("\e$,1?G\e(B" . "\\e$,45&4~\e(B")
 201     ("\e$,1?H\e(B" . "\\e$,45&5'\e(B")
 202     ("\e$,1>r\e(B" . "\e$,43J\e(B")
 203     ("\e$,1?J\e(B" . "\e$,45&5 \e(B")
 204     ("\e$,1>s\e(B" . "\e$,43K\e(B")
 205     ("\e$,1?K\e(B" . "\\e$,45&5 4~\e(B")
 206     ("\e$,1>t\e(B" . "\e$,43L\e(B")
 207     ("\e$,1?L\e(B" . "\\e$,45(\e(B")
 208     ("\e$,1>b\e(B" . "\e$,43M\e(B")
 209     ("\e$,1>c\e(B" . "\e$,43N\e(B")
 210     ("\e$,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$,43P4~\e(B")
 211     ("\e$,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$,43S5(\e(B")
 212     ("\e$,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$,43W4~\e(B")
 213     ("\e$,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$,43Z4~\e(B")
 214     ("\e$,1>y?M\e(B" . "\e$,43\5)\e(B") ("\e$,1>y\e(B" . "\e$,43\\e(B")
 215     ("\e$,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$,43_4~\e(B")
 216     ("\e$,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$,43b4~\e(B")
 217     ("\e$,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$,43e5(\e(B")
 218     ("\e$,1>}?M\e(B" . "\e$,44a4z3h4\x7f5)\e(B") ("\e$,1>}\e(B" . "\e$,44a4z3h4\x7f\e(B") ("\e$,1>}??\e(B" . "\e$,44b3h4\x7f\e(B") ("\e$,1>}?@\e(B" . "\e$,44b3h4\x7f4~\e(B") ("\e$,1>}?B\e(B". "\e$,44a4z3h5 \e(B") ("\e$,1>}?J\e(B". "\e$,44a5&3h5 \e(B") ("\e$,1>}?K\e(B". "\e$,44a5&3h5 4~\e(B")
 219     ("\e$,1>~?M\e(B" . "\e$,43j5)\e(B") ("\e$,1>~\e(B" . "\e$,43j\e(B")
 220     ("\e$,1>\x7f?M\e(B" . "\e$,43m5)\e(B") ("\e$,1>\x7f\e(B" . "\e$,43l\e(B") ("\e$,1?#?>\e(B" . "\e$,43m4{\e(B") ("\e$,1>\x7f??\e(B" . "\e$,43n\e(B") ("\e$,1>\x7f?@\e(B" . "\e$,43n4~\e(B") ("\e$,1>\x7f?F\e(B" . "\e$,43m5&\e(B") ("\e$,1>\x7f?G\e(B" . "\e$,43m5&4~\e(B") ("\e$,1>\x7f?H\e(B" . "\e$,43m5&5'\e(B") ("\e$,1>\x7f?J\e(B" .  "\e$,43m5&5&5 \e(B") ("\e$,1>\x7f?K\e(B" . "\e$,43m5&5&5 4~\e(B") ("\e$,1>\x7f?L\e(B" . "\e$,43m5(\e(B")
 221     ("\e$,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$,43q4~\e(B")
 222     ("\e$,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$,43t4~\e(B")
 223     ("\e$,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$,43w4~\e(B")
 224     ("\e$,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$,43z5(\e(B")
 225     ("\e$,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$,43~4~\e(B")
 226     ("\e$,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$,44C4~\e(B")
 227     ("\e$,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$,44F4~\e(B")
 228     ("\e$,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$,44I4~\e(B")
 229     ("\e$,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$,44L4~\e(B")
 230     ("\e$,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$,44N5&5#4~\e(B")
 231     ("\e$,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$,44Q5&5#4~\e(B")
 232     ("\e$,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$,44W5(\e(B")
 233     ("\e$,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$,44[4~\e(B")
 234     ("\e$,1?.?M\e(B" . "\e$,44h5!5)\e(B") ("\e$,1?.\e(B" . "\e$,44h4z4\x7f\e(B") ("\e$,1?.?>\e(B" . "\e$,44h4z5!4{\e(B") ("\e$,1?.??\e(B" . "\e$,44i4\x7f\e(B") ("\e$,1?.?@\e(B" . "\e$,44i4\x7f4~\e(B") ("\e$,1?.?J\e(B". "\e$,44h5&5 \e(B") ("\e$,1?.?K\e(B". "\e$,44h5&5 4~\e(B")
 235     ("\e$,1?/?M\e(B" . "\e$,44^4z5!5)\e(B") ("\e$,1?/\e(B" . "\e$,44^4z4\x7f\e(B") ("\e$,1?/?>\e(B" . "\e$,44^4z5!4{\e(B")("\e$,1?/??\e(B" . "\e$,44_4\x7f\e(B") ("\e$,1?/?@\e(B" . "\e$,44_4\x7f4~\e(B") ("\e$,1?/?J\e(B" . "\e$,44^5&5 \e(B") ("\e$,1?/?K\e(B" . "\e$,44^5&5 4~\e(B")
 236     ("\e$,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$,44b4~\e(B")
 237     ("\e$,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$,44b4~\e(B")
 238     ("\e$,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$,44e5(\e(B")
 239     ("\e$,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$,44h5&5#4~\e(B")
 240     ("\e$,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$,44l4~\e(B")
 241     ("\e$,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$,44o4~\e(B")
 242     ("\e$,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$,44r4~\e(B")
 243     ("\e$,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$,44u4~\e(B")
 244     ("\e$,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$,44x4~\e(B"))
 245 "Kannada characters to glyphs conversion table.
