src/charset.h

   1 /* Header for multilingual character handler.
   2    Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN.
   3    Licensed to the Free Software Foundation.
   4
   5 This file is part of GNU Emacs.
   6
   7 GNU Emacs is free software; you can redistribute it and/or modify
   8 it under the terms of the GNU General Public License as published by
   9 the Free Software Foundation; either version 2, or (at your option)
  10 any later version.
  11
  12 GNU Emacs is distributed in the hope that it will be useful,
  13 but WITHOUT ANY WARRANTY; without even the implied warranty of
  14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15 GNU General Public License for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with GNU Emacs; see the file COPYING.  If not, write to
  19 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  20 Boston, MA 02111-1307, USA.  */
  21
  22 #ifndef _CHARSET_H
  23 #define _CHARSET_H
  24
  25 /*** GENERAL NOTE on CHARACTER SET (CHARSET) ***
  26
  27   A character set ("charset" hereafter) is a meaningful collection
  28   (i.e. language, culture, functionality, etc) of characters.  Emacs
  29   handles multiple charsets at once.  Each charset corresponds to one
  30   of ISO charsets (except for a special charset for composition
  31   characters).  Emacs identifies a charset by a unique identification
  32   number, whereas ISO identifies a charset by a triplet of DIMENSION,
  33   CHARS and FINAL-CHAR.  So, hereafter, just saying "charset" means an
  34   identification number (integer value).
  35
  36   The value range of charset is 0x00, 0x80..0xFE.  There are four
  37   kinds of charset depending on DIMENSION (1 or 2) and CHARS (94 or
  38   96).  For instance, a charset of DIMENSION2_CHARS94 contains 94x94
  39
  40
  41   Within Emacs Lisp, a charset is treated as a symbol which has a
  42   property `charset'.  The property value is a vector containing
  43   various information about the charset.  For readability of C codes,
  44   we use the following convention on C variable names:
  45         charset_symbol: Emacs Lisp symbol of a charset
  46         charset_id: Emacs Lisp integer of an identification number of a charset
  47         charset: C integer of an identification number of a charset
  48
  49   Each charset (except for ASCII) is assigned a base leading-code
  50   (range 0x80..0x9D).  In addition, a charset of greater than 0xA0
  51   (whose base leading-code is 0x9A..0x9D) is assigned an extended
  52   leading-code (range 0xA0..0xFE).  In this case, each base
  53   leading-code specify the allowable range of extended leading-code as
  54   shown in the table below.  A leading-code is used to represent a
  55   character in Emacs' buffer and string.
  56
  57   We call a charset which has extended leading-code as "private
  58   charset" because those are mainly for a charset which is not
  59   registered by ISO.  On the contrary, we call a charset which does
  60   not have extended leading-code as "official charset".
  61
  62   ---------------------------------------------------------------------------
  63   charset       dimension        base leading-code      extended leading-code
  64   ---------------------------------------------------------------------------
  65   0x00          official dim1    -- none --             -- none --
  66                 (ASCII)
  67   0x01..0x7F    --never used--
  68   0x80          COMPOSITION      same as charset        -- none --
  69   0x81..0x8F    official dim1    same as charset        -- none --
  70   0x90..0x99    official dim2    same as charset        -- none --
  71   0x9A..0x9F    --never used--
  72   0xA0..0xDF    private dim1        0x9A                same as charset
  73                 of 1-column width
  74   0xE0..0xEF    private dim1        0x9B                same as charset
  75                 of 2-column width
  76   0xF0..0xF4    private dim2        0x9C                same as charset
  77                 of 1-column width
  78   0xF5..0xFE    private dim2        0x9D                same as charset
  79                 of 2-column width
  80   0xFF          --never used--
  81   ---------------------------------------------------------------------------
  82
  83   In the table, "COMPOSITION" means a charset for a composite
  84   character which is a character composed from several (up to 16)
  85   non-composite characters (components).  Although a composite
  86   character can contain components of many charsets, a composite
  87   character itself belongs to the charset CHARSET-COMPOSITION.  See
  88   the document "GENERAL NOTE on COMPOSITE CHARACTER" below for more
  89   detail.
