Commit | Line | Data |
---|---|---|
b7b65b15 EZ |
1 | /* Low-level bidirectional buffer-scanning functions for GNU Emacs. |
2 | Copyright (C) 2000, 2001, 2004, 2005 Free Software Foundation, Inc. | |
3 | ||
4 | This file is part of GNU Emacs. | |
5 | ||
6 | GNU Emacs is free software; you can redistribute it and/or modify | |
7 | it under the terms of the GNU General Public License as published by | |
8 | the Free Software Foundation; either version 2, or (at your option) | |
9 | any later version. | |
10 | ||
11 | GNU Emacs is distributed in the hope that it will be useful, | |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | GNU General Public License for more details. | |
15 | ||
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU Emacs; see the file COPYING. If not, write to | |
19 | the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, | |
20 | Boston, MA 02110-1301, USA. */ | |
21 | ||
22 | /* A sequential implementation of the Unicode Bidirectional algorithm, | |
23 | as per UAX#9, a part of the Unicode Standard. | |
24 | ||
25 | Unlike the reference and most other implementations, this one is | |
26 | designed to be called once for every character in the buffer. | |
27 | ||
28 | The main entry point is bidi_get_next_char_visually. Each time it | |
29 | is called, it finds the next character in the visual order, and | |
30 | returns its information in a special structure. The caller is then | |
31 | expected to process this character for display or any other | |
32 | purposes, and call bidi_get_next_char_visually for the next | |
33 | character. See the comments in bidi_get_next_char_visually for | |
34 | more details about its algorithm that finds the next visual-order | |
35 | character by resolving their levels on the fly. | |
36 | ||
37 | A note about references to UAX#9 rules: if the reference says | |
38 | something like "X9/Retaining", it means that you need to refer to | |
39 | rule X9 and to its modifications decribed in the "Implementation | |
40 | Notes" section of UAX#9, under "Retaining Format Codes". */ | |
41 | ||
42 | #ifdef HAVE_CONFIG_H | |
43 | #include <config.h> | |
44 | #endif | |
45 | ||
46 | #include <stdio.h> | |
47 | ||
48 | #ifdef HAVE_STRING_H | |
49 | #include <string.h> | |
50 | #endif | |
51 | ||
52 | #include "lisp.h" | |
53 | #include "buffer.h" | |
54 | #include "character.h" | |
55 | #include "dispextern.h" | |
56 | ||
57 | static int bidi_initialized = 0; | |
58 | ||
59 | static Lisp_Object bidi_type_table; | |
60 | ||
61 | #define LRM_CHAR 0x200E | |
62 | #define RLM_CHAR 0x200F | |
63 | #define LRE_CHAR 0x202A | |
64 | #define RLE_CHAR 0x202B | |
65 | #define PDF_CHAR 0x202C | |
66 | #define LRO_CHAR 0x202D | |
67 | #define RLO_CHAR 0x202E | |
68 | ||
69 | #define CHARSET_HEBREW 0x88 | |
70 | #define CHARSET_ARABIC 0x87 | |
71 | #define CHARSET_SYRIAC -1 /* these are undefined yet, -1 is invalid */ | |
72 | #define CHARSET_THAANA -1 | |
73 | ||
74 | /* FIXME: need to define wrappers for FETCH_CHAR etc. that return | |
75 | BIDI_EOB when they hit ZV. */ | |
76 | #define BIDI_EOB -1 | |
77 | #define BIDI_BOB -2 | |
78 | ||
79 | #ifdef TEST_STANDALONE | |
80 | /* Testing. */ | |
81 | ||
82 | static unsigned char *input_buf; | |
83 | static size_t input_buf_size; | |
84 | ||
85 | int _fetch_multibyte_char_len, _c_c_; | |
86 | ||
87 | #undef FETCH_CHAR_ADVANCE | |
88 | #define FETCH_CHAR_ADVANCE(ch, cp, bp) \ | |
89 | do { \ | |
90 | ch = input_buf[cp]; \ | |
91 | (cp)++; \ | |
92 | (bp)++; \ | |
93 | if (ch == '\0') \ | |
94 | ch = BIDI_EOB; \ | |
95 | } while (0) | |
96 | ||
97 | #undef FETCH_CHAR | |
98 | #define FETCH_CHAR(n) ((_c_c_ = input_buf[n]) ? _c_c_ : BIDI_EOB) | |
99 | ||
100 | #undef CHAR_CHARSET | |
101 | #define CHAR_CHARSET(c) \ | |
102 | (((c) >= 128 || ((c) < 8 && (c)) || ((c) >= 'A' && (c) < 'X')) \ | |
103 | ? CHARSET_HEBREW \ | |
104 | : ((((c) >= 'X' && (c) <= 'Z') || ((c) >= '6' && (c) <= '9')) \ | |
105 | ? CHARSET_ARABIC \ | |
106 | : CHARSET_ASCII)) | |
107 | ||
108 | #undef CHAR_TO_BYTE | |
109 | #define CHAR_TO_BYTE(pos) (pos) | |
110 | ||
111 | #define char_bytes(ch) 1 | |
112 | ||
113 | #undef LRE_CHAR | |
114 | #undef LRO_CHAR | |
115 | #undef RLE_CHAR | |
116 | #undef RLO_CHAR | |
117 | #undef PDF_CHAR | |
118 | #undef RLM_CHAR | |
119 | #undef LRM_CHAR | |
120 | ||
121 | #define LRE_CHAR 1 | |
122 | #define LRO_CHAR 2 | |
123 | #define RLE_CHAR 3 | |
124 | #define RLO_CHAR 4 | |
125 | #define PDF_CHAR 5 | |
126 | #define RLM_CHAR 6 | |
127 | #define LRM_CHAR 7 | |
128 | ||
129 | static const char *bidi_name[] = | |
130 | { | |
131 | "[???]", "[LRE]", "[LRO]", "[RLE]", "[RLO]", "[PDF]", "[RLM]", "[LRM]" | |
132 | }; | |
133 | ||
134 | #endif /* TEST_STANDALONE */ | |
135 | ||
136 | /* Local data structures. (Look in dispextern.h for the rest.) */ | |
137 | ||
138 | /* What we need to know about the current paragraph. */ | |
139 | struct bidi_paragraph_info { | |
140 | int start_bytepos; /* byte position where it begins */ | |
141 | int end_bytepos; /* byte position where it ends */ | |
142 | int embedding_level; /* its basic embedding level */ | |
143 | bidi_dir_t base_dir; /* its base direction */ | |
144 | }; | |
145 | ||
146 | /* Data type for describing the bidirectional character categories. */ | |
147 | typedef enum { | |
148 | UNKNOWN_BC, | |
149 | NEUTRAL, | |
150 | WEAK, | |
151 | STRONG | |
152 | } bidi_category_t; | |
153 | ||
154 | int bidi_ignore_explicit_marks_for_paragraph_level = 1; | |
155 | ||
156 | bidi_dir_t bidi_overriding_paragraph_direction = NEUTRAL_DIR; | |
157 | ||
158 | #define ASCII_BIDI_TYPE_SET(STR, TYPE) \ | |
159 | do { \ | |
160 | unsigned char *p; \ | |
161 | for (p = (STR); *p; p++) \ | |
162 | CHAR_TABLE_SET (bidi_type_table, *p, (TYPE)); \ | |
163 | } while (0) | |
164 | ||
165 | static void | |
166 | bidi_initialize () | |
167 | { | |
168 | /* FIXME: This should come from the Unicode Database. */ | |
169 | struct { | |
170 | int from, to; | |
171 | bidi_type_t type; | |
172 | } bidi_type[] = | |
173 | { { 0x0000, 0x0008, WEAK_BN }, | |
174 | { 0x0009, 0x0000, NEUTRAL_S }, | |
175 | { 0x000A, 0x0000, NEUTRAL_B }, | |
176 | { 0x000B, 0x0000, NEUTRAL_S }, | |
177 | { 0x000C, 0x0000, NEUTRAL_WS }, | |
178 | { 0x000D, 0x0000, NEUTRAL_B }, | |
179 | { 0x000E, 0x001B, WEAK_BN }, | |
180 | { 0x001C, 0x001E, NEUTRAL_B }, | |
181 | { 0x001F, 0x0000, NEUTRAL_S }, | |
182 | { 0x0020, 0x0000, NEUTRAL_WS }, | |
183 | { 0x0021, 0x0022, NEUTRAL_ON }, | |
184 | { 0x0023, 0x0025, WEAK_ET }, | |
185 | { 0x0026, 0x002A, NEUTRAL_ON }, | |
186 | { 0x002B, 0x0000, WEAK_ET }, | |
187 | { 0x002C, 0x0000, WEAK_CS }, | |
188 | { 0x002D, 0x0000, WEAK_ET }, | |
189 | { 0x002E, 0x0000, WEAK_CS }, | |
190 | { 0x002F, 0x0000, WEAK_ES }, | |
191 | { 0x0030, 0x0039, WEAK_EN }, | |
192 | { 0x003A, 0x0000, WEAK_CS }, | |
193 | { 0x003B, 0x0040, NEUTRAL_ON }, | |
194 | { 0x005B, 0x0060, NEUTRAL_ON }, | |
195 | { 0x007B, 0x007E, NEUTRAL_ON }, | |
196 | { 0x007F, 0x0084, WEAK_BN }, | |
197 | { 0x0085, 0x0000, NEUTRAL_B }, | |
198 | { 0x0086, 0x009F, WEAK_BN }, | |
199 | { 0x00A0, 0x0000, WEAK_CS }, | |
200 | { 0x00A1, 0x0000, NEUTRAL_ON }, | |
201 | { 0x00A2, 0x00A5, WEAK_ET }, | |
202 | { 0x00A6, 0x00A9, NEUTRAL_ON }, | |
203 | { 0x00AB, 0x00AF, NEUTRAL_ON }, | |
204 | { 0x00B0, 0x00B1, WEAK_ET }, | |
205 | { 0x00B2, 0x00B3, WEAK_EN }, | |
206 | { 0x00B4, 0x0000, NEUTRAL_ON }, | |
207 | { 0x00B6, 0x00B8, NEUTRAL_ON }, | |
208 | { 0x00B9, 0x0000, WEAK_EN }, | |
209 | { 0x00BB, 0x00BF, NEUTRAL_ON }, | |
210 | { 0x00D7, 0x0000, NEUTRAL_ON }, | |
211 | { 0x00F7, 0x0000, NEUTRAL_ON }, | |
212 | { 0x02B9, 0x02BA, NEUTRAL_ON }, | |
213 | { 0x02C2, 0x02CF, NEUTRAL_ON }, | |
214 | { 0x02D2, 0x02DF, NEUTRAL_ON }, | |
215 | { 0x02E5, 0x02ED, NEUTRAL_ON }, | |
216 | { 0x0300, 0x036F, WEAK_NSM }, | |
217 | { 0x0374, 0x0375, NEUTRAL_ON }, | |
218 | { 0x037E, 0x0385, NEUTRAL_ON }, | |
219 | { 0x0387, 0x0000, NEUTRAL_ON }, | |
220 | { 0x03F6, 0x0000, NEUTRAL_ON }, | |
221 | { 0x0483, 0x0489, WEAK_NSM }, | |
222 | { 0x058A, 0x0000, NEUTRAL_ON }, | |
223 | { 0x0591, 0x05BD, WEAK_NSM }, | |
224 | { 0x05BE, 0x0000, STRONG_R }, | |
225 | { 0x05BF, 0x0000, WEAK_NSM }, | |
226 | { 0x05C0, 0x0000, STRONG_R }, | |
227 | { 0x05C1, 0x05C2, WEAK_NSM }, | |
228 | { 0x05C3, 0x0000, STRONG_R }, | |
229 | { 0x05C4, 0x0000, WEAK_NSM }, | |
230 | { 0x05D0, 0x05F4, STRONG_R }, | |
231 | { 0x060C, 0x0000, WEAK_CS }, | |
