src/charset.h

   1 /* Header for multilingual character handler.
   2    Copyright (C) 1995, 1997, 1998 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 DIMENSION2 of 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 CH is the head of multi-byte form, i.e.,
 138    an ASCII character or a base leading-code.  */
 139 #define CHAR_HEAD_P(ch) ((unsigned char) (ch) < 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 /* 1 if BYTE is a character in itself, in multibyte mode.  */
 254 #define ASCII_BYTE_P(byte) ((byte) < 0x80)
 255
 256 /* A char-table containing information of each character set.
 257
 258    Unlike ordinary char-tables, this doesn't contain any nested table.
 259    Only the top level elements are used.  Each element is a vector of
 260    the following information:
 261         CHARSET-ID, BYTES, DIMENSION, CHARS, WIDTH, DIRECTION,
 262         LEADING-CODE-BASE, LEADING-CODE-EXT,
 263         ISO-FINAL-CHAR, ISO-GRAPHIC-PLANE,
 264         REVERSE-CHARSET, SHORT-NAME, LONG-NAME, DESCRIPTION,
 265         PLIST.
 266
 267    CHARSET-ID (integer) is the identification number of the charset.
 268
 269    BYTE (integer) is the length of multi-byte form of a character in
 270    the charset: one of 1, 2, 3, and 4.
 271
 272    DIMENSION (integer) is the number of bytes to represent a character: 1 or 2.
 273
 274    CHARS (integer) is the number of characters in a dimension: 94 or 96.
 275
 276    WIDTH (integer) is the number of columns a character in the charset
 277    occupies on the screen: one of 0, 1, and 2.
 278
 279    DIRECTION (integer) is the rendering direction of characters in the
 280    charset when rendering.  If 0, render from right to left, else
 281    render from left to right.
 282
 283    LEADING-CODE-BASE (integer) is the base leading-code for the
 284    charset.
 285
 286    LEADING-CODE-EXT (integer) is the extended leading-code for the
 287    charset.  All charsets of less than 0xA0 has the value 0.
 288
 289    ISO-FINAL-CHAR (character) is the final character of the
 290    corresponding ISO 2022 charset.
 291
 292    ISO-GRAPHIC-PLANE (integer) is the graphic plane to be invoked
 293    while encoding to variants of ISO 2022 coding system, one of the
 294    following: 0/graphic-plane-left(GL), 1/graphic-plane-right(GR).
 295
 296    REVERSE-CHARSET (integer) is the charset which differs only in
 297    LEFT-TO-RIGHT value from the charset.  If there's no such a
 298    charset, the value is -1.
 299
 300    SHORT-NAME (string) is the short name to refer to the charset.
 301
 302    LONG-NAME (string) is the long name to refer to the charset.
 303
 304    DESCRIPTION (string) is the description string of the charset.
 305
 306    PLIST (property list) may contain any type of information a user
 307    want to put and get by functions `put-charset-property' and
 308    `get-charset-property' respectively.  */
 309 extern Lisp_Object Vcharset_table;
 310
 311 /* Macros to access various information of CHARSET in Vcharset_table.
