src/charset.h

   1 /* Header for multibyte character handler.
   2    Copyright (C) 1995, 1997, 1998 Electrotechnical Laboratory, JAPAN.
   3    Licensed to the Free Software Foundation.
   4    Copyright (C) 2001 Free Software Foundation, Inc.
   5
   6 This file is part of GNU Emacs.
   7
   8 GNU Emacs is free software; you can redistribute it and/or modify
   9 it under the terms of the GNU General Public License as published by
  10 the Free Software Foundation; either version 2, or (at your option)
  11 any later version.
  12
  13 GNU Emacs is distributed in the hope that it will be useful,
  14 but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 GNU General Public License for more details.
  17
  18 You should have received a copy of the GNU General Public License
  19 along with GNU Emacs; see the file COPYING.  If not, write to
  20 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  21 Boston, MA 02111-1307, USA.  */
  22
  23 #ifndef EMACS_CHARSET_H
  24 #define EMACS_CHARSET_H
  25
  26 /* #define BYTE_COMBINING_DEBUG */
  27
  28 /*** GENERAL NOTE on CHARACTER SET (CHARSET) ***
  29
  30   A character set ("charset" hereafter) is a meaningful collection
  31   (i.e. language, culture, functionality, etc) of characters.  Emacs
  32   handles multiple charsets at once.  Each charset corresponds to one
  33   of the ISO charsets.  Emacs identifies a charset by a unique
  34   identification number, whereas ISO identifies a charset by a triplet
  35   of DIMENSION, CHARS and FINAL-CHAR.  So, hereafter, just saying
  36   "charset" means an identification number (integer value).
  37
  38   The value range of charsets is 0x00, 0x81..0xFE.  There are four
  39   kinds of charset depending on DIMENSION (1 or 2) and CHARS (94 or
  40   96).  For instance, a charset of DIMENSION2_CHARS94 contains 94x94
  41   characters.
  42
  43   Within Emacs Lisp, a charset is treated as a symbol which has a
  44   property `charset'.  The property value is a vector containing
  45   various information about the charset.  For readability of C code,
  46   we use the following convention for C variable names:
  47         charset_symbol: Emacs Lisp symbol of a charset
  48         charset_id: Emacs Lisp integer of an identification number of a charset
  49         charset: C integer of an identification number of a charset
  50
  51   Each charset (except for ascii) is assigned a base leading-code
  52   (range 0x80..0x9E).  In addition, a charset of greater than 0xA0
  53   (whose base leading-code is 0x9A..0x9D) is assigned an extended
  54   leading-code (range 0xA0..0xFE).  In this case, each base
  55   leading-code specifies the allowable range of extended leading-code
  56   as shown in the table below.  A leading-code is used to represent a
  57   character in Emacs' buffer and string.
  58
  59   We call a charset which has extended leading-code a "private
  60   charset" because those are mainly for a charset which is not yet
  61   registered by ISO.  On the contrary, we call a charset which does
  62   not have extended leading-code an "official charset".
  63
  64   ---------------------------------------------------------------------------
  65   charset       dimension        base leading-code      extended leading-code
  66   ---------------------------------------------------------------------------
  67   0x00          official dim1    -- none --             -- none --
  68                 (ASCII)
  69   0x01..0x7F    --never used--
  70   0x80          official dim1    -- none --             -- none --
  71                 (eight-bit-graphic)
  72   0x81..0x8F    official dim1    same as charset        -- none --
  73   0x90..0x99    official dim2    same as charset        -- none --
  74   0x9A..0x9D    --never used--
  75   0x9E          official dim1    same as charset        -- none --
  76                 (eight-bit-control)
  77   0x9F          --never used--
  78   0xA0..0xDF    private dim1        0x9A                same as charset
  79                 of 1-column width
  80   0xE0..0xEF    private dim1        0x9B                same as charset
  81                 of 2-column width
  82   0xF0..0xF4    private dim2        0x9C                same as charset
  83                 of 1-column width
  84   0xF5..0xFE    private dim2        0x9D                same as charset
  85                 of 2-column width
  86   0xFF          --never used--
  87   ---------------------------------------------------------------------------
  88
  89 */
  90
  91 /* Definition of special leading-codes.  */
  92 /* Leading-code followed by extended leading-code.  */
  93 #define LEADING_CODE_PRIVATE_11 0x9A /* for private DIMENSION1 of 1-column */
  94 #define LEADING_CODE_PRIVATE_12 0x9B /* for private DIMENSION1 of 2-column */
  95 #define LEADING_CODE_PRIVATE_21 0x9C /* for private DIMENSION2 of 1-column */
  96 #define LEADING_CODE_PRIVATE_22 0x9D /* for private DIMENSION2 of 2-column */
  97
  98 #define LEADING_CODE_8_BIT_CONTROL 0x9E /* for `eight-bit-control' */
  99
 100 /* Extended leading-code.  */
 101 /* Start of each extended leading-codes.  */
 102 #define LEADING_CODE_EXT_11 0xA0 /* follows LEADING_CODE_PRIVATE_11 */
 103 #define LEADING_CODE_EXT_12 0xE0 /* follows LEADING_CODE_PRIVATE_12 */
 104 #define LEADING_CODE_EXT_21 0xF0 /* follows LEADING_CODE_PRIVATE_21 */
 105 #define LEADING_CODE_EXT_22 0xF5 /* follows LEADING_CODE_PRIVATE_22 */
 106 /* Maximum value of extended leading-codes.  */
 107 #define LEADING_CODE_EXT_MAX 0xFE
 108
 109 /* Definition of minimum/maximum charset of each DIMENSION.  */
 110 #define MIN_CHARSET_OFFICIAL_DIMENSION1 0x80
 111 #define MAX_CHARSET_OFFICIAL_DIMENSION1 0x8F
 112 #define MIN_CHARSET_OFFICIAL_DIMENSION2 0x90
 113 #define MAX_CHARSET_OFFICIAL_DIMENSION2 0x99
 114 #define MIN_CHARSET_PRIVATE_DIMENSION1  LEADING_CODE_EXT_11
 115 #define MIN_CHARSET_PRIVATE_DIMENSION2  LEADING_CODE_EXT_21
 116
 117 /* Maximum value of overall charset identification number.  */
 118 #define MAX_CHARSET 0xFE
 119
 120 /* Definition of special charsets.  */
 121 #define CHARSET_ASCII           0       /* 0x00..0x7F */
 122 #define CHARSET_8_BIT_CONTROL   0x9E    /* 0x80..0x9F */
 123 #define CHARSET_8_BIT_GRAPHIC   0x80    /* 0xA0..0xFF */
 124
 125 extern int charset_latin_iso8859_1; /* ISO8859-1 (Latin-1) */
 126 extern int charset_jisx0208_1978; /* JISX0208.1978 (Japanese Kanji old set) */
 127 extern int charset_jisx0208;    /* JISX0208.1983 (Japanese Kanji) */
 128 extern int charset_katakana_jisx0201; /* JISX0201.Kana (Japanese Katakana) */
 129 extern int charset_latin_jisx0201; /* JISX0201.Roman (Japanese Roman) */
 130 extern int charset_big5_1;      /* Big5 Level 1 (Chinese Traditional) */
 131 extern int charset_big5_2;      /* Big5 Level 2 (Chinese Traditional) */
 132
 133 /* Check if CH is an ASCII character or a base leading-code.