  90
  91 */
  92
  93 /* Definition of special leading-codes.  */
  94 /* Base leading-code.  */
  95 /* Special leading-code followed by components of a composite character.  */
  96 #define LEADING_CODE_COMPOSITION        0x80
  97 /* Leading-code followed by extended leading-code.  */
  98 #define LEADING_CODE_PRIVATE_11 0x9A /* for private DIMENSION1 of 1-column */
  99 #define LEADING_CODE_PRIVATE_12 0x9B /* for private DIMENSION1 of 2-column */
 100 #define LEADING_CODE_PRIVATE_21 0x9C /* for private DIMENSION2 of 1-column */
 101 #define LEADING_CODE_PRIVATE_22 0x9D /* for private DIMENSION2o f 2-column */
 102
 103 /* Extended leading-code.  */
 104 /* Start of each extended leading-codes.  */
 105 #define LEADING_CODE_EXT_11 0xA0 /* follows LEADING_CODE_PRIVATE_11 */
 106 #define LEADING_CODE_EXT_12 0xE0 /* follows LEADING_CODE_PRIVATE_12 */
 107 #define LEADING_CODE_EXT_21 0xF0 /* follows LEADING_CODE_PRIVATE_21 */
 108 #define LEADING_CODE_EXT_22 0xF5 /* follows LEADING_CODE_PRIVATE_22 */
 109 /* Maximum value of extended leading-codes.  */
 110 #define LEADING_CODE_EXT_MAX 0xFE
 111
 112 /* Definition of minimum/maximum charset of each DIMENSION.  */
 113 #define MIN_CHARSET_OFFICIAL_DIMENSION1 0x81
 114 #define MAX_CHARSET_OFFICIAL_DIMENSION1 0x8F
 115 #define MIN_CHARSET_OFFICIAL_DIMENSION2 0x90
 116 #define MAX_CHARSET_OFFICIAL_DIMENSION2 0x99
 117 #define MIN_CHARSET_PRIVATE_DIMENSION1  LEADING_CODE_EXT_11
 118 #define MIN_CHARSET_PRIVATE_DIMENSION2  LEADING_CODE_EXT_21
 119
 120 /* Maximum value of overall charset identification number.  */
 121 #define MAX_CHARSET 0xFE
 122
 123 /* Definition of special charsets.  */
 124 #define CHARSET_ASCII           0
 125 #define CHARSET_COMPOSITION     0x80
 126
 127 extern int charset_ascii;       /* ASCII */
 128 extern int charset_composition; /* for a composite character */
 129 extern int charset_latin_iso8859_1; /* ISO8859-1 (Latin-1) */
 130 extern int charset_jisx0208_1978; /* JISX0208.1978 (Japanese Kanji old set) */
 131 extern int charset_jisx0208;    /* JISX0208.1983 (Japanese Kanji) */
 132 extern int charset_katakana_jisx0201; /* JISX0201.Kana (Japanese Katakana) */
 133 extern int charset_latin_jisx0201; /* JISX0201.Roman (Japanese Roman) */
 134 extern int charset_big5_1;      /* Big5 Level 1 (Chinese Traditional) */
 135 extern int charset_big5_2;      /* Big5 Level 2 (Chinese Traditional) */
 136
 137 /* Check if STR points the head of multi-byte form, i.e. *STR is an
 138    ASCII character or a base leading-code.  */
 139 #define CHAR_HEAD_P(str) ((unsigned char) *(str) < 0xA0)
 140
 141 /*** GENERAL NOTE on CHARACTER REPRESENTATION ***
 142
 143   At first, the term "character" or "char" is used for a multilingual
 144   character (of course, including ASCII character), not for a byte in
 145   computer memory.  We use the term "code" or "byte" for the latter
 146   case.
 147
 148   A character is identified by charset and one or two POSITION-CODEs.
 149   POSITION-CODE is the position of the character in the charset.  A
 150   character of DIMENSION1 charset has one POSITION-CODE: POSITION-CODE-1.
 151   A character of DIMENSION2 charset has two POSITION-CODE:
 152   POSITION-CODE-1 and POSITION-CODE-2.  The code range of
 153   POSITION-CODE is 0x20..0x7F.
 154
 155   Emacs has two kinds of representation of a character: multi-byte
 156   form (for buffer and string) and single-word form (for character
 157   object in Emacs Lisp).  The latter is called "character code" here
 158   after.  Both representation encode the information of charset and
 159   POSITION-CODE but in a different way (for instance, MSB of
 160   POSITION-CODE is set in multi-byte form).
 161
 162   For details of multi-byte form, see the section "2. Emacs internal
 163   format handlers" of `coding.c'.
 164
 165   Emacs uses 19 bits for a character code.  The bits are divided into
 166   3 fields: FIELD1(5bits):FIELD2(7bits):FIELD3(7bits).
