Mercurial > emacs

comparison src/coding.h @ 89483:2f877ed80fa6

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
*** empty log message ***
author Kenichi Handa <handa@m17n.org>
date Mon, 08 Sep 2003 12:53:41 +0000
parents 375f2633d815 e8f4dff2bfc1
children 4df223da38ce
comparison
equal deleted inserted replaced
88123:375f2633d815 89483:2f877ed80fa6
1 /* Header for coding system handler. 1 /* Header for coding system handler.
2 Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN. 2 Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN.
3 Licensed to the Free Software Foundation. 3 Licensed to the Free Software Foundation.
4 Copyright (C) 2003
5 National Institute of Advanced Industrial Science and Technology (AIST)
6 Registration Number H13PRO009
4 7
5 This file is part of GNU Emacs. 8 This file is part of GNU Emacs.
6 9
7 GNU Emacs is free software; you can redistribute it and/or modify 10 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 11 it under the terms of the GNU General Public License as published by
20 Boston, MA 02111-1307, USA. */ 23 Boston, MA 02111-1307, USA. */
21 24
22 #ifndef EMACS_CODING_H 25 #ifndef EMACS_CODING_H
23 #define EMACS_CODING_H 26 #define EMACS_CODING_H
24 27
25 #include "ccl.h" 28 /* Index to arguments of Fdefine_coding_system_internal. */
26 29
27 /*** EMACS' INTERNAL FORMAT (emacs-mule) section ***/ 30 enum define_coding_system_arg_index
28 31 {
29 /* All code (1-byte) of Emacs' internal format is classified into one 32 coding_arg_name,
30 of the followings. See also `charset.h'. */ 33 coding_arg_mnemonic,
31 enum emacs_code_class_type 34 coding_arg_coding_type,
32 { 35 coding_arg_charset_list,
33 EMACS_control_code, /* Control codes in the range 36 coding_arg_ascii_compatible_p,
34 0x00..0x1F and 0x7F except for the 37 coding_arg_decode_translation_table,
35 following two codes. */ 38 coding_arg_encode_translation_table,
36 EMACS_linefeed_code, /* 0x0A (linefeed) to denote 39 coding_arg_post_read_conversion,
37 end-of-line. */ 40 coding_arg_pre_write_conversion,
38 EMACS_carriage_return_code, /* 0x0D (carriage-return) to be used 41 coding_arg_default_char,
39 in selective display mode. */ 42 coding_arg_for_unibyte,
40 EMACS_ascii_code, /* ASCII characters. */ 43 coding_arg_plist,
41 EMACS_leading_code_2, /* Base leading code of official 44 coding_arg_eol_type,
42 TYPE9N character. */ 45 coding_arg_max
43 EMACS_leading_code_3, /* Base leading code of private TYPE9N 46 };
44 or official TYPE9Nx9N character. */ 47
45 EMACS_leading_code_4, /* Base leading code of private 48 enum define_coding_iso2022_arg_index
46 TYPE9Nx9N character. */ 49 {
47 EMACS_invalid_code /* Invalid code, i.e. a base leading 50 coding_arg_iso2022_initial = coding_arg_max,
48 code not yet assigned to any 51 coding_arg_iso2022_reg_usage,
49 charset, or a code of the range 52 coding_arg_iso2022_request,
50 0xA0..0xFF. */ 53 coding_arg_iso2022_flags,
51 }; 54 coding_arg_iso2022_max
52 55 };
53 extern enum emacs_code_class_type emacs_code_class[256]; 56
54 57 enum define_coding_utf16_arg_index
55 /*** ISO2022 section ***/ 58 {
56 59 coding_arg_utf16_bom = coding_arg_max,
57 /* Macros to define code of control characters for ISO2022's functions. */ 60 coding_arg_utf16_endian,
58 /* code */ /* function */ 61 coding_arg_utf16_max
59 #define ISO_CODE_LF 0x0A /* line-feed */ 62 };
60 #define ISO_CODE_CR 0x0D /* carriage-return */ 63
61 #define ISO_CODE_SO 0x0E /* shift-out */ 64 enum define_coding_ccl_arg_index
62 #define ISO_CODE_SI 0x0F /* shift-in */ 65 {
63 #define ISO_CODE_SS2_7 0x19 /* single-shift-2 for 7-bit code */ 66 coding_arg_ccl_decoder = coding_arg_max,
64 #define ISO_CODE_ESC 0x1B /* escape */ 67 coding_arg_ccl_encoder,
65 #define ISO_CODE_SS2 0x8E /* single-shift-2 */ 68 coding_arg_ccl_valids,
66 #define ISO_CODE_SS3 0x8F /* single-shift-3 */ 69 coding_arg_ccl_max
67 #define ISO_CODE_CSI 0x9B /* control-sequence-introduce */ 70 };
68 71
69 /* All code (1-byte) of ISO2022 is classified into one of the 72 extern Lisp_Object Vcoding_system_hash_table;
70 followings. */ 73
71 enum iso_code_class_type 74 /* Enumeration of coding system type. */
72 { 75
73 ISO_control_0, /* Control codes in the range 76 enum coding_system_type
74 0x00..0x1F and 0x7F, except for the 77 {
75 following 5 codes. */ 78 coding_type_charset,
76 ISO_carriage_return, /* ISO_CODE_CR (0x0D) */ 79 coding_type_utf_8,
77 ISO_shift_out, /* ISO_CODE_SO (0x0E) */ 80 coding_type_utf_16,
78 ISO_shift_in, /* ISO_CODE_SI (0x0F) */ 81 coding_type_iso_2022,
79 ISO_single_shift_2_7, /* ISO_CODE_SS2_7 (0x19) */ 82 coding_type_emacs_mule,
80 ISO_escape, /* ISO_CODE_SO (0x1B) */ 83 coding_type_sjis,
81 ISO_control_1, /* Control codes in the range 84 coding_type_ccl,
82 0x80..0x9F, except for the 85 coding_type_raw_text,
83 following 3 codes. */ 86 coding_type_undecided,
84 ISO_single_shift_2, /* ISO_CODE_SS2 (0x8E) */ 87 coding_type_max
85 ISO_single_shift_3, /* ISO_CODE_SS3 (0x8F) */ 88 };
86 ISO_control_sequence_introducer, /* ISO_CODE_CSI (0x9B) */ 89
87 ISO_0x20_or_0x7F, /* Codes of the values 0x20 or 0x7F. */ 90
88 ISO_graphic_plane_0, /* Graphic codes in the range 0x21..0x7E. */ 91 /* Enumeration of end-of-line format type. */
89 ISO_0xA0_or_0xFF, /* Codes of the values 0xA0 or 0xFF. */ 92
90 ISO_graphic_plane_1 /* Graphic codes in the range 0xA1..0xFE. */ 93 enum end_of_line_type
91 }; 94 {
92 95 eol_lf, /* Line-feed only, same as Emacs' internal
93 /** The macros CODING_FLAG_ISO_XXX defines a flag bit of the `flags' 96 format. */
94 element in the structure `coding_system'. This information is used 97 eol_crlf, /* Sequence of carriage-return and
95 while encoding a text to ISO2022. **/ 98 line-feed. */
96 99 eol_cr, /* Carriage-return only. */
97 /* If set, produce short-form designation sequence (e.g. ESC $ A) 100 eol_any, /* Accept any of above. Produce line-feed
98 instead of long-form sequence (e.g. ESC $ ( A). */ 101 only. */
99 #define CODING_FLAG_ISO_SHORT_FORM 0x0001 102 eol_undecided, /* This value is used to denote that the
100 103 eol-type is not yet undecided. */
101 /* If set, reset graphic planes and registers at end-of-line to the 104 eol_type_max
102 initial state. */ 105 };
103 #define CODING_FLAG_ISO_RESET_AT_EOL 0x0002 106
104 107 /* Enumeration of index to an attribute vector of a coding system. */