 134    Nowadays, any byte can be the first byte of a character in a
 135    multibyte buffer/string.  So this macro name is not appropriate.  */
 136 #define CHAR_HEAD_P(ch) ((unsigned char) (ch) < 0xA0)
 137
 138 /*** GENERAL NOTE on CHARACTER REPRESENTATION ***
 139
 140   Firstly, the term "character" or "char" is used for a multilingual
 141   character (of course, including ASCII characters), not for a byte in
 142   computer memory.  We use the term "code" or "byte" for the latter
 143   case.
 144
 145   A character is identified by charset and one or two POSITION-CODEs.
 146   POSITION-CODE is the position of the character in the charset.  A
 147   character of DIMENSION1 charset has one POSITION-CODE: POSITION-CODE-1.
 148   A character of DIMENSION2 charset has two POSITION-CODE:
 149   POSITION-CODE-1 and POSITION-CODE-2.  The code range of
 150   POSITION-CODE is 0x20..0x7F.
 151
 152   Emacs has two kinds of representation of a character: multi-byte
 153   form (for buffers and strings) and single-word form (for character
 154   objects in Emacs Lisp).  The latter is called "character code"
 155   hereafter.  Both representations encode the information of charset
 156   and POSITION-CODE but in a different way (for instance, the MSB of
 157   POSITION-CODE is set in multi-byte form).
 158
 159   For details of the multi-byte form, see the section "2. Emacs
 160   internal format handlers" of `coding.c'.
 161
 162   Emacs uses 19 bits for a character code.  The bits are divided into
 163   3 fields: FIELD1(5bits):FIELD2(7bits):FIELD3(7bits).
 164
 165   A character code of DIMENSION1 character uses FIELD2 to hold charset
 166   and FIELD3 to hold POSITION-CODE-1.  A character code of DIMENSION2
 167   character uses FIELD1 to hold charset, FIELD2 and FIELD3 to hold
 168   POSITION-CODE-1 and POSITION-CODE-2 respectively.
 169
 170   More precisely...
 171
 172   FIELD2 of DIMENSION1 character (except for ascii, eight-bit-control,
 173   and eight-bit-graphic) is "charset - 0x70".  This is to make all
 174   character codes except for ASCII and 8-bit codes greater than 256.
 175   So, the range of FIELD2 of DIMENSION1 character is 0, 1, or
 176   0x11..0x7F.
 177
 178   FIELD1 of DIMENSION2 character is "charset - 0x8F" for official
 179   charset and "charset - 0xE0" for private charset.  So, the range of
 180   FIELD1 of DIMENSION2 character is 0x01..0x1E.
 181
 182   -----------------------------------------------------------------------------
 183   charset               FIELD1 (5-bit)      FIELD2 (7-bit)      FIELD3 (7-bit)
 184   -----------------------------------------------------------------------------
 185   ascii                 0                   0                   0x00..0x7F
 186   eight-bit-control     0                   1                   0x00..0x1F
 187   eight-bit-graphic     0                   1                   0x20..0x7F
 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 /* Masks of each field of character code.  */
 197 #define CHAR_FIELD1_MASK (0x1F << 14)
 198 #define CHAR_FIELD2_MASK (0x7F << 7)
 199 #define CHAR_FIELD3_MASK 0x7F
 200
 201 /* Macros to access each field of character C.  */
 202 #define CHAR_FIELD1(c) (((c) & CHAR_FIELD1_MASK) >> 14)
 203 #define CHAR_FIELD2(c) (((c) & CHAR_FIELD2_MASK) >> 7)
 204 #define CHAR_FIELD3(c) ((c) & CHAR_FIELD3_MASK)
 205
 206 /* Minimum character code of character of each DIMENSION.  */
 207 #define MIN_CHAR_OFFICIAL_DIMENSION1 \
 208   ((0x81 - 0x70) << 7)
 209 #define MIN_CHAR_PRIVATE_DIMENSION1 \
 210   ((MIN_CHARSET_PRIVATE_DIMENSION1 - 0x70) << 7)
 211 #define MIN_CHAR_OFFICIAL_DIMENSION2 \
 212   ((MIN_CHARSET_OFFICIAL_DIMENSION2 - 0x8F) << 14)
 213 #define MIN_CHAR_PRIVATE_DIMENSION2 \
 214   ((MIN_CHARSET_PRIVATE_DIMENSION2 - 0xE0) << 14)
 215 /* Maximum character code currently used plus 1.  */
 216 #define MAX_CHAR (0x1F << 14)
 217
 218 /* 1 if C is a single byte character, else 0.  */
 219 #define SINGLE_BYTE_CHAR_P(c) ((unsigned) (c) < 0x100)
 220
 221 /* 1 if BYTE is an ASCII character in itself, in multibyte mode.  */
 222 #define ASCII_BYTE_P(byte) ((byte) < 0x80)
 223
 224 /* A char-table containing information on each character set.
