lisp/international/fontset.el

   1 ;;; fontset.el --- commands for handling fontset
   2
   3 ;; Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
   4 ;;   2005, 2006, 2007, 2008  Free Software Foundation, Inc.
   5 ;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
   6 ;;   2005, 2006, 2007, 2008
   7 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
   8 ;;   Registration Number H14PRO021
   9 ;; Copyright (C) 2003, 2006
  10 ;;   National Institute of Advanced Industrial Science and Technology (AIST)
  11 ;;   Registration Number H13PRO009
  12
  13 ;; Keywords: mule, i18n, fontset
  14
  15 ;; This file is part of GNU Emacs.
  16
  17 ;; GNU Emacs is free software; you can redistribute it and/or modify
  18 ;; it under the terms of the GNU General Public License as published by
  19 ;; the Free Software Foundation; either version 3, or (at your option)
  20 ;; any later version.
  21
  22 ;; GNU Emacs is distributed in the hope that it will be useful,
  23 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  24 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25 ;; GNU General Public License for more details.
  26
  27 ;; You should have received a copy of the GNU General Public License
  28 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  29 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  30 ;; Boston, MA 02110-1301, USA.
  31
  32 ;;; Commentary:
  33
  34 ;;; Code:
  35
  36 ;; Setup font-encoding-alist for all known encodings.
  37
  38 (setq font-encoding-alist
  39       '(("iso8859-1$" . iso-8859-1)
  40         ("iso8859-2$" . iso-8859-2)
  41         ("iso8859-3$" . iso-8859-3)
  42         ("iso8859-4$" . iso-8859-4)
  43         ("iso8859-5$" . iso-8859-5)
  44         ("iso8859-6$" . iso-8859-6)
  45         ("iso8859-7$" . iso-8859-7)
  46         ("iso8859-8$" . iso-8859-8)
  47         ("iso8859-9$" . iso-8859-9)
  48         ("iso8859-10$" . iso-8859-10)
  49         ("iso8859-11$" . iso-8859-11)
  50         ("iso8859-13$" . iso-8859-13)
  51         ("iso8859-14$" . iso-8859-14)
  52         ("iso8859-15$" . iso-8859-15)
  53         ("iso8859-16$" . iso-8859-16)
  54         ("gb2312.1980" . chinese-gb2312)
  55         ("gbk" . chinese-gbk)
  56         ("gb18030" . (unicode . nil))
  57         ("jisx0208.1978" . japanese-jisx0208-1978)
  58         ("jisx0208" . japanese-jisx0208)
  59         ("jisx0201" . jisx0201)
  60         ("jisx0212" . japanese-jisx0212)
  61         ("ksc5601.1987" . korean-ksc5601)
  62         ("cns11643.1992.*1" . chinese-cns11643-1)
  63         ("cns11643.1992.*2" . chinese-cns11643-2)
  64         ("cns11643.1992.*3" . chinese-cns11643-3)
  65         ("cns11643.1992.*4" . chinese-cns11643-4)
  66         ("cns11643.1992.*5" . chinese-cns11643-5)
  67         ("cns11643.1992.*6" . chinese-cns11643-6)
  68         ("cns11643.1992.*7" . chinese-cns11643-7)
  69         ("cns11643.92p1-0" . chinese-cns11643-1)
  70         ("cns11643.92p2-0" . chinese-cns11643-2)
  71         ("cns11643.92p3-0" . chinese-cns11643-3)
  72         ("cns11643.92p4-0" . chinese-cns11643-4)
  73         ("cns11643.92p5-0" . chinese-cns11643-5)
  74         ("cns11643.92p6-0" . chinese-cns11643-6)
  75         ("cns11643.92p7-0" . chinese-cns11643-7)
  76         ("big5" . big5)
  77         ("viscii" . viscii)
  78         ("tis620" . tis620-2533)
  79         ("microsoft-cp1251" . windows-1251)
  80         ("koi8-r" . koi8-r)
  81         ("jisx0213.2000-1" . japanese-jisx0213-1)
  82         ("jisx0213.2000-2" . japanese-jisx0213-2)
  83         ("jisx0213.2004-1" . japanese-jisx0213.2004-1)
  84         ("iso10646-1$" . (unicode-bmp . nil))
  85         ("iso10646.indian-1" . (unicode-bmp . nil))
  86         ("unicode-bmp" . (unicode-bmp . nil))
  87         ("abobe-symbol" . symbol)
  88         ("sisheng_cwnn" . chinese-sisheng)
  89         ("mulearabic-0" . arabic-digit)
  90         ("mulearabic-1" . arabic-1-column)
  91         ("mulearabic-2" . arabic-2-column)
  92         ("muleipa" . ipa)
  93         ("ethiopic-unicode" . (unicode-bmp . ethiopic))
  94         ("is13194-devanagari" . indian-is13194)
  95         ("Devanagari-CDAC" . devanagari-cdac)
  96         ("Sanskrit-CDAC" . sanskrit-cdac)
  97         ("Bengali-CDAC" . bengali-cdac)
  98         ("Assamese-CDAC" . assamese-cdac)
  99         ("Punjabi-CDAC" . punjabi-cdac)
 100         ("Gujarati-CDAC" . gujarati-cdac)
 101         ("Oriya-CDAC" . oriya-cdac)
 102         ("Tamil-CDAC" . tamil-cdac)
 103         ("Telugu-CDAC" . telugu-cdac)
 104         ("Kannada-CDAC" . kannada-cdac)
 105         ("Malayalam-CDAC" . malayalam-cdac)
 106         ("Devanagari-Akruti" . devanagari-akruti)
 107         ("Bengali-Akruti" . bengali-akruti)
 108         ("Punjabi-Akruti" . punjabi-akruti)
 109         ("Gujarati-Akruti" . gujarati-akruti)
 110         ("Oriya-Akruti" . oriya-akruti)
 111         ("Tamil-Akruti" . tamil-akruti)
 112         ("Telugu-Akruti" . telugu-akruti)
 113         ("Kannada-Akruti" . kannada-akruti)
 114         ("Malayalam-Akruti" . malayalam-akruti)
 115         ("muleindian-2" . indian-2-column)
 116         ("muleindian-1" . indian-1-column)
 117         ("mulelao-1" . mule-lao)