105 /* If set, reset graphic planes and registers before any control 108
106 characters to the initial state. */ 109 enum coding_attr_index
107 #define CODING_FLAG_ISO_RESET_AT_CNTL 0x0004 110 {
108 111 coding_attr_base_name,
109 /* If set, encode by 7-bit environment. */ 112 coding_attr_docstring,
110 #define CODING_FLAG_ISO_SEVEN_BITS 0x0008 113 coding_attr_mnemonic,
111 114 coding_attr_type,
112 /* If set, use locking-shift function. */ 115 coding_attr_charset_list,
113 #define CODING_FLAG_ISO_LOCKING_SHIFT 0x0010 116 coding_attr_ascii_compat,
114 117 coding_attr_decode_tbl,
115 /* If set, use single-shift function. Overwrite 118 coding_attr_encode_tbl,
116 CODING_FLAG_ISO_LOCKING_SHIFT. */ 119 coding_attr_post_read,
117 #define CODING_FLAG_ISO_SINGLE_SHIFT 0x0020 120 coding_attr_pre_write,
118 121 coding_attr_default_char,
119 /* If set, designate JISX0201-Roman instead of ASCII. */ 122 coding_attr_for_unibyte,
120 #define CODING_FLAG_ISO_USE_ROMAN 0x0040 123 coding_attr_plist,
121 124
122 /* If set, designate JISX0208-1978 instead of JISX0208-1983. */ 125 coding_attr_category,
123 #define CODING_FLAG_ISO_USE_OLDJIS 0x0080 126 coding_attr_safe_charsets,
124 127
125 /* If set, do not produce ISO6429's direction specifying sequence. */ 128 /* The followings are extra attributes for each type. */
126 #define CODING_FLAG_ISO_NO_DIRECTION 0x0100 129 coding_attr_charset_valids,
127 130
128 /* If set, assume designation states are reset at beginning of line on 131 coding_attr_ccl_decoder,
129 output. */ 132 coding_attr_ccl_encoder,
130 #define CODING_FLAG_ISO_INIT_AT_BOL 0x0200 133 coding_attr_ccl_valids,
131 134
132 /* If set, designation sequence should be placed at beginning of line 135 coding_attr_iso_initial,
133 on output. */ 136 coding_attr_iso_usage,
134 #define CODING_FLAG_ISO_DESIGNATE_AT_BOL 0x0400 137 coding_attr_iso_request,
135 138 coding_attr_iso_flags,
136 /* If set, do not encode unsafe characters on output. */ 139
137 #define CODING_FLAG_ISO_SAFE 0x0800 140 coding_attr_utf_16_bom,
138 141 coding_attr_utf_16_endian,
139 /* If set, extra latin codes (128..159) are accepted as a valid code 142
140 on input. */ 143 coding_attr_emacs_mule_full,
141 #define CODING_FLAG_ISO_LATIN_EXTRA 0x1000 144
142 145 coding_attr_last_index
143 /* If set, use designation escape sequence. */ 146 };
144 #define CODING_FLAG_ISO_DESIGNATION 0x10000 147
145 148
146 /* A character to be produced on output if encoding of the original 149 #define CODING_ATTR_BASE_NAME(attrs) AREF (attrs, coding_attr_base_name)
147 character is inhibitted by CODING_MODE_INHIBIT_UNENCODABLE_CHAR. 150 #define CODING_ATTR_TYPE(attrs) AREF (attrs, coding_attr_type)
148 It must be an ASCII character. */ 151 #define CODING_ATTR_CHARSET_LIST(attrs) AREF (attrs, coding_attr_charset_list)
149 #define CODING_REPLACEMENT_CHARACTER '?' 152 #define CODING_ATTR_MNEMONIC(attrs) AREF (attrs, coding_attr_mnemonic)
150 153 #define CODING_ATTR_DOCSTRING(attrs) AREF (attrs, coding_attr_docstring)
151 /* Structure of the field `spec.iso2022' in the structure `coding_system'. */ 154 #define CODING_ATTR_ASCII_COMPAT(attrs) AREF (attrs, coding_attr_ascii_compat)
152 struct iso2022_spec 155 #define CODING_ATTR_DECODE_TBL(attrs) AREF (attrs, coding_attr_decode_tbl)
156 #define CODING_ATTR_ENCODE_TBL(attrs) AREF (attrs, coding_attr_encode_tbl)
157 #define CODING_ATTR_POST_READ(attrs) AREF (attrs, coding_attr_post_read)
158 #define CODING_ATTR_PRE_WRITE(attrs) AREF (attrs, coding_attr_pre_write)
159 #define CODING_ATTR_DEFAULT_CHAR(attrs) AREF (attrs, coding_attr_default_char)
160 #define CODING_ATTR_FOR_UNIBYTE(attrs) AREF (attrs, coding_attr_for_unibyte)
161 #define CODING_ATTR_FLUSHING(attrs) AREF (attrs, coding_attr_flushing)
162 #define CODING_ATTR_PLIST(attrs) AREF (attrs, coding_attr_plist)
163 #define CODING_ATTR_CATEGORY(attrs) AREF (attrs, coding_attr_category)
164 #define CODING_ATTR_SAFE_CHARSETS(attrs)AREF (attrs, coding_attr_safe_charsets)
165
166
167 #define CODING_ID_ATTRS(id) \
168 (AREF (HASH_VALUE (XHASH_TABLE (Vcoding_system_hash_table), id), 0))
169
170 #define CODING_ID_ALIASES(id) \
171 (AREF (HASH_VALUE (XHASH_TABLE (Vcoding_system_hash_table), id), 1))
172
173 #define CODING_ID_EOL_TYPE(id) \
174 (AREF (HASH_VALUE (XHASH_TABLE (Vcoding_system_hash_table), id), 2))
175
176 #define CODING_ID_NAME(id) \
177 (HASH_KEY (XHASH_TABLE (Vcoding_system_hash_table), id))
178
179 #define CODING_SYSTEM_SPEC(coding_system_symbol) \
180 (Fgethash (coding_system_symbol, Vcoding_system_hash_table, Qnil))
181
182 #define CODING_SYSTEM_ID(coding_system_symbol) \
183 hash_lookup (XHASH_TABLE (Vcoding_system_hash_table), \
184 coding_system_symbol, NULL)
185
186 #define CODING_SYSTEM_P(coding_system_symbol) \
187 (! NILP (CODING_SYSTEM_SPEC (coding_system_symbol)))
188
189 #define CHECK_CODING_SYSTEM(x) \
190 do { \
191 if (!CODING_SYSTEM_P (x)) \
192 wrong_type_argument (Qcoding_system_p, (x)); \
193 } while (0)
194
195
196 #define CHECK_CODING_SYSTEM_GET_SPEC(x, spec) \
197 do { \
198 spec = CODING_SYSTEM_SPEC (x); \
199 if (NILP (spec)) \
200 x = wrong_type_argument (Qcoding_system_p, (x)); \
201 } while (0)
202
203
204 #define CHECK_CODING_SYSTEM_GET_ID(x, id) \
205 do \
206 { \
207 id = CODING_SYSTEM_ID (x); \
208 if (id < 0) \
209 x = wrong_type_argument (Qcoding_system_p, (x)); \
210 } while (0)
211
212
213 /*** GENERAL section ***/
214
215 /* Enumeration of result code of code conversion. */
216 enum coding_result_code
217 {
218 CODING_RESULT_SUCCESS,
219 CODING_RESULT_INSUFFICIENT_SRC,
220 CODING_RESULT_INSUFFICIENT_DST,
221 CODING_RESULT_INCONSISTENT_EOL,
222 CODING_RESULT_INSUFFICIENT_CMP,
223 CODING_RESULT_INTERRUPT,
224 CODING_RESULT_INSUFFICIENT_MEM
225 };
226
227
228 /* Macros used for the member `mode' of the struct coding_system. */
229
230 /* If set, recover the original CR or LF of the already decoded text
231 when the decoding routine encounters an inconsistent eol format. */
232 #define CODING_MODE_INHIBIT_INCONSISTENT_EOL 0x01
233
234 /* If set, the decoding/encoding routines treat the current data as
235 the last block of the whole text to be converted, and do the
236 appropriate finishing job. */
237 #define CODING_MODE_LAST_BLOCK 0x02
238
239 /* If set, it means that the current source text is in a buffer which