 225
 226    Unlike ordinary char-tables, this doesn't contain any nested tables.
 227    Only the top level elements are used.  Each element is a vector of
 228    the following information:
 229         CHARSET-ID, BYTES, DIMENSION, CHARS, WIDTH, DIRECTION,
 230         LEADING-CODE-BASE, LEADING-CODE-EXT,
 231         ISO-FINAL-CHAR, ISO-GRAPHIC-PLANE,
 232         REVERSE-CHARSET, SHORT-NAME, LONG-NAME, DESCRIPTION,
 233         PLIST.
 234
 235    CHARSET-ID (integer) is the identification number of the charset.
 236
 237    BYTES (integer) is the length of the multi-byte form of a character
 238    in the charset: one of 1, 2, 3, and 4.
 239
 240    DIMENSION (integer) is the number of bytes to represent a character: 1 or 2.
 241
 242    CHARS (integer) is the number of characters in a dimension: 94 or 96.
 243
 244    WIDTH (integer) is the number of columns a character in the charset
 245    occupies on the screen: one of 0, 1, and 2..
 246
 247    DIRECTION (integer) is the rendering direction of characters in the
 248    charset when rendering.  If 0, render from left to right, else
 249    render from right to left.
 250
 251    LEADING-CODE-BASE (integer) is the base leading-code for the
 252    charset.
 253
 254    LEADING-CODE-EXT (integer) is the extended leading-code for the
 255    charset.  All charsets of less than 0xA0 have the value 0.
 256
 257    ISO-FINAL-CHAR (character) is the final character of the
 258    corresponding ISO 2022 charset.  It is -1 for such a character
 259    that is used only internally (e.g. `eight-bit-control').
 260
 261    ISO-GRAPHIC-PLANE (integer) is the graphic plane to be invoked
 262    while encoding to variants of ISO 2022 coding system, one of the
 263    following: 0/graphic-plane-left(GL), 1/graphic-plane-right(GR).  It
 264    is -1 for such a character that is used only internally
 265    (e.g. `eight-bit-control').
 266
 267    REVERSE-CHARSET (integer) is the charset which differs only in
 268    LEFT-TO-RIGHT value from the charset.  If there's no such a
 269    charset, the value is -1.
 270
 271    SHORT-NAME (string) is the short name to refer to the charset.
 272
 273    LONG-NAME (string) is the long name to refer to the charset.
 274
 275    DESCRIPTION (string) is the description string of the charset.
 276
 277    PLIST (property list) may contain any type of information a user
 278    wants to put and get by functions `put-charset-property' and
 279    `get-charset-property' respectively.  */
 280 extern Lisp_Object Vcharset_table;
 281
 282 /* Macros to access various information of CHARSET in Vcharset_table.
 283    We provide these macros for efficiency.  No range check of CHARSET.  */
 284
 285 /* Return entry of CHARSET (C integer) in Vcharset_table.  */
 286 #define CHARSET_TABLE_ENTRY(charset)                                    \
 287   XCHAR_TABLE (Vcharset_table)->contents[((charset) == CHARSET_ASCII    \
 288                                           ? 0 : (charset) + 128)]
 289
 290 /* Return information INFO-IDX of CHARSET.  */
 291 #define CHARSET_TABLE_INFO(charset, info_idx) \
 292   XVECTOR (CHARSET_TABLE_ENTRY (charset))->contents[info_idx]
 293
 294 #define CHARSET_ID_IDX (0)
 295 #define CHARSET_BYTES_IDX (1)
 296 #define CHARSET_DIMENSION_IDX (2)
 297 #define CHARSET_CHARS_IDX (3)
 298 #define CHARSET_WIDTH_IDX (4)
 299 #define CHARSET_DIRECTION_IDX (5)
 300 #define CHARSET_LEADING_CODE_BASE_IDX (6)
 301 #define CHARSET_LEADING_CODE_EXT_IDX (7)
 302 #define CHARSET_ISO_FINAL_CHAR_IDX (8)
 303 #define CHARSET_ISO_GRAPHIC_PLANE_IDX (9)
 304 #define CHARSET_REVERSE_CHARSET_IDX (10)
 305 #define CHARSET_SHORT_NAME_IDX (11)
 306 #define CHARSET_LONG_NAME_IDX (12)
 307 #define CHARSET_DESCRIPTION_IDX (13)
 308 #define CHARSET_PLIST_IDX (14)
 309 /* Size of a vector of each entry of Vcharset_table.  */
 310 #define CHARSET_MAX_IDX (15)
 311
 312 /* And several more macros to be used frequently.  */
 313 #define CHARSET_BYTES(charset) \
 314   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_BYTES_IDX))
 315 #define CHARSET_DIMENSION(charset) \
 316   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_DIMENSION_IDX))
 317 #define CHARSET_CHARS(charset) \
 318   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_CHARS_IDX))
 319 #define CHARSET_WIDTH(charset) \
 320   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_WIDTH_IDX))
 321 #define CHARSET_DIRECTION(charset) \
 322   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_DIRECTION_IDX))
 323 #define CHARSET_LEADING_CODE_BASE(charset) \
 324   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_BASE_IDX))
 325 #define CHARSET_LEADING_CODE_EXT(charset) \