 167
 168   A character code of DIMENSION1 character uses FIELD2 to hold charset
 169   and FIELD3 to hold POSITION-CODE-1.  A character code of DIMENSION2
 170   character uses FIELD1 to hold charset, FIELD2 and FIELD3 to hold
 171   POSITION-CODE-1 and POSITION-CODE-2 respectively.
 172
 173   More precisely...
 174
 175   FIELD2 of DIMENSION1 character (except for ASCII) is "charset - 0x70".
 176   This is to make all character codes except for ASCII greater than
 177   256 (ASCII's FIELD2 is 0).  So, the range of FIELD2 of DIMENSION1
 178   character is 0 or 0x11..0x7F.
 179
 180   FIELD1 of DIMENSION2 character is "charset - 0x8F" for official
 181   charset and "charset - 0xE0" for private charset.  So, the range of
 182   FIELD1 of DIMENSION2 character is 0x01..0x1E.
 183
 184   -----------------------------------------------------------------------
 185   charset       FIELD1 (5-bit)      FIELD2 (7-bit)      FIELD3 (7-bit)
 186   -----------------------------------------------------------------------
 187   ASCII         0                   0                   POSITION-CODE-1
 188   DIMENSION1    0                   charset - 0x70      POSITION-CODE-1
 189   DIMENSION2(o) charset - 0x8F      POSITION-CODE-1     POSITION-CODE-2
 190   DIMENSION2(p) charset - 0xE0      POSITION-CODE-1     POSITION-CODE-2
 191   -----------------------------------------------------------------------
 192   "(o)": official, "(p)": private
 193   -----------------------------------------------------------------------
 194
 195 */
 196
 197 /*** GENERAL NOTE on COMPOSITE CHARACTER ***
 198
 199   A composite character is a character composed from several (up to
 200   16) non-composite characters (components).  Although each components
 201   can belong to any charset, a composite character itself belongs to
 202   the charset `charset-composition' and is assigned a special
 203   leading-code `LEADING_CODE_COMPOSITION' for multi-byte form.  See
 204   the document "2. Emacs internal format handlers" in `coding.c' for
 205   more detail about multi-byte form.
 206
 207   A character code of composite character has special format.  In the
 208   above document, FIELD1 of a composite character is 0x1F.  Each
 209   composite character is assigned a sequential number CMPCHAR-ID.
 210   FIELD2 and FIELD3 are combined to make 14bits field for holding
 211   CMPCHAR-ID, which means that Emacs can handle at most 2^14 (= 16384)
 212   composite characters at once.
 213
 214   -----------------------------------------------------------------------
 215   charset               FIELD1 (5-bit)      FIELD2&3 (14-bit)
 216   -----------------------------------------------------------------------
 217   CHARSET-COMPOSITION   0x1F                CMPCHAR-ID
 218   -----------------------------------------------------------------------
 219
 220   Emacs assigns CMPCHAR-ID to a composite character only when it
 221   requires the character code of the composite character (e.g. while
 222   displaying the composite character).
 223
 224 */
 225
 226 /* Masks of each field of character code.  */
 227 #define CHAR_FIELD1_MASK (0x1F << 14)
 228 #define CHAR_FIELD2_MASK (0x7F << 7)
 229 #define CHAR_FIELD3_MASK 0x7F
 230
 231 /* Macros to access each field of character C.  */
 232 #define CHAR_FIELD1(c) (((c) & CHAR_FIELD1_MASK) >> 14)
 233 #define CHAR_FIELD2(c) (((c) & CHAR_FIELD2_MASK) >> 7)
 234 #define CHAR_FIELD3(c) ((c) & CHAR_FIELD3_MASK)
 235
 236 /* Minimum character code of character of each DIMENSION.  */
 237 #define MIN_CHAR_OFFICIAL_DIMENSION1 \
 238   ((MIN_CHARSET_OFFICIAL_DIMENSION1 - 0x70) << 7)
 239 #define MIN_CHAR_PRIVATE_DIMENSION1 \
 240   ((MIN_CHARSET_PRIVATE_DIMENSION1 - 0x70) << 7)
 241 #define MIN_CHAR_OFFICIAL_DIMENSION2 \
 242   ((MIN_CHARSET_OFFICIAL_DIMENSION2 - 0x8F) << 14)
 243 #define MIN_CHAR_PRIVATE_DIMENSION2 \
 244   ((MIN_CHARSET_PRIVATE_DIMENSION2 - 0xE0) << 14)
 245 #define MIN_CHAR_COMPOSITION \
 246   (0x1F << 14)
 247
 248 /* 1 if C is an ASCII character, else 0.  */
 249 #define SINGLE_BYTE_CHAR_P(c) ((c) < 0x100)
 250 /* 1 if C is an composite character, else 0.  */
 251 #define COMPOSITE_CHAR_P(c) ((c) >= MIN_CHAR_COMPOSITION)
 252
 253 /* A char-table containing information of each character set.