 312    We provide these macros for efficiency.  No range check of CHARSET.  */
 313
 314 /* Return entry of CHARSET (lisp integer) in Vcharset_table.  */
 315 #define CHARSET_TABLE_ENTRY(charset)                                    \
 316   XCHAR_TABLE (Vcharset_table)->contents[((charset) == CHARSET_ASCII    \
 317                                           ? 0 : (charset) + 128)]
 318
 319 /* Return information INFO-IDX of CHARSET.  */
 320 #define CHARSET_TABLE_INFO(charset, info_idx) \
 321   XVECTOR (CHARSET_TABLE_ENTRY (charset))->contents[info_idx]
 322
 323 #define CHARSET_ID_IDX (0)
 324 #define CHARSET_BYTES_IDX (1)
 325 #define CHARSET_DIMENSION_IDX (2)
 326 #define CHARSET_CHARS_IDX (3)
 327 #define CHARSET_WIDTH_IDX (4)
 328 #define CHARSET_DIRECTION_IDX (5)
 329 #define CHARSET_LEADING_CODE_BASE_IDX (6)
 330 #define CHARSET_LEADING_CODE_EXT_IDX (7)
 331 #define CHARSET_ISO_FINAL_CHAR_IDX (8)
 332 #define CHARSET_ISO_GRAPHIC_PLANE_IDX (9)
 333 #define CHARSET_REVERSE_CHARSET_IDX (10)
 334 #define CHARSET_SHORT_NAME_IDX (11)
 335 #define CHARSET_LONG_NAME_IDX (12)
 336 #define CHARSET_DESCRIPTION_IDX (13)
 337 #define CHARSET_PLIST_IDX (14)
 338 /* Size of a vector of each entry of Vcharset_table.  */
 339 #define CHARSET_MAX_IDX (15)
 340
 341 /* And several more macros to be used frequently.  */
 342 #define CHARSET_BYTES(charset) \
 343   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_BYTES_IDX))
 344 #define CHARSET_DIMENSION(charset) \
 345   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_DIMENSION_IDX))
 346 #define CHARSET_CHARS(charset) \
 347   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_CHARS_IDX))
 348 #define CHARSET_WIDTH(charset) \
 349   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_WIDTH_IDX))
 350 #define CHARSET_DIRECTION(charset) \
 351   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX))
 352 #define CHARSET_LEADING_CODE_BASE(charset) \
 353   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_BASE_IDX))
 354 #define CHARSET_LEADING_CODE_EXT(charset) \
 355   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_EXT_IDX))
 356 #define CHARSET_ISO_FINAL_CHAR(charset) \
 357   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_ISO_FINAL_CHAR_IDX))
 358 #define CHARSET_ISO_GRAPHIC_PLANE(charset) \
 359   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_ISO_GRAPHIC_PLANE_IDX))
 360 #define CHARSET_REVERSE_CHARSET(charset) \
 361   XINT (CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX))
 362
 363 /* Macros to specify direction of a charset.  */
 364 #define CHARSET_DIRECTION_LEFT_TO_RIGHT 0
 365 #define CHARSET_DIRECTION_RIGHT_TO_LEFT 1
 366
 367 /* A vector of charset symbol indexed by charset-id.  This is used
 368    only for returning charset symbol from C functions.  */
 369 extern Lisp_Object Vcharset_symbol_table;
 370
 371 /* Return symbol of CHARSET.  */
 372 #define CHARSET_SYMBOL(charset) \
 373   XVECTOR (Vcharset_symbol_table)->contents[charset]
 374
 375 /* 1 if CHARSET is valid, else 0.  */
 376 #define CHARSET_VALID_P(charset)                                         \
 377   ((charset) == 0                                                        \
 378    || ((charset) >= 0x80 && (charset) <= MAX_CHARSET_OFFICIAL_DIMENSION2) \
 379    || ((charset) >= MIN_CHARSET_PRIVATE_DIMENSION1 && (charset) <= MAX_CHARSET))