232 | { 0x061B, 0x064A, STRONG_AL }, | |
233 | { 0x064B, 0x0655, WEAK_NSM }, | |
234 | { 0x0660, 0x0669, WEAK_AN }, | |
235 | { 0x066A, 0x0000, WEAK_ET }, | |
236 | { 0x066B, 0x066C, WEAK_AN }, | |
237 | { 0x066D, 0x066F, STRONG_AL }, | |
238 | { 0x0670, 0x0000, WEAK_NSM }, | |
239 | { 0x0671, 0x06D5, STRONG_AL }, | |
240 | { 0x06D6, 0x06DC, WEAK_NSM }, | |
241 | { 0x06DD, 0x0000, STRONG_AL }, | |
242 | { 0x06DE, 0x06E4, WEAK_NSM }, | |
243 | { 0x06E5, 0x06E6, STRONG_AL }, | |
244 | { 0x06E7, 0x06E8, WEAK_NSM }, | |
245 | { 0x06E9, 0x0000, NEUTRAL_ON }, | |
246 | { 0x06EA, 0x06ED, WEAK_NSM }, | |
247 | { 0x06F0, 0x06F9, WEAK_EN }, | |
248 | { 0x06FA, 0x070D, STRONG_AL }, | |
249 | { 0x070F, 0x0000, WEAK_BN }, | |
250 | { 0x0710, 0x0000, STRONG_AL }, | |
251 | { 0x0711, 0x0000, WEAK_NSM }, | |
252 | { 0x0712, 0x072C, STRONG_AL }, | |
253 | { 0x0730, 0x074A, WEAK_NSM }, | |
254 | { 0x0780, 0x07A5, STRONG_AL }, | |
255 | { 0x07A6, 0x07B0, WEAK_NSM }, | |
256 | { 0x07B1, 0x0000, STRONG_AL }, | |
257 | { 0x0901, 0x0902, WEAK_NSM }, | |
258 | { 0x093C, 0x0000, WEAK_NSM }, | |
259 | { 0x0941, 0x0948, WEAK_NSM }, | |
260 | { 0x094D, 0x0000, WEAK_NSM }, | |
261 | { 0x0951, 0x0954, WEAK_NSM }, | |
262 | { 0x0962, 0x0963, WEAK_NSM }, | |
263 | { 0x0981, 0x0000, WEAK_NSM }, | |
264 | { 0x09BC, 0x0000, WEAK_NSM }, | |
265 | { 0x09C1, 0x09C4, WEAK_NSM }, | |
266 | { 0x09CD, 0x0000, WEAK_NSM }, | |
267 | { 0x09E2, 0x09E3, WEAK_NSM }, | |
268 | { 0x09F2, 0x09F3, WEAK_ET }, | |
269 | { 0x0A02, 0x0000, WEAK_NSM }, | |
270 | { 0x0A3C, 0x0000, WEAK_NSM }, | |
271 | { 0x0A41, 0x0A4D, WEAK_NSM }, | |
272 | { 0x0A70, 0x0A71, WEAK_NSM }, | |
273 | { 0x0A81, 0x0A82, WEAK_NSM }, | |
274 | { 0x0ABC, 0x0000, WEAK_NSM }, | |
275 | { 0x0AC1, 0x0AC8, WEAK_NSM }, | |
276 | { 0x0ACD, 0x0000, WEAK_NSM }, | |
277 | { 0x0B01, 0x0000, WEAK_NSM }, | |
278 | { 0x0B3C, 0x0000, WEAK_NSM }, | |
279 | { 0x0B3F, 0x0000, WEAK_NSM }, | |
280 | { 0x0B41, 0x0B43, WEAK_NSM }, | |
281 | { 0x0B4D, 0x0B56, WEAK_NSM }, | |
282 | { 0x0B82, 0x0000, WEAK_NSM }, | |
283 | { 0x0BC0, 0x0000, WEAK_NSM }, | |
284 | { 0x0BCD, 0x0000, WEAK_NSM }, | |
285 | { 0x0C3E, 0x0C40, WEAK_NSM }, | |
286 | { 0x0C46, 0x0C56, WEAK_NSM }, | |
287 | { 0x0CBF, 0x0000, WEAK_NSM }, | |
288 | { 0x0CC6, 0x0000, WEAK_NSM }, | |
289 | { 0x0CCC, 0x0CCD, WEAK_NSM }, | |
290 | { 0x0D41, 0x0D43, WEAK_NSM }, | |
291 | { 0x0D4D, 0x0000, WEAK_NSM }, | |
292 | { 0x0DCA, 0x0000, WEAK_NSM }, | |
293 | { 0x0DD2, 0x0DD6, WEAK_NSM }, | |
294 | { 0x0E31, 0x0000, WEAK_NSM }, | |
295 | { 0x0E34, 0x0E3A, WEAK_NSM }, | |
296 | { 0x0E3F, 0x0000, WEAK_ET }, | |
297 | { 0x0E47, 0x0E4E, WEAK_NSM }, | |
298 | { 0x0EB1, 0x0000, WEAK_NSM }, | |
299 | { 0x0EB4, 0x0EBC, WEAK_NSM }, | |
300 | { 0x0EC8, 0x0ECD, WEAK_NSM }, | |
301 | { 0x0F18, 0x0F19, WEAK_NSM }, | |
302 | { 0x0F35, 0x0000, WEAK_NSM }, | |
303 | { 0x0F37, 0x0000, WEAK_NSM }, | |
304 | { 0x0F39, 0x0000, WEAK_NSM }, | |
305 | { 0x0F3A, 0x0F3D, NEUTRAL_ON }, | |
306 | { 0x0F71, 0x0F7E, WEAK_NSM }, | |
307 | { 0x0F80, 0x0F84, WEAK_NSM }, | |
308 | { 0x0F86, 0x0F87, WEAK_NSM }, | |
309 | { 0x0F90, 0x0FBC, WEAK_NSM }, | |
310 | { 0x0FC6, 0x0000, WEAK_NSM }, | |
311 | { 0x102D, 0x1030, WEAK_NSM }, | |
312 | { 0x1032, 0x1037, WEAK_NSM }, | |
313 | { 0x1039, 0x0000, WEAK_NSM }, | |
314 | { 0x1058, 0x1059, WEAK_NSM }, | |
315 | { 0x1680, 0x0000, NEUTRAL_WS }, | |
316 | { 0x169B, 0x169C, NEUTRAL_ON }, | |
317 | { 0x1712, 0x1714, WEAK_NSM }, | |
318 | { 0x1732, 0x1734, WEAK_NSM }, | |
319 | { 0x1752, 0x1753, WEAK_NSM }, | |
320 | { 0x1772, 0x1773, WEAK_NSM }, | |
321 | { 0x17B7, 0x17BD, WEAK_NSM }, | |
322 | { 0x17C6, 0x0000, WEAK_NSM }, | |
323 | { 0x17C9, 0x17D3, WEAK_NSM }, | |
324 | { 0x17DB, 0x0000, WEAK_ET }, | |
325 | { 0x1800, 0x180A, NEUTRAL_ON }, | |
326 | { 0x180B, 0x180D, WEAK_NSM }, | |
327 | { 0x180E, 0x0000, WEAK_BN }, | |
328 | { 0x18A9, 0x0000, WEAK_NSM }, | |
329 | { 0x1FBD, 0x0000, NEUTRAL_ON }, | |
330 | { 0x1FBF, 0x1FC1, NEUTRAL_ON }, | |
331 | { 0x1FCD, 0x1FCF, NEUTRAL_ON }, | |
332 | { 0x1FDD, 0x1FDF, NEUTRAL_ON }, | |
333 | { 0x1FED, 0x1FEF, NEUTRAL_ON }, | |
334 | { 0x1FFD, 0x1FFE, NEUTRAL_ON }, | |
335 | { 0x2000, 0x200A, NEUTRAL_WS }, | |
336 | { 0x200B, 0x200D, WEAK_BN }, | |
337 | { 0x200F, 0x0000, STRONG_R }, | |
338 | { 0x2010, 0x2027, NEUTRAL_ON }, | |
339 | { 0x2028, 0x0000, NEUTRAL_WS }, | |
340 | { 0x2029, 0x0000, NEUTRAL_B }, | |
341 | { 0x202A, 0x0000, LRE }, | |
342 | { 0x202B, 0x0000, RLE }, | |
343 | { 0x202C, 0x0000, PDF }, | |
344 | { 0x202D, 0x0000, LRO }, | |
345 | { 0x202E, 0x0000, RLO }, | |
346 | { 0x202F, 0x0000, NEUTRAL_WS }, | |
347 | { 0x2030, 0x2034, WEAK_ET }, | |
348 | { 0x2035, 0x2057, NEUTRAL_ON }, | |
349 | { 0x205F, 0x0000, NEUTRAL_WS }, | |
350 | { 0x2060, 0x206F, WEAK_BN }, | |
351 | { 0x2070, 0x0000, WEAK_EN }, | |
352 | { 0x2074, 0x2079, WEAK_EN }, | |
353 | { 0x207A, 0x207B, WEAK_ET }, | |
354 | { 0x207C, 0x207E, NEUTRAL_ON }, | |
355 | { 0x2080, 0x2089, WEAK_EN }, | |
356 | { 0x208A, 0x208B, WEAK_ET }, | |
357 | { 0x208C, 0x208E, NEUTRAL_ON }, | |
358 | { 0x20A0, 0x20B1, WEAK_ET }, | |
359 | { 0x20D0, 0x20EA, WEAK_NSM }, | |
360 | { 0x2100, 0x2101, NEUTRAL_ON }, | |
361 | { 0x2103, 0x2106, NEUTRAL_ON }, | |
362 | { 0x2108, 0x2109, NEUTRAL_ON }, | |
363 | { 0x2114, 0x0000, NEUTRAL_ON }, | |
364 | { 0x2116, 0x2118, NEUTRAL_ON }, | |
365 | { 0x211E, 0x2123, NEUTRAL_ON }, | |
366 | { 0x2125, 0x0000, NEUTRAL_ON }, | |
367 | { 0x2127, 0x0000, NEUTRAL_ON }, | |
368 | { 0x2129, 0x0000, NEUTRAL_ON }, | |
369 | { 0x212E, 0x0000, WEAK_ET }, | |
370 | { 0x2132, 0x0000, NEUTRAL_ON }, | |
371 | { 0x213A, 0x0000, NEUTRAL_ON }, | |
372 | { 0x2140, 0x2144, NEUTRAL_ON }, | |
373 | { 0x214A, 0x215F, NEUTRAL_ON }, | |
374 | { 0x2190, 0x2211, NEUTRAL_ON }, | |
375 | { 0x2212, 0x2213, WEAK_ET }, | |
376 | { 0x2214, 0x2335, NEUTRAL_ON }, | |
377 | { 0x237B, 0x2394, NEUTRAL_ON }, | |
378 | { 0x2396, 0x244A, NEUTRAL_ON }, | |
379 | { 0x2460, 0x249B, WEAK_EN }, | |
380 | { 0x24EA, 0x0000, WEAK_EN }, | |
381 | { 0x24EB, 0x2FFB, NEUTRAL_ON }, | |
382 | { 0x3000, 0x0000, NEUTRAL_WS }, | |
383 | { 0x3001, 0x3004, NEUTRAL_ON }, | |
384 | { 0x3008, 0x3020, NEUTRAL_ON }, | |
385 | { 0x302A, 0x302F, WEAK_NSM }, | |
386 | { 0x3030, 0x0000, NEUTRAL_ON }, | |
387 | { 0x3036, 0x3037, NEUTRAL_ON }, | |
388 | { 0x303D, 0x303F, NEUTRAL_ON }, | |
389 | { 0x3099, 0x309A, WEAK_NSM }, | |
390 | { 0x309B, 0x309C, NEUTRAL_ON }, | |
391 | { 0x30A0, 0x0000, NEUTRAL_ON }, | |
392 | { 0x30FB, 0x0000, NEUTRAL_ON }, | |
393 | { 0x3251, 0x325F, NEUTRAL_ON }, | |
394 | { 0x32B1, 0x32BF, NEUTRAL_ON }, | |
395 | { 0xA490, 0xA4C6, NEUTRAL_ON }, | |
396 | { 0xFB1D, 0x0000, STRONG_R }, | |
397 | { 0xFB1E, 0x0000, WEAK_NSM }, | |
398 | { 0xFB1F, 0xFB28, STRONG_R }, | |
399 | { 0xFB29, 0x0000, WEAK_ET }, | |
400 | { 0xFB2A, 0xFB4F, STRONG_R }, | |
401 | { 0xFB50, 0xFD3D, STRONG_AL }, | |
402 | { 0xFD3E, 0xFD3F, NEUTRAL_ON }, | |
403 | { 0xFD50, 0xFDFC, STRONG_AL }, | |
404 | { 0xFE00, 0xFE23, WEAK_NSM }, | |
405 | { 0xFE30, 0xFE4F, NEUTRAL_ON }, | |
406 | { 0xFE50, 0x0000, WEAK_CS }, | |
407 | { 0xFE51, 0x0000, NEUTRAL_ON }, | |
408 | { 0xFE52, 0x0000, WEAK_CS }, | |
409 | { 0xFE54, 0x0000, NEUTRAL_ON }, | |
410 | { 0xFE55, 0x0000, WEAK_CS }, | |
411 | { 0xFE56, 0xFE5E, NEUTRAL_ON }, | |
412 | { 0xFE5F, 0x0000, WEAK_ET }, | |
413 | { 0xFE60, 0xFE61, NEUTRAL_ON }, | |
414 | { 0xFE62, 0xFE63, WEAK_ET }, | |
415 | { 0xFE64, 0xFE68, NEUTRAL_ON }, | |
416 | { 0xFE69, 0xFE6A, WEAK_ET }, | |
417 | { 0xFE6B, 0x0000, NEUTRAL_ON }, | |
418 | { 0xFE70, 0xFEFC, STRONG_AL }, | |
419 | { 0xFEFF, 0x0000, WEAK_BN }, | |
420 | { 0xFF01, 0xFF02, NEUTRAL_ON }, | |
421 | { 0xFF03, 0xFF05, WEAK_ET }, | |
422 | { 0xFF06, 0xFF0A, NEUTRAL_ON }, | |
423 | { 0xFF0B, 0x0000, WEAK_ET }, | |
424 | { 0xFF0C, 0x0000, WEAK_CS }, | |
425 | { 0xFF0D, 0x0000, WEAK_ET }, | |
426 | { 0xFF0E, 0x0000, WEAK_CS }, | |
427 | { 0xFF0F, 0x0000, WEAK_ES }, | |
428 | { 0xFF10, 0xFF19, WEAK_EN }, | |
429 | { 0xFF1A, 0x0000, WEAK_CS }, | |
430 | { 0xFF1B, 0xFF20, NEUTRAL_ON }, | |
431 | { 0xFF3B, 0xFF40, NEUTRAL_ON }, | |
432 | { 0xFF5B, 0xFF65, NEUTRAL_ON }, | |
433 | { 0xFFE0, 0xFFE1, WEAK_ET }, | |
434 | { 0xFFE2, 0xFFE4, NEUTRAL_ON }, | |
435 | { 0xFFE5, 0xFFE6, WEAK_ET }, | |
436 | { 0xFFE8, 0xFFEE, NEUTRAL_ON }, | |
437 | { 0xFFF9, 0xFFFB, WEAK_BN }, | |
438 | { 0xFFFC, 0xFFFD, NEUTRAL_ON }, | |
439 | { 0x1D167, 0x1D169, WEAK_NSM }, | |
440 | { 0x1D173, 0x1D17A, WEAK_BN }, | |