240 enables selective display. */
241 #define CODING_MODE_SELECTIVE_DISPLAY 0x04
242
243 /* This flag is used by the decoding/encoding routines on the fly. If
244 set, it means that right-to-left text is being processed. */
245 #define CODING_MODE_DIRECTION 0x08
246
247 #define CODING_MODE_FIXED_DESTINATION 0x10
248
249 #define CODING_MODE_SAFE_ENCODING 0x20
250
251 /* Structure of the field `spec.iso_2022' in the structure
252 `coding_system'. */
253 struct iso_2022_spec
153 { 254 {
255 /* Bit-wise-or of CODING_ISO_FLAG_XXX. */
256 unsigned flags;
257
154 /* The current graphic register invoked to each graphic plane. */ 258 /* The current graphic register invoked to each graphic plane. */
155 int current_invocation[2]; 259 int current_invocation[2];
156 260
157 /* The current charset designated to each graphic register. */ 261 /* The current charset designated to each graphic register. The
262 value -1 means that not charset is designated, -2 means that
263 there was an invalid designation previously. */
158 int current_designation[4]; 264 int current_designation[4];
159
160 /* A charset initially designated to each graphic register. */
161 int initial_designation[4];
162
163 /* If not -1, it is a graphic register specified in an invalid
164 designation sequence. */
165 int last_invalid_designation_register;
166
167 /* A graphic register to which each charset should be designated. */
168 unsigned char requested_designation[MAX_CHARSET + 1];
169
170 /* A revision number to be specified for each charset on encoding.
171 The value 255 means no revision number for the corresponding
172 charset. */
173 unsigned char charset_revision_number[MAX_CHARSET + 1];
174 265
175 /* Set to 1 temporarily only when graphic register 2 or 3 is invoked 266 /* Set to 1 temporarily only when graphic register 2 or 3 is invoked
176 by single-shift while encoding. */ 267 by single-shift while encoding. */
177 int single_shifting; 268 int single_shifting;
178 269
179 /* Set to 1 temporarily only when processing at beginning of line. */ 270 /* Set to 1 temporarily only when processing at beginning of line. */
180 int bol; 271 int bol;
181 }; 272 };
182 273
183 /* Macros to access each field in the structure `spec.iso2022'. */ 274 struct ccl_spec;
184 #define CODING_SPEC_ISO_INVOCATION(coding, plane) \ 275
185 (coding)->spec.iso2022.current_invocation[plane] 276 enum utf_16_bom_type
186 #define CODING_SPEC_ISO_DESIGNATION(coding, reg) \ 277 {
187 (coding)->spec.iso2022.current_designation[reg] 278 utf_16_detect_bom,
188 #define CODING_SPEC_ISO_INITIAL_DESIGNATION(coding, reg) \ 279 utf_16_without_bom,
189 (coding)->spec.iso2022.initial_designation[reg] 280 utf_16_with_bom
190 #define CODING_SPEC_ISO_REQUESTED_DESIGNATION(coding, charset) \ 281 };
191 (coding)->spec.iso2022.requested_designation[charset] 282
192 #define CODING_SPEC_ISO_REVISION_NUMBER(coding, charset) \ 283 enum utf_16_endian_type
193 (coding)->spec.iso2022.charset_revision_number[charset] 284 {
194 #define CODING_SPEC_ISO_SINGLE_SHIFTING(coding) \ 285 utf_16_big_endian,
195 (coding)->spec.iso2022.single_shifting 286 utf_16_little_endian
196 #define CODING_SPEC_ISO_BOL(coding) \ 287 };
197 (coding)->spec.iso2022.bol 288
198 289 struct utf_16_spec
199 /* A value which may appear in
200 coding->spec.iso2022.requested_designation indicating that the
201 corresponding charset does not request any graphic register to be
202 designated. */
203 #define CODING_SPEC_ISO_NO_REQUESTED_DESIGNATION 4
204
205 /* Return a charset which is currently designated to the graphic plane
206 PLANE in the coding-system CODING. */
207 #define CODING_SPEC_ISO_PLANE_CHARSET(coding, plane) \
208 ((CODING_SPEC_ISO_INVOCATION (coding, plane) < 0) \
209 ? -1 \
210 : CODING_SPEC_ISO_DESIGNATION (coding, \
211 CODING_SPEC_ISO_INVOCATION (coding, plane)))
212
213 /*** BIG5 section ***/
214
215 /* Macros to denote each type of BIG5 coding system. */
216 #define CODING_FLAG_BIG5_HKU 0x00 /* BIG5-HKU is one of variants of
217 BIG5 developed by Hong Kong
218 University. */
219 #define CODING_FLAG_BIG5_ETEN 0x01 /* BIG5_ETen is one of variants
220 of BIG5 developed by the
221 company ETen in Taiwan. */
222
223 /*** GENERAL section ***/
224
225 /* Types of coding system. */
226 enum coding_type
227 {
228 coding_type_no_conversion, /* A coding system which requires no
229 conversion for reading and writing
230 including end-of-line format. */
231 coding_type_emacs_mule, /* A coding system used in Emacs'
232 buffer and string. Requires no
233 conversion for reading and writing
234 except for end-of-line format. */
235 coding_type_undecided, /* A coding system which requires
236 automatic detection of a real
237 coding system. */
238 coding_type_sjis, /* SJIS coding system for Japanese. */
239 coding_type_iso2022, /* Any coding system of ISO2022
240 variants. */
241 coding_type_big5, /* BIG5 coding system for Chinese. */
242 coding_type_ccl, /* The coding system of which decoder
243 and encoder are written in CCL. */
244 coding_type_raw_text /* A coding system for a text
245 containing random 8-bit code which
246 does not require code conversion
247 except for end-of-line format. */
248 };
249
250 /* Formats of end-of-line. */
251 #define CODING_EOL_LF 0 /* Line-feed only, same as Emacs'
252 internal format. */
253 #define CODING_EOL_CRLF 1 /* Sequence of carriage-return and
254 line-feed. */
255 #define CODING_EOL_CR 2 /* Carriage-return only. */
256 #define CODING_EOL_UNDECIDED 3 /* This value is used to denote the
257 eol-type is not yet decided. */
258 #define CODING_EOL_INCONSISTENT 4 /* This value is used to denote the
259 eol-type is not consistent
260 through the file. */
261
262 /* 1 iff composing. */
263 #define COMPOSING_P(coding) ((int) coding->composing > (int) COMPOSITION_NO)
264
265 #define COMPOSITION_DATA_SIZE 4080
266 #define COMPOSITION_DATA_MAX_BUNCH_LENGTH (4 + MAX_COMPOSITION_COMPONENTS*2)
267
268 /* Data structure to hold information about compositions of text that
269 is being decoded or encode. ISO 2022 base code conversion routines
270 handle special ESC sequences for composition specification. But,
271 they can't get/put such information directly from/to a buffer in
272 the deepest place. So, they store or retrieve the information
273 through this structure.