 380
 381 /* 1 if CHARSET is already defined, else 0.  */
 382 #define CHARSET_DEFINED_P(charset)                      \
 383   (((charset) >= 0) && ((charset) <= MAX_CHARSET)       \
 384    && !NILP (CHARSET_TABLE_ENTRY (charset)))
 385
 386 /* Since the information CHARSET-BYTES and CHARSET-WIDTH of
 387    Vcharset_table can be retrieved only from the first byte of
 388    multi-byte form (an ASCII code or a base leading-code), we provide
 389    here tables to be used by macros BYTES_BY_CHAR_HEAD and
 390    WIDTH_BY_CHAR_HEAD for faster information retrieval.  */
 391 extern int bytes_by_char_head[256];
 392 extern int width_by_char_head[256];
 393
 394 #define BYTES_BY_CHAR_HEAD(char_head) bytes_by_char_head[char_head]
 395 #define WIDTH_BY_CHAR_HEAD(char_head) width_by_char_head[char_head]
 396
 397 /* Charset of the character C.  */
 398 #define CHAR_CHARSET(c)                         \
 399   (SINGLE_BYTE_CHAR_P (c)                       \
 400    ? CHARSET_ASCII                              \
 401    : ((c) < MIN_CHAR_OFFICIAL_DIMENSION2        \
 402       ? CHAR_FIELD2 (c) + 0x70                  \
 403       : ((c) < MIN_CHAR_PRIVATE_DIMENSION2      \
 404          ? CHAR_FIELD1 (c) + 0x8F               \
 405          : ((c) < MIN_CHAR_COMPOSITION          \
 406             ? CHAR_FIELD1 (c) + 0xE0            \
 407             : CHARSET_COMPOSITION))))
 408
 409 /* Return charset at the place pointed by P.  */
 410 #define CHARSET_AT(p)                           \
 411   (*(p) < 0x80                                  \
 412    ? CHARSET_ASCII                              \
 413    : (*(p) == LEADING_CODE_COMPOSITION          \
 414       ? CHARSET_COMPOSITION                     \
 415       : (*(p) < LEADING_CODE_PRIVATE_11         \
 416          ? (int)*(p)                            \
 417          : (*(p) <= LEADING_CODE_PRIVATE_22     \
 418             ? (int)*((p) + 1)                   \
 419             : -1))))
 420
 421 /* Same as `CHARSET_AT ()' but perhaps runs faster because of an
 422    additional argument C which is the code (byte) at P.  */
 423 #define FIRST_CHARSET_AT(p, c)                  \
 424   ((c) < 0x80                                   \
 425    ? CHARSET_ASCII                              \
 426    : ((c) == LEADING_CODE_COMPOSITION           \
 427       ? CHARSET_COMPOSITION                     \
 428       : ((c) < LEADING_CODE_PRIVATE_11          \
 429          ? (int)(c)                             \
 430          : ((c) <= LEADING_CODE_PRIVATE_22      \
 431             ? (int)*((p) + 1)                   \
 432             : -1))))
 433
 434 /* Check if two characters C1 and C2 belong to the same charset.
 435    Always return 0 for composite characters.  */
 436 #define SAME_CHARSET_P(c1, c2)                                  \
 437   (c1 < MIN_CHAR_COMPOSITION                                    \
 438    && (SINGLE_BYTE_CHAR_P (c1)                                  \
 439        ? SINGLE_BYTE_CHAR_P (c2)                                \
 440        : (c1 < MIN_CHAR_OFFICIAL_DIMENSION2                     \
 441           ? (c1 & CHAR_FIELD2_MASK) == (c2 & CHAR_FIELD2_MASK)  \
 442           : (c1 & CHAR_FIELD1_MASK) == (c2 & CHAR_FIELD1_MASK))))
 443
 444 /* Return a non-ASCII character of which charset is CHARSET and