 118         ("muletibetan-2" . tibetan)
 119         ("muletibetan-1" . tibetan-1-column)))
 120
 121 (setq font-encoding-charset-alist
 122       '((latin-iso8859-1 . iso-8859-1)
 123         (latin-iso8859-2 . iso-8859-2)
 124         (latin-iso8859-3 . iso-8859-3)
 125         (latin-iso8859-4 . iso-8859-4)
 126         (latin-iso8859-9 . iso-8859-9)
 127         (latin-iso8859-10 . iso-8859-10)
 128         (latin-iso8859-13 . iso-8859-13)
 129         (latin-iso8859-14 . iso-8859-14)
 130         (latin-iso8859-15 . iso-8859-15)
 131         (latin-iso8859-16 . iso-8859-16)
 132         (cyrillic-iso8859-5 . iso-8859-5)
 133         (greek-iso8859-7 . iso-8859-7)
 134         (arabic-iso8859-6 . iso-8859-6)
 135         (thai-tis620 . tis620-2533)
 136         (latin-jisx0201 . jisx0201)
 137         (katakana-jisx0201 . jisx0201)
 138         (chinese-big5-1 . big5)
 139         (chinese-big5-2 . big5)
 140         (vietnamese-viscii-lower . viscii)
 141         (vietnamese-viscii-upper . viscii)
 142         (tibetan . unicode-bmp)))
 143
 144 (setq script-representative-chars
 145       '((latin ?A ?Z ?a ?z)
 146         (greek #x3A9)
 147         (coptic #x3E2)
 148         (cyrillic #x42F)
 149         (armenian #x531)
 150         (hebrew #x5D0)
 151         (arabic #x628)
 152         (syriac #x710)
 153         (thaana #x78C)
 154         (devanagari #x915)
 155         (bengali #x995)
 156         (gurmukhi #xA15)
 157         (gujarati #xA95)
 158         (oriya #xB15)
 159         (tamil #xB95)
 160         (telugu #xC15)
 161         (kannada #xC95)
 162         (malayalam #xD15)
 163         (sinhala #xD95)
 164         (thai #xE17)
 165         (lao #xEA5)
 166         (tibetan #xF40)
 167         (myanmar #x1000)
 168         (georgian #x10D3)
 169         (ethiopic #x1208)
 170         (cherokee #x13B6)
 171         (canadian-aboriginal #x14C0)
 172         (ogham #x168F)
 173         (runic #x16A0)
 174         (khmer #x1780)
 175         (mongolian #x1826)
 176         (braille #x2800)
 177         (ideographic-description #x2FF0)
 178         (cjk-misc #x300E)
 179         (kana #x304B)
 180         (bopomofo #x3105)
 181         (kanbun #x319D)
 182         (han #x5B57)
 183         (yi #xA288)
 184         (cham #xAA00)
 185         (tai-viet #xAA80)
 186         (hangul #xAC00)))
 187
 188 (setq otf-script-alist
 189       '((arab . arabic)
 190         (armn . armenian)
 191         (bali . balinese)
 192         (beng . bengali)
 193         (bopo . bopomofo)
 194         (brai . braille)
 195         (bugi . buginese)
 196         (buhd . buhid)
 197         (byzm . byzantine-musical-symbol)
 198         (cans . canadian-aboriginal)
 199         (cher . cherokee)
 200         (copt . coptic)
 201         (xsux . cuneiform)
 202         (cyrl . cyrillic)
 203         (cprt . cypriot)
 204         (dsrt . deseret)
 205         (deva . devanagari)
 206         (ethi . ethiopic)
 207         (geor . georgian)
 208         (glag . glagolitic)
 209         (goth . gothic)
 210         (grek . greek)
 211         (gujr . gujarati)
 212         (guru . gurmukhi)
 213         (hani . han)
 214         (hang . hangul)
 215         (hano . hanunoo)
 216         (hebr . hebrew)
 217         (kana . kana)
 218         (knda . kannada)
 219         (khar . kharoshthi)
 220         (khmr . khmer)
 221         (lao\  . lao)
 222         (latn . latin)
 223         (limb . limbu)
 224         (linb . linear_b)
 225         (mlym . malayalam)
 226         (math . mathematical)
 227         (mong . mongolian)
 228         (musc . musical-symbol)
 229         (mymr . myanmar)
 230         (nko\  . nko)
 231         (ogam . ogham)
 232         (ital . old_italic)
 233         (xpeo . old_persian)
 234         (orya . oriya)
 235         (osma . osmanya)
 236         (phag . phags-pa)
 237         (phnx . phoenician)
 238         (runr . runic)
 239         (shaw . shavian)
 240         (sinh . sinhala)
 241         (sylo . syloti_nagri)
 242         (syrc . syriac)
 243         (tglg . tagalog)
 244         (tagb . tagbanwa)
 245         (taml . tamil)
 246         (tale . tai_le)
 247         (telu . telugu)
 248         (thaa . thaana)
 249         (thai . thai)
 250         (tibt . tibetan)
 251         (tfng . tifinagh)
 252         (ugar . ugaritic)
 253         (yi\ \   . yi)))
 254
 255 ;; Set standard fontname specification of characters in the default
 256 ;; fontset to find an appropriate font for each script/charset.  The
 257 ;; specification has the form ((SCRIPT FONT-SPEC ...) ...), where
 258 ;; FONT-SPEC is:
 259 ;;      a cons (FAMILY . REGISTRY),
 260 ;;      or a string FONT-NAME,
 261 ;;      or an object created by `font-spec'.
 262 ;;
 263 ;; FAMILY may be nil, in which case, the the corresponding name of
 264 ;; default face is used.  If REGISTRY contains a character `-', the
 265 ;; string before that is embedded in `CHARSET_REGISTRY' field, and the
 266 ;; string after that is embedded in `CHARSET_ENCODING' field.  If it
 267 ;; does not contain `-', the whole string is embedded in
 268 ;; `CHARSET_REGISTRY' field, and a wild card character `*' is embedded
 269 ;; in `CHARSET_ENCODING' field.