274
275 The encoder stores the information in this structure when it meets
276 ESC sequences for composition while encoding codes, then, after all
277 text codes are encoded, puts `composition' properties on the text
278 by referring to the structure.
279
280 The decoder at first stores the information of a text to be
281 decoded, then, while decoding codes, generates ESC sequences for
282 composition at proper places by referring to the structure. */
283
284 struct composition_data
285 { 290 {
286 /* The character position of the first character to be encoded or 291 enum utf_16_bom_type bom;
287 decoded. START and END (see below) are relative to this 292 enum utf_16_endian_type endian;
288 position. */ 293 int surrogate;
289 int char_offset;
290
291 /* The composition data. These elements are repeated for each
292 composition:
293 LENGTH START END METHOD [ COMPONENT ... ]
294 where,
295 LENGTH is the number of elements for this composition.
296
297 START and END are starting and ending character positions of
298 the composition relative to `char_offset'.
299
300 METHOD is one of `enum composing_status' specifying the way of
301 composition.
302
303 COMPONENT is a character or an encoded composition rule. */
304 int data[COMPOSITION_DATA_SIZE];
305
306 /* The number of elements in `data' currently used. */
307 int used;
308
309 /* Pointers to the previous and next structures. When `data' is
310 filled up, another structure is allocated and linked in `next'.
311 The new structure has backward link to this structure in `prev'.
312 The number of chained structures depends on how many compositions
313 the text being encoded or decoded contains. */
314 struct composition_data *prev, *next;
315 }; 294 };
316 295
317 /* Macros used for the member `result' of the struct 296 struct coding_detection_info
318 coding_system. */ 297 {
319 #define CODING_FINISH_NORMAL 0 298 /* Values of these members are bitwise-OR of CATEGORY_MASK_XXXs. */
320 #define CODING_FINISH_INSUFFICIENT_SRC 1 299 /* Which categories are already checked. */
321 #define CODING_FINISH_INSUFFICIENT_DST 2 300 int checked;
322 #define CODING_FINISH_INCONSISTENT_EOL 3 301 /* Which categories are strongly found. */
323 #define CODING_FINISH_INSUFFICIENT_CMP 4 302 int found;
324 #define CODING_FINISH_INTERRUPT 5 303 /* Which categories are rejected. */
325 304 int rejected;
326 /* Macros used for the member `mode' of the struct coding_system. */ 305 };
327 306
328 /* If set, recover the original CR or LF of the already decoded text
329 when the decoding routine encounters an inconsistent eol format. */
330 #define CODING_MODE_INHIBIT_INCONSISTENT_EOL 0x01
331
332 /* If set, the decoding/encoding routines treat the current data as
333 the last block of the whole text to be converted, and do
334 appropriate finishing job. */
335 #define CODING_MODE_LAST_BLOCK 0x02
336
337 /* If set, it means that the current source text is in a buffer which
338 enables selective display. */
339 #define CODING_MODE_SELECTIVE_DISPLAY 0x04
340
341 /* If set, replace unencodabae characters by `?' on encoding. */
342 #define CODING_MODE_INHIBIT_UNENCODABLE_CHAR 0x08
343
344 /* This flag is used by the decoding/encoding routines on the fly. If
345 set, it means that right-to-left text is being processed. */
346 #define CODING_MODE_DIRECTION 0x10
347 307
348 struct coding_system 308 struct coding_system
349 { 309 {
350 /* Type of the coding system. */ 310 /* ID number of the coding system. This is an index to
351 enum coding_type type; 311 Vcoding_system_hash_table. This value is set by
352 312 setup_coding_system. At the early stage of building time, this
353 /* Type of end-of-line format (LF, CRLF, or CR) of the coding system. */ 313 value is -1 in the array coding_categories to indicate that no
354 int eol_type; 314 coding-system of that category is yet defined. */
315 int id;
355 316
356 /* Flag bits of the coding system. The meaning of each bit is common 317 /* Flag bits of the coding system. The meaning of each bit is common
357 to all types of coding systems. */ 318 to all types of coding systems. */
358 unsigned int common_flags; 319 int common_flags;
359
360 /* Flag bits of the coding system. The meaning of each bit depends
361 on the type of the coding system. */
362 unsigned int flags;
363 320
364 /* Mode bits of the coding system. See the comments of the macros 321 /* Mode bits of the coding system. See the comments of the macros
365 CODING_MODE_XXX. */ 322 CODING_MODE_XXX. */
366 unsigned int mode; 323 unsigned int mode;
367 324
368 /* The current status of composition handling. */
369 int composing;
370
371 /* 1 iff the next character is a composition rule. */
372 int composition_rule_follows;
373
374 /* Information of compositions are stored here on decoding and set
375 in advance on encoding. */
376 struct composition_data *cmp_data;
377
378 /* Index to cmp_data->data for the first element for the current
379 composition. */
380 int cmp_data_start;
381
382 /* Index to cmp_data->data for the current element for the current
383 composition. */
384 int cmp_data_index;
385
386 /* Detailed information specific to each type of coding system. */ 325 /* Detailed information specific to each type of coding system. */
387 union spec 326 union
388 { 327 {
389 struct iso2022_spec iso2022; 328 struct iso_2022_spec iso_2022;
390 struct ccl_spec ccl; /* Defined in ccl.h. */ 329 struct ccl_spec *ccl; /* Defined in ccl.h. */
330 struct utf_16_spec utf_16;
331 int emacs_mule_full_support;
391 } spec; 332 } spec;
392 333
393 /* Index number of coding category of the coding system. */ 334 int max_charset_id;
394 int category_idx; 335 char *safe_charsets;
395 336
396 /* The following two members specify how characters 128..159 are 337 /* The following two members specify how binary 8-bit code 128..255