 445    position-codes are C1 and C2.  DIMENSION1 character ignores C2.  */
 446 #define MAKE_NON_ASCII_CHAR(charset, c1, c2)                    \
 447   ((charset) == CHARSET_COMPOSITION                             \
 448    ? MAKE_COMPOSITE_CHAR (((c1) << 7) + (c2))                   \
 449    : (CHARSET_DIMENSION (charset) == 1                          \
 450       ? (((charset) - 0x70) << 7) | (c1)                        \
 451       : ((charset) < MIN_CHARSET_PRIVATE_DIMENSION2             \
 452          ? (((charset) - 0x8F) << 14) | ((c1) << 7) | (c2)      \
 453          : (((charset) - 0xE0) << 14) | ((c1) << 7) | (c2))))
 454
 455 /* Return a composite character of which CMPCHAR-ID is ID.  */
 456 #define MAKE_COMPOSITE_CHAR(id) (MIN_CHAR_COMPOSITION + (id))
 457
 458 /* Return CMPCHAR-ID of a composite character C.  */
 459 #define COMPOSITE_CHAR_ID(c) ((c) - MIN_CHAR_COMPOSITION)
 460
 461 /* Return a character of which charset is CHARSET and position-codes
 462    are C1 and C2.  DIMENSION1 character ignores C2.  */
 463 #define MAKE_CHAR(charset, c1, c2)                      \
 464   ((charset) == CHARSET_ASCII                           \
 465    ? (c1)                                               \
 466    : MAKE_NON_ASCII_CHAR ((charset), (c1) & 0x7F, (c2) & 0x7F))
 467
 468 /* If GENERICP is nonzero, return nonzero iff C is a valid normal or
 469    generic character.  If GENERICP is zero, return nonzero iff C is a
 470    valid normal character.  */
 471 #define CHAR_VALID_P(c, genericp)       \
 472   ((c) >= 0                             \
 473    && (SINGLE_BYTE_CHAR_P (c) || char_valid_p (c, genericp)))
 474
 475 /* The charset of non-ASCII character C is stored in CHARSET, and the
 476    position-codes of C are stored in C1 and C2.
 477    We store -1 in C2 if the character is just 2 bytes.
 478
 479    Do not use this macro for an ASCII character.  */
 480
 481 #define SPLIT_NON_ASCII_CHAR(c, charset, c1, c2)                         \
 482   ((c) < MIN_CHAR_OFFICIAL_DIMENSION2                                    \
 483    ? (charset = CHAR_FIELD2 (c) + 0x70,                                  \
 484       c1 = CHAR_FIELD3 (c),                                              \
 485       c2 = -1)                                                           \
 486    : (charset = ((c) < MIN_CHAR_COMPOSITION                              \
 487                  ? (CHAR_FIELD1 (c)                                      \
 488                     + ((c) < MIN_CHAR_PRIVATE_DIMENSION2 ? 0x8F : 0xE0)) \
 489                  : CHARSET_COMPOSITION),                                 \
 490       c1 = CHAR_FIELD2 (c),                                              \
 491       c2 = CHAR_FIELD3 (c)))
 492
 493 /* The charset of character C is stored in CHARSET, and the
 494    position-codes of C are stored in C1 and C2.
 495    We store -1 in C2 if the character is just 2 bytes.  */
 496
 497 #define SPLIT_CHAR(c, charset, c1, c2)                  \
 498   (SINGLE_BYTE_CHAR_P (c)                               \
 499    ? charset = CHARSET_ASCII, c1 = (c), c2 = -1         \
 500    : SPLIT_NON_ASCII_CHAR (c, charset, c1, c2))
 501
 502 /* The charset of the character at STR is stored in CHARSET, and the
 503    position-codes are stored in C1 and C2.