441 | { 0x1D17B, 0x1D182, WEAK_NSM }, | |
442 | { 0x1D185, 0x1D18B, WEAK_NSM }, | |
443 | { 0x1D1AA, 0x1D1AD, WEAK_NSM }, | |
444 | { 0x1D7CE, 0x1D7FF, WEAK_EN }, | |
445 | { 0xE0001, 0xE007F, WEAK_BN } }; | |
446 | int i; | |
447 | ||
448 | bidi_type_table = Fmake_char_table (Qnil, make_number (STRONG_L)); | |
449 | ||
450 | for (i = 0; i < sizeof bidi_type / sizeof bidi_type[0]; i++) | |
451 | char_table_set_range (bidi_type_table, bidi_type[i].from, bidi_type[i].to, | |
452 | make_number (bidi_type[i].type)); | |
453 | bidi_initialized = 1; | |
454 | } | |
455 | ||
456 | static int | |
457 | bidi_is_arabic_number (int ch) | |
458 | { | |
459 | #ifdef TEST_STANDALONE | |
460 | return ch >= '6' && ch <= '9'; | |
461 | #else | |
462 | return 0; /* FIXME! */ | |
463 | #endif | |
464 | } | |
465 | ||
466 | /* Return the bidi type of a character CH. */ | |
467 | bidi_type_t | |
468 | bidi_get_type (int ch) | |
469 | { | |
470 | return (bidi_type_t) XINT (CHAR_TABLE_REF (bidi_type_table, ch)); | |
471 | } | |
472 | ||
473 | /* Given a bidi TYPE of a character, return its category. */ | |
474 | bidi_category_t | |
475 | bidi_get_category (bidi_type_t type) | |
476 | { | |
477 | switch (type) | |
478 | { | |
479 | case UNKNOWN_BT: | |
480 | return UNKNOWN_BC; | |
481 | case STRONG_L: | |
482 | case STRONG_R: | |
483 | case STRONG_AL: | |
484 | case LRE: | |
485 | case LRO: | |
486 | case RLE: | |
487 | case RLO: | |
488 | return STRONG; | |
489 | case PDF: /* ??? really?? */ | |
490 | case WEAK_EN: | |
491 | case WEAK_ES: | |
492 | case WEAK_ET: | |
493 | case WEAK_AN: | |
494 | case WEAK_CS: | |
495 | case WEAK_NSM: | |
496 | case WEAK_BN: | |
497 | return WEAK; | |
498 | case NEUTRAL_B: | |
499 | case NEUTRAL_S: | |
500 | case NEUTRAL_WS: | |
501 | case NEUTRAL_ON: | |
502 | return NEUTRAL; | |
503 | default: | |
504 | abort (); | |
505 | } | |
506 | } | |
507 | ||
508 | /* FIXME: exceedingly temporary! Should consult the Unicode database | |
509 | of character properties. */ | |
510 | int | |
511 | bidi_mirror_char (int c) | |
512 | { | |
513 | static const char mirrored_pairs[] = "()<>[]{}"; | |
514 | const char *p = strchr (mirrored_pairs, c); | |
515 | ||
516 | if (p) | |
517 | { | |
518 | size_t i = p - mirrored_pairs; | |
519 | ||
520 | if ((i & 1) == 0) | |
521 | return mirrored_pairs[i + 1]; | |
522 | else | |
523 | return mirrored_pairs[i - 1]; | |
524 | } | |
525 | return c; | |
526 | } | |
527 | ||
528 | /* Copy the bidi iterator from FROM to TO. To save cycles, this only | |
529 | copies the part of the level stack that is actually in use. */ | |
530 | static inline void | |
531 | bidi_copy_it (struct bidi_it *to, struct bidi_it *from) | |
532 | { | |
533 | int i; | |
534 | ||
535 | /* Copy everything except the level stack. */ | |
536 | memcpy (to, from, ((int)&((struct bidi_it *)0)->level_stack[0])); | |
537 | ||
538 | /* Copy the active part of the level stack. */ | |
539 | to->level_stack[0] = from->level_stack[0]; /* level zero is always in use */ | |
540 | for (i = 1; i <= from->stack_idx; i++) | |
541 | to->level_stack[i] = from->level_stack[i]; | |
542 | } | |
543 | ||
544 | /* Caching the bidi iterator states. */ | |
545 | ||
546 | static struct bidi_it bidi_cache[1000]; /* FIXME: make this dynamically allocated! */ | |
547 | static int bidi_cache_idx; | |
548 | static int bidi_cache_last_idx; | |
549 | ||
550 | static inline void | |
551 | bidi_cache_reset (void) | |
552 | { | |
553 | bidi_cache_idx = 0; | |
554 | bidi_cache_last_idx = -1; | |
555 | } | |
556 | ||
557 | static inline void | |
558 | bidi_cache_fetch_state (int idx, struct bidi_it *bidi_it) | |
559 | { | |
560 | int current_scan_dir = bidi_it->scan_dir; | |
561 | ||
562 | if (idx < 0 || idx >= bidi_cache_idx) | |
563 | abort (); | |
564 | ||
565 | bidi_copy_it (bidi_it, &bidi_cache[idx]); | |
566 | bidi_it->scan_dir = current_scan_dir; | |
567 | bidi_cache_last_idx = idx; | |
568 | } | |
569 | ||
570 | /* Find a cached state with a given CHARPOS and resolved embedding | |
571 | level less or equal to LEVEL. if LEVEL is -1, disregard the | |
572 | resolved levels in cached states. DIR, if non-zero, means search | |
573 | in that direction from the last cache hit. */ | |
574 | static inline int | |
575 | bidi_cache_search (int charpos, int level, int dir) | |
576 | { | |
577 | int i, i_start; | |
578 | ||
579 | if (bidi_cache_idx) | |
580 | { | |
581 | if (charpos < bidi_cache[bidi_cache_last_idx].charpos) | |
582 | dir = -1; | |
583 | else if (charpos > bidi_cache[bidi_cache_last_idx].charpos) | |
584 | dir = 1; | |
585 | if (dir) | |
586 | i_start = bidi_cache_last_idx; | |
587 | else | |
588 | { | |
589 | dir = -1; | |
590 | i_start = bidi_cache_idx - 1; | |
591 | } | |
592 | ||
593 | if (dir < 0) | |
594 | { | |
595 | /* Linear search for now; FIXME! */ | |
596 | for (i = i_start; i >= 0; i--) | |
597 | if (bidi_cache[i].charpos == charpos | |
598 | && (level == -1 || bidi_cache[i].resolved_level <= level)) | |
599 | return i; | |
600 | } | |
601 | else | |
602 | { | |
603 | for (i = i_start; i < bidi_cache_idx; i++) | |
604 | if (bidi_cache[i].charpos == charpos | |
605 | && (level == -1 || bidi_cache[i].resolved_level <= level)) | |
606 | return i; | |
607 | } | |
608 | } | |
609 | ||
610 | return -1; | |
611 | } | |
612 | ||
613 | /* Find a cached state where the resolved level changes to a value | |
614 | that is lower than LEVEL, and return its cache slot index. DIR is | |
615 | the direction to search, starting with the last used cache slot. | |
616 | BEFORE, if non-zero, means return the index of the slot that is | |
617 | ``before'' the level change in the search direction. That is, | |
618 | given the cached levels like this: | |
619 | ||
620 | 1122333442211 | |
621 | AB C | |
622 | ||
623 | and assuming we are at the position cached at the slot marked with | |
624 | C, searching backwards (DIR = -1) for LEVEL = 2 will return the | |
625 | index of slot B or A, depending whether BEFORE is, respectively, | |
626 | non-zero or zero. */ | |
627 | static int | |
628 | bidi_cache_find_level_change (int level, int dir, int before) | |
629 | { | |
630 | if (bidi_cache_idx) | |
631 | { | |
632 | int i = dir ? bidi_cache_last_idx : bidi_cache_idx - 1; | |
633 | int incr = before ? 1 : 0; | |
634 | ||
635 | if (!dir) | |
636 | dir = -1; | |
637 | else if (!incr) | |
638 | i += dir; | |
639 | ||
640 | if (dir < 0) | |
641 | { | |
642 | while (i >= incr) | |
643 | { | |
644 | if (bidi_cache[i - incr].resolved_level >= 0 | |
645 | && bidi_cache[i - incr].resolved_level < level) | |
646 | return i; | |
647 | i--; | |
648 | } | |
649 | } | |
650 | else | |
651 | { | |
652 | while (i < bidi_cache_idx - incr) | |
653 | { | |
654 | if (bidi_cache[i + incr].resolved_level >= 0 | |
655 | && bidi_cache[i + incr].resolved_level < level) | |
656 | return i; | |
657 | i++; | |
658 | } | |
659 | } | |
660 | } | |
661 | ||
662 | return -1; | |
663 | } | |
664 | ||
665 | static inline void | |
666 | bidi_cache_iterator_state (struct bidi_it *bidi_it, int resolved) | |
667 | { | |
668 | int idx; | |
669 | ||
670 | /* We should never cache on backward scans. */ | |
671 | if (bidi_it->scan_dir == -1) | |
672 | abort (); | |
673 | idx = bidi_cache_search (bidi_it->charpos, -1, 1); | |
674 | ||
675 | if (idx < 0) | |
676 | { | |
677 | idx = bidi_cache_idx; | |
678 | if (idx > sizeof (bidi_cache) / sizeof (bidi_cache[0]) - 1) | |
679 | abort (); | |
680 | bidi_copy_it (&bidi_cache[idx], bidi_it); | |
681 | if (!resolved) | |
682 | bidi_cache[idx].resolved_level = -1; | |
683 | } | |
684 | else | |
685 | { | |
686 | /* Copy only the members which could have changed, to avoid | |
687 | costly copying of the entire struct. */ | |
688 | bidi_cache[idx].type = bidi_it->type; | |
689 | bidi_cache[idx].orig_type = bidi_it->orig_type; | |
690 | if (resolved) | |
691 | bidi_cache[idx].resolved_level = bidi_it->resolved_level; | |
692 | else | |
693 | bidi_cache[idx].resolved_level = -1; | |
694 | bidi_cache[idx].invalid_levels = bidi_it->invalid_levels; | |
695 | bidi_cache[idx].invalid_rl_levels = bidi_it->invalid_rl_levels; | |
696 | bidi_cache[idx].next_for_neutral = bidi_it->next_for_neutral; | |
697 | bidi_cache[idx].next_for_ws = bidi_it->next_for_ws; | |
698 | bidi_cache[idx].ignore_bn_limit = bidi_it->ignore_bn_limit; | |
699 | } | |
700 | ||
701 | bidi_cache_last_idx = idx; | |
702 | if (idx >= bidi_cache_idx) | |