397 represented in source and destination text respectively. 1 means 338 are represented in source and destination text respectively. 1
398 they are represented by 2-byte sequence, 0 means they are 339 means they are represented by 2-byte sequence, 0 means they are
399 represented by 1-byte as is (see the comment in charset.h). */ 340 represented by 1-byte as is (see the comment in character.h). */
400 unsigned src_multibyte : 1; 341 unsigned src_multibyte : 1;
401 unsigned dst_multibyte : 1; 342 unsigned dst_multibyte : 1;
402 343
403 /* How may heading bytes we can skip for decoding. This is set to 344 /* How may heading bytes we can skip for decoding. This is set to
404 -1 in setup_coding_system, and updated by detect_coding. So, 345 -1 in setup_coding_system, and updated by detect_coding. So,
405 when this is equal to the byte length of the text being 346 when this is equal to the byte length of the text being
406 converted, we can skip the actual conversion process. */ 347 converted, we can skip the actual conversion process. */
407 int heading_ascii; 348 int head_ascii;
408 349
409 /* The following members are set by encoding/decoding routine. */ 350 /* The following members are set by encoding/decoding routine. */
410 int produced, produced_char, consumed, consumed_char; 351 EMACS_INT produced, produced_char, consumed, consumed_char;
411 352
412 /* Number of error source data found in a decoding routine. */ 353 /* Number of error source data found in a decoding routine. */
413 int errors; 354 int errors;
414 355
415 /* Finish status of code conversion. It should be one of macros 356 /* Store the positions of error source data. */
416 CODING_FINISH_XXXX. */ 357 EMACS_INT *error_positions;
417 int result; 358
418 359 /* Finish status of code conversion. */
419 /* If nonzero, suppress error notification. */ 360 enum coding_result_code result;
420 int suppress_error;
421 361
422 /* The following members are all Lisp symbols. We don't have to 362 /* The following members are all Lisp symbols. We don't have to
423 protect them from GC because the current garbage collection 363 protect them from GC because the current garbage collection
424 doesn't relocate Lisp symbols. But, when it is changed, we must 364 doesn't relocate Lisp symbols. But, when it is changed, we must
425 find a way to protect them. */ 365 find a way to protect them. */
426 366
427 /* Backward pointer to the Lisp symbol of the coding system. */ 367 EMACS_INT src_pos, src_pos_byte, src_chars, src_bytes;
428 Lisp_Object symbol; 368 Lisp_Object src_object;
429 369 const unsigned char *source;
430 /* Lisp function (symbol) to be called after decoding to do 370
431 additional conversion, or nil. */ 371 EMACS_INT dst_pos, dst_pos_byte, dst_bytes;
432 Lisp_Object post_read_conversion; 372 Lisp_Object dst_object;
433 373 unsigned char *destination;
434 /* Lisp function (symbol) to be called before encoding to do 374
435 additional conversion, or nil. */ 375 int chars_at_source;
436 Lisp_Object pre_write_conversion; 376
437 377 /* If an element is non-negative, it is a character code.
438 /* Character translation tables to look up, or nil. */ 378
439 Lisp_Object translation_table_for_decode; 379 If it is in the range -128..-1, it is a 8-bit character code
440 Lisp_Object translation_table_for_encode; 380 minus 256.
381
382 If it is less than -128, it specifies the start of an annotation
383 chunk. The length of the chunk is -128 minus the value of the
384 element. The following elements are OFFSET, ANNOTATION-TYPE, and
385 a sequence of actual data for the annotation. OFFSET is a
386 character position offset from dst_pos or src_pos,
387 ANNOTATION-TYPE specfies the meaning of the annotation and how to
388 handle the following data.. */
389 int *charbuf;
390 int charbuf_size, charbuf_used;
391
392 /* Set to 1 if charbuf contains an annotation. */
393 int annotated;
394
395 unsigned char carryover[64];
396 int carryover_bytes;
397
398 int default_char;
399
400 int (*detector) P_ ((struct coding_system *,
401 struct coding_detection_info *));
402 void (*decoder) P_ ((struct coding_system *));
403 int (*encoder) P_ ((struct coding_system *));
441 }; 404 };
442 405
443 #define CODING_REQUIRE_FLUSHING_MASK 1 406 /* Meanings of bits in the member `common_flags' of the structure
444 #define CODING_REQUIRE_DECODING_MASK 2 407 coding_system. The lowest 8 bits are reserved for various kind of
445 #define CODING_REQUIRE_ENCODING_MASK 4 408 annotations (currently two of them are used). */
446 #define CODING_REQUIRE_DETECTION_MASK 8 409 #define CODING_ANNOTATION_MASK 0x00FF
447 410 #define CODING_ANNOTATE_COMPOSITION_MASK 0x0001
448 /* Return 1 if the coding system CODING requires specific code to be 411 #define CODING_ANNOTATE_DIRECTION_MASK 0x0002
412 #define CODING_ANNOTATE_CHARSET_MASK 0x0003
413 #define CODING_FOR_UNIBYTE_MASK 0x0100
414 #define CODING_REQUIRE_FLUSHING_MASK 0x0200
415 #define CODING_REQUIRE_DECODING_MASK 0x0400
416 #define CODING_REQUIRE_ENCODING_MASK 0x0800
417 #define CODING_REQUIRE_DETECTION_MASK 0x1000
418 #define CODING_RESET_AT_BOL_MASK 0x2000
419
420 /* Return 1 if the coding context CODING requires annotaion
421 handling. */
422 #define CODING_REQUIRE_ANNOTATION(coding) \
423 ((coding)->common_flags & CODING_ANNOTATION_MASK)
424
425 /* Return 1 if the coding context CODING prefers decoding into unibyte. */
426 #define CODING_FOR_UNIBYTE(coding) \
427 ((coding)->common_flags & CODING_FOR_UNIBYTE_MASK)
428
429 /* Return 1 if the coding context CODING requires specific code to be
449 attached at the tail of converted text. */ 430 attached at the tail of converted text. */
450 #define CODING_REQUIRE_FLUSHING(coding) \ 431 #define CODING_REQUIRE_FLUSHING(coding) \