 270 ;;
 271 ;; SCRIPT is a symbol that appears as an element of the char table
 272 ;; `char-script-table'.  SCRIPT may be a charset specifying the range
 273 ;; of characters.
 274
 275 (defun setup-default-fontset ()
 276   "Setup the default fontset."
 277   (new-fontset
 278    "fontset-default"
 279    `(;; for each script
 280      (latin (nil . "ISO8859-1")
 281             (nil . "ISO8859-2")
 282             (nil . "ISO8859-3")
 283             (nil . "ISO8859-4")
 284             (nil . "ISO8859-9")
 285             (nil . "ISO8859-10")
 286             (nil . "ISO8859-13")
 287             (nil . "ISO8859-14")
 288             (nil . "ISO8859-15")
 289             (nil . "ISO8859-16")
 290             (nil . "VISCII1.1-1")
 291             (nil . "ISO10646-1"))
 292
 293      (thai  ,(font-spec :registry "iso10646-1" :otf '(thai nil nil (mark)))
 294             (nil . "TIS620*")
 295             (nil . "ISO8859-11"))
 296
 297      (devanagari ,(font-spec :registry "iso10646-1" :otf '(deva nil (rphf)))
 298                  (nil . "iso10646.indian-1"))
 299      (bengali ,(font-spec :registry "iso10646-1" :otf '(beng nil (rphf))))
 300      (gurmukhi ,(font-spec :registry "iso10646-1" :otf '(guru nil (blwf))))
 301      (gujarati ,(font-spec :registry "iso10646-1" :otf '(gujr nil (rphf))))
 302      (oriya ,(font-spec :registry "iso10646-1" :otf '(orya nil (rphf))))
 303      (tamil ,(font-spec :registry "iso10646-1" :otf '(taml nil (akhn))))
 304      (telugu ,(font-spec :registry "iso10646-1" :otf '(telu nil (blwf))))
 305      (kannada ,(font-spec :registry "iso10646-1" :otf '(knda nil (rphf))))
 306      (sinhala ,(font-spec :registry "iso10646-1" :otf '(sinh nil (akhn))))
 307      (malayalam ,(font-spec :registry "iso10646-1" :otf '(mlym nil (akhn))))
 308
 309      (lao ,(font-spec :registry "iso10646-1" :otf '(lao\  nil nil (mark)))
 310           ,(font-spec :registry "iso10646-1" :script 'lao)
 311           (nil . "MuleLao-1"))
 312
 313      (tai-viet ("TaiViet" . "iso10646-1"))
 314
 315      ;; both for script and charset.
 316      (tibetan ,(font-spec :registry "iso10646-1"
 317                           :otf '(tibt nil (ccmp blws abvs)))
 318               ,(font-spec :family "mtib" :registry "iso10646-1")
 319               (nil . "muletibetan-2"))
 320
 321      ;; both for script and charset.
 322      (ethiopic ,(font-spec :registry "iso10646-1" :script 'ethiopic)
 323                (nil . "ethiopic-unicode"))
 324
 325      (greek ,(font-spec :registry "iso10646-1" :script 'greek)
 326             (nil . "ISO8859-7"))
 327
 328      (cyrillic ,(font-spec :registry "iso10646-1" :script 'cyrillic)
 329                (nil . "ISO8859-5")
 330                (nil . "microsoft-cp1251")
 331                (nil . "koi8-r"))
 332
 333      (arabic ,(font-spec :registry "iso10646-1"
 334                          :otf '(arab nil (init medi fini liga)))
 335              (nil . "MuleArabic-0")
 336              (nil . "MuleArabic-1")
 337              (nil . "MuleArabic-2")
 338              (nil . "ISO8859-6"))
 339
 340      (hebrew ,(font-spec :registry "iso10646-1" :script 'hebrew)
 341              (nil . "ISO8859-8"))
 342
 343      (syriac ,(font-spec :registry "iso10646-1" :script 'syriac))
 344
 345      (thaana ,(font-spec :registry "iso10646-1" :otf '(thaa nil nil)))
 346
 347      (myanmar ,(font-spec :registry "iso10646-1" :script 'myanmar))
 348
 349      (georgian ,(font-spec :registry "iso10646-1" :script 'georgian))
 350
 351      (cherokee ,(font-spec :registry "iso10646-1" :script 'cherokee))
 352
 353      (canadian-aboriginal ,(font-spec :registry "iso10646-1"
 354                                       :script 'canadian-aboriginal))
 355
 356      (ogham ,(font-spec :registry "iso10646-1" :script 'ogham))
 357
 358      (runic ,(font-spec :registry "iso10646-1" :script 'runic))
 359
 360      (khmer ,(font-spec :registry "iso10646-1" :otf '(khmr nil (pres))))
 361
 362      (yi ,(font-spec :registry "iso10646-1" :script 'yi))
 363
 364      (kana ,(font-spec :registry "iso10646-1" :script 'kana)
 365            (nil . "JISX0208*")
 366            (nil . "GB2312.1980-0")
 367            (nil . "KSC5601.1987*")
 368            (nil . "JISX0201*")
 369            (nil . "JISX0213.2000-1")
 370            (nil . "JISX0213.2004-1"))
 371
 372      (bopomofo ,(font-spec :registry "iso10646-1" :script 'bopomofo)
 373                (nil . "sisheng_cwnn-0"))
 374
 375      (han ,(font-spec :registry "iso10646-1" :language 'ja)