703 | bidi_cache_idx = idx + 1; | |
704 | } | |
705 | ||
706 | static inline bidi_type_t | |
707 | bidi_cache_find (int charpos, int level, struct bidi_it *bidi_it) | |
708 | { | |
709 | int i = bidi_cache_search (charpos, level, bidi_it->scan_dir); | |
710 | ||
711 | if (i >= 0) | |
712 | { | |
713 | bidi_dir_t current_scan_dir = bidi_it->scan_dir; | |
714 | ||
715 | *bidi_it = bidi_cache[i]; | |
716 | bidi_cache_last_idx = i; | |
717 | /* Don't let scan direction from from the cached state override | |
718 | the current scan direction. */ | |
719 | bidi_it->scan_dir = current_scan_dir; | |
720 | return bidi_it->type; | |
721 | } | |
722 | ||
723 | return UNKNOWN_BT; | |
724 | } | |
725 | ||
726 | static inline int | |
727 | bidi_peek_at_next_level (struct bidi_it *bidi_it) | |
728 | { | |
729 | if (bidi_cache_idx == 0 || bidi_cache_last_idx == -1) | |
730 | abort (); | |
731 | return bidi_cache[bidi_cache_last_idx + bidi_it->scan_dir].resolved_level; | |
732 | } | |
733 | ||
734 | /* Return non-zero if buffer's byte position POS is the last character | |
735 | of a paragraph. THIS_CH is the character preceding the one at POS in | |
736 | the buffer. */ | |
737 | int | |
738 | bidi_at_paragraph_end (int this_ch, int pos) | |
739 | { | |
740 | int next_ch = FETCH_CHAR (pos); | |
741 | ||
742 | /* FIXME: This should support all Unicode characters that can end a | |
743 | paragraph. */ | |
744 | return (this_ch == '\n' && next_ch == '\n') || this_ch == BIDI_EOB; | |
745 | } | |
746 | ||
747 | /* Determine the start-of-run (sor) directional type given the two | |
748 | embedding levels on either side of the run boundary. Also, update | |
749 | the saved info about previously seen characters, since that info is | |
750 | generally valid for a single level run. */ | |
751 | static inline void | |
752 | bidi_set_sor_type (struct bidi_it *bidi_it, int level_before, int level_after) | |
753 | { | |
754 | int higher_level = level_before > level_after ? level_before : level_after; | |
755 | ||
756 | /* The prev_was_pdf gork is required for when we have several PDFs | |
757 | in a row. In that case, we want to compute the sor type for the | |
758 | next level run only once: when we see the first PDF. That's | |
759 | because the sor type depends only on the higher of the two levels | |
760 | that we find on the two sides of the level boundary (see UAX#9, | |
761 | clause X10), and so we don't need to know the final embedding | |
762 | level to which we descend after processing all the PDFs. */ | |
763 | if (level_before < level_after || !bidi_it->prev_was_pdf) | |
764 | /* FIXME: should the default sor direction be user selectable? */ | |
765 | bidi_it->sor = (higher_level & 1) != 0 ? R2L : L2R; | |
766 | if (level_before > level_after) | |
767 | bidi_it->prev_was_pdf = 1; | |
768 | ||
769 | bidi_it->prev.type = UNKNOWN_BT; | |
770 | bidi_it->last_strong.type = bidi_it->last_strong.orig_type = | |
771 | bidi_it->last_strong.pristine_type = UNKNOWN_BT; | |
772 | bidi_it->prev_for_neutral.type = bidi_it->sor == R2L ? STRONG_R : STRONG_L; | |
773 | bidi_it->prev_for_neutral.charpos = bidi_it->charpos; | |
774 | bidi_it->prev_for_neutral.bytepos = bidi_it->bytepos; | |
775 | bidi_it->next_for_neutral.type = bidi_it->next_for_neutral.orig_type = | |
776 | bidi_it->next_for_neutral.pristine_type = UNKNOWN_BT; | |
777 | bidi_it->ignore_bn_limit = 0; /* meaning it's unknown */ | |
778 | } | |
779 | ||
780 | void | |
781 | bidi_paragraph_init (bidi_dir_t dir, struct bidi_it *bidi_it) | |
782 | { | |
783 | bidi_it->level_stack[0].level = 0; | |
784 | if (dir == R2L) | |
785 | bidi_it->level_stack[0].level = 1; | |
786 | else if (dir == NEUTRAL_DIR) /* P2 */ | |
787 | { | |
788 | bidi_type_t type; | |
789 | int pos = bidi_it->charpos, bytepos = bidi_it->bytepos; | |
790 | int ch; | |
791 | ||
792 | if (pos < 0) | |
793 | pos = bytepos = 0; | |
794 | else if (bidi_it->ch != BIDI_EOB) | |
795 | { | |
796 | pos++; | |
797 | bytepos += bidi_it->ch_len; | |
798 | } | |
799 | ||
800 | ch = FETCH_CHAR (bytepos); | |
801 | pos++; | |
802 | bytepos += CHAR_BYTES (ch); | |
803 | ||
804 | /* FIXME: should actually go to where the paragraph begins and | |
805 | start the loop below from there, since UAX#9 says to find the | |
806 | first strong directional character in the paragraph. */ | |
807 | ||
808 | for (type = bidi_get_type (ch); | |
809 | /* NOTE: UAX#9 says to search only for L, AL, or R types of | |
810 | characters, and ignore RLE, RLO, LRE, and LRO. However, | |
811 | I'm not sure it makes sense to omit those 4; should try | |
812 | with and without that to see the effect. */ | |
813 | (bidi_get_category (type) != STRONG) | |
814 | || (bidi_ignore_explicit_marks_for_paragraph_level | |
815 | && (type == RLE || type == RLO | |
816 | || type == LRE || type == LRO)); | |
817 | type = bidi_get_type (ch)) | |
818 | { | |
819 | if (type == NEUTRAL_B || bidi_at_paragraph_end (ch, bytepos)) | |
820 | break; | |
821 | FETCH_CHAR_ADVANCE (ch, pos, bytepos); | |
822 | } | |
823 | if (type == STRONG_R || type == STRONG_AL) /* P3 */ | |
824 | bidi_it->level_stack[0].level = 1; | |
825 | } | |
826 | bidi_it->scan_dir = 1; /* FIXME: do we need to have control on this? */ | |
827 | bidi_it->resolved_level = bidi_it->level_stack[0].level; | |
828 | bidi_it->level_stack[0].override = NEUTRAL_DIR; /* X1 */ | |
829 | bidi_it->invalid_levels = 0; | |
830 | bidi_it->invalid_rl_levels = -1; | |
831 | bidi_it->new_paragraph = 0; | |
832 | bidi_it->next_en_pos = -1; | |
833 | bidi_it->next_for_ws.type = UNKNOWN_BT; | |
834 | bidi_set_sor_type (bidi_it, bidi_it->level_stack[0].level, 0); /* X10 */ | |
835 | ||
836 | bidi_cache_reset (); | |
837 | } | |
838 | ||
839 | /* Do whatever UAX#9 clause X8 says should be done at paragraph's end, | |
840 | and set the new paragraph flag in the iterator. */ | |
841 | static inline void | |
842 | bidi_set_paragraph_end (struct bidi_it *bidi_it) | |
843 | { | |
844 | bidi_it->invalid_levels = 0; | |
845 | bidi_it->invalid_rl_levels = -1; | |
846 | bidi_it->stack_idx = 0; | |
847 | bidi_it->resolved_level = bidi_it->level_stack[0].level; | |
848 | bidi_it->new_paragraph = 1; | |
849 | } | |
850 | ||
851 | /* Initialize the bidi iterator from buffer position POS for paragraph | |
852 | direction DIR. Return the embedding level at POS. */ | |
853 | void | |
854 | bidi_init_it (int pos, bidi_dir_t dir, struct bidi_it *bidi_it) | |
855 | { | |
856 | if (! bidi_initialized) | |
857 | bidi_initialize (); | |
858 | bidi_set_paragraph_end (bidi_it); | |
859 | bidi_it->charpos = pos; | |
860 | if (pos <= 0) | |
861 | { | |
862 | bidi_it->bytepos = bidi_it->charpos; | |
863 | bidi_it->ch_len = 1; /* so that incrementing bytepos works */ | |
864 | } | |
865 | else | |
866 | { | |
867 | bidi_it->bytepos = CHAR_TO_BYTE (pos); | |
868 | bidi_it->ch_len | |
869 | = MULTIBYTE_FORM_LENGTH (BYTE_POS_ADDR (bidi_it->bytepos), | |
870 | MAX_MULTIBYTE_LENGTH); | |
871 | } | |
872 | bidi_it->ch = '\x1d'; /* FIXME: should be U+2029 */ | |
873 | bidi_it->type = NEUTRAL_B; | |
874 | bidi_it->orig_type = UNKNOWN_BT; | |
875 | bidi_it->pristine_type = UNKNOWN_BT; | |
876 | bidi_it->prev_was_pdf = 0; | |
877 | bidi_it->prev.type = bidi_it->prev.orig_type = UNKNOWN_BT; | |
878 | bidi_it->last_strong.type = bidi_it->last_strong.orig_type = | |
879 | bidi_it->last_strong.pristine_type = UNKNOWN_BT; | |
880 | bidi_it->next_for_neutral.charpos = -1; | |
881 | bidi_it->next_for_neutral.type = | |
882 | bidi_it->next_for_neutral.orig_type = | |
883 | bidi_it->next_for_neutral.pristine_type = UNKNOWN_BT; | |
884 | bidi_it->prev_for_neutral.charpos = -1; | |
885 | bidi_it->prev_for_neutral.type = | |
886 | bidi_it->prev_for_neutral.orig_type = | |
887 | bidi_it->prev_for_neutral.pristine_type = UNKNOWN_BT; | |
888 | bidi_it->sor = L2R; /* FIXME: should it be user-selectable? */ | |
889 | bidi_paragraph_init (dir, bidi_it); | |
890 | } | |
891 | ||
892 | /* Push the current embedding level and override status; reset the | |
893 | current level to LEVEL and the current override status to OVERRIDE. */ | |
894 | static inline void | |
895 | bidi_push_embedding_level (struct bidi_it *bidi_it, | |
896 | int level, bidi_dir_t override) | |
897 | { | |
898 | bidi_it->stack_idx++; | |
899 | if (bidi_it->stack_idx >= BIDI_MAXLEVEL) | |
900 | abort (); | |
901 | bidi_it->level_stack[bidi_it->stack_idx].level = level; | |