451 ((coding)->common_flags & CODING_REQUIRE_FLUSHING_MASK) 432 ((coding)->common_flags & CODING_REQUIRE_FLUSHING_MASK)
452 433
453 /* Return 1 if the coding system CODING requires code conversion on 434 /* Return 1 if the coding context CODING requires code conversion on
454 decoding. */ 435 decoding. */
455 #define CODING_REQUIRE_DECODING(coding) \ 436 #define CODING_REQUIRE_DECODING(coding) \
456 ((coding)->dst_multibyte \ 437 ((coding)->dst_multibyte \
457 || (coding)->common_flags & CODING_REQUIRE_DECODING_MASK) 438 || (coding)->common_flags & CODING_REQUIRE_DECODING_MASK)
458 439
459 /* Return 1 if the coding system CODING requires code conversion on 440
441 /* Return 1 if the coding context CODING requires code conversion on
460 encoding. */ 442 encoding. */
461 #define CODING_REQUIRE_ENCODING(coding) \ 443 #define CODING_REQUIRE_ENCODING(coding) \
462 ((coding)->src_multibyte \ 444 ((coding)->src_multibyte \
463 || (coding)->common_flags & CODING_REQUIRE_ENCODING_MASK) 445 || (coding)->common_flags & CODING_REQUIRE_ENCODING_MASK \
464 446 || (coding)->mode & CODING_MODE_SELECTIVE_DISPLAY)
465 /* Return 1 if the coding system CODING requires some kind of code 447
448
449 /* Return 1 if the coding context CODING requires some kind of code
466 detection. */ 450 detection. */
467 #define CODING_REQUIRE_DETECTION(coding) \ 451 #define CODING_REQUIRE_DETECTION(coding) \
468 ((coding)->common_flags & CODING_REQUIRE_DETECTION_MASK) 452 ((coding)->common_flags & CODING_REQUIRE_DETECTION_MASK)
469 453
470 /* Return 1 if the coding system CODING requires code conversion on 454 /* Return 1 if the coding context CODING requires code conversion on
471 decoding or some kind of code detection. */ 455 decoding or some kind of code detection. */
472 #define CODING_MAY_REQUIRE_DECODING(coding) \ 456 #define CODING_MAY_REQUIRE_DECODING(coding) \
473 (CODING_REQUIRE_DECODING (coding) \ 457 (CODING_REQUIRE_DECODING (coding) \
474 || CODING_REQUIRE_DETECTION (coding)) 458 || CODING_REQUIRE_DETECTION (coding))
475 459
476 /* Index for each coding category in `coding_category_table' */
477 #define CODING_CATEGORY_IDX_EMACS_MULE 0
478 #define CODING_CATEGORY_IDX_SJIS 1
479 #define CODING_CATEGORY_IDX_ISO_7 2
480 #define CODING_CATEGORY_IDX_ISO_7_TIGHT 3
481 #define CODING_CATEGORY_IDX_ISO_8_1 4
482 #define CODING_CATEGORY_IDX_ISO_8_2 5
483 #define CODING_CATEGORY_IDX_ISO_7_ELSE 6
484 #define CODING_CATEGORY_IDX_ISO_8_ELSE 7
485 #define CODING_CATEGORY_IDX_CCL 8
486 #define CODING_CATEGORY_IDX_BIG5 9
487 #define CODING_CATEGORY_IDX_UTF_8 10
488 #define CODING_CATEGORY_IDX_UTF_16_BE 11
489 #define CODING_CATEGORY_IDX_UTF_16_LE 12
490 #define CODING_CATEGORY_IDX_RAW_TEXT 13
491 #define CODING_CATEGORY_IDX_BINARY 14
492 #define CODING_CATEGORY_IDX_MAX 15
493
494 /* Definitions of flag bits returned by the function
495 detect_coding_mask (). */
496 #define CODING_CATEGORY_MASK_EMACS_MULE (1 << CODING_CATEGORY_IDX_EMACS_MULE)
497 #define CODING_CATEGORY_MASK_SJIS (1 << CODING_CATEGORY_IDX_SJIS)
498 #define CODING_CATEGORY_MASK_ISO_7 (1 << CODING_CATEGORY_IDX_ISO_7)
499 #define CODING_CATEGORY_MASK_ISO_7_TIGHT (1 << CODING_CATEGORY_IDX_ISO_7_TIGHT)
500 #define CODING_CATEGORY_MASK_ISO_8_1 (1 << CODING_CATEGORY_IDX_ISO_8_1)
501 #define CODING_CATEGORY_MASK_ISO_8_2 (1 << CODING_CATEGORY_IDX_ISO_8_2)
502 #define CODING_CATEGORY_MASK_ISO_7_ELSE (1 << CODING_CATEGORY_IDX_ISO_7_ELSE)
503 #define CODING_CATEGORY_MASK_ISO_8_ELSE (1 << CODING_CATEGORY_IDX_ISO_8_ELSE)
504 #define CODING_CATEGORY_MASK_CCL (1 << CODING_CATEGORY_IDX_CCL)
505 #define CODING_CATEGORY_MASK_BIG5 (1 << CODING_CATEGORY_IDX_BIG5)
506 #define CODING_CATEGORY_MASK_UTF_8 (1 << CODING_CATEGORY_IDX_UTF_8)
507 #define CODING_CATEGORY_MASK_UTF_16_BE (1 << CODING_CATEGORY_IDX_UTF_16_BE)
508 #define CODING_CATEGORY_MASK_UTF_16_LE (1 << CODING_CATEGORY_IDX_UTF_16_LE)
509 #define CODING_CATEGORY_MASK_RAW_TEXT (1 << CODING_CATEGORY_IDX_RAW_TEXT)
510 #define CODING_CATEGORY_MASK_BINARY (1 << CODING_CATEGORY_IDX_BINARY)
511
512 /* This value is returned if detect_coding_mask () find nothing other
513 than ASCII characters. */
514 #define CODING_CATEGORY_MASK_ANY \
515 ( CODING_CATEGORY_MASK_EMACS_MULE \
516 | CODING_CATEGORY_MASK_SJIS \
517 | CODING_CATEGORY_MASK_ISO_7 \
518 | CODING_CATEGORY_MASK_ISO_7_TIGHT \
519 | CODING_CATEGORY_MASK_ISO_8_1 \
520 | CODING_CATEGORY_MASK_ISO_8_2 \
521 | CODING_CATEGORY_MASK_ISO_7_ELSE \
522 | CODING_CATEGORY_MASK_ISO_8_ELSE \
523 | CODING_CATEGORY_MASK_CCL \
524 | CODING_CATEGORY_MASK_BIG5 \
525 | CODING_CATEGORY_MASK_UTF_8 \
526 | CODING_CATEGORY_MASK_UTF_16_BE \
527 | CODING_CATEGORY_MASK_UTF_16_LE)
528
529 #define CODING_CATEGORY_MASK_ISO_7BIT \
530 (CODING_CATEGORY_MASK_ISO_7 | CODING_CATEGORY_MASK_ISO_7_TIGHT)
531
532 #define CODING_CATEGORY_MASK_ISO_8BIT \
533 (CODING_CATEGORY_MASK_ISO_8_1 | CODING_CATEGORY_MASK_ISO_8_2)
534
535 #define CODING_CATEGORY_MASK_ISO_SHIFT \
536 (CODING_CATEGORY_MASK_ISO_7_ELSE | CODING_CATEGORY_MASK_ISO_8_ELSE)
537
538 #define CODING_CATEGORY_MASK_ISO \
539 ( CODING_CATEGORY_MASK_ISO_7BIT \
540 | CODING_CATEGORY_MASK_ISO_SHIFT \
541 | CODING_CATEGORY_MASK_ISO_8BIT)
542
543 #define CODING_CATEGORY_MASK_UTF_16_BE_LE \
544 (CODING_CATEGORY_MASK_UTF_16_BE | CODING_CATEGORY_MASK_UTF_16_LE)
545
546 /* Macros to decode or encode a character of JISX0208 in SJIS. S1 and 460 /* Macros to decode or encode a character of JISX0208 in SJIS. S1 and
547 S2 are the 1st and 2nd position-codes of JISX0208 in SJIS coding 461 S2 are the 1st and 2nd position-codes of JISX0208 in SJIS coding
548 system. C1 and C2 are the 1st and 2nd position codes of Emacs' 462 system. C1 and C2 are the 1st and 2nd position codes of Emacs'
549 internal format. */ 463 internal format. */
550 464
551 #define DECODE_SJIS(s1, s2, c1, c2) \ 465 #define SJIS_TO_JIS(code) \
552 do { \ 466 do { \