 246 Default value contains only the basic rules.")
 247
 248 (defvar knd-char-glyph-hash
 249   (let* ((hash (make-hash-table :test 'equal)))
 250     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 251           knd-char-glyph)
 252     hash))
 253
 254 (defvar knd-char-glyph-regexp
 255   (kannada-regexp-of-hashtbl-keys knd-char-glyph-hash))
 256
 257 (defvar knd-conjunct-glyph
 258   '(("\e$,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$,43]\e(B")
 259     ("\e$,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$,43k\e(B")
 260     ("\e$,1>\x7f\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$,43|\e(B")
 261     ("\e$,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")
 262     ("\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")
 263     ("\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")
 264     ("\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"))
 265 "Kannada characters to conjunct glyphs conversion table.")
 266
 267 (defvar knd-conjunct-glyph-hash
 268   (let* ((hash (make-hash-table :test 'equal)))
 269     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 270           knd-conjunct-glyph)
 271     hash))
 272
 273 (defvar knd-conjunct-glyph-regexp
 274   (kannada-regexp-of-hashtbl-vals knd-conjunct-glyph-hash))
 275
 276 (mapc
 277   (function (lambda (x)
 278     (put-char-code-property (aref (cdr x) 0) 'reference-point '(5 . 3))))
 279   knd-conjunct-glyph)
 280
 281 ;; glyph-to-glyph conversion table.
 282 ;; it is supposed that glyphs are ordered in
 283 ;;   [consonant/nukta] - [matra/virama] - [preceding-r] - [anuswar].
 284
 285 (defvar knd-glyph-glyph
 286   '(("\e$,45$4A\e(B" . "\e$,45*\e(B")
 287     ("\e$,45'4A\e(B" . "\e$,45+\e(B")
 288     ("\e$,44A3g\e(B" . "\e$,45,\e(B")
 289     ("\e$,45$3Q\e(B" . "\e$,45-\e(B")))
 290
 291 (defvar knd-glyph-glyph-hash
 292   (let* ((hash (make-hash-table :test 'equal)))
 293     (mapc (function (lambda (x) (puthash (car x) (cdr x) hash)))
 294           knd-glyph-glyph)
 295     hash))
 296 (defvar knd-glyph-glyph-regexp
 297   (kannada-regexp-of-hashtbl-keys knd-glyph-glyph-hash))
 298
 299 (defun knd-charseq (from &optional to)
 300   (if (null to) (setq to from))
 301   (mapcar (function (lambda (x) (indian-glyph-char x 'kannada)))
 302           (kannada-range from to)))
 303
 304 (defvar knd-glyph-cv
 305   (append
 306    (knd-charseq #x40 #x50)
 307    (knd-charseq #x52 #x54)
 308    (knd-charseq #x56 #x57)
 309    (knd-charseq #x59 #x5a)
 310    (knd-charseq #x5c)
 311    (knd-charseq #x5e #x5f)
 312    (knd-charseq #x61 #x62)
 313    (knd-charseq #x64 #x66)
 314    (knd-charseq #x6a)
 315    (knd-charseq #x6c #x6e)
 316    (knd-charseq #x70 #x71)
 317    (knd-charseq #x73 #x74)
 318    (knd-charseq #x76 #x77)
 319    (knd-charseq #x79 #x7b)
 320    (knd-charseq #x7d #x7e)
 321    (knd-charseq #xa2 #xa3)
 322    (knd-charseq #xa5 #xa6)
 323    (knd-charseq #xa8 #xa9)
 324    (knd-charseq #xab #xac)
 325    (knd-charseq #xae #xaf)
 326    (knd-charseq #xb1 #xb2)
 327    (knd-charseq #xb6 #xb8)
 328    (knd-charseq #xb6 #xb8)
 329    (knd-charseq #xba #xbb)
 330    (knd-charseq #xbe #xbf)
 331    (knd-charseq #xc1 #xc2)
 332    (knd-charseq #xc4 #xc6)
 333    (knd-charseq #xc8 #xc9)
 334    (knd-charseq #xcb #xcc)
 335    (knd-charseq #xce #xcf)
 336    (knd-charseq #xd1 #xd2)
 337    (knd-charseq #xd4 #xd5)
 338    (knd-charseq #xd7 #xd8)
 339    (knd-charseq #xc3))
 340   "Kannada Consonants/Vowels/Nukta Glyphs")
 341
 342 (defvar knd-glyph-space
 343   (knd-charseq #xb3 #xb4)
 344   "Kannada Spacing Glyphs")
 345
 346 (defvar knd-glyph-right-modifier
 347   (append
 348    (knd-charseq #xdb #xdd)
 349    (knd-charseq #xdf)
 350    (knd-charseq #xe0 #xe3)
 351    (knd-charseq #xe9))
 352   "Kannada Modifiers attached at the right side.")