 504    We store -1 in C2 if the character is just 2 bytes.
 505
 506    If the character is a composite character, the upper 7-bit and
 507    lower 7-bit of CMPCHAR-ID are set in C1 and C2 respectively.  No
 508    range checking.  */
 509
 510 #define SPLIT_STRING(str, len, charset, c1, c2)                         \
 511   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) < 2                     \
 512     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > len                \
 513     || split_non_ascii_string (str, len, &charset, &c1, &c2) < 0)       \
 514    ? c1 = *(str), charset = CHARSET_ASCII                               \
 515    : charset)
 516
 517 /* Mapping table from ISO2022's charset (specified by DIMENSION,
 518    CHARS, and FINAL_CHAR) to Emacs' charset.  Should be accessed by
 519    macro ISO_CHARSET_TABLE (DIMENSION, CHARS, FINAL_CHAR).  */
 520 extern int iso_charset_table[2][2][128];
 521
 522 #define ISO_CHARSET_TABLE(dimension, chars, final_char) \
 523   iso_charset_table[XINT (dimension) - 1][XINT (chars) > 94][XINT (final_char)]
 524
 525 #define BASE_LEADING_CODE_P(c) (BYTES_BY_CHAR_HEAD ((unsigned char) (c)) > 1)
 526
 527 /* The following two macros CHAR_STRING and STRING_CHAR are the main
 528    entry points to convert between Emacs two types of character
 529    representations: multi-byte form and single-word form (character
 530    code).  */
 531
 532 /* Set STR a pointer to the multi-byte form of the character C.  If C
 533    is not a composite character, the multi-byte form is set in WORKBUF
 534    and STR points WORKBUF.  The caller should allocate at least 4-byte
 535    area at WORKBUF in advance.  Returns the length of the multi-byte
 536    form.  If C is an invalid character code, signal an error.  */
 537
 538 #define CHAR_STRING(c, workbuf, str)                    \
 539   (SINGLE_BYTE_CHAR_P (c)                               \
 540    ? *(str = workbuf) = (unsigned char)(c), 1           \
 541    : non_ascii_char_to_string (c, workbuf, &str))
 542
 543 /* Return a character code of the character of which multi-byte form
 544    is at STR and the length is LEN.  If STR doesn't contain valid
 545    multi-byte form, only the first byte in STR is returned.  */
 546
 547 #define STRING_CHAR(str, len)                                   \
 548   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) == 1            \
 549     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > (len))     \
 550    ? (unsigned char) *(str)                                     \
 551    : string_to_non_ascii_char (str, len, 0))
 552
 553 /* This is like STRING_CHAR but the third arg ACTUAL_LEN is set to
 554    the length of the multi-byte form.  Just to know the length, use
 555    MULTIBYTE_FORM_LENGTH.  */
 556
 557 #define STRING_CHAR_AND_LENGTH(str, len, actual_len)            \
 558   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) == 1            \
 559     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > (len))     \
 560    ? (actual_len = 1), (unsigned char) *(str)                   \
 561    : string_to_non_ascii_char (str, len, &actual_len))
 562
 563 /* Fetch the "next" multibyte character from Lisp string STRING
 564    at byte position BYTEIDX, character position CHARIDX.
 565    Store it into OUTPUT.
 566
 567    All the args must be side-effect-free.
 568    BYTEIDX and CHARIDX must be lvalues;
 569    we increment them past the character fetched.  */
 570
 571 #define FETCH_STRING_CHAR_ADVANCE(OUTPUT, STRING, CHARIDX, BYTEIDX)           \
 572 if (1)                                                                        \
 573   {                                                                           \
 574     unsigned char *fetch_string_char_ptr = &XSTRING (STRING)->data[BYTEIDX];  \
 575     int fetch_string_char_space_left = XSTRING (STRING)->size_byte - BYTEIDX; \
 576     int actual_len;                                                           \
 577                                                                               \
 578     OUTPUT                                                                    \
 579       = STRING_CHAR_AND_LENGTH (fetch_string_char_ptr,                        \
 580                                 fetch_string_char_space_left, actual_len);    \
 581                                                                               \
 582     BYTEIDX += actual_len;                                                    \