 326   XFASTINT (CHARSET_TABLE_INFO (charset, CHARSET_LEADING_CODE_EXT_IDX))
 327 #define CHARSET_ISO_FINAL_CHAR(charset) \
 328   XINT (CHARSET_TABLE_INFO (charset, CHARSET_ISO_FINAL_CHAR_IDX))
 329 #define CHARSET_ISO_GRAPHIC_PLANE(charset) \
 330   XINT (CHARSET_TABLE_INFO (charset, CHARSET_ISO_GRAPHIC_PLANE_IDX))
 331 #define CHARSET_REVERSE_CHARSET(charset) \
 332   XINT (CHARSET_TABLE_INFO (charset, CHARSET_REVERSE_CHARSET_IDX))
 333
 334 /* Macros to specify direction of a charset.  */
 335 #define CHARSET_DIRECTION_LEFT_TO_RIGHT 0
 336 #define CHARSET_DIRECTION_RIGHT_TO_LEFT 1
 337
 338 /* A vector of charset symbol indexed by charset-id.  This is used
 339    only for returning charset symbol from C functions.  */
 340 extern Lisp_Object Vcharset_symbol_table;
 341
 342 /* Return symbol of CHARSET.  */
 343 #define CHARSET_SYMBOL(charset) \
 344   XVECTOR (Vcharset_symbol_table)->contents[charset]
 345
 346 /* 1 if CHARSET is in valid value range, else 0.  */
 347 #define CHARSET_VALID_P(charset)                                         \
 348   ((charset) == 0                                                        \
 349    || ((charset) > 0x80 && (charset) <= MAX_CHARSET_OFFICIAL_DIMENSION2) \
 350    || ((charset) >= MIN_CHARSET_PRIVATE_DIMENSION1                       \
 351        && (charset) <= MAX_CHARSET)                                      \
 352    || ((charset) == CHARSET_8_BIT_CONTROL)                               \
 353    || ((charset) == CHARSET_8_BIT_GRAPHIC))
 354
 355 /* 1 if CHARSET is already defined, else 0.  */
 356 #define CHARSET_DEFINED_P(charset)                      \
 357   (((charset) >= 0) && ((charset) <= MAX_CHARSET)       \
 358    && !NILP (CHARSET_TABLE_ENTRY (charset)))
 359
 360 /* Since the information CHARSET-BYTES and CHARSET-WIDTH of
 361    Vcharset_table can be retrieved only by the first byte of
 362    multi-byte form (an ASCII code or a base leading-code), we provide
 363    here tables to be used by macros BYTES_BY_CHAR_HEAD and
 364    WIDTH_BY_CHAR_HEAD for faster information retrieval.  */
 365 extern int bytes_by_char_head[256];
 366 extern int width_by_char_head[256];
 367
 368 #define BYTES_BY_CHAR_HEAD(char_head)   \
 369   (ASCII_BYTE_P (char_head) ? 1 : bytes_by_char_head[char_head])
 370 #define WIDTH_BY_CHAR_HEAD(char_head)   \
 371   (ASCII_BYTE_P (char_head) ? 1 : width_by_char_head[char_head])
 372
 373 /* Charset of the character C.  */
 374 #define CHAR_CHARSET(c)                                                 \
 375   (SINGLE_BYTE_CHAR_P (c)                                               \
 376    ? (ASCII_BYTE_P (c)                                                  \
 377       ? CHARSET_ASCII                                                   \
 378       : (c) < 0xA0 ? CHARSET_8_BIT_CONTROL : CHARSET_8_BIT_GRAPHIC)     \
 379    : ((c) < MIN_CHAR_OFFICIAL_DIMENSION2                                \
 380       ? CHAR_FIELD2 (c) + 0x70                                          \
 381       : ((c) < MIN_CHAR_PRIVATE_DIMENSION2                              \
 382          ? CHAR_FIELD1 (c) + 0x8F                                       \
 383          : CHAR_FIELD1 (c) + 0xE0)))
 384
 385 /* Check if two characters C1 and C2 belong to the same charset.  */
 386 #define SAME_CHARSET_P(c1, c2)                          \
 387   (c1 < MIN_CHAR_OFFICIAL_DIMENSION2                    \
 388    ? (c1 & CHAR_FIELD2_MASK) == (c2 & CHAR_FIELD2_MASK) \
 389    : (c1 & CHAR_FIELD1_MASK) == (c2 & CHAR_FIELD1_MASK))
 390
 391 /* Return a character of which charset is CHARSET and position-codes
 392    are C1 and C2.  DIMENSION1 character ignores C2.  */
 393 #define MAKE_CHAR(charset, c1, c2)                                          \
 394   ((charset) == CHARSET_ASCII                                               \
 395    ? (c1) & 0x7F                                                            \
 396    : (((charset) == CHARSET_8_BIT_CONTROL                                   \
 397        || (charset) == CHARSET_8_BIT_GRAPHIC)                               \
 398       ? ((c1) & 0x7F) | 0x80                                                \
 399       : ((CHARSET_DEFINED_P (charset)                                       \
 400           ? CHARSET_DIMENSION (charset) == 1                                \
 401           : (charset) < MIN_CHARSET_PRIVATE_DIMENSION2)                     \
 402          ? (((charset) - 0x70) << 7) | ((c1) <= 0 ? 0 : ((c1) & 0x7F))      \
 403          : ((((charset)                                                     \
 404               - ((charset) < MIN_CHARSET_PRIVATE_DIMENSION2 ? 0x8F : 0xE0)) \