 254
 255    Unlike ordinary char-tables, this doesn't contain any nested table.
 256    Only the top level elements are used.  Each element is a vector of
 257    the following information:
 258         CHARSET-ID, BYTES, DIMENSION, CHARS, WIDTH, DIRECTION,
 259         LEADING-CODE-BASE, LEADING-CODE-EXT,
 260         ISO-FINAL-CHAR, ISO-GRAPHIC-PLANE,
 261         REVERSE-CHARSET, SHORT-NAME, LONG-NAME, DESCRIPTION,
 262         PLIST.
 263
 264    CHARSET-ID (integer) is the identification number of the charset.
 265
 266    BYTE (integer) is the length of multi-byte form of a character in
 267    the charset: one of 1, 2, 3, and 4.
 268
 269    DIMENSION (integer) is the number of bytes to represent a character: 1 or 2.
 270
 271    CHARS (integer) is the number of characters in a dimension: 94 or 96.
 272
 273    WIDTH (integer) is the number of columns a character in the charset
 274    occupies on the screen: one of 0, 1, and 2.
 275
 276    DIRECTION (integer) is the rendering direction of characters in the
 277    charset when rendering.  If 0, render from right to left, else
 278    render from left to right.
 279
 280    LEADING-CODE-BASE (integer) is the base leading-code for the
 281    charset.
 282
 283    LEADING-CODE-EXT (integer) is the extended leading-code for the
 284    charset.  All charsets of less than 0xA0 has the value 0.
 285
 286    ISO-FINAL-CHAR (character) is the final character of the
 287    corresponding ISO 2022 charset.
 288
 289    ISO-GRAPHIC-PLANE (integer) is the graphic plane to be invoked
 290    while encoding to variants of ISO 2022 coding system, one of the
 291    following: 0/graphic-plane-left(GL), 1/graphic-plane-right(GR).
 292
 293    REVERSE-CHARSET (integer) is the charset which differs only in
 294    LEFT-TO-RIGHT value from the charset.  If there's no such a
 295    charset, the value is -1.
 296
 297    SHORT-NAME (string) is the short name to refer to the charset.
 298
 299    LONG-NAME (string) is the long name to refer to the charset.
 300
 301    DESCRIPTION (string) is the description string of the charset.
 302
 303    PLIST (property list) may contain any type of information a user
 304    want to put and get by functions `put-charset-property' and
 305    `get-charset-property' respectively.  */
 306 extern Lisp_Object Vcharset_table;
 307
 308 /* Macros to access various information of CHARSET in Vcharset_table.
 309    We provide these macros for efficiency.  No range check of CHARSET.  */
 310
 311 /* Return entry of CHARSET (lisp integer) in Vcharset_table.  */
 312 #define CHARSET_TABLE_ENTRY(charset)                                    \
 313   XCHAR_TABLE (Vcharset_table)->contents[((charset) == CHARSET_ASCII    \
 314                                           ? 0 : (charset) + 128)]
 315
 316 /* Return information INFO-IDX of CHARSET.  */
 317 #define CHARSET_TABLE_INFO(charset, info_idx) \
 318   XVECTOR (CHARSET_TABLE_ENTRY (charset))->contents[info_idx]
 319
 320 #define CHARSET_ID_IDX (0)
 321 #define CHARSET_BYTES_IDX (1)
 322 #define CHARSET_DIMENSION_IDX (2)
 323 #define CHARSET_CHARS_IDX (3)
 324 #define CHARSET_WIDTH_IDX (4)
 325 #define CHARSET_DIRECTION_IDX (5)
 326 #define CHARSET_LEADING_CODE_BASE_IDX (6)
 327 #define CHARSET_LEADING_CODE_EXT_IDX (7)
 328 #define CHARSET_ISO_FINAL_CHAR_IDX (8)
 329 #define CHARSET_ISO_GRAPHIC_PLANE_IDX (9)
 330 #define CHARSET_REVERSE_CHARSET_IDX (10)
 331 #define CHARSET_SHORT_NAME_IDX (11)
 332 #define CHARSET_LONG_NAME_IDX (12)
 333 #define CHARSET_DESCRIPTION_IDX (13)
 334 #define CHARSET_PLIST_IDX (14)
 335 /* Size of a vector of each entry of Vcharset_table.  */
 336 #define CHARSET_MAX_IDX (15)
 337
 338 /* And several more macros to be used frequently.  */
 339 #define CHARSET_BYTES(charset) \
 340   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_BYTES_IDX))
 341 #define CHARSET_DIMENSION(charset) \
 342   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_DIMENSION_IDX))
 343 #define CHARSET_CHARS(charset) \
 344   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_CHARS_IDX))
 345 #define CHARSET_WIDTH(charset) \
 346   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_WIDTH_IDX))
 347 #define CHARSET_DIRECTION(charset) \