 353
 354 (defvar knd-glyph-right-modifier-regexp
 355   (concat "[" knd-glyph-right-modifier "]"))
 356
 357 (defvar knd-glyph-jha-tail
 358    (knd-charseq #x68)
 359   "Kannada tail for jha.")
 360
 361 (defvar knd-glyph-top-matra
 362   (append
 363    (knd-charseq #xda)
 364    (knd-charseq #xdd)
 365    (knd-charseq #xe6)
 366    (knd-charseq #xe8))
 367   "Kannada Matras attached at the top side.")
 368
 369 (defvar knd-glyph-bottom-matra
 370   (append
 371    (knd-charseq #xe4 #xe5)
 372    (knd-charseq #xe7))
 373   "Kannada Matras attached at the bottom.")
 374
 375 (defvar knd-glyph-end-marks
 376   (append
 377    (knd-charseq #x25)
 378    (knd-charseq #x4d #x4e)
 379    (knd-charseq #xde))
 380   "Kannada end marks: arkavattu, virama, au and diirghaa.")
 381
 382 (defvar knd-glyph-bottom-modifier
 383   (append
 384    (knd-charseq #x51)
 385    (knd-charseq #x55)
 386    (knd-charseq #x58)
 387    (knd-charseq #x5b)
 388    (knd-charseq #x5d)
 389    (knd-charseq #x60)
 390    (knd-charseq #x63)
 391    (knd-charseq #x67)
 392    (knd-charseq #x69)
 393    (knd-charseq #x6b)
 394    (knd-charseq #x6f)
 395    (knd-charseq #x72)
 396    (knd-charseq #x75)
 397    (knd-charseq #x78)
 398    (knd-charseq #x7c)
 399    (knd-charseq #xa1)
 400    (knd-charseq #xa4)
 401    (knd-charseq #xa7)
 402    (knd-charseq #xaa)
 403    (knd-charseq #xad)
 404    (knd-charseq #xb0)
 405    (knd-charseq #xb5)
 406    (knd-charseq #xb9)
 407    (knd-charseq #xbc #xbd)
 408    (knd-charseq #xc0)
 409    (knd-charseq #xc3)
 410    (knd-charseq #xc7)
 411    (knd-charseq #xca)
 412    (knd-charseq #xcd)
 413    (knd-charseq #xd0)
 414    (knd-charseq #xd3)
 415    (knd-charseq #xd6)
 416    (knd-charseq #xd9)
 417    (knd-charseq #xea #xef))
 418   "Kannada Modifiers attached at the bottom.")
 419
 420 (defvar knd-glyph-order
 421   `((,knd-glyph-cv . 1)
 422     (,knd-glyph-top-matra . 2)
 423     (,knd-glyph-jha-tail . 3)
 424     (,knd-glyph-right-modifier . 4)
 425     (,knd-glyph-space . 5)
 426     (,knd-glyph-bottom-modifier . 5)
 427     (,knd-glyph-bottom-matra . 6)
 428     (,knd-glyph-end-marks . 7)
 429     ))
 430
 431 (mapc
 432  (function (lambda (x)
 433    (mapc
 434      (function (lambda (y)
 435        (put-char-code-property y 'composition-order (cdr x))))
 436      (car x))))
 437   knd-glyph-order)
 438
 439 (defun kannada-compose-syllable-string (string)
 440   (with-temp-buffer
 441     (insert (decompose-string string))
 442     (kannada-compose-syllable-region (point-min) (point-max))
 443     (buffer-string)))
 444
 445 ;; kch
 446 (defun kannada-compose-syllable-region (from to)
 447   "Compose kannada syllable in region FROM to TO."