902 | bidi_it->level_stack[bidi_it->stack_idx].override = override; | |
903 | } | |
904 | ||
905 | /* Pop the embedding level and directional override status from the | |
906 | stack, and return the new level. */ | |
907 | static inline int | |
908 | bidi_pop_embedding_level (struct bidi_it *bidi_it) | |
909 | { | |
910 | /* UAX#9 says to ignore invalid PDFs. */ | |
911 | if (bidi_it->stack_idx > 0) | |
912 | bidi_it->stack_idx--; | |
913 | return bidi_it->level_stack[bidi_it->stack_idx].level; | |
914 | } | |
915 | ||
916 | /* Record in SAVED_INFO the information about the current character. */ | |
917 | static inline void | |
918 | bidi_remember_char (struct bidi_saved_info *saved_info, | |
919 | struct bidi_it *bidi_it) | |
920 | { | |
921 | saved_info->charpos = bidi_it->charpos; | |
922 | saved_info->bytepos = bidi_it->bytepos; | |
923 | saved_info->type = bidi_it->type; | |
924 | saved_info->orig_type = bidi_it->orig_type; | |
925 | saved_info->pristine_type = bidi_it->pristine_type; | |
926 | } | |
927 | ||
928 | /* Resolve the type of a neutral character according to the type of | |
929 | surrounding strong text and the current embedding level. */ | |
930 | static inline bidi_type_t | |
931 | bidi_resolve_neutral_1 (bidi_type_t prev_type, bidi_type_t next_type, int lev) | |
932 | { | |
933 | /* N1: European and Arabic numbers are treated as though they were R. */ | |
934 | if (next_type == WEAK_EN || next_type == WEAK_AN) | |
935 | next_type = STRONG_R; | |
936 | if (prev_type == WEAK_EN || prev_type == WEAK_AN) | |
937 | prev_type = STRONG_R; | |
938 | ||
939 | if (next_type == prev_type) /* N1 */ | |
940 | return next_type; | |
941 | else if ((lev & 1) == 0) /* N2 */ | |
942 | return STRONG_L; | |
943 | else | |
944 | return STRONG_R; | |
945 | } | |
946 | ||
947 | static inline int | |
948 | bidi_explicit_dir_char (int c) | |
949 | { | |
950 | /* FIXME: this should be replaced with a lookup table with suitable | |
951 | bits set, like standard C ctype macros do. */ | |
952 | return (c == LRE_CHAR || c == LRO_CHAR | |
953 | || c == RLE_CHAR || c == RLO_CHAR || c == PDF_CHAR); | |
954 | } | |
955 | ||
956 | /* A helper function for bidi_resolve_explicit. It advances to the | |
957 | next character in logical order and determines the new embedding | |
958 | level and directional override, but does not take into account | |
959 | empty embeddings. */ | |
960 | static int | |
961 | bidi_resolve_explicit_1 (struct bidi_it *bidi_it) | |
962 | { | |
963 | int curchar; | |
964 | bidi_type_t type; | |
965 | int current_level; | |
966 | int new_level; | |
967 | bidi_dir_t override; | |
968 | ||
969 | if (bidi_it->charpos < 0) | |
970 | bidi_it->charpos = bidi_it->bytepos = 0; | |
971 | else | |
972 | { | |
973 | bidi_it->charpos++; | |
974 | bidi_it->bytepos += bidi_it->ch_len; | |
975 | } | |
976 | ||
977 | current_level = bidi_it->level_stack[bidi_it->stack_idx].level; /* X1 */ | |
978 | override = bidi_it->level_stack[bidi_it->stack_idx].override; | |
979 | new_level = current_level; | |
980 | ||
981 | /* in case it is a unibyte character (not yet implemented) */ | |
982 | /* _fetch_multibyte_char_len = 1; */ | |
983 | curchar = FETCH_CHAR (bidi_it->bytepos); | |
984 | bidi_it->ch = curchar; | |
985 | bidi_it->ch_len = CHAR_BYTES (curchar); | |
986 | ||
987 | type = bidi_get_type (curchar); | |
988 | bidi_it->pristine_type = type; | |
989 | ||
990 | if (type != PDF) | |
991 | bidi_it->prev_was_pdf = 0; | |
992 | ||
993 | bidi_it->orig_type = UNKNOWN_BT; | |
994 | ||
995 | switch (type) | |
996 | { | |
997 | case RLE: /* X2 */ | |
998 | case RLO: /* X4 */ | |
999 | bidi_it->orig_type = type; | |
1000 | type = WEAK_BN; /* X9/Retaining */ | |
1001 | if (bidi_it->ignore_bn_limit <= 0) | |
1002 | { | |
1003 | if (current_level <= BIDI_MAXLEVEL - 4) | |
1004 | { | |
1005 | /* Compute the least odd embedding level greater than | |
1006 | the current level. */ | |
1007 | new_level = ((current_level + 1) & ~1) + 1; | |
1008 | if (bidi_it->orig_type == RLE) | |
1009 | override = NEUTRAL_DIR; | |
1010 | else | |
1011 | override = R2L; | |
1012 | if (current_level == BIDI_MAXLEVEL - 4) | |
1013 | bidi_it->invalid_rl_levels = 0; | |
1014 | bidi_push_embedding_level (bidi_it, new_level, override); | |
1015 | } | |
1016 | else | |
1017 | { | |
1018 | bidi_it->invalid_levels++; | |
1019 | /* See the commentary about invalid_rl_levels below. */ | |
1020 | if (bidi_it->invalid_rl_levels < 0) | |
1021 | bidi_it->invalid_rl_levels = 0; | |
1022 | bidi_it->invalid_rl_levels++; | |
1023 | } | |
1024 | } | |
1025 | else if (bidi_it->prev.orig_type == WEAK_EN /* W5/Retaining */ | |
1026 | || bidi_it->next_en_pos > bidi_it->charpos) | |
1027 | type = WEAK_EN; | |
1028 | break; | |
1029 | case LRE: /* X3 */ | |
1030 | case LRO: /* X5 */ | |
1031 | bidi_it->orig_type = type; | |
1032 | type = WEAK_BN; /* X9/Retaining */ | |
1033 | if (bidi_it->ignore_bn_limit <= 0) | |
1034 | { | |
1035 | if (current_level <= BIDI_MAXLEVEL - 5) | |
1036 | { | |
1037 | /* Compute the least even embedding level greater than | |
1038 | the current level. */ | |
1039 | new_level = ((current_level + 2) & ~1); | |
1040 | if (bidi_it->orig_type == LRE) | |
1041 | override = NEUTRAL_DIR; | |
1042 | else | |
1043 | override = L2R; | |
1044 | bidi_push_embedding_level (bidi_it, new_level, override); | |
1045 | } | |
1046 | else | |
1047 | { | |
1048 | bidi_it->invalid_levels++; | |
1049 | /* invalid_rl_levels counts invalid levels encountered | |
1050 | while the embedding level was already too high for | |
1051 | LRE/LRO, but not for RLE/RLO. That is because | |
1052 | there may be exactly one PDF which we should not | |
1053 | ignore even though invalid_levels is non-zero. | |
1054 | invalid_rl_levels helps to know what PDF is | |
1055 | that. */ | |
1056 | if (bidi_it->invalid_rl_levels >= 0) | |
1057 | bidi_it->invalid_rl_levels++; | |
1058 | } | |
1059 | } | |
1060 | else if (bidi_it->prev.orig_type == WEAK_EN /* W5/Retaining */ | |
1061 | || bidi_it->next_en_pos > bidi_it->charpos) | |
1062 | type = WEAK_EN; | |
1063 | break; | |
1064 | case PDF: /* X7 */ | |
1065 | bidi_it->orig_type = type; | |
1066 | type = WEAK_BN; /* X9/Retaining */ | |
1067 | if (bidi_it->ignore_bn_limit <= 0) | |
1068 | { | |
1069 | if (!bidi_it->invalid_rl_levels) | |
1070 | { | |
1071 | new_level = bidi_pop_embedding_level (bidi_it); | |
1072 | bidi_it->invalid_rl_levels = -1; | |
1073 | if (bidi_it->invalid_levels) | |
1074 | bidi_it->invalid_levels--; | |
1075 | /* else nothing: UAX#9 says to ignore invalid PDFs */ | |
1076 | } | |
1077 | if (!bidi_it->invalid_levels) | |
1078 | new_level = bidi_pop_embedding_level (bidi_it); | |
1079 | else | |
1080 | { | |
1081 | bidi_it->invalid_levels--; | |
1082 | bidi_it->invalid_rl_levels--; | |
1083 | } | |
1084 | } | |
1085 | else if (bidi_it->prev.orig_type == WEAK_EN /* W5/Retaining */ | |
1086 | || bidi_it->next_en_pos > bidi_it->charpos) | |
1087 | type = WEAK_EN; | |
1088 | break; | |
1089 | default: | |
1090 | /* Nothing. */ | |
1091 | break; | |
1092 | } | |
1093 | ||
1094 | bidi_it->type = type; | |
1095 | ||
1096 | return new_level; | |
1097 | } | |
1098 | ||
1099 | /* Given an iterator state in BIDI_IT, advance one character position | |
1100 | in the buffer to the next character (in the logical order), resolve | |
1101 | any explicit embeddings and directional overrides, and return the | |
1102 | embedding level of the character after resolving explicit | |
1103 | directives and ignoring empty embeddings. */ | |
1104 | static int | |
1105 | bidi_resolve_explicit (struct bidi_it *bidi_it) | |
1106 | { | |
1107 | int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level; | |
1108 | int new_level = bidi_resolve_explicit_1 (bidi_it); | |
1109 | ||
1110 | if (prev_level < new_level | |
1111 | && bidi_it->type == WEAK_BN | |
1112 | && bidi_it->ignore_bn_limit == 0 /* only if not already known */ | |
1113 | && bidi_explicit_dir_char (FETCH_CHAR (bidi_it->bytepos | |
1114 | + bidi_it->ch_len))) | |
1115 | { | |
1116 | /* Avoid pushing and popping embedding levels if the level run | |
1117 | is empty, as this breaks level runs where it shouldn't. | |
1118 | UAX#9 removes all the explicit embedding and override codes, | |
1119 | so empty embeddings disappear without a trace. We need to | |
1120 | behave as if we did the same. */ | |
1121 | struct bidi_it saved_it; | |
1122 | int level = prev_level; | |