 376           ,(font-spec :registry "iso10646-1" :language 'zh)
 377           (nil . "GB2312.1980-0")
 378           (nil . "JISX0208*")
 379           (nil . "JISX0212*")
 380           (nil . "big5*")
 381           (nil . "KSC5601.1987*")
 382           (nil . "CNS11643.1992-1")
 383           (nil . "CNS11643.1992-2")
 384           (nil . "CNS11643.1992-3")
 385           (nil . "CNS11643.1992-4")
 386           (nil . "CNS11643.1992-5")
 387           (nil . "CNS11643.1992-6")
 388           (nil . "CNS11643.1992-7")
 389           (nil . "gbk-0")
 390           (nil . "gb18030")
 391           (nil . "JISX0213.2000-1")
 392           (nil . "JISX0213.2000-2")
 393           (nil . "JISX0213.2004-1"))
 394
 395      (cjk-misc ,(font-spec :registry "iso10646-1" :language 'ja)
 396                ,(font-spec :registry "iso10646-1" :language 'zh)
 397                (nil . "GB2312.1980-0")
 398                (nil . "JISX0208*")
 399                (nil . "JISX0212*")
 400                (nil . "big5*")
 401                (nil . "KSC5601.1987*")
 402                (nil . "CNS11643.1992-1")
 403                (nil . "CNS11643.1992-2")
 404                (nil . "CNS11643.1992-3")
 405                (nil . "CNS11643.1992-4")
 406                (nil . "CNS11643.1992-5")
 407                (nil . "CNS11643.1992-6")
 408                (nil . "CNS11643.1992-7")
 409                (nil . "gbk-0")
 410                (nil . "gb18030")
 411                (nil . "JISX0213.2000-1")
 412                (nil . "JISX0213.2000-2"))
 413
 414      (hangul ,(font-spec :registry "iso10646-1" :language 'ko)
 415              (nil . "KSC5601.1987-0"))
 416
 417      (braille ,(font-spec :registry "iso10646-1" :script 'braille))
 418
 419      (mathematical ,(font-spec :registry "iso10646-1" :script 'mathematical))
 420
 421      ;; for each charset
 422      (ascii (nil . "ISO8859-1"))
 423      (arabic-digit ("*" . "MuleArabic-0"))
 424      (arabic-1-column ("*" . "MuleArabic-1"))
 425      (arabic-2-column ("*" . "MuleArabic-2"))
 426      (indian-is13194 (nil . "is13194-devanagari"))
 427      (indian-1-column ("*" . "muleindian-2"))
 428      ;; Indian CDAC
 429      (devanagari-cdac (nil . "Devanagari-CDAC"))
 430      (sanskrit-cdac (nil . "Sanskrit-CDAC"))
 431      (bengali-cdac (nil . "Bengali-CDAC"))
 432      (assamese-cdac (nil . "Assamese-CDAC"))
 433      (punjabi-cdac (nil . "Punjabi-CDAC"))
 434      (gujarati-cdac (nil . "Gujarati-CDAC"))
 435      (oriya-cdac (nil . "Oriya-CDAC"))
 436      (tamil-cdac (nil . "Tamil-CDAC"))
 437      (telugu-cdac (nil . "Telugu-CDAC"))
 438      (kannada-cdac (nil . "Kannada-CDAC"))
 439      (malayalam-cdac (nil . "Malayalam-CDAC"))
 440      ;; Indian AKRUTI
 441      (devanagari-akruti (nil . "Devanagari-Akruti"))
 442      (bengali-akruti (nil . "Bengali-Akruti"))
 443      (punjabi-akruti (nil . "Punjabi-Akruti"))
 444      (gujarati-akruti (nil . "Gujarati-Akruti"))
 445      (oriya-akruti (nil . "Oriya-Akruti"))
 446      (tamil-akruti (nil . "Tamil-Akruti"))
 447      (telugu-akruti (nil . "Telugu-Akruti"))
 448      (kannada-akruti (nil . "Kannada-Akruti"))
 449      (malayalam-akruti (nil . "Malayalam-Akruti"))
 450      ;;(devanagari-glyph ("altsys-dv_ttsurekh" . "devanagari-cdac"))
 451      ;;(malayalam-glyph ("altsys-ml_ttkarthika" . "malayalam-cdac"))
 452      (ipa ,(font-spec :registry "iso10646-1" :script 'phonetic)
 453           (nil . "MuleIPA-1")
 454           (nil . "iso10646-1"))
 455
 456      ;; Fallback fonts
 457      (nil (nil . "gb2312.1980")
 458           (nil . "gbk-0")
 459           (nil . "gb18030")
 460           (nil . "jisx0208")
 461           (nil . "ksc5601.1987")
 462           (nil . "CNS11643.1992-1")
 463           (nil . "CNS11643.1992-2")
 464           (nil . "CNS11643.1992-3")
 465           (nil . "CNS11643.1992-4")
 466           (nil . "CNS11643.1992-5")
 467           (nil . "CNS11643.1992-6")
 468           (nil . "CNS11643.1992-7")
 469           (nil . "big5")
 470           (nil . "jisx0213.2000-1")
 471           (nil . "jisx0213.2004-1")
 472           (nil . "jisx0212"))
 473      ))
 474
 475   ;; Append Unicode fonts.
 476   ;; This may find fonts with more variants (bold, italic) but which
 477   ;; don't cover many characters.
 478   (set-fontset-font "fontset-default" nil
 479                     '(nil . "iso10646-1") nil 'append)
 480   ;; These may find fonts that cover many characters but with fewer
 481   ;; variants.