553 if (s2 >= 0x9F) \ 467 int s1, s2, j1, j2; \
554 c1 = s1 * 2 - (s1 >= 0xE0 ? 0x160 : 0xE0), \ 468 \
555 c2 = s2 - 0x7E; \ 469 s1 = (code) >> 8, s2 = (code) & 0xFF; \
556 else \ 470 \
557 c1 = s1 * 2 - ((s1 >= 0xE0) ? 0x161 : 0xE1), \ 471 if (s2 >= 0x9F) \
558 c2 = s2 - ((s2 >= 0x7F) ? 0x20 : 0x1F); \ 472 (j1 = s1 * 2 - (s1 >= 0xE0 ? 0x160 : 0xE0), \
473 j2 = s2 - 0x7E); \
474 else \
475 (j1 = s1 * 2 - ((s1 >= 0xE0) ? 0x161 : 0xE1), \
476 j2 = s2 - ((s2 >= 0x7F) ? 0x20 : 0x1F)); \
477 (code) = (j1 << 8) | j2; \
559 } while (0) 478 } while (0)
560 479
561 #define ENCODE_SJIS(c1, c2, s1, s2) \ 480
481 #define JIS_TO_SJIS(code) \
562 do { \ 482 do { \
563 if (c1 & 1) \ 483 int s1, s2, j1, j2; \
564 s1 = c1 / 2 + ((c1 < 0x5F) ? 0x71 : 0xB1), \ 484 \
565 s2 = c2 + ((c2 >= 0x60) ? 0x20 : 0x1F); \ 485 j1 = (code) >> 8, j2 = (code) & 0xFF; \
486 if (j1 & 1) \
487 (s1 = j1 / 2 + ((j1 < 0x5F) ? 0x71 : 0xB1), \
488 s2 = j2 + ((j2 >= 0x60) ? 0x20 : 0x1F)); \
566 else \ 489 else \
567 s1 = c1 / 2 + ((c1 < 0x5F) ? 0x70 : 0xB0), \ 490 (s1 = j1 / 2 + ((j1 < 0x5F) ? 0x70 : 0xB0), \
568 s2 = c2 + 0x7E; \ 491 s2 = j2 + 0x7E); \
492 (code) = (s1 << 8) | s2; \
569 } while (0) 493 } while (0)
494
570 495
571 /* Encode the file name NAME using the specified coding system 496 /* Encode the file name NAME using the specified coding system
572 for file names, if any. */ 497 for file names, if any. */
573 #define ENCODE_FILE(name) \ 498 #define ENCODE_FILE(name) \
574 (! NILP (Vfile_name_coding_system) \ 499 (! NILP (Vfile_name_coding_system) \
577 : (! NILP (Vdefault_file_name_coding_system) \ 502 : (! NILP (Vdefault_file_name_coding_system) \
578 && XFASTINT (Vdefault_file_name_coding_system) != 0 \ 503 && XFASTINT (Vdefault_file_name_coding_system) != 0 \
579 ? code_convert_string_norecord (name, Vdefault_file_name_coding_system, 1) \ 504 ? code_convert_string_norecord (name, Vdefault_file_name_coding_system, 1) \
580 : name)) 505 : name))
581 506
507
582 /* Decode the file name NAME using the specified coding system 508 /* Decode the file name NAME using the specified coding system
583 for file names, if any. */ 509 for file names, if any. */
584 #define DECODE_FILE(name) \ 510 #define DECODE_FILE(name) \
585 (! NILP (Vfile_name_coding_system) \ 511 (! NILP (Vfile_name_coding_system) \
586 && XFASTINT (Vfile_name_coding_system) != 0 \ 512 && XFASTINT (Vfile_name_coding_system) != 0 \
588 : (! NILP (Vdefault_file_name_coding_system) \ 514 : (! NILP (Vdefault_file_name_coding_system) \
589 && XFASTINT (Vdefault_file_name_coding_system) != 0 \ 515 && XFASTINT (Vdefault_file_name_coding_system) != 0 \
590 ? code_convert_string_norecord (name, Vdefault_file_name_coding_system, 0) \ 516 ? code_convert_string_norecord (name, Vdefault_file_name_coding_system, 0) \
591 : name)) 517 : name))
592 518
519
593 #ifdef WINDOWSNT 520 #ifdef WINDOWSNT
594 /* Encode the string STR using the specified coding system 521 /* Encode the string STR using the specified coding system
595 for w32 system functions, if any. */ 522 for w32 system functions, if any. */
596 #define ENCODE_SYSTEM(str) \ 523 #define ENCODE_SYSTEM(str) \
597 (! NILP (Vlocale_coding_system) \ 524 (! NILP (Vlocale_coding_system) \
598 && XFASTINT (Vlocale_coding_system) != 0 \ 525 && XFASTINT (Vlocale_coding_system) != 0 \
599 ? code_convert_string_norecord (str, Vlocale_coding_system, 1) \ 526 ? code_convert_string_norecord (str, Vlocale_coding_system, 1) \
600 : str) 527 : str)
601 528
602 /* Decode the string STR using the specified coding system 529 /* Decode the string STR using the specified coding system
603 for w32 system functions, if any. */ 530 for w32 system functions, if any. */
604 #define DECODE_SYSTEM(name) \ 531 #define DECODE_SYSTEM(name) \
605 (! NILP (Vlocale_coding_system) \ 532 (! NILP (Vlocale_coding_system) \
606 && XFASTINT (Vlocale_coding_system) != 0 \ 533 && XFASTINT (Vlocale_coding_system) != 0 \
607 ? code_convert_string_norecord (str, Vlocale_coding_system, 0) \ 534 ? code_convert_string_norecord (str, Vlocale_coding_system, 0) \
608 : str) 535 : str)
609 536
610 #else /* WINDOWSNT */ 537 #else /* WINDOWSNT */
611 538
612 #define ENCODE_SYSTEM(str) string_make_unibyte(str) 539 #define ENCODE_SYSTEM(str) string_make_unibyte(str)
613 #define DECODE_SYSTEM(name) name 540 #define DECODE_SYSTEM(name) name
614 541
615 #endif /* !WINDOWSNT */ 542 #endif /* !WINDOWSNT */
616 543
617 #define ENCODE_UTF_8(str) code_convert_string_norecord (str, Qutf_8, 1)
618
619 /* Extern declarations. */ 544 /* Extern declarations. */
620 extern int decode_coding P_ ((struct coding_system *, const unsigned char *, 545 extern Lisp_Object make_conversion_work_buffer P_ ((int, int));
621 unsigned char *, int, int));
622 extern int encode_coding P_ ((struct coding_system *, const unsigned char *,
623 unsigned char *, int, int));
624 extern void coding_save_composition P_ ((struct coding_system *, int, int,
625 Lisp_Object));
626 extern void coding_free_composition_data P_ ((struct coding_system *));
627 extern void coding_adjust_composition_offset P_ ((struct coding_system *,
628 int));
629 extern void coding_allocate_composition_data P_ ((struct coding_system *,
630 int));
631 extern void coding_restore_composition P_ ((struct coding_system *,
632 Lisp_Object));
633 extern int code_convert_region P_ ((int, int, int, int, struct coding_system *,
634 int, int));
635 extern Lisp_Object run_pre_post_conversion_on_str P_ ((Lisp_Object,
636 struct coding_system *,
637 int));
638 extern int decoding_buffer_size P_ ((struct coding_system *, int)); 546 extern int decoding_buffer_size P_ ((struct coding_system *, int));
639 extern int encoding_buffer_size P_ ((struct coding_system *, int)); 547 extern int encoding_buffer_size P_ ((struct coding_system *, int));