 583     CHARIDX++;                                                                \
 584   }                                                                           \
 585 else
 586
 587   /* Nonzero enables use of dialog boxes for questions
 588      asked by mouse commands.  */
 589   int use_dialog_box;
 590
 591 /* Return the length of the multi-byte form at string STR of length LEN.  */
 592
 593 #define MULTIBYTE_FORM_LENGTH(str, len)                         \
 594   ((BYTES_BY_CHAR_HEAD (*(unsigned char *)(str)) == 1           \
 595     || BYTES_BY_CHAR_HEAD (*(unsigned char *)(str)) > (len))    \
 596    ? 1                                                          \
 597    : multibyte_form_length (str, len))
 598
 599 /* Set C a (possibly multibyte) character at P.  P points into a
 600    string which is the virtual concatenation of STR1 (which ends at
 601    END1) or STR2 (which ends at END2).  */
 602
 603 #define GET_CHAR_AFTER_2(c, p, str1, end1, str2, end2)                      \
 604   do {                                                                      \
 605     const char *dtemp = (p) == (end1) ? (str2) : (p);                       \
 606     const char *dlimit = ((p) >= (str1) && (p) < (end1)) ? (end1) : (end2); \
 607     c = STRING_CHAR (dtemp, dlimit - dtemp);                                \
 608   } while (0)
 609
 610 /* Set C a (possibly multibyte) character before P.  P points into a
 611    string which is the virtual concatenation of STR1 (which ends at
 612    END1) or STR2 (which ends at END2).  */
 613
 614 #define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2)                     \
 615   do {                                                                      \
 616     const char *dtemp = (p);                                                \
 617     const char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
 618     while (dtemp-- > dlimit && (unsigned char) *dtemp >= 0xA0);             \
 619     c = STRING_CHAR (dtemp, p - dtemp);                                     \
 620   } while (0)
 621
 622 #ifdef emacs
 623
 624 /* Increase the buffer point POS of the current buffer to the next
 625    character boundary.  This macro relies on the fact that *GPT_ADDR
 626    and *Z_ADDR are always accessible and the values are '\0'.  No
 627    range checking of POS.  */
 628 #define INC_POS(pos)                            \
 629   do {                                          \
 630     unsigned char *p = BYTE_POS_ADDR (pos);     \
 631     pos++;                                      \
 632     if (*p++ >= 0x80)                           \
 633       while (!CHAR_HEAD_P (*p)) p++, pos++;     \
 634   } while (0)
 635
 636 /* Decrease the buffer point POS of the current buffer to the previous
 637    character boundary.  No range checking of POS.  */
 638 #define DEC_POS(pos)                                            \
 639   do {                                                          \
 640     unsigned char *p, *p_min;                                   \
 641     int pos_saved = --pos;                                      \
 642     if (pos < GPT_BYTE)                                         \
 643       p = BEG_ADDR + pos - 1, p_min = BEG_ADDR;                 \
 644     else                                                        \
 645       p = BEG_ADDR + GAP_SIZE + pos - 1, p_min = GAP_END_ADDR;  \
 646     while (p > p_min && !CHAR_HEAD_P (*p)) p--, pos--;          \
 647     if (*p < 0x80 && pos != pos_saved) pos = pos_saved;         \
 648   } while (0)
 649
 650 /* Increment both CHARPOS and BYTEPOS, each in the appropriate way.  */
 651
 652 #define INC_BOTH(charpos, bytepos)              \
 653 do                                              \
 654   {                                             \
 655     (charpos)++;                                \
 656     INC_POS ((bytepos));                        \
 657   }                                             \
 658 while (0)
 659
 660 /* Decrement both CHARPOS and BYTEPOS, each in the appropriate way.  */
 661
 662 #define DEC_BOTH(charpos, bytepos)              \
 663 do                                              \
 664   {                                             \
 665     (charpos)--;                                \
 666     DEC_POS ((bytepos));                        \
 667   }                                             \
 668 while (0)
 669
 670 /* Increase the buffer point POS of the current buffer to the next
 671    character boundary.  This macro relies on the fact that *GPT_ADDR