 348   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX))
 349 #define CHARSET_LEADING_CODE_BASE(charset) \
 350   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_BASE_IDX))
 351 #define CHARSET_LEADING_CODE_EXT(charset) \
 352   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_EXT_IDX))
 353 #define CHARSET_ISO_FINAL_CHAR(charset) \
 354   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_ISO_FINAL_CHAR_IDX))
 355 #define CHARSET_ISO_GRAPHIC_PLANE(charset) \
 356   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_ISO_GRAPHIC_PLANE_IDX))
 357 #define CHARSET_REVERSE_CHARSET(charset) \
 358   XINT (CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX))
 359
 360 /* Macros to specify direction of a charset.  */
 361 #define CHARSET_DIRECTION_LEFT_TO_RIGHT 0
 362 #define CHARSET_DIRECTION_RIGHT_TO_LEFT 1
 363
 364 /* A vector of charset symbol indexed by charset-id.  This is used
 365    only for returning charset symbol from C functions.  */
 366 extern Lisp_Object Vcharset_symbol_table;
 367
 368 /* Return symbol of CHARSET.  */
 369 #define CHARSET_SYMBOL(charset) \
 370   XVECTOR (Vcharset_symbol_table)->contents[charset]
 371
 372 /* 1 if CHARSET is valid, else 0.  */
 373 #define CHARSET_VALID_P(charset)                                         \
 374   ((charset) == 0                                                        \
 375    || ((charset) >= 0x80 && (charset) <= MAX_CHARSET_OFFICIAL_DIMENSION2) \
 376    || ((charset) >= MIN_CHARSET_PRIVATE_DIMENSION1 && (charset) <= MAX_CHARSET))
 377
 378 /* 1 if CHARSET is already defined, else 0.  */
 379 #define CHARSET_DEFINED_P(charset)                      \
 380   (((charset) >= 0) && ((charset) <= MAX_CHARSET)       \
 381    && !NILP (CHARSET_TABLE_ENTRY (charset)))
 382
 383 /* Since the information CHARSET-BYTES and CHARSET-WIDTH of
 384    Vcharset_table can be retrieved only from the first byte of
 385    multi-byte form (an ASCII code or a base leading-code), we provide
 386    here tables to be used by macros BYTES_BY_CHAR_HEAD and
 387    WIDTH_BY_CHAR_HEAD for faster information retrieval.  */
 388 extern int bytes_by_char_head[256];
 389 extern int width_by_char_head[256];
 390
 391 #define BYTES_BY_CHAR_HEAD(char_head) bytes_by_char_head[char_head]
 392 #define WIDTH_BY_CHAR_HEAD(char_head) width_by_char_head[char_head]
 393
 394 /* Charset of the character C.  */
 395 #define CHAR_CHARSET(c)                         \
 396   (SINGLE_BYTE_CHAR_P (c)                       \
 397    ? CHARSET_ASCII                              \
 398    : ((c) < MIN_CHAR_OFFICIAL_DIMENSION2        \
 399       ? CHAR_FIELD2 (c) + 0x70                  \
 400       : ((c) < MIN_CHAR_PRIVATE_DIMENSION2      \
 401          ? CHAR_FIELD1 (c) + 0x8F               \
 402          : ((c) < MIN_CHAR_COMPOSITION          \
 403             ? CHAR_FIELD1 (c) + 0xE0            \
 404             : CHARSET_COMPOSITION))))
 405
 406 /* Return charset at the place pointed by P.  */
 407 #define CHARSET_AT(p)                           \
 408   (*(p) < 0x80                                  \
 409    ? CHARSET_ASCII                              \
 410    : (*(p) == LEADING_CODE_COMPOSITION          \
 411       ? CHARSET_COMPOSITION                     \
 412       : (*(p) < LEADING_CODE_PRIVATE_11         \
 413          ? (int)*(p)                            \
 414          : (*(p) <= LEADING_CODE_PRIVATE_22     \
 415             ? (int)*((p) + 1)                   \
 416             : -1))))
 417
 418 /* Same as `CHARSET_AT ()' but perhaps runs faster because of an
 419    additional argument C which is the code (byte) at P.  */
 420 #define FIRST_CHARSET_AT(p, c)                  \
 421   ((c) < 0x80                                   \
 422    ? CHARSET_ASCII                              \
 423    : ((c) == LEADING_CODE_COMPOSITION           \
 424       ? CHARSET_COMPOSITION                     \
 425       : ((c) < LEADING_CODE_PRIVATE_11          \
 426          ? (int)(c)                             \
 427          : ((c) <= LEADING_CODE_PRIVATE_22      \
 428             ? (int)*((p) + 1)                   \
 429             : -1))))
 430
 431 /* Check if two characters C1 and C2 belong to the same charset.