640 extern void detect_coding P_ ((struct coding_system *, const unsigned char *, 548 extern void setup_coding_system P_ ((Lisp_Object, struct coding_system *));
641 int)); 549 extern void detect_coding P_ ((struct coding_system *));
642 extern void detect_eol P_ ((struct coding_system *, const unsigned char *, 550 extern Lisp_Object code_convert_region P_ ((Lisp_Object, Lisp_Object,
643 int)); 551 Lisp_Object, Lisp_Object,
644 extern int setup_coding_system P_ ((Lisp_Object, struct coding_system *)); 552 int, int));
645 extern Lisp_Object code_convert_string P_ ((Lisp_Object, 553 extern Lisp_Object code_convert_string P_ ((Lisp_Object, Lisp_Object,
646 struct coding_system *, int, int)); 554 Lisp_Object, int, int, int));
647 extern Lisp_Object code_convert_string1 P_ ((Lisp_Object, Lisp_Object,
648 Lisp_Object, int));
649 extern Lisp_Object code_convert_string_norecord P_ ((Lisp_Object, Lisp_Object, 555 extern Lisp_Object code_convert_string_norecord P_ ((Lisp_Object, Lisp_Object,
650 int)); 556 int));
651 extern void setup_raw_text_coding_system P_ ((struct coding_system *)); 557 extern Lisp_Object raw_text_coding_system P_ ((Lisp_Object));
652 extern Lisp_Object encode_coding_string P_ ((Lisp_Object, 558 extern Lisp_Object coding_inherit_eol_type P_ ((Lisp_Object, Lisp_Object));
653 struct coding_system *, int)); 559
654 extern Lisp_Object decode_coding_string P_ ((Lisp_Object, 560 extern int decode_coding_gap P_ ((struct coding_system *,
655 struct coding_system *, int)); 561 EMACS_INT, EMACS_INT));
562 extern int encode_coding_gap P_ ((struct coding_system *,
563 EMACS_INT, EMACS_INT));
564 extern void decode_coding_object P_ ((struct coding_system *,
565 Lisp_Object, EMACS_INT, EMACS_INT,
566 EMACS_INT, EMACS_INT, Lisp_Object));
567 extern void encode_coding_object P_ ((struct coding_system *,
568 Lisp_Object, EMACS_INT, EMACS_INT,
569 EMACS_INT, EMACS_INT, Lisp_Object));
570
571 #define decode_coding_region(coding, from, to) \
572 decode_coding_object (coding, Fcurrent_buffer (), \
573 from, CHAR_TO_BYTE (from), \
574 to, CHAR_TO_BYTE (to), Fcurrent_buffer ())
575
576
577 #define encode_coding_region(coding, from, to) \
578 encode_coding_object (coding, Fcurrent_buffer (), \
579 from, CHAR_TO_BYTE (from), \
580 to, CHAR_TO_BYTE (to), Fcurrent_buffer ())
581
582
583 #define decode_coding_string(coding, string, nocopy) \
584 decode_coding_object (coding, string, 0, 0, XSTRING (string)->size, \
585 STRING_BYTES (XSTRING (string)), Qt)
586
587 #define encode_coding_string(coding, string, nocopy) \
588 (encode_coding_object (coding, string, 0, 0, XSTRING (string)->size, \
589 STRING_BYTES (XSTRING (string)), Qt), \
590 (coding)->dst_object)
591
592
593 #define decode_coding_c_string(coding, src, bytes, dst_object) \
594 do { \
595 (coding)->source = (src); \
596 (coding)->src_chars = (coding)->src_bytes = (bytes); \
597 decode_coding_object ((coding), Qnil, 0, 0, (bytes), (bytes), \
598 (dst_object)); \
599 } while (0)
600
601
602 extern Lisp_Object preferred_coding_system P_ (());
603
604
605 extern Lisp_Object Qutf_8, Qutf_8_emacs;
606
656 extern Lisp_Object Qcoding_system, Qeol_type, Qcoding_category_index; 607 extern Lisp_Object Qcoding_system, Qeol_type, Qcoding_category_index;
657 extern Lisp_Object Qraw_text, Qemacs_mule; 608 extern Lisp_Object Qcoding_system_p;
609 extern Lisp_Object Qraw_text, Qemacs_mule, Qno_conversion, Qundecided;
610 extern Lisp_Object Qiso_2022;
658 extern Lisp_Object Qbuffer_file_coding_system; 611 extern Lisp_Object Qbuffer_file_coding_system;
659 extern Lisp_Object Vcoding_category_list; 612
660 extern Lisp_Object Qutf_8; 613 extern Lisp_Object Qunix, Qdos, Qmac;
661 614
662 extern Lisp_Object Qtranslation_table; 615 extern Lisp_Object Qtranslation_table;
663 extern Lisp_Object Qtranslation_table_id; 616 extern Lisp_Object Qtranslation_table_id;
664 617
665 /* Mnemonic strings to indicate each type of end-of-line. */ 618 /* Mnemonic strings to indicate each type of end-of-line. */
700 /* Coding-system of what is sent from terminal keyboard. This 653 /* Coding-system of what is sent from terminal keyboard. This
701 structure contains information of a coding-system specified by the 654 structure contains information of a coding-system specified by the
702 function `set-keyboard-coding-system'. */ 655 function `set-keyboard-coding-system'. */
703 extern struct coding_system keyboard_coding; 656 extern struct coding_system keyboard_coding;
704 657
705 /* Default coding system to be used to write a file. */
706 extern struct coding_system default_buffer_file_coding;
707
708 /* Default coding systems used for process I/O. */ 658 /* Default coding systems used for process I/O. */
709 extern Lisp_Object Vdefault_process_coding_system; 659 extern Lisp_Object Vdefault_process_coding_system;
710 660
711 /* Function to call to force a user to force select a proper coding 661 /* Function to call to force a user to force select a propert coding
712 system. */ 662 system. */
713 extern Lisp_Object Vselect_safe_coding_system_function; 663 extern Lisp_Object Vselect_safe_coding_system_function;
714 664
715 /* If nonzero, on writing a file, Vselect_safe_coding_system_function 665 /* If nonzero, on writing a file, Vselect_safe_coding_system_function
716 is called even if Vcoding_system_for_write is non-nil. */ 666 is called even if Vcoding_system_for_write is non-nil. */
726 #endif 676 #endif
727 677
728 /* Error signaled when there's a problem with detecting coding system */ 678 /* Error signaled when there's a problem with detecting coding system */
729 extern Lisp_Object Qcoding_system_error; 679 extern Lisp_Object Qcoding_system_error;
730 680
681 extern char emacs_mule_bytes[256];
682 extern int emacs_mule_string_char P_ ((unsigned char *));
683
731 #endif /* EMACS_CODING_H */ 684 #endif /* EMACS_CODING_H */