 482   (set-fontset-font "fontset-default" nil
 483                     '("gnu-unifont" . "iso10646-1") nil 'append)
 484   (set-fontset-font "fontset-default" nil
 485                     '("mutt-clearlyu" . "iso10646-1") nil 'append)
 486
 487   (set-fontset-font "fontset-default" '(#xE000 . #xF8FF) nil))
 488
 489 ;; These are the registered registries/encodings from
 490 ;; ftp://ftp.x.org/pub/DOCS/registry 2001/06/01
 491
 492 ;; Name                                            Reference
 493 ;; ----                                            ---------
 494 ;; "DEC"                                           [27]
 495 ;;         registry prefix
 496 ;; "DEC.CNS11643.1986-2"                           [53]
 497 ;;         CNS11643 2-plane using the encoding
 498 ;;         suggested in that standard
 499 ;; "DEC.DTSCS.1990-2"                              [54]
 500 ;;         DEC Taiwan Supplemental Character Set
 501 ;; "fujitsu.u90x01.1991-0"                         [87]
 502 ;; "fujitsu.u90x03.1991-0"                         [87]
 503 ;; "GB2312.1980-0"                                 [39],[12]
 504 ;;         China (PRC) Hanzi, GL encoding
 505 ;; "GB2312.1980-1"                                 [39]
 506 ;;         (deprecated)
 507 ;;         China (PRC) Hanzi, GR encoding
 508 ;; "HP-Arabic8"                                    [36]
 509 ;;         HPARABIC8 8-bit character set
 510 ;; "HP-East8"                                      [36]
 511 ;;         HPEAST8 8-bit character set
 512 ;; "HP-Greek8"                                     [36]
 513 ;;         HPGREEK8 8-bit character set
 514 ;; "HP-Hebrew8"                                    [36]
 515 ;;         HPHEBREW8 8-bit character set
 516 ;; "HP-Japanese15"                                 [36]
 517 ;;         HPJAPAN15 15-bit characer set,
 518 ;;         modified from industry defacto
 519 ;;         standard Shift-JIS
 520 ;; "HP-Kana8"                                      [36]
 521 ;;         HPKANA8 8-bit character set
 522 ;; "HP-Korean15"                                   [36]
 523 ;;         HPKOREAN15 15-bit character set
 524 ;; "HP-Roman8"                                     [36]
 525 ;;         HPROMAN8 8-bit character set
 526 ;; "HP-SChinese15"                                 [36]
 527 ;;         HPSCHINA15 15-bit character set for
 528 ;;         support of Simplified Chinese
 529 ;; "HP-TChinese15"                                 [36]
 530 ;;         HPTCHINA15 15-bit character set for
 531 ;;         support of Traditional Chinese
 532 ;; "HP-Turkish8"                                   [36]
 533 ;;         HPTURKISH8 8-bit character set
 534 ;; "IPSYS"                                         [59]
 535 ;;         registry prefix
 536 ;; "IPSYS.IE-1"                                    [59]
 537 ;; "ISO2022"<REG>"-"<ENC>                          [44]
 538 ;; "ISO646.1991-IRV"                               [107]
 539 ;;         ISO 646 International Reference Version
 540 ;; "ISO8859-1"                                     [15],[12]
 541 ;;         ISO Latin alphabet No. 1
 542 ;; "ISO8859-2"                                     [15],[12]
 543 ;;         ISO Latin alphabet No. 2
 544 ;; "ISO8859-3"                                     [15],[12]
 545 ;;         ISO Latin alphabet No. 3
 546 ;; "ISO8859-4"                                     [15],[12]
 547 ;;         ISO Latin alphabet No. 4
 548 ;; "ISO8859-5"                                     [15],[12]
 549 ;;         ISO Latin/Cyrillic alphabet
 550 ;; "ISO8859-6"                                     [15],[12]
 551 ;;         ISO Latin/Arabic alphabet
 552 ;; "ISO8859-7"                                     [15],[12]
 553 ;;         ISO Latin/Greek alphabet
 554 ;; "ISO8859-8"                                     [15],[12]
 555 ;;         ISO Latin/Hebrew alphabet
 556 ;; "ISO8859-9"                                     [15],[12]
 557 ;;         ISO Latin alphabet No. 5
 558 ;; "ISO8859-10"                                    [15],[12]
 559 ;;         ISO Latin alphabet No. 6
 560 ;; "ISO8859-13"                                    [15],[12]
 561 ;;         ISO Latin alphabet No. 7
 562 ;; "ISO8859-14"                                    [15],[12]
 563 ;;         ISO Latin alphabet No. 8
 564 ;; "ISO8859-15"                                    [15],[12]
 565 ;;         ISO Latin alphabet No. 9
 566 ;; "FCD8859-15"                                    [7]
 567 ;;         (deprecated)
 568 ;;         ISO Latin alphabet No. 9, Final Committee Draft
 569 ;; "ISO10646-1"                                    [133]
 570 ;;         Unicode Universal Multiple-Octet Coded Character Set
 571 ;; "ISO10646-MES"                                  [133]
 572 ;;         (deprecated)
 573 ;;         Unicode Minimum European Subset
 574 ;; "JISX0201.1976-0"                               [38],[12]
 575 ;;         8-Bit Alphanumeric-Katakana Code
 576 ;; "JISX0208.1983-0"                               [40],[12]
 577 ;;         Japanese Graphic Character Set,
 578 ;;         GL encoding
 579 ;; "JISX0208.1990-0"                               [71]
 580 ;;         Japanese Graphic Character Set,
 581 ;;         GL encoding
 582 ;; "JISX0208.1983-1"                               [40]
 583 ;;         (deprecated)
 584 ;;         Japanese Graphic Character Set,
 585 ;;         GR encoding
 586 ;; "JISX0212.1990-0"                               [72]
 587 ;;         Supplementary Japanese Graphic Character Set,
 588 ;;         GL encoding
 589 ;; "KOI8-R"                                        [119]
 590 ;;         Cyrillic alphabet
 591 ;; "KSC5601.1987-0"                                [41],[12]
 592 ;;         Korean Graphic Character Set,
 593 ;;         GL encoding
 594 ;; "KSC5601.1987-1"                                [41]
 595 ;;         (deprecated)
 596 ;;         Korean Graphic Character Set,
 597 ;;         GR encoding
 598 ;; "omron_CNS11643-0"                              [45]
 599 ;; "omron_CNS11643-1"                              [45]
 600 ;; "omron_BIG5-0"                                  [45]
 601 ;; "omron_BIG5-1"                                  [45]
 602 ;; "wn.tamil.1993"                                 [103]
 603
 604 (defun set-font-encoding (pattern charset)
 605   "Set arguments in `font-encoding-alist' (which see)."
 606   (let ((slot (assoc pattern font-encoding-alist)))
 607     (if slot
 608         (setcdr slot charset)
 609       (setq font-encoding-alist
 610             (cons (cons pattern charset) font-encoding-alist)))))
 611
 612 ;; Setting for suppressing XLoadQueryFont on big fonts.