 672    and *Z_ADDR are always accessible and the values are '\0'.  No
 673    range checking of POS.  */
 674 #define BUF_INC_POS(buf, pos)                   \
 675   do {                                          \
 676     unsigned char *p = BUF_BYTE_ADDRESS (buf, pos);     \
 677     pos++;                                      \
 678     if (*p++ >= 0x80)                           \
 679       while (!CHAR_HEAD_P (*p)) p++, pos++;     \
 680   } while (0)
 681
 682 /* Decrease the buffer point POS of the current buffer to the previous
 683    character boundary.  No range checking of POS.  */
 684 #define BUF_DEC_POS(buf, pos)                                   \
 685   do {                                                          \
 686     unsigned char *p, *p_min;                                   \
 687     int pos_saved = --pos;                                      \
 688     if (pos < BUF_GPT_BYTE (buf))                               \
 689       {                                                         \
 690         p = BUF_BEG_ADDR (buf) + pos - 1;                       \
 691         p_min = BUF_BEG_ADDR (buf);                             \
 692       }                                                         \
 693     else                                                        \
 694       {                                                         \
 695         p = BUF_BEG_ADDR (buf) + BUF_GAP_SIZE (buf) + pos - 1;  \
 696         p_min = BUF_GAP_END_ADDR (buf);                         \
 697       }                                                         \
 698     while (p > p_min && !CHAR_HEAD_P (*p)) p--, pos--;          \
 699     if (*p < 0x80 && pos != pos_saved) pos = pos_saved;         \
 700   } while (0)
 701
 702 #endif /* emacs */
 703
 704 /* Maximum counts of components in one composite character.  */
 705 #define MAX_COMPONENT_COUNT 16
 706
 707 /* Structure to hold information of a composite character.  */
 708 struct cmpchar_info {
 709   /* Byte length of the composite character.  */
 710   int len;
 711
 712   /* Multi-byte form of the composite character.  */
 713   unsigned char *data;
 714
 715   /* Length of glyph codes.  */
 716   int glyph_len;
 717
 718   /* Width of the overall glyph of the composite character.  */
 719   int width;
 720
 721   /* Pointer to an array of glyph codes of the composite character.
 722      This actually contains only character code, no face.  */
 723   GLYPH *glyph;
 724
 725   /* Pointer to an array of composition rules.  The value has the form:
 726         (0xA0 + ((GLOBAL-REF-POINT << 2) | NEW-REF-POINT))
 727      where each XXX-REF-POINT is 0..8.  */
 728   unsigned char *cmp_rule;
 729
 730   /* Pointer to an array of x-axis offset of left edge of glyphs
 731      relative to the left of of glyph[0] except for the first element
 732      which is the absolute offset from the left edge of overall glyph.
 733      The actual pixel offset should be calculated by multiplying each
 734      frame's one column width by this value:
 735         (i.e. FONT_WIDTH (f->output_data.x->font) * col_offset[N]).  */
 736   float *col_offset;
 737
 738   /* Work slot used by `dumpglyphs' (xterm.c).  */
 739   int face_work;
 740 };
 741
 742 /* Table of pointers to the structure `cmpchar_info' indexed by
 743    CMPCHAR-ID.  */
 744 extern struct cmpchar_info **cmpchar_table;
 745 /* Number of the current composite characters.  */
 746 extern int n_cmpchars;
 747
 748 /* This is the maximum length of multi-byte form.  */
 749 #define MAX_LENGTH_OF_MULTI_BYTE_FORM (MAX_COMPONENT_COUNT * 6)
 750
 751 /* Maximum character code currently used.  */
 752 #define MAX_CHAR (MIN_CHAR_COMPOSITION + n_cmpchars)
 753
 754 extern int unify_char P_ ((Lisp_Object, int, int, int, int));
 755 extern int split_non_ascii_string P_ ((unsigned char *, int, int *,
 756                                        unsigned char *, unsigned char *));
 757 extern int string_to_non_ascii_char P_ ((unsigned char *, int, int *));
 758 extern int non_ascii_char_to_string P_ ((int, unsigned char *, unsigned char **));
 759 extern int multibyte_form_length P_ ((unsigned char *, int));
 760 extern int str_cmpchar_id P_ ((unsigned char *, int));
 761 extern int get_charset_id P_ ((Lisp_Object));
 762 extern int cmpchar_component P_ ((unsigned int, unsigned int));
 763 extern int find_charset_in_str P_ ((unsigned char *, int, int *, Lisp_Object));
 764 extern int strwidth P_ ((unsigned char *, int));
 765
 766 #endif /* _CHARSET_H */