 405              << 14)                                                         \
 406             | ((c2) <= 0 ? 0 : ((c2) & 0x7F))                               \
 407             | ((c1) <= 0 ? 0 : (((c1) & 0x7F) << 7))))))
 408
 409
 410 /* If GENERICP is nonzero, return nonzero iff C is a valid normal or
 411    generic character.  If GENERICP is zero, return nonzero iff C is a
 412    valid normal character.  */
 413 #define CHAR_VALID_P(c, genericp)       \
 414   ((c) >= 0                             \
 415    && (SINGLE_BYTE_CHAR_P (c) || char_valid_p (c, genericp)))
 416
 417 /* This default value is used when nonascii-translation-table or
 418    nonascii-insert-offset fail to convert unibyte character to a valid
 419    multibyte character.  This makes a Latin-1 character.  */
 420
 421 #define DEFAULT_NONASCII_INSERT_OFFSET 0x800
 422
 423 /* Parse multibyte string STR of length LENGTH and set BYTES to the
 424    byte length of a character at STR.  */
 425
 426 #ifdef BYTE_COMBINING_DEBUG
 427
 428 #define PARSE_MULTIBYTE_SEQ(str, length, bytes)                 \
 429   do {                                                          \
 430     int i = 1;                                                  \
 431     while (i < (length) && ! CHAR_HEAD_P ((str)[i])) i++;       \
 432     (bytes) = BYTES_BY_CHAR_HEAD ((str)[0]);                    \
 433     if ((bytes) > i)                                            \
 434       abort ();                                                 \
 435   } while (0)
 436
 437 #else  /* not BYTE_COMBINING_DEBUG */
 438
 439 #define PARSE_MULTIBYTE_SEQ(str, length, bytes) \
 440   ((void)(length), (bytes) = BYTES_BY_CHAR_HEAD ((str)[0]))
 441
 442 #endif /* not BYTE_COMBINING_DEBUG */
 443
 444 /* Return 1 iff the byte sequence at unibyte string STR (LENGTH bytes)
 445    is valid as a multibyte form.  If valid, by a side effect, BYTES is
 446    set to the byte length of the multibyte form.  */
 447
 448 #define UNIBYTE_STR_AS_MULTIBYTE_P(str, length, bytes)  \
 449   (((str)[0] < 0x80 || (str)[0] >= 0xA0)                \
 450    ? ((bytes) = 1)                                      \
 451    : (((bytes) = BYTES_BY_CHAR_HEAD ((str)[0])),        \
 452       ((bytes) > 1 && (bytes) <= (length)               \
 453        && (str)[0] != LEADING_CODE_8_BIT_CONTROL        \
 454        && !CHAR_HEAD_P ((str)[1])                       \
 455        && ((bytes) == 2                                 \
 456            || (!CHAR_HEAD_P ((str)[2])                  \
 457                && ((bytes) == 3                         \
 458                    || !CHAR_HEAD_P ((str)[3])))))))
 459
 460 /* Return 1 iff the byte sequence at multibyte string STR is valid as
 461    a unibyte form.  By a side effect, BYTES is set to the byte length
 462    of one character at STR.  */
 463
 464 #define MULTIBYTE_STR_AS_UNIBYTE_P(str, bytes)  \
 465   ((bytes) = BYTES_BY_CHAR_HEAD ((str)[0]),     \
 466    (str)[0] != LEADING_CODE_8_BIT_CONTROL)
 467
 468 /* The charset of character C is stored in CHARSET, and the
 469    position-codes of C are stored in C1 and C2.
 470    We store -1 in C2 if the dimension of the charset is 1.  */
 471
 472 #define SPLIT_CHAR(c, charset, c1, c2)                                      \
 473   (SINGLE_BYTE_CHAR_P (c)                                                   \
 474    ? ((charset                                                              \
 475        = (ASCII_BYTE_P (c)                                                  \
 476           ? CHARSET_ASCII                                                   \
 477           : ((c) < 0xA0 ? CHARSET_8_BIT_CONTROL : CHARSET_8_BIT_GRAPHIC))), \
 478       c1 = (c), c2 = -1)                                                    \
 479    : ((c) & CHAR_FIELD1_MASK                                                \
 480       ? (charset = (CHAR_FIELD1 (c)                                         \
 481                     + ((c) < MIN_CHAR_PRIVATE_DIMENSION2 ? 0x8F : 0xE0)),   \
 482          c1 = CHAR_FIELD2 (c),                                              \
 483          c2 = CHAR_FIELD3 (c))                                              \
 484       : (charset = CHAR_FIELD2 (c) + 0x70,                                  \
 485          c1 = CHAR_FIELD3 (c),                                              \
 486          c2 = -1)))
 487
 488 /* Return 1 iff character C has valid printable glyph.  */
 489 #define CHAR_PRINTABLE_P(c) (ASCII_BYTE_P (c) || char_printable_p (c))
 490
 491 /* The charset of the character at STR is stored in CHARSET, and the
 492    position-codes are stored in C1 and C2.