 613 (setq x-pixel-size-width-font-regexp
 614       "gb2312\\|gbk\\|gb18030\\|jisx0208\\|ksc5601\\|cns11643\\|big5")
 615
 616 ;; These fonts require vertical centering.
 617 (setq vertical-centering-font-regexp
 618       "gb2312\\|gbk\\|gb18030\\|jisx0208\\|jisx0212\\|ksc5601\\|cns11643\\|big5")
 619
 620 ;; CDAC fonts are actually smaller than their design sizes.
 621 (setq face-font-rescale-alist
 622       '(("-cdac$" . 1.3)))
 623
 624 (defvar x-font-name-charset-alist nil
 625   "This variable has no meaning now.  Just kept for backward compatibility.")
 626
 627 ;;; XLFD (X Logical Font Description) format handler.
 628
 629 ;; Define XLFD's field index numbers.           ; field name
 630 (defconst xlfd-regexp-family-subnum 0)          ; FOUNDRY and FAMILY
 631 (defconst xlfd-regexp-weight-subnum 1)          ; WEIGHT_NAME
 632 (defconst xlfd-regexp-slant-subnum 2)           ; SLANT
 633 (defconst xlfd-regexp-swidth-subnum 3)          ; SETWIDTH_NAME
 634 (defconst xlfd-regexp-adstyle-subnum 4)         ; ADD_STYLE_NAME
 635 (defconst xlfd-regexp-pixelsize-subnum 5)       ; PIXEL_SIZE
 636 (defconst xlfd-regexp-pointsize-subnum 6)       ; POINT_SIZE
 637 (defconst xlfd-regexp-resx-subnum 7)            ; RESOLUTION_X
 638 (defconst xlfd-regexp-resy-subnum 8)            ; RESOLUTION_Y
 639 (defconst xlfd-regexp-spacing-subnum 8)         ; SPACING
 640 (defconst xlfd-regexp-avgwidth-subnum 10)       ; AVERAGE_WIDTH
 641 (defconst xlfd-regexp-registry-subnum 11)       ; REGISTRY and ENCODING
 642
 643 ;; Regular expression matching against a fontname which conforms to
 644 ;; XLFD (X Logical Font Description).  All fields in XLFD should be
 645 ;; not be omitted (but can be a wild card) to be matched.
 646 (defconst xlfd-tight-regexp
 647   "^\
 648 -\\([^-]*-[^-]*\\)-\\([^-]*\\)-\\([^-]*\\)-\\([^-]*\\)\
 649 -\\([^-]*\\)-\\([^-]*\\)-\\([^-]*\\)-\\([^-]*\\)-\\([^-]*\\)\
 650 -\\([^-]*\\)-\\([^-]*\\)-\\([^-]*-[^-]*\\)$")
 651
 652 ;; Regular expression matching against a fontname which conforms to
 653 ;; XLFD (X Logical Font Description).  All fields in XLFD from FOUNDRY
 654 ;; to ADSTYLE, REGISTRY, and ENCODING should be not be omitted (but
 655 ;; can be a wild card) to be matched.
 656 (defconst xlfd-style-regexp
 657   "^\
 658 -\\([^-]*-[^-]*\\)-\\([^-]*\\)-\\([^-]*\\)-\\([^-]*\\)-\\([^-]*\\)-.*\
 659 -\\([^-]*-[^-]*\\)$")
 660
 661 ;; List of field numbers of XLFD whose values are numeric.
 662 (defconst xlfd-regexp-numeric-subnums
 663   (list xlfd-regexp-pixelsize-subnum    ;5
 664         xlfd-regexp-pointsize-subnum    ;6
 665         xlfd-regexp-resx-subnum         ;7
 666         xlfd-regexp-resy-subnum         ;8
 667         xlfd-regexp-avgwidth-subnum     ;10
 668         ))
 669
 670 (defun x-decompose-font-name (pattern)
 671   "Decompose PATTERN into XLFD fields and return a vector of the fields.
 672 The length of the vector is 12.
 673 The FOUNDRY and FAMILY fields are concatinated and stored in the first
 674 element of the vector.
 675 The REGISTRY and ENCODING fields are concatinated and stored in the last
 676 element of the vector.
 677
 678 Return nil if PATTERN doesn't conform to XLFD."
 679   (if (string-match xlfd-tight-regexp pattern)
 680       (let ((xlfd-fields (make-vector 12 nil)))
 681         (dotimes (i 12)
 682           (aset xlfd-fields i (match-string (1+ i) pattern)))
 683         (dotimes (i 12)
 684           (if (string-match "^[*-]+$" (aref xlfd-fields i))
 685               (aset xlfd-fields i nil)))
 686         xlfd-fields)))
 687
 688 (defun x-compose-font-name (fields &optional reduce)
 689   "Compose X fontname from FIELDS.
 690 FIELDS is a vector of XLFD fields, of length 12.
 691 If a field is nil, wild-card letter `*' is embedded.
 692 Optional argument REDUCE exists just for backward compatibility,
 693 and is always ignored."
 694   (concat "-" (mapconcat (lambda (x) (or x "*")) fields "-")))
 695
 696
 697 (defun x-must-resolve-font-name (xlfd-fields)
 698   "Like `x-resolve-font-name', but always return a font name.
 699 XLFD-FIELDS is a vector of XLFD (X Logical Font Description) fields.
 700 If no font matching XLFD-FIELDS is available, successively replace
 701 parts of the font name pattern with \"*\" until some font is found.
 702 Value is name of that font."
 703   (let ((ascii-font nil) (index 0))
 704     (while (and (null ascii-font) (<= index xlfd-regexp-registry-subnum))
 705       (let ((pattern (x-compose-font-name xlfd-fields)))
 706         (condition-case nil
 707             (setq ascii-font (x-resolve-font-name pattern))
 708           (error
 709            (message "Warning: no fonts matching `%s' available" pattern)
 710            (aset xlfd-fields index "*")
 711            (setq index (1+ index))))))
 712     (unless ascii-font
 713       (error "No fonts found"))
 714     ascii-font))
 715
 716
 717 (defun x-complement-fontset-spec (default-spec fontlist)
 718   "Complement elements of FONTLIST based on DEFAULT-SPEC.