 432    Always return 0 for composite characters.  */
 433 #define SAME_CHARSET_P(c1, c2)                                  \
 434   (c1 < MIN_CHAR_COMPOSITION                                    \
 435    && (SINGLE_BYTE_CHAR_P (c1)                                  \
 436        ? SINGLE_BYTE_CHAR_P (c2)                                \
 437        : (c1 < MIN_CHAR_OFFICIAL_DIMENSION2                     \
 438           ? (c1 & CHAR_FIELD2_MASK) == (c2 & CHAR_FIELD2_MASK)  \
 439           : (c1 & CHAR_FIELD1_MASK) == (c2 & CHAR_FIELD1_MASK))))
 440
 441 /* Return a non-ASCII character of which charset is CHARSET and
 442    position-codes are C1 and C2.  DIMENSION1 character ignores C2.  */
 443 #define MAKE_NON_ASCII_CHAR(charset, c1, c2)                    \
 444   ((charset) == CHARSET_COMPOSITION                             \
 445    ? MAKE_COMPOSITE_CHAR (((c1) << 7) + (c2))                   \
 446    : (CHARSET_DIMENSION (charset) == 1                          \
 447       ? (((charset) - 0x70) << 7) | (c1)                        \
 448       : ((charset) < MIN_CHARSET_PRIVATE_DIMENSION2             \
 449          ? (((charset) - 0x8F) << 14) | ((c1) << 7) | (c2)      \
 450          : (((charset) - 0xE0) << 14) | ((c1) << 7) | (c2))))
 451
 452 /* Return a composite character of which CMPCHAR-ID is ID.  */
 453 #define MAKE_COMPOSITE_CHAR(id) (MIN_CHAR_COMPOSITION + (id))
 454
 455 /* Return CMPCHAR-ID of a composite character C.  */
 456 #define COMPOSITE_CHAR_ID(c) ((c) - MIN_CHAR_COMPOSITION)
 457
 458 /* Return a character of which charset is CHARSET and position-codes
 459    are C1 and C2.  DIMENSION1 character ignores C2.  */
 460 #define MAKE_CHAR(charset, c1, c2)                      \
 461   ((charset) == CHARSET_ASCII                           \
 462    ? (c1)                                               \
 463    : MAKE_NON_ASCII_CHAR ((charset), (c1) & 0x7F, (c2) & 0x7F))
 464
 465 /* 1 if C is in the range of possible character code Emacs can have.  */
 466 #define VALID_CHAR_P(c)                                                 \
 467   ((c) >= 0                                                             \
 468    && (SINGLE_BYTE_CHAR_P (c)                                           \
 469        || ((c) < MIN_CHAR_COMPOSITION                                   \
 470            ? ((c) & CHAR_FIELD1_MASK                                    \
 471               ? (CHAR_FIELD2 (c) >= 32 && CHAR_FIELD3 (c) >= 32)        \
 472               : (CHAR_FIELD2 (c) >= 16 && CHAR_FIELD3 (c) >= 32))       \
 473            : (c) < MIN_CHAR_COMPOSITION + n_cmpchars)))
 474
 475 /* The charset of non-ASCII character C is set to CHARSET, and the
 476    position-codes of C are set to C1 and C2.  C2 of DIMENSION1 character
 477    is -1.  */
 478 #define SPLIT_NON_ASCII_CHAR(c, charset, c1, c2)                         \
 479   ((c) < MIN_CHAR_OFFICIAL_DIMENSION2                                    \
 480    ? (charset = CHAR_FIELD2 (c) + 0x70,                                  \
 481       c1 = CHAR_FIELD3 (c),                                              \
 482       c2 = -1)                                                           \
 483    : (charset = ((c) < MIN_CHAR_COMPOSITION                              \
 484                  ? (CHAR_FIELD1 (c)                                      \
 485                     + ((c) < MIN_CHAR_PRIVATE_DIMENSION2 ? 0x8F : 0xE0)) \
 486                  : CHARSET_COMPOSITION),                                 \
 487       c1 = CHAR_FIELD2 (c),                                              \
 488       c2 = CHAR_FIELD3 (c)))
 489
 490 /* The charset of character C is set to CHARSET, and the
 491    position-codes of C are set to C1 and C2.  C2 of DIMENSION1 character
 492    is -1.  */
 493 #define SPLIT_CHAR(c, charset, c1, c2)                  \
 494   (SINGLE_BYTE_CHAR_P (c)                               \
 495    ? charset = CHARSET_ASCII, c1 = (c), c2 = -1         \
 496    : SPLIT_NON_ASCII_CHAR (c, charset, c1, c2))
 497
 498 /* The charset of the character at STR is set to CHARSET, and the
 499    position-codes are set to C1 and C2.  C2 of DIMENSION1 character is -1.