1123 | ||
1124 | bidi_copy_it (&saved_it, bidi_it); | |
1125 | ||
1126 | while (bidi_explicit_dir_char (FETCH_CHAR (bidi_it->bytepos | |
1127 | + bidi_it->ch_len))) | |
1128 | { | |
1129 | level = bidi_resolve_explicit_1 (bidi_it); | |
1130 | } | |
1131 | ||
1132 | if (level == prev_level) /* empty embedding */ | |
1133 | saved_it.ignore_bn_limit = bidi_it->charpos + 1; | |
1134 | else /* this embedding is non-empty */ | |
1135 | saved_it.ignore_bn_limit = -1; | |
1136 | ||
1137 | bidi_copy_it (bidi_it, &saved_it); | |
1138 | if (bidi_it->ignore_bn_limit > 0) | |
1139 | { | |
1140 | /* We pushed a level, but we shouldn't have. Undo that. */ | |
1141 | if (!bidi_it->invalid_rl_levels) | |
1142 | { | |
1143 | new_level = bidi_pop_embedding_level (bidi_it); | |
1144 | bidi_it->invalid_rl_levels = -1; | |
1145 | if (bidi_it->invalid_levels) | |
1146 | bidi_it->invalid_levels--; | |
1147 | } | |
1148 | if (!bidi_it->invalid_levels) | |
1149 | new_level = bidi_pop_embedding_level (bidi_it); | |
1150 | else | |
1151 | { | |
1152 | bidi_it->invalid_levels--; | |
1153 | bidi_it->invalid_rl_levels--; | |
1154 | } | |
1155 | } | |
1156 | } | |
1157 | ||
1158 | /* For when the paragraph end is defined by anything other than a | |
1159 | special Unicode character (a.k.a. ``higher protocols''). */ | |
1160 | if (bidi_it->type != NEUTRAL_B) | |
1161 | if (bidi_at_paragraph_end (bidi_it->ch, | |
1162 | bidi_it->bytepos + bidi_it->ch_len)) | |
1163 | bidi_it->type = NEUTRAL_B; | |
1164 | ||
1165 | if (bidi_it->type == NEUTRAL_B) /* X8 */ | |
1166 | { | |
1167 | bidi_set_paragraph_end (bidi_it); | |
1168 | bidi_it->orig_type = bidi_it->type; /* needed below and in L1 */ | |
1169 | } | |
1170 | ||
1171 | return new_level; | |
1172 | } | |
1173 | ||
1174 | /* Advance in the buffer, resolve weak types and return the type of | |
1175 | the next character after weak type resolution. */ | |
1176 | bidi_type_t | |
1177 | bidi_resolve_weak (struct bidi_it *bidi_it) | |
1178 | { | |
1179 | bidi_type_t type; | |
1180 | bidi_dir_t override; | |
1181 | int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level; | |
1182 | int new_level = bidi_resolve_explicit (bidi_it); | |
1183 | int next_char; | |
1184 | bidi_type_t type_of_next; | |
1185 | struct bidi_it saved_it; | |
1186 | ||
1187 | type = bidi_it->type; | |
1188 | override = bidi_it->level_stack[bidi_it->stack_idx].override; | |
1189 | ||
1190 | if (type == UNKNOWN_BT | |
1191 | || type == LRE | |
1192 | || type == LRO | |
1193 | || type == RLE | |
1194 | || type == RLO | |
1195 | || type == PDF) | |
1196 | abort (); | |
1197 | ||
1198 | if (new_level != prev_level | |
1199 | || bidi_it->type == NEUTRAL_B) | |
1200 | { | |
1201 | /* We've got a new embedding level run, compute the directional | |
1202 | type of sor and initialize per-run variables (UAX#9, clause | |
1203 | X10). */ | |
1204 | bidi_set_sor_type (bidi_it, prev_level, new_level); | |
1205 | } | |
1206 | else if (type == NEUTRAL_S || type == NEUTRAL_WS | |
1207 | || type == WEAK_BN || type == STRONG_AL) | |
1208 | bidi_it->orig_type = type; /* needed in L1 */ | |
1209 | ||
1210 | /* Level and directional override status are already recorded in | |
1211 | bidi_it, and do not need any change; see X6. */ | |
1212 | if (override == R2L) /* X6 */ | |
1213 | type = STRONG_R; | |
1214 | else if (override == L2R) | |
1215 | type = STRONG_L; | |
1216 | else if (type == STRONG_AL) | |
1217 | type = STRONG_R; /* W3 */ | |
1218 | else if (type == WEAK_NSM) /* W1 */ | |
1219 | { | |
1220 | /* Note that we don't need to consider the case where the prev | |
1221 | character has its type overridden by an RLO or LRO: such | |
1222 | characters are outside the current level run, and thus not | |
1223 | relevant to this NSM. Thus, NSM gets the pristine_type of | |
1224 | the previous character. */ | |
1225 | if (bidi_it->prev.type != UNKNOWN_BT) | |
1226 | type = bidi_it->prev.pristine_type; | |
1227 | else if (bidi_it->sor == R2L) | |
1228 | type = STRONG_R; | |
1229 | else if (bidi_it->sor == L2R) | |
1230 | type = STRONG_L; | |
1231 | else /* shouldn't happen! */ | |
1232 | abort (); | |
1233 | if (type == WEAK_EN /* W2 after W1 */ | |
1234 | && bidi_it->last_strong.orig_type == STRONG_AL) | |
1235 | type = WEAK_AN; | |
1236 | } | |
1237 | else if (type == WEAK_EN /* W2 */ | |
1238 | && bidi_it->last_strong.orig_type == STRONG_AL) | |
1239 | type = WEAK_AN; | |
1240 | else if ((type == WEAK_ES | |
1241 | && (bidi_it->prev.orig_type == WEAK_EN /* W4 */ | |
1242 | && (bidi_it->prev.pristine_type == WEAK_EN | |
1243 | || bidi_it->prev.pristine_type == WEAK_NSM))) /* aft W1 */ | |
1244 | || (type == WEAK_CS | |
1245 | && ((bidi_it->prev.orig_type == WEAK_EN | |
1246 | && (bidi_it->prev.pristine_type == WEAK_EN /* W4 */ | |
1247 | || bidi_it->prev.pristine_type == WEAK_NSM)) /* a/W1 */ | |
1248 | || bidi_it->prev.orig_type == WEAK_AN))) /* W4 */ | |
1249 | { | |
1250 | next_char = FETCH_CHAR (bidi_it->bytepos + bidi_it->ch_len); | |
1251 | type_of_next = bidi_get_type (next_char); | |
1252 | ||
1253 | if (type_of_next == WEAK_BN | |
1254 | || bidi_explicit_dir_char (next_char)) | |
1255 | { | |
1256 | bidi_copy_it (&saved_it, bidi_it); | |
1257 | while (bidi_resolve_explicit (bidi_it) == new_level | |
1258 | && bidi_it->type == WEAK_BN) | |
1259 | ; | |
1260 | type_of_next = bidi_it->type; | |
1261 | bidi_copy_it (bidi_it, &saved_it); | |
1262 | } | |
1263 | ||
1264 | /* If the next character is EN, but the last strong-type | |
1265 | character is AL, that next EN will be changed to AN when we | |
1266 | process it in W2 above. So in that case, this ES should not | |
1267 | be changed into EN. */ | |
1268 | if (type == WEAK_ES | |
1269 | && type_of_next == WEAK_EN | |
1270 | && bidi_it->last_strong.orig_type != STRONG_AL) | |
1271 | type = WEAK_EN; | |
1272 | else if (type == WEAK_CS) | |
1273 | { | |
1274 | if (bidi_it->prev.orig_type == WEAK_AN | |
1275 | && (type_of_next == WEAK_AN | |
1276 | /* If the next character is EN, but the last | |
1277 | strong-type character is AL, EN will be later | |
1278 | changed to AN when we process it in W2 above. So | |
1279 | in that case, this ES should not be changed into | |
1280 | EN. */ | |
1281 | || (type_of_next == WEAK_EN | |
1282 | && bidi_it->last_strong.orig_type == STRONG_AL))) | |
1283 | type = WEAK_AN; | |
1284 | else if (bidi_it->prev.orig_type == WEAK_EN | |
1285 | && type_of_next == WEAK_EN | |
1286 | && bidi_it->last_strong.orig_type != STRONG_AL) | |
1287 | type = WEAK_EN; | |
1288 | } | |
1289 | } | |
1290 | else if (type == WEAK_ET /* W5: ET with EN before or after it */ | |
1291 | || type == WEAK_BN) /* W5/Retaining */ | |
1292 | { | |
1293 | if (bidi_it->prev.orig_type == WEAK_EN /* ET/BN with EN before it */ | |
1294 | || bidi_it->next_en_pos > bidi_it->charpos) | |
1295 | type = WEAK_EN; | |
1296 | /* W5: ET with EN after it. */ | |
1297 | else | |
1298 | { | |
1299 | int en_pos = bidi_it->charpos + 1; | |
1300 | ||
1301 | next_char = FETCH_CHAR (bidi_it->bytepos + bidi_it->ch_len); | |
1302 | type_of_next = bidi_get_type (next_char); | |
1303 | ||
1304 | if (type_of_next == WEAK_ET | |
1305 | || type_of_next == WEAK_BN | |
1306 | || bidi_explicit_dir_char (next_char)) | |
1307 | { | |
1308 | bidi_copy_it (&saved_it, bidi_it); | |
1309 | while (bidi_resolve_explicit (bidi_it) == new_level | |
1310 | && (bidi_it->type == WEAK_BN || bidi_it->type == WEAK_ET)) | |
1311 | ; | |
1312 | type_of_next = bidi_it->type; | |
1313 | en_pos = bidi_it->charpos; | |
1314 | bidi_copy_it (bidi_it, &saved_it); | |
1315 | } | |
1316 | if (type_of_next == WEAK_EN) | |
1317 | { | |
1318 | /* If the last strong character is AL, the EN we've | |
1319 | found will become AN when we get to it (W2). */ | |
1320 | if (bidi_it->last_strong.orig_type != STRONG_AL) | |
1321 | { | |
1322 | type = WEAK_EN; | |
1323 | /* Remember this EN position, to speed up processing | |
1324 | of the next ETs. */ | |
1325 | bidi_it->next_en_pos = en_pos; | |
1326 | } | |
1327 | else if (type == WEAK_BN) | |
1328 | type = NEUTRAL_ON; /* W6/Retaining */ | |
1329 | } | |
1330 | } | |
1331 | } | |
1332 | ||
1333 | if (type == WEAK_ES || type == WEAK_ET || type == WEAK_CS /* W6 */ | |
1334 | || (type == WEAK_BN && (bidi_it->prev.orig_type == WEAK_CS /* W6/Ret. */ | |
1335 | || bidi_it->prev.orig_type == WEAK_ES | |
1336 | || bidi_it->prev.orig_type == WEAK_ET))) | |