 493    We store -1 in C2 if the character is just 2 bytes.  */
 494
 495 #define SPLIT_STRING(str, len, charset, c1, c2)                 \
 496   ((BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) < 2             \
 497     || BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) > len        \
 498     || split_string (str, len, &charset, &c1, &c2) < 0)         \
 499    ? c1 = *(str), charset = CHARSET_ASCII                       \
 500    : charset)
 501
 502 /* Mapping table from ISO2022's charset (specified by DIMENSION,
 503    CHARS, and FINAL_CHAR) to Emacs' charset.  Should be accessed by
 504    macro ISO_CHARSET_TABLE (DIMENSION, CHARS, FINAL_CHAR).  */
 505 extern int iso_charset_table[2][2][128];
 506
 507 #define ISO_CHARSET_TABLE(dimension, chars, final_char) \
 508   iso_charset_table[XINT (dimension) - 1][XINT (chars) > 94][XINT (final_char)]
 509
 510 #define BASE_LEADING_CODE_P(c) (BYTES_BY_CHAR_HEAD ((unsigned char) (c)) > 1)
 511
 512 /* Return how many bytes C will occupy in a multibyte buffer.  */
 513 #define CHAR_BYTES(c)                                   \
 514   (SINGLE_BYTE_CHAR_P (c)                               \
 515    ? ((ASCII_BYTE_P (c) || (c) >= 0xA0) ? 1 : 2)        \
 516    : char_bytes (c))
 517
 518 /* The following two macros CHAR_STRING and STRING_CHAR are the main
 519    entry points to convert between Emacs's two types of character
 520    representations: multi-byte form and single-word form (character
 521    code).  */
 522
 523 /* Store multi-byte form of the character C in STR.  The caller should
 524    allocate at least MAX_MULTIBYTE_LENGTH bytes area at STR in
 525    advance.  Returns the length of the multi-byte form.  If C is an
 526    invalid character code, signal an error.  */
 527
 528 #define CHAR_STRING(c, str)                                               \
 529   (SINGLE_BYTE_CHAR_P (c)                                                 \
 530    ? ((ASCII_BYTE_P (c) || c >= 0xA0)                                     \
 531       ? (*(str) = (unsigned char)(c), 1)                                  \
 532       : (*(str) = LEADING_CODE_8_BIT_CONTROL, *((str)+ 1) = c + 0x20, 2)) \
 533    : char_to_string (c, (unsigned char *) str))
 534
 535 /* Like CHAR_STRING but don't signal an error if C is invalid.
 536    Value is -1 in this case.  */
 537
 538 #define CHAR_STRING_NO_SIGNAL(c, str)                                     \
 539   (SINGLE_BYTE_CHAR_P (c)                                                 \
 540    ? ((ASCII_BYTE_P (c) || c >= 0xA0)                                     \
 541       ? (*(str) = (unsigned char)(c), 1)                                  \
 542       : (*(str) = LEADING_CODE_8_BIT_CONTROL, *((str)+ 1) = c + 0x20, 2)) \
 543    : char_to_string_1 (c, (unsigned char *) str))
 544
 545 /* Return a character code of the character of which multi-byte form
 546    is at STR and the length is LEN.  If STR doesn't contain valid
 547    multi-byte form, only the first byte in STR is returned.  */
 548
 549 #define STRING_CHAR(str, len)                           \
 550   (BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) == 1     \
 551    ? (unsigned char) *(str)                             \
 552    : string_to_char (str, len, 0))
 553
 554 /* This is like STRING_CHAR but the third arg ACTUAL_LEN is set to the
 555    length of the multi-byte form.  Just to know the length, use
 556    MULTIBYTE_FORM_LENGTH.  */
 557
 558 #define STRING_CHAR_AND_LENGTH(str, len, actual_len)    \
 559   (BYTES_BY_CHAR_HEAD ((unsigned char) *(str)) == 1     \
 560    ? ((actual_len) = 1), (unsigned char) *(str)         \
 561    : string_to_char (str, len, &(actual_len)))
 562
 563 /* Fetch the "next" character from Lisp string STRING at byte position
 564    BYTEIDX, character position CHARIDX.  Store it into OUTPUT.
 565
 566    All the args must be side-effect-free.
 567    BYTEIDX and CHARIDX must be lvalues;
 568    we increment them past the character fetched.  */
 569
 570 #define FETCH_STRING_CHAR_ADVANCE(OUTPUT, STRING, CHARIDX, BYTEIDX)        \
 571 if (1)                                                                     \
 572   {                                                                        \
 573     CHARIDX++;                                                             \
 574     if (STRING_MULTIBYTE (STRING))                                         \
 575       {                                                                    \
 576         const unsigned char *ptr = SDATA (STRING) + BYTEIDX;               \
 577         int space_left = SBYTES (STRING) - BYTEIDX;                        \
 578         int actual_len;                                                    \
 579                                                                            \
 580         OUTPUT = STRING_CHAR_AND_LENGTH (ptr, space_left, actual_len);     \
 581         BYTEIDX += actual_len;                                             \
 582       }                                                                    \
 583     else                                                                   \
 584       OUTPUT = SREF (STRING, BYTEIDX++);                                   \
 585   }                                                                        \
 586 else
 587
 588 /* Like FETCH_STRING_CHAR_ADVANCE but assume STRING is multibyte.  */
 589
 590 #define FETCH_STRING_CHAR_ADVANCE_NO_CHECK(OUTPUT, STRING, CHARIDX, BYTEIDX)  \
 591 if (1)                                                                        \
 592   {                                                                           \
 593     const unsigned char *fetch_string_char_ptr = SDATA (STRING) + BYTEIDX;    \
 594     int fetch_string_char_space_left = SBYTES (STRING) - BYTEIDX;             \
 595     int actual_len;                                                           \
 596                                                                               \
 597     OUTPUT                                                                    \
 598       = STRING_CHAR_AND_LENGTH (fetch_string_char_ptr,                        \
 599                                 fetch_string_char_space_left, actual_len);    \
 600                                                                               \
 601     BYTEIDX += actual_len;                                                    \
 602     CHARIDX++;                                                                \
 603   }                                                                           \
 604 else
 605
 606 /* Like FETCH_STRING_CHAR_ADVANCE but fetch character from the current
 607    buffer.  */
 608
 609 #define FETCH_CHAR_ADVANCE(OUTPUT, CHARIDX, BYTEIDX)                      \
 610 if (1)                                                                    \
 611   {                                                                       \
 612     CHARIDX++;                                                            \
 613     if (!NILP (current_buffer->enable_multibyte_characters))              \
 614       {                                                                   \
 615         unsigned char *ptr = BYTE_POS_ADDR (BYTEIDX);                     \
 616         int space_left = ((CHARIDX < GPT ? GPT_BYTE : Z_BYTE) - BYTEIDX); \
 617         int actual_len;                                                   \
 618                                                                           \
 619         OUTPUT= STRING_CHAR_AND_LENGTH (ptr, space_left, actual_len);     \
 620         BYTEIDX += actual_len;                                            \
 621       }                                                                   \
 622     else                                                                  \
 623       {                                                                   \
 624         OUTPUT = *(BYTE_POS_ADDR (BYTEIDX));                              \
 625         BYTEIDX++;                                                        \
 626       }                                                                   \
 627   }                                                                       \
 628 else
 629
 630 /* Return the length of the multi-byte form at string STR of length LEN.  */
 631
 632 #define MULTIBYTE_FORM_LENGTH(str, len)                 \
 633   (BYTES_BY_CHAR_HEAD (*(unsigned char *)(str)) == 1    \
 634    ? 1                                                  \
 635    : multibyte_form_length (str, len))
 636
 637 /* If P is before LIMIT, advance P to the next character boundary.  It
 638    assumes that P is already at a character boundary of the sane
 639    mulitbyte form whose end address is LIMIT.  */
 640
 641 #define NEXT_CHAR_BOUNDARY(p, limit)    \
 642   do {                                  \
 643     if ((p) < (limit))                  \
 644       (p) += BYTES_BY_CHAR_HEAD (*(p)); \
 645   } while (0)
 646
 647
 648 /* If P is after LIMIT, advance P to the previous character boundary.