 500    If the character is a composite character, the upper 7-bit and
 501    lower 7-bit of CMPCHAR-ID are set in C1 and C2 respectively.  No
 502    range checking.  */
 503 #define SPLIT_STRING(str, len, charset, c1, c2)                         \
 504   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) < 2                     \
 505     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > len                \
 506     || split_non_ascii_string (str, len, &charset, &c1, &c2, 0) < 0)    \
 507    ? c1 = *(str), charset = CHARSET_ASCII                               \
 508    : charset)
 509
 510 /* Mapping table from ISO2022's charset (specified by DIMENSION,
 511    CHARS, and FINAL_CHAR) to Emacs' charset.  Should be accessed by
 512    macro ISO_CHARSET_TABLE (DIMENSION, CHARS, FINAL_CHAR).  */
 513 extern int iso_charset_table[2][2][128];
 514
 515 #define ISO_CHARSET_TABLE(dimension, chars, final_char) \
 516   iso_charset_table[XINT (dimension) - 1][XINT (chars) > 94][XINT (final_char)]
 517
 518 #define BASE_LEADING_CODE_P(c) (BYTES_BY_CHAR_HEAD ((unsigned char) (c)) > 1)
 519
 520 /* The following two macros CHAR_STRING and STRING_CHAR are the main
 521    entry points to convert between Emacs two types of character
 522    representations: multi-byte form and single-word form (character
 523    code).  */
 524
 525 /* Set STR a pointer to the multi-byte form of the character C.  If C
 526    is not a composite character, the multi-byte form is set in WORKBUF
 527    and STR points WORKBUF.  The caller should allocate at least 4-byte
 528    area at WORKBUF in advance.  Returns the length of the multi-byte
 529    form.  If C is an invalid character code, signal an error.  */
 530
 531 #define CHAR_STRING(c, workbuf, str)                    \
 532   (SINGLE_BYTE_CHAR_P (c)                               \
 533    ? *(str = workbuf) = (unsigned char)(c), 1           \
 534    : non_ascii_char_to_string (c, workbuf, &str))
 535
 536 /* Return a character code of the character of which multi-byte form
 537    is at STR and the length is LEN.  If STR doesn't contain valid
 538    multi-byte form, only the first byte in STR is returned.  */
 539
 540 #define STRING_CHAR(str, len)                                   \
 541   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) == 1            \
 542     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > (len))     \
 543    ? (unsigned char) *(str)                                     \
 544    : string_to_non_ascii_char (str, len, 0))
 545
 546 /* This is like STRING_CHAR but the third arg ACTUAL_LEN is set to
 547    the length of the multi-byte form.  Just to know the length, use
 548    MULTIBYTE_FORM_LENGTH.  */
 549
 550 #define STRING_CHAR_AND_LENGTH(str, len, actual_len)            \
 551   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) == 1            \
 552     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > (len))     \
 553    ? (actual_len = 1), (unsigned char) *(str)                   \
 554    : string_to_non_ascii_char (str, len, &actual_len))
 555
 556 /* Return the length of the multi-byte form at string STR of length LEN.  */
 557
 558 #define MULTIBYTE_FORM_LENGTH(str, len)                         \
 559   ((BYTES_BY_CHAR_HEAD (*(unsigned char *)(str)) == 1           \
 560     || BYTES_BY_CHAR_HEAD (*(unsigned char *)(str)) > (len))    \
 561    ? 1                                                          \
 562    : multibyte_form_length (str, len))
 563
 564 /* Set C a (possibly multibyte) character at P.  P points into a
 565    string which is the virtual concatenation of STR1 (which ends at
 566    END1) or STR2 (which ends at END2).  */
 567
 568 #define GET_CHAR_AFTER_2(c, p, str1, end1, str2, end2)                      \
 569   do {                                                                      \
 570     const char *dtemp = (p) == (end1) ? (str2) : (p);                       \
 571     const char *dlimit = ((p) >= (str1) && (p) < (end1)) ? (end1) : (end2); \
 572     c = STRING_CHAR (dtemp, dlimit - dtemp);                                \
 573   } while (0)
 574
 575 /* Set C a (possibly multibyte) character before P.  P points into a