1337 | type = NEUTRAL_ON; | |
1338 | ||
1339 | /* Store the type we've got so far, before we clobber it with strong | |
1340 | types in W7 and while resolving neutral types. But leave alone | |
1341 | the original types that were recorded above, because we will need | |
1342 | them for the L1 clause. */ | |
1343 | if (bidi_it->orig_type == UNKNOWN_BT) | |
1344 | bidi_it->orig_type = type; | |
1345 | ||
1346 | if (type == WEAK_EN) /* W7 */ | |
1347 | { | |
1348 | if ((bidi_it->last_strong.orig_type == STRONG_L) | |
1349 | || (bidi_it->last_strong.type == UNKNOWN_BT && bidi_it->sor == L2R)) | |
1350 | type = STRONG_L; | |
1351 | } | |
1352 | ||
1353 | bidi_it->type = type; | |
1354 | return type; | |
1355 | } | |
1356 | ||
1357 | bidi_type_t | |
1358 | bidi_resolve_neutral (struct bidi_it *bidi_it) | |
1359 | { | |
1360 | int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level; | |
1361 | bidi_type_t type = bidi_resolve_weak (bidi_it); | |
1362 | int current_level = bidi_it->level_stack[bidi_it->stack_idx].level; | |
1363 | ||
1364 | if (!(type == STRONG_R | |
1365 | || type == STRONG_L | |
1366 | || type == WEAK_BN | |
1367 | || type == WEAK_EN | |
1368 | || type == WEAK_AN | |
1369 | || type == NEUTRAL_B | |
1370 | || type == NEUTRAL_S | |
1371 | || type == NEUTRAL_WS | |
1372 | || type == NEUTRAL_ON)) | |
1373 | abort (); | |
1374 | ||
1375 | if (bidi_get_category (type) == NEUTRAL | |
1376 | || (type == WEAK_BN && prev_level == current_level)) | |
1377 | { | |
1378 | if (bidi_it->next_for_neutral.type != UNKNOWN_BT) | |
1379 | type = bidi_resolve_neutral_1 (bidi_it->prev_for_neutral.type, | |
1380 | bidi_it->next_for_neutral.type, | |
1381 | current_level); | |
1382 | else | |
1383 | { | |
1384 | /* Arrrgh!! The UAX#9 algorithm is too deeply entrenched in | |
1385 | the assumption of batch-style processing; see clauses W4, | |
1386 | W5, and especially N1, which require to look far forward | |
1387 | (as well as back) in the buffer. May the fleas of a | |
1388 | thousand camels infest the armpits of those who design | |
1389 | supposedly general-purpose algorithms by looking at their | |
1390 | own implementations, and fail to consider other possible | |
1391 | implementations! */ | |
1392 | struct bidi_it saved_it; | |
1393 | bidi_type_t next_type; | |
1394 | ||
1395 | if (bidi_it->scan_dir == -1) | |
1396 | abort (); | |
1397 | ||
1398 | bidi_copy_it (&saved_it, bidi_it); | |
1399 | /* Scan the text forward until we find the first non-neutral | |
1400 | character, and then use that to resolve the neutral we | |
1401 | are dealing with now. We also cache the scanned iterator | |
1402 | states, to salvage some of the effort later. */ | |
1403 | bidi_cache_iterator_state (bidi_it, 0); | |
1404 | do { | |
1405 | /* Record the info about the previous character, so that | |
1406 | it will be cached below with this state. */ | |
1407 | if (bidi_it->orig_type != WEAK_BN /* W1/Retaining */ | |
1408 | && bidi_it->type != WEAK_BN) | |
1409 | bidi_remember_char (&bidi_it->prev, bidi_it); | |
1410 | type = bidi_resolve_weak (bidi_it); | |
1411 | /* Paragraph separators have their levels fully resolved | |
1412 | at this point, so cache them as resolved. */ | |
1413 | bidi_cache_iterator_state (bidi_it, type == NEUTRAL_B); | |
1414 | /* FIXME: implement L1 here, by testing for a newline and | |
1415 | resetting the level for any sequence of whitespace | |
1416 | characters adjacent to it. */ | |
1417 | } while (!(type == NEUTRAL_B | |
1418 | || (type != WEAK_BN | |
1419 | && bidi_get_category (type) != NEUTRAL) | |
1420 | /* This is all per level run, so stop when we | |
1421 | reach the end of this level run. */ | |
1422 | || bidi_it->level_stack[bidi_it->stack_idx].level != | |
1423 | current_level)); | |
1424 | ||
1425 | bidi_remember_char (&saved_it.next_for_neutral, bidi_it); | |
1426 | ||
1427 | switch (type) | |
1428 | { | |
1429 | case STRONG_L: | |
1430 | case STRONG_R: | |
1431 | case STRONG_AL: | |
1432 | next_type = type; | |
1433 | break; | |
1434 | case WEAK_EN: | |
1435 | case WEAK_AN: | |
1436 | /* N1: ``European and Arabic numbers are treated as | |
1437 | though they were R.'' */ | |
1438 | next_type = STRONG_R; | |
1439 | saved_it.next_for_neutral.type = STRONG_R; | |
1440 | break; | |
1441 | case WEAK_BN: | |
1442 | if (!bidi_explicit_dir_char (bidi_it->ch)) | |
1443 | abort (); /* can't happen: BNs are skipped */ | |
1444 | /* FALLTHROUGH */ | |
1445 | case NEUTRAL_B: | |
1446 | /* Marched all the way to the end of this level run. | |
1447 | We need to use the eor type, whose information is | |
1448 | stored by bidi_set_sor_type in the prev_for_neutral | |
1449 | member. */ | |
1450 | if (saved_it.type != WEAK_BN | |
1451 | || bidi_get_category (bidi_it->prev.orig_type) == NEUTRAL) | |
1452 | { | |
1453 | next_type = bidi_it->prev_for_neutral.type; | |
1454 | saved_it.next_for_neutral.type = next_type; | |
1455 | } | |
1456 | else | |
1457 | { | |
1458 | /* This is a BN which does not adjoin neutrals. | |
1459 | Leave its type alone. */ | |
1460 | bidi_copy_it (bidi_it, &saved_it); | |
1461 | return bidi_it->type; | |
1462 | } | |
1463 | break; | |
1464 | default: | |
1465 | abort (); | |
1466 | } | |
1467 | type = bidi_resolve_neutral_1 (saved_it.prev_for_neutral.type, | |
1468 | next_type, current_level); | |
1469 | saved_it.type = type; | |
1470 | bidi_copy_it (bidi_it, &saved_it); | |
1471 | } | |
1472 | } | |
1473 | return type; | |
1474 | } | |
1475 | ||
1476 | /* Given an iterator state in BIDI_IT, advance one character position | |
1477 | in the buffer to the next character (in the logical order), resolve | |
1478 | the bidi type of that next character, and return that type. */ | |
1479 | bidi_type_t | |
1480 | bidi_type_of_next_char (struct bidi_it *bidi_it) | |
1481 | { | |
1482 | bidi_type_t type; | |
1483 | ||
1484 | /* This should always be called during a forward scan. */ | |
1485 | if (bidi_it->scan_dir != 1) | |
1486 | abort (); | |
1487 | ||
1488 | /* Reset the limit until which to ignore BNs if we step out of the | |
1489 | area where we found only empty levels. */ | |
1490 | if ((bidi_it->ignore_bn_limit > 0 | |
1491 | && bidi_it->ignore_bn_limit <= bidi_it->charpos) | |
1492 | || (bidi_it->ignore_bn_limit == -1 | |
1493 | && !bidi_explicit_dir_char (bidi_it->ch))) | |
1494 | bidi_it->ignore_bn_limit = 0; | |
1495 | ||
1496 | type = bidi_resolve_neutral (bidi_it); | |
1497 | ||
1498 | return type; | |
1499 | } | |
1500 | ||
1501 | /* Given an iterator state BIDI_IT, advance one character position in | |
1502 | the buffer to the next character (in the logical order), resolve | |
1503 | the embedding and implicit levels of that next character, and | |
1504 | return the resulting level. */ | |
1505 | int | |
1506 | bidi_level_of_next_char (struct bidi_it *bidi_it) | |
1507 | { | |
1508 | bidi_type_t type; | |
1509 | int level, prev_level = -1; | |
1510 | struct bidi_saved_info next_for_neutral; | |
1511 | ||
1512 | if (bidi_it->scan_dir == 1) | |
1513 | { | |
1514 | /* There's no sense in trying to advance if we hit end of text. */ | |
1515 | if (bidi_it->ch == BIDI_EOB) | |
1516 | return bidi_it->resolved_level; | |
1517 | ||
1518 | /* Record the info about the previous character. */ | |
1519 | if (bidi_it->orig_type != WEAK_BN /* W1/Retaining */ | |
1520 | && bidi_it->type != WEAK_BN) | |
1521 | bidi_remember_char (&bidi_it->prev, bidi_it); | |
1522 | if (bidi_it->orig_type == STRONG_R || bidi_it->orig_type == STRONG_L | |
1523 | || bidi_it->orig_type == STRONG_AL) | |
1524 | bidi_remember_char (&bidi_it->last_strong, bidi_it); | |
1525 | /* FIXME: it sounds like we don't need both prev and | |
1526 | prev_for_neutral members, but I'm leaving them both for now. */ | |
1527 | if (bidi_it->type == STRONG_R || bidi_it->type == STRONG_L | |
1528 | || bidi_it->type == WEAK_EN || bidi_it->type == WEAK_AN) | |
1529 | bidi_remember_char (&bidi_it->prev_for_neutral, bidi_it); | |
1530 | ||
1531 | /* If we overstepped the characters used for resolving neutrals | |
1532 | and whitespace, invalidate their info in the iterator. */ | |
1533 | if (bidi_it->charpos >= bidi_it->next_for_neutral.charpos) | |
1534 | bidi_it->next_for_neutral.type = UNKNOWN_BT; | |
1535 | if (bidi_it->next_en_pos >= 0 | |
1536 | && bidi_it->charpos >= bidi_it->next_en_pos) | |
1537 | bidi_it->next_en_pos = -1; | |
1538 | if (bidi_it->next_for_ws.type != UNKNOWN_BT | |