 649    It assumes that P is already at a character boundary of the sane
 650    mulitbyte form whose beginning address is LIMIT.  */
 651
 652 #define PREV_CHAR_BOUNDARY(p, limit)                                    \
 653   do {                                                                  \
 654     if ((p) > (limit))                                                  \
 655       {                                                                 \
 656         const unsigned char *p0 = (p);                                  \
 657         do {                                                            \
 658           p0--;                                                         \
 659         } while (p0 >= limit && ! CHAR_HEAD_P (*p0));                   \
 660         (p) = (BYTES_BY_CHAR_HEAD (*p0) == (p) - p0) ? p0 : (p) - 1;    \
 661       }                                                                 \
 662   } while (0)
 663
 664
 665 #ifdef emacs
 666
 667 /* Increase the buffer byte position POS_BYTE of the current buffer to
 668    the next character boundary.  This macro relies on the fact that
 669    *GPT_ADDR and *Z_ADDR are always accessible and the values are
 670    '\0'.  No range checking of POS.  */
 671
 672 #ifdef BYTE_COMBINING_DEBUG
 673
 674 #define INC_POS(pos_byte)                               \
 675   do {                                                  \
 676     unsigned char *p = BYTE_POS_ADDR (pos_byte);        \
 677     if (BASE_LEADING_CODE_P (*p))                       \
 678       {                                                 \
 679         int len, bytes;                                 \
 680         len = Z_BYTE - pos_byte;                        \
 681         PARSE_MULTIBYTE_SEQ (p, len, bytes);            \
 682         pos_byte += bytes;                              \
 683       }                                                 \
 684     else                                                \
 685       pos_byte++;                                       \
 686   } while (0)
 687
 688 #else  /* not BYTE_COMBINING_DEBUG */
 689
 690 #define INC_POS(pos_byte)                               \
 691   do {                                                  \
 692     unsigned char *p = BYTE_POS_ADDR (pos_byte);        \
 693     pos_byte += BYTES_BY_CHAR_HEAD (*p);                \
 694   } while (0)
 695
 696 #endif /* not BYTE_COMBINING_DEBUG */
 697
 698 /* Decrease the buffer byte position POS_BYTE of the current buffer to
 699    the previous character boundary.  No range checking of POS.  */
 700 #define DEC_POS(pos_byte)                                               \
 701   do {                                                                  \
 702     unsigned char *p, *p_min;                                           \
 703                                                                         \
 704     pos_byte--;                                                         \
 705     if (pos_byte < GPT_BYTE)                                            \
 706       p = BEG_ADDR + pos_byte - BEG_BYTE, p_min = BEG_ADDR;             \
 707     else                                                                \
 708       p = BEG_ADDR + GAP_SIZE + pos_byte - BEG_BYTE, p_min = GAP_END_ADDR;\
 709     if (p > p_min && !CHAR_HEAD_P (*p))                                 \
 710       {                                                                 \
 711         unsigned char *pend = p--;                                      \
 712         int len, bytes;                                                 \
 713         if (p_min < p - MAX_MULTIBYTE_LENGTH)                           \
 714           p_min = p - MAX_MULTIBYTE_LENGTH;                             \
 715         while (p > p_min && !CHAR_HEAD_P (*p)) p--;                     \
 716         len = pend + 1 - p;                                             \
 717         PARSE_MULTIBYTE_SEQ (p, len, bytes);                            \
 718         if (bytes == len)                                               \
 719           pos_byte -= len - 1;                                          \
 720       }                                                                 \
 721   } while (0)
 722
 723 /* Increment both CHARPOS and BYTEPOS, each in the appropriate way.  */
 724
 725 #define INC_BOTH(charpos, bytepos)                              \
 726 do                                                              \
 727   {                                                             \
 728     (charpos)++;                                                \
 729     if (NILP (current_buffer->enable_multibyte_characters))     \
 730       (bytepos)++;                                              \
 731     else                                                        \
 732       INC_POS ((bytepos));                                      \
 733   }                                                             \
 734 while (0)
 735
 736 /* Decrement both CHARPOS and BYTEPOS, each in the appropriate way.  */
 737
 738 #define DEC_BOTH(charpos, bytepos)                              \
 739 do                                                              \
 740   {                                                             \
 741     (charpos)--;                                                \
 742     if (NILP (current_buffer->enable_multibyte_characters))     \
 743       (bytepos)--;                                              \
 744     else                                                        \
 745       DEC_POS ((bytepos));                                      \
 746   }                                                             \
 747 while (0)
 748
 749 /* Increase the buffer byte position POS_BYTE of the current buffer to
 750    the next character boundary.  This macro relies on the fact that
 751    *GPT_ADDR and *Z_ADDR are always accessible and the values are
 752    '\0'.  No range checking of POS_BYTE.  */
 753
 754 #ifdef BYTE_COMBINING_DEBUG
 755
 756 #define BUF_INC_POS(buf, pos_byte)                              \