 448   (let ((glyph-str nil) (cons-num 0) (glyph-str-list nil)
 449         (last-virama nil) (preceding-r nil) (last-modifier nil)
 450         (last-char (char-before to)) match-str pos
 451         glyph-block split-pos (conj nil) (rest nil))
 452     (save-excursion
 453       (save-restriction
 454           ;;; *** char-to-glyph conversion ***
 455         ;; Special rule 1. -- Last virama must be preserved.
 456         (if (eq last-char ?\e$,1?M\e(B)
 457             (progn
 458               (setq last-virama t)
 459               (narrow-to-region from (1- to)))
 460           (narrow-to-region from to))
 461         (goto-char (point-min))
 462         ;; Special rule 2. -- preceding "r virama" must be modifier.
 463         (when (looking-at "\e$,1?0?M\e(B.")
 464           (setq preceding-r t)
 465           (goto-char (+ 2 (point))))
 466         ;; remove conjunct consonants
 467         (while (re-search-forward knd-char-glyph-regexp nil t)
 468           (setq match-str (match-string 0))
 469           (if (and (string-match kannada-consonant match-str)
 470                    (> cons-num 0))
 471               (progn
 472                 (setq conj (concat conj (gethash (match-string 0 match-str)
 473                                                  knd-conjunct-glyph-hash)))
 474                 (setq match-str (replace-match "" t nil match-str))
 475                 (if (string-match "\e$,1?M\e(B" rest)
 476                     (setq rest (replace-match "" t nil rest)))))
 477           (setq rest (concat rest match-str))
 478           ;; count the number of consonant-glyhs.
 479           (if (string-match kannada-consonant match-str)
 480               (setq cons-num (1+ cons-num))))
 481         ;; translate the rest characters into glyphs
 482         (setq pos 0)
 483         (while (string-match knd-char-glyph-regexp rest pos)
 484           (setq match-str (match-string 0 rest))
 485           (setq pos (match-end 0))
 486           (setq glyph-str
 487                 (concat glyph-str (gethash match-str knd-char-glyph-hash))))
 488
 489         (if conj (setq glyph-str (concat glyph-str conj)))
 490         (if last-virama (setq glyph-str (concat glyph-str "\e$,45)\e(B"))
 491           (goto-char (point-min))
 492           (if (re-search-forward kannada-consonant-needs-twirl nil t)
 493               (progn
 494                 (setq match-str (match-string 0))
 495                 (setq glyph-str (concat glyph-str "\e$,44z\e(B")))))
 496         ;; preceding-r must be attached
 497         (if preceding-r
 498               (setq glyph-str (concat glyph-str "\e$,43%\e(B")))
 499           ;;; *** glyph-to-glyph conversion ***
 500         (when (string-match knd-glyph-glyph-regexp glyph-str)
 501           (setq glyph-str
 502                 (replace-match (gethash (match-string 0 glyph-str)
 503                                         knd-glyph-glyph-hash)
 504                                nil t glyph-str)))
 505           ;;; *** glyph reordering ***
 506         (while (setq split-pos (string-match "\e$,45)\e(B\\|.$" glyph-str))
 507           (setq glyph-block (substring glyph-str 0 (1+ split-pos)))
 508           (setq glyph-str (substring glyph-str (1+ split-pos)))
 509           (setq
 510            glyph-block
 511                (sort (string-to-list glyph-block)
 512                      (function (lambda (x y)
 513                         (< (get-char-code-property x 'composition-order)
 514                            (get-char-code-property y 'composition-order))))))
 515           (setq glyph-str-list (nconc glyph-str-list glyph-block)))
 516           ;;; *** insert space glyphs for kerning ***
 517         (if (> cons-num 0)
 518             (let ((curr glyph-str-list) (prev nil) (last-bott nil) bott co)
 519               (while curr
 520                 (setq co (get-char-code-property
 521                                 (car curr) 'composition-order)
 522                       bott (or (eq co 5) (eq co 6)))
 523                 (if (and bott last-bott)
 524                     (setcdr prev (cons ?\e$,44T\e(B curr)))
 525                 (setq last-bott bott prev curr curr (cdr curr)))))
 526           ;; concatenate and attach reference-points.
 527         (setq glyph-str
 528               (cdr
 529                (apply
 530                 'nconc
 531                 (mapcar
 532                  (function (lambda (x)
 533                    (list
 534                     (or (get-char-code-property x 'reference-point)
 535                     '(5 . 3) ;; default reference point.
 536                      )
 537                     x)))
 538                  glyph-str-list))))))
 539       (compose-region from to glyph-str)))
 540
 541 (provide 'knd-util)
 542
 543 ;;; arch-tag: 78d32230-a960-46a5-b622-61ed6ffcf8fc
 544 ;;; knd-util.el ends here