 719 DEFAULT-SPEC is a font-spec object providing default font properties.
 720 FONTLIST is an alist of script names vs the corresponding font names.
 721
 722 The font names are parsed and unspecified font properties are
 723 given from DEFAULT-SPEC."
 724   (let ((prop-list '(:foundry :family :weight :slant :width :adstyle :size)))
 725     (dolist (elt fontlist)
 726       (let ((spec (font-spec :name (cadr elt))))
 727         (dolist (prop prop-list)
 728           (let ((val (font-get spec prop)))
 729             (or val
 730                 (font-put spec prop (font-get default-spec prop)))))
 731         (setcar (cdr elt) spec)))
 732     fontlist))
 733
 734 (defun fontset-name-p (fontset)
 735   "Return non-nil if FONTSET is valid as fontset name.
 736 A valid fontset name should conform to XLFD (X Logical Font Description)
 737 with \"fontset\" in `<CHARSET_REGISTRY>' field."
 738   (and (string-match xlfd-tight-regexp fontset)
 739        (string= (match-string (1+ xlfd-regexp-registry-subnum) fontset)
 740                 "fontset")))
 741
 742 (defun generate-fontset-menu ()
 743   "Return list to be appended to `x-fixed-font-alist'.
 744 Done when `mouse-set-font' is called."
 745   (let (l)
 746     (dolist (fontset (fontset-list))
 747       (or (string-match "fontset-default$" fontset)
 748           (push (list (fontset-plain-name fontset) fontset) l)))
 749     (cons "Fontset"
 750           (sort l #'(lambda (x y) (string< (car x) (car y)))))))
 751
 752 (defun fontset-plain-name (fontset)
 753   "Return a plain and descriptive name of FONTSET."
 754   (if (not (setq fontset (query-fontset fontset)))
 755       (error "Invalid fontset: %s" fontset))
 756   (let ((xlfd-fields (x-decompose-font-name fontset)))
 757     (if xlfd-fields
 758         (let ((family (aref xlfd-fields xlfd-regexp-family-subnum))
 759               (weight (aref xlfd-fields xlfd-regexp-weight-subnum))
 760               (slant  (aref xlfd-fields xlfd-regexp-slant-subnum))
 761               (swidth (aref xlfd-fields xlfd-regexp-swidth-subnum))
 762               (size   (aref xlfd-fields xlfd-regexp-pixelsize-subnum))
 763               (nickname (aref xlfd-fields xlfd-regexp-registry-subnum))
 764               name)
 765           (if (not (string-match "^fontset-\\(.*\\)$" nickname))
 766               (setq nickname family)
 767             (setq nickname (match-string 1 nickname)))
 768           (if (and size (> (string-to-number size) 0))
 769               (setq name (format "%s: %s-dot" nickname size))
 770             (setq name nickname))
 771           (and weight
 772                (cond ((string-match "^medium$" weight)
 773                       (setq name (concat name " " "medium")))
 774                      ((string-match "^bold$\\|^demibold$" weight)
 775                       (setq name (concat name " " weight)))))
 776           (and slant
 777                (cond ((string-match "^i$" slant)
 778                       (setq name (concat name " " "italic")))
 779                      ((string-match "^o$" slant)
 780                       (setq name (concat name " " "slant")))
 781                      ((string-match "^ri$" slant)
 782                       (setq name (concat name " " "reverse italic")))
 783                      ((string-match "^ro$" slant)
 784                       (setq name (concat name " " "reverse slant")))))
 785           name)
 786       fontset)))
 787
 788 (defvar charset-script-alist
 789   '((ascii . latin)
 790     (latin-iso8859-1 . latin)
 791     (latin-iso8859-2 . latin)
 792     (latin-iso8859-3 . latin)
 793     (latin-iso8859-4 . latin)
 794     (latin-iso8859-9 . latin)
 795     (latin-iso8859-10 . latin)
 796     (latin-iso8859-13 . latin)
 797     (latin-iso8859-14 . latin)
 798     (latin-iso8859-15 . latin)
 799     (latin-iso8859-16 . latin)
 800     (latin-jisx0201 . latin)
 801     (thai-tis620 . thai)
 802     (cyrillic-iso8859-5 . cyrillic)
 803     (arabic-iso8859-6 . arabic)
 804     (greek-iso8859-7 . latin)
 805     (hebrew-iso8859-8 . latin)
 806     (katakana-jisx0201 . kana)
 807     (chinese-gb2312 . han)
 808     (chinese-gbk . han)
 809     (gb18030-2-byte . han)
 810     (gb18030-4-byte-bmp . han)
 811     (gb18030-4-byte-ext-1 . han)
 812     (gb18030-4-byte-ext-2 . han)
 813     (gb18030-4-byte-smp . han)
 814     (chinese-big5-1 . han)
 815     (chinese-big5-2 . han)
 816     (chinese-cns11643-1 . han)
 817     (chinese-cns11643-2 . han)
 818     (chinese-cns11643-3 . han)
 819     (chinese-cns11643-4 . han)
 820     (chinese-cns11643-5 . han)
 821     (chinese-cns11643-6 . han)
 822     (chinese-cns11643-7 . han)
 823     (japanese-jisx0208 . han)
 824     (japanese-jisx0208-1978 . han)
 825     (japanese-jisx0212 . han)
 826     (japanese-jisx0213-1 . han)
 827     (japanese-jisx0213-2 . han)
 828     (korean-ksc5601 . hangul)
 829     (chinese-sisheng . bopomofo)
 830     (vietnamese-viscii-lower . latin)
 831     (vietnamese-viscii-upper . latin)
 832     (arabic-digit . arabic)
 833     (arabic-1-column . arabic)
 834     (arabic-2-column . arabic)
 835     (indian-is13194 . devanagari)
 836     (indian-glyph . devanagari)
 837     (indian-1-column . devanagari)
 838     (indian-2-column . devanagari)
 839     (tibetan-1-column . tibetan))
 840   "Alist of charsets vs the corresponding most appropriate scripts.