 757   do {                                                          \
 758     unsigned char *p = BUF_BYTE_ADDRESS (buf, pos_byte);        \
 759     if (BASE_LEADING_CODE_P (*p))                               \
 760       {                                                         \
 761         int len, bytes;                                         \
 762         len = BUF_Z_BYTE (buf) - pos_byte;                      \
 763         PARSE_MULTIBYTE_SEQ (p, len, bytes);                    \
 764         pos_byte += bytes;                                      \
 765       }                                                         \
 766     else                                                        \
 767       pos_byte++;                                               \
 768   } while (0)
 769
 770 #else  /* not BYTE_COMBINING_DEBUG */
 771
 772 #define BUF_INC_POS(buf, pos_byte)                              \
 773   do {                                                          \
 774     unsigned char *p = BUF_BYTE_ADDRESS (buf, pos_byte);        \
 775     pos_byte += BYTES_BY_CHAR_HEAD (*p);                        \
 776   } while (0)
 777
 778 #endif /* not BYTE_COMBINING_DEBUG */
 779
 780 /* Decrease the buffer byte position POS_BYTE of the current buffer to
 781    the previous character boundary.  No range checking of POS_BYTE.  */
 782 #define BUF_DEC_POS(buf, pos_byte)                                      \
 783   do {                                                                  \
 784     unsigned char *p, *p_min;                                           \
 785     pos_byte--;                                                         \
 786     if (pos_byte < BUF_GPT_BYTE (buf))                                  \
 787       {                                                                 \
 788         p = BUF_BEG_ADDR (buf) + pos_byte - BEG_BYTE;                   \
 789         p_min = BUF_BEG_ADDR (buf);                                     \
 790       }                                                                 \
 791     else                                                                \
 792       {                                                                 \
 793         p = BUF_BEG_ADDR (buf) + BUF_GAP_SIZE (buf) + pos_byte - BEG_BYTE;\
 794         p_min = BUF_GAP_END_ADDR (buf);                                 \
 795       }                                                                 \
 796     if (p > p_min && !CHAR_HEAD_P (*p))                                 \
 797       {                                                                 \
 798         unsigned char *pend = p--;                                      \
 799         int len, bytes;                                                 \
 800         if (p_min < p - MAX_MULTIBYTE_LENGTH)                           \
 801           p_min = p - MAX_MULTIBYTE_LENGTH;                             \
 802         while (p > p_min && !CHAR_HEAD_P (*p)) p--;                     \
 803         len = pend + 1 - p;                                             \
 804         PARSE_MULTIBYTE_SEQ (p, len, bytes);                            \
 805         if (bytes == len)                                               \
 806           pos_byte -= len - 1;                                          \
 807       }                                                                 \
 808   } while (0)
 809
 810 #endif /* emacs */
 811
 812 /* This is the maximum byte length of multi-byte sequence.  */
 813 #define MAX_MULTIBYTE_LENGTH 4
 814
 815 extern void invalid_character P_ ((int));
 816
 817 extern int translate_char P_ ((Lisp_Object, int, int, int, int));
 818 extern int split_string P_ ((const unsigned char *, int, int *,
 819                                        unsigned char *, unsigned char *));
 820 extern int char_to_string P_ ((int, unsigned char *));
 821 extern int char_to_string_1 P_ ((int, unsigned char *));
 822 extern int string_to_char P_ ((const unsigned char *, int, int *));
 823 extern int char_printable_p P_ ((int c));
 824 extern int multibyte_form_length P_ ((const unsigned char *, int));
 825 extern void parse_str_as_multibyte P_ ((const unsigned char *, int, int *,
 826                                         int *));
 827 extern int str_as_multibyte P_ ((unsigned char *, int, int, int *));
 828 extern int parse_str_to_multibyte P_ ((unsigned char *, int));
 829 extern int str_to_multibyte P_ ((unsigned char *, int, int));
 830 extern int str_as_unibyte P_ ((unsigned char *, int));
 831 extern int get_charset_id P_ ((Lisp_Object));
 832 extern int find_charset_in_text P_ ((const unsigned char *, int, int, int *,
 833                                     Lisp_Object));
 834 extern int strwidth P_ ((unsigned char *, int));
 835 extern int c_string_width P_ ((const unsigned char *, int, int, int *, int *));
 836 extern int lisp_string_width P_ ((Lisp_Object, int, int *, int *));
 837 extern int char_bytes P_ ((int));
 838 extern int char_valid_p P_ ((int, int));
 839
 840 EXFUN (Funibyte_char_to_multibyte, 1);
 841
 842 extern Lisp_Object Vtranslation_table_vector;
 843
 844 /* Return a translation table of id number ID.  */
 845 #define GET_TRANSLATION_TABLE(id) \
 846   (XCDR(XVECTOR(Vtranslation_table_vector)->contents[(id)]))
 847
 848 /* A char-table for characters which may invoke auto-filling.  */
 849 extern Lisp_Object Vauto_fill_chars;
 850
 851 /* Copy LEN bytes from FROM to TO.  This macro should be used only
 852    when a caller knows that LEN is short and the obvious copy loop is
 853    faster than calling bcopy which has some overhead.  Copying a
 854    multibyte sequence of a multibyte character is the typical case.  */
 855
 856 #define BCOPY_SHORT(from, to, len)              \
 857   do {                                          \
 858     int i = len;                                \
 859     const unsigned char *from_p = from;         \
 860     unsigned char *to_p = to;                   \
 861     while (i--) *to_p++ = *from_p++;            \
 862   } while (0)
 863
 864 #endif /* EMACS_CHARSET_H */