 576    string which is the virtual concatenation of STR1 (which ends at
 577    END1) or STR2 (which ends at END2).  */
 578
 579 #define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2)                     \
 580   do {                                                                      \
 581     const char *dtemp = (p);                                                \
 582     const char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
 583     while (dtemp-- > dlimit && (unsigned char) *dtemp >= 0xA0);             \
 584     c = STRING_CHAR (dtemp, p - dtemp);                                     \
 585   } while (0)
 586
 587 #ifdef emacs
 588
 589 /* Increase the buffer point POS of the current buffer to the next
 590    character boundary.  This macro relies on the fact that *GPT_ADDR
 591    and *Z_ADDR are always accessible and the values are '\0'.  No
 592    range checking of POS.  */
 593 #define INC_POS(pos)                            \
 594   do {                                          \
 595     unsigned char *p = POS_ADDR (pos);          \
 596     pos++;                                      \
 597     if (*p++ >= 0x80)                           \
 598       while (!CHAR_HEAD_P (p)) p++, pos++;      \
 599   } while (0)
 600
 601 /* Decrease the buffer point POS of the current buffer to the previous
 602    character boundary.  No range checking of POS.  */
 603 #define DEC_POS(pos)                                            \
 604   do {                                                          \
 605     unsigned char *p, *p_min;                                   \
 606     int pos_saved = --pos;                                      \
 607     if (pos < GPT)                                              \
 608       p = BEG_ADDR + pos - 1, p_min = BEG_ADDR;                 \
 609     else                                                        \
 610       p = BEG_ADDR + GAP_SIZE + pos - 1, p_min = GAP_END_ADDR;  \
 611     while (p > p_min && !CHAR_HEAD_P (p)) p--, pos--;           \
 612     if (*p < 0x80 && pos != pos_saved) pos = pos_saved;         \
 613   } while (0)
 614
 615 #endif /* emacs */
 616
 617 /* Maximum counts of components in one composite character.  */
 618 #define MAX_COMPONENT_COUNT 16
 619
 620 /* Structure to hold information of a composite character.  */
 621 struct cmpchar_info {
 622   /* Byte length of the composite character.  */
 623   int len;
 624
 625   /* Multi-byte form of the composite character.  */
 626   unsigned char *data;
 627
 628   /* Length of glyph codes.  */
 629   int glyph_len;
 630
 631   /* Width of the overall glyph of the composite character.  */
 632   int width;
 633
 634   /* Pointer to an array of glyph codes of the composite character.
 635      This actually contains only character code, no face.  */
 636   GLYPH *glyph;
 637
 638   /* Pointer to an array of composition rules.  The value has the form:
 639         (0xA0 + ((GLOBAL-REF-POINT << 2) | NEW-REF-POINT))
 640      where each XXX-REF-POINT is 0..8.  */
 641   unsigned char *cmp_rule;
 642
 643   /* Pointer to an array of x-axis offset of left edge of glyphs
 644      relative to the left of of glyph[0] except for the first element
 645      which is the absolute offset from the left edge of overall glyph.
 646      The actual pixel offset should be calculated by multiplying each
 647      frame's one column width by this value:
 648         (i.e. FONT_WIDTH (f->output_data.x->font) * col_offset[N]).  */
 649   float *col_offset;
 650
 651   /* Work slot used by `dumpglyphs' (xterm.c).  */
 652   int face_work;
 653 };
 654
 655 /* Table of pointers to the structure `cmpchar_info' indexed by
 656    CMPCHAR-ID.  */
 657 extern struct cmpchar_info **cmpchar_table;
 658 /* Number of the current composite characters.  */
 659 extern int n_cmpchars;
 660
 661 /* This is the maximum length of multi-byte form.  */
 662 #define MAX_LENGTH_OF_MULTI_BYTE_FORM (MAX_COMPONENT_COUNT * 6)
 663
 664 /* Maximum character code currently used.  */
 665 #define MAX_CHAR (MIN_CHAR_COMPOSITION + n_cmpchars)
 666
 667 extern int unify_char ();
 668
 669 #endif /* _CHARSET_H */