 841
 842 This alist is used by the function `create-fontset-from-fontset-spec'
 843 to map charsets to scripts.")
 844
 845 (defun create-fontset-from-fontset-spec (fontset-spec
 846                                          &optional style-variant noerror)
 847   "Create a fontset from fontset specification string FONTSET-SPEC.
 848 FONTSET-SPEC is a string of the format:
 849         FONTSET-NAME,SCRIPT0:FONT0,SCRIPT1:FONT1, ...
 850 Any number of SPACE, TAB, and NEWLINE can be put before and after commas.
 851
 852 When a frame uses the fontset as the `font' parameter, the frame's
 853 default font name is derived from FONTSET-NAME by substituting
 854 \"iso8859-1\" for the tail part \"fontset-XXX\".  But, if SCRIPT-NAMEn
 855 is \"ascii\", use the corresponding FONT-NAMEn as the default font
 856 name.
 857
 858 Optional 2nd and 3rd arguments exist just for backward compatibility,
 859 and are ignored.
 860
 861 It returns a name of the created fontset.
 862
 863 For backward compatibility, SCRIPT-NAME may be a charset name, in
 864 which case, the corresponding script is decided by the variable
 865 `charset-script-alist' (which see)."
 866   (or (string-match "^[^,]+" fontset-spec)
 867       (error "Invalid fontset spec: %s" fontset-spec))
 868   (let ((idx (match-end 0))
 869         (name (match-string 0 fontset-spec))
 870         default-spec target script fontlist)
 871     (or (string-match xlfd-tight-regexp name)
 872         (error "Fontset name \"%s\" not conforming to XLFD" name))
 873     (setq default-spec (font-spec :name name))
 874     ;; At first, extract pairs of charset and fontname from FONTSET-SPEC.
 875     (while (string-match "[, \t\n]*\\([^:]+\\):[ \t]*\\([^,]+\\)"
 876                          fontset-spec idx)
 877       (setq idx (match-end 0))
 878       (setq target (intern (match-string 1 fontset-spec)))
 879       (cond ((or (eq target 'ascii)
 880                  (memq target (char-table-extra-slot char-script-table 0)))
 881              (push (list target (match-string 2 fontset-spec)) fontlist))
 882             ((setq script (cdr (assq target charset-script-alist)))
 883              (push (list script (match-string 2 fontset-spec)) fontlist))
 884             ((charsetp target)
 885              (push (list target (match-string 2 fontset-spec)) fontlist))))
 886
 887     ;; Complement FONTLIST.
 888     (setq fontlist (x-complement-fontset-spec default-spec fontlist))
 889
 890     ;; Create a fontset.
 891     (new-fontset name (nreverse fontlist))))
 892
 893 (defun create-fontset-from-ascii-font (font &optional resolved-font
 894                                             fontset-name)
 895   "Create a fontset from an ASCII font FONT.
 896
 897 Optional 2nd arg RESOLVED-FONT is a resolved name of FONT.
 898 If omitted, `x-resolve-font-name' is called to get the resolved name.
 899 At this time, if FONT is not available, an error is signaled.
 900
 901 Optional 3rd arg FONTSET-NAME is a string to be used in
 902 `<CHARSET_ENCODING>' fields of a new fontset name.  If it is omitted,
 903 an appropriate name is generated automatically.
 904
 905 It returns a name of the created fontset."
 906   (setq font (downcase font))
 907   (setq resolved-font
 908         (downcase (or resolved-font (x-resolve-font-name font))))
 909   (let ((xlfd (x-decompose-font-name resolved-font))
 910         fontset)
 911     (if fontset-name
 912         (setq fontset-name (downcase fontset-name))
 913       (if (query-fontset "fontset-startup")
 914           (setq fontset-name
 915                 (subst-char-in-string
 916                  ?- ?_ (aref xlfd xlfd-regexp-registry-subnum) t))
 917         (setq fontset-name "startup")))
 918     (aset xlfd xlfd-regexp-registry-subnum
 919           (format "fontset-%s" fontset-name))
 920     (setq fontset (x-compose-font-name xlfd))
 921     (or (query-fontset fontset)
 922         (create-fontset-from-fontset-spec (concat fontset ", ascii:" font)))))
 923
 924 \f
 925 ;; Create standard fontset from 16 dots fonts which are the most widely
 926 ;; installed fonts.  Fonts for Chinese-GB, Korean, and Chinese-CNS are
 927 ;; specified here because FAMILY of those fonts are not "fixed" in
 928 ;; many cases.
 929 (defvar standard-fontset-spec
 930   (purecopy "-*-fixed-medium-r-normal-*-16-*-*-*-*-*-fontset-standard")
 931   "String of fontset spec of the standard fontset.
 932 You have the biggest chance to display international characters
 933 with correct glyphs by using the standard fontset.
 934 See the documentation of `create-fontset-from-fontset-spec' for the format.")
 935
 936
 937 ;; Create fontsets from X resources of the name `fontset-N (class
 938 ;; Fontset-N)' where N is integer 0, 1, ...
 939 ;; The values of the resources the string of the same format as
 940 ;; `standard-fontset-spec'.
 941
 942 (defun create-fontset-from-x-resource ()
 943   (let ((idx 0)
 944         fontset-spec)
 945     (while (setq fontset-spec (x-get-resource (format "fontset-%d" idx)
 946                                               (format "Fontset-%d" idx)))
 947       (condition-case nil
 948           (create-fontset-from-fontset-spec fontset-spec t 'noerror)
 949         (error (message "Fontset-%d: invalid specification in X resource" idx)))
 950       (setq idx (1+ idx)))))
 951
 952 ;;
 953 (provide 'fontset)
 954
 955 ;; arch-tag: bb53e629-0234-403c-950e-551e61554849
 956 ;;; fontset.el ends here