1539 | && bidi_it->charpos >= bidi_it->next_for_ws.charpos) | |
1540 | bidi_it->next_for_ws.type = UNKNOWN_BT; | |
1541 | ||
1542 | /* This must be taken before we fill the iterator with the info | |
1543 | about the next char. If we scan backwards, the iterator | |
1544 | state must be already cached, so there's no need to know the | |
1545 | embedding level of the previous character, since we will be | |
1546 | returning to our caller shortly. */ | |
1547 | prev_level = bidi_it->level_stack[bidi_it->stack_idx].level; | |
1548 | } | |
1549 | next_for_neutral = bidi_it->next_for_neutral; | |
1550 | ||
1551 | /* Perhaps it is already cached. */ | |
1552 | type = bidi_cache_find (bidi_it->charpos + bidi_it->scan_dir, -1, bidi_it); | |
1553 | if (type != UNKNOWN_BT) | |
1554 | { | |
1555 | /* Don't lose the information for resolving neutrals! The | |
1556 | cached states could have been cached before their | |
1557 | next_for_neutral member was computed. If we are on our way | |
1558 | forward, we can simply take the info from the previous | |
1559 | state. */ | |
1560 | if (bidi_it->scan_dir == 1 | |
1561 | && bidi_it->next_for_neutral.type == UNKNOWN_BT) | |
1562 | bidi_it->next_for_neutral = next_for_neutral; | |
1563 | ||
1564 | /* If resolved_level is -1, it means this state was cached | |
1565 | before it was completely resolved, so we cannot return | |
1566 | it. */ | |
1567 | if (bidi_it->resolved_level != -1) | |
1568 | return bidi_it->resolved_level; | |
1569 | } | |
1570 | if (bidi_it->scan_dir == -1) | |
1571 | /* If we are going backwards, the iterator state is already cached | |
1572 | from previous scans, and should be fully resolved. */ | |
1573 | abort (); | |
1574 | ||
1575 | if (type == UNKNOWN_BT) | |
1576 | type = bidi_type_of_next_char (bidi_it); | |
1577 | ||
1578 | if (type == NEUTRAL_B) | |
1579 | return bidi_it->resolved_level; | |
1580 | ||
1581 | level = bidi_it->level_stack[bidi_it->stack_idx].level; | |
1582 | if ((bidi_get_category (type) == NEUTRAL /* && type != NEUTRAL_B */) | |
1583 | || (type == WEAK_BN && prev_level == level)) | |
1584 | { | |
1585 | if (bidi_it->next_for_neutral.type == UNKNOWN_BT) | |
1586 | abort (); | |
1587 | ||
1588 | /* If the cached state shows a neutral character, it was not | |
1589 | resolved by bidi_resolve_neutral, so do it now. */ | |
1590 | type = bidi_resolve_neutral_1 (bidi_it->prev_for_neutral.type, | |
1591 | bidi_it->next_for_neutral.type, | |
1592 | level); | |
1593 | } | |
1594 | ||
1595 | if (!(type == STRONG_R | |
1596 | || type == STRONG_L | |
1597 | || type == WEAK_BN | |
1598 | || type == WEAK_EN | |
1599 | || type == WEAK_AN)) | |
1600 | abort (); | |
1601 | bidi_it->type = type; | |
1602 | ||
1603 | /* For L1 below, we need to know, for each WS character, whether | |
1604 | it belongs to a sequence of WS characters preceeding a newline | |
1605 | or a TAB or a paragraph separator. */ | |
1606 | if (bidi_it->pristine_type == NEUTRAL_WS | |
1607 | && bidi_it->next_for_ws.type == UNKNOWN_BT) | |
1608 | { | |
1609 | int ch; | |
1610 | int clen = bidi_it->ch_len; | |
1611 | int bpos = bidi_it->bytepos; | |
1612 | int cpos = bidi_it->charpos; | |
1613 | bidi_type_t chtype; | |
1614 | ||
1615 | do { | |
1616 | /*_fetch_multibyte_char_len = 1;*/ | |
1617 | ch = FETCH_CHAR (bpos + clen); | |
1618 | bpos += clen; | |
1619 | cpos++; | |
1620 | clen = CHAR_BYTES (ch); | |
1621 | if (ch == '\n' /* || ch == LINESEP_CHAR */) | |
1622 | chtype = NEUTRAL_B; | |
1623 | else | |
1624 | chtype = bidi_get_type (ch); | |
1625 | } while (chtype == NEUTRAL_WS || chtype == WEAK_BN | |
1626 | || bidi_explicit_dir_char (ch)); /* L1/Retaining */ | |
1627 | bidi_it->next_for_ws.type = chtype; | |
1628 | bidi_it->next_for_ws.charpos = cpos; | |
1629 | bidi_it->next_for_ws.bytepos = bpos; | |
1630 | } | |
1631 | ||
1632 | /* Resolve implicit levels, with a twist: PDFs get the embedding | |
1633 | level of the enbedding they terminate. See below for the | |
1634 | reason. */ | |
1635 | if (bidi_it->pristine_type == PDF | |
1636 | /* Don't do this if this formatting code didn't change the | |
1637 | embedding level due to invalid or empty embeddings. */ | |
1638 | && prev_level != level) | |
1639 | { | |
1640 | /* Don't look in UAX#9 for the reason for this: it's our own | |
1641 | private quirk. The reason is that we want the formatting | |
1642 | codes to be delivered so that they bracket the text of their | |
1643 | embedding. For example, given the text | |
1644 | ||
1645 | {RLO}teST{PDF} | |
1646 | ||
1647 | we want it to be displayed as | |
1648 | ||
1649 | {RLO}STet{PDF} | |
1650 | ||
1651 | not as | |
1652 | ||
1653 | STet{RLO}{PDF} | |
1654 | ||
1655 | which will result because we bump up the embedding level as | |
1656 | soon as we see the RLO and pop it as soon as we see the PDF, | |
1657 | so RLO itself has the same embedding level as "teST", and | |
1658 | thus would be normally delivered last, just before the PDF. | |
1659 | The switch below fiddles with the level of PDF so that this | |
1660 | ugly side effect does not happen. | |
1661 | ||
1662 | (This is, of course, only important if the formatting codes | |
1663 | are actually displayed, but Emacs does display them if the | |
1664 | user wants to.) */ | |
1665 | level = prev_level; | |
1666 | } | |
1667 | else if (bidi_it->pristine_type == NEUTRAL_B /* L1 */ | |
1668 | || bidi_it->pristine_type == NEUTRAL_S | |
1669 | || bidi_it->ch == '\n' /* || bidi_it->ch == LINESEP_CHAR */ | |
1670 | || (bidi_it->pristine_type == NEUTRAL_WS | |
1671 | && (bidi_it->next_for_ws.type == NEUTRAL_B | |
1672 | || bidi_it->next_for_ws.type == NEUTRAL_S))) | |
1673 | level = bidi_it->level_stack[0].level; | |
1674 | else if ((level & 1) == 0) /* I1 */ | |
1675 | { | |
1676 | if (type == STRONG_R) | |
1677 | level++; | |
1678 | else if (type == WEAK_EN || type == WEAK_AN) | |
1679 | level += 2; | |
1680 | } | |
1681 | else /* I2 */ | |
1682 | { | |
1683 | if (type == STRONG_L || type == WEAK_EN || type == WEAK_AN) | |
1684 | level++; | |
1685 | } | |
1686 | ||
1687 | bidi_it->resolved_level = level; | |
1688 | return level; | |
1689 | } | |
1690 | ||
1691 | /* Move to the other edge of a level given by LEVEL. If END_FLAG is | |
1692 | non-zero, we are at the end of a level, and we need to prepare to | |
1693 | resume the scan of the lower level. | |
1694 | ||
1695 | If this level's other edge is cached, we simply jump to it, filling | |
1696 | the iterator structure with the iterator state on the other edge. | |
1697 | Otherwise, we walk the buffer until we come back to the same level | |
1698 | as LEVEL. | |
1699 | ||
1700 | Note: we are not talking here about a ``level run'' in the UAX#9 | |
1701 | sense of the term, but rather about a ``level'' which includes | |
1702 | all the levels higher than it. In other words, given the levels | |
1703 | like this: | |
1704 | ||
1705 | 11111112222222333333334443343222222111111112223322111 | |
1706 | A B C | |
1707 | ||
1708 | and assuming we are at point A scanning left to right, this | |
1709 | function moves to point C, whereas the UAX#9 ``level 2 run'' ends | |
1710 | at point B. */ | |
1711 | static void | |
1712 | bidi_find_other_level_edge (struct bidi_it *bidi_it, int level, int end_flag) | |
1713 | { | |
1714 | int dir = end_flag ? -bidi_it->scan_dir : bidi_it->scan_dir; | |
1715 | int idx; | |
1716 | ||
1717 | /* Try the cache first. */ | |
1718 | if ((idx = bidi_cache_find_level_change (level, dir, end_flag)) >= 0) | |
1719 | bidi_cache_fetch_state (idx, bidi_it); | |
1720 | else | |
1721 | { | |
1722 | int new_level; | |
1723 | ||
1724 | if (end_flag) | |
1725 | abort (); /* if we are at end of level, its edges must be cached */ | |
1726 | ||
1727 | bidi_cache_iterator_state (bidi_it, 1); | |
1728 | do { | |
1729 | new_level = bidi_level_of_next_char (bidi_it); | |
1730 | bidi_cache_iterator_state (bidi_it, 1); | |
1731 | } while (new_level >= level); | |
1732 | } | |
1733 | } | |
1734 | ||
1735 | void | |
1736 | bidi_get_next_char_visually (struct bidi_it *bidi_it) | |
1737 | { | |
1738 | int old_level, new_level, next_level; | |
1739 | struct bidi_it prev_bidi_it; | |
1740 | ||
1741 | if (bidi_it->scan_dir == 0) | |
1742 | { | |
1743 | bidi_it->scan_dir = 1; /* default to logical order */ | |
1744 | } | |
1745 | ||
1746 | if (bidi_it->new_paragraph) | |
1747 | bidi_paragraph_init (bidi_overriding_paragraph_direction, bidi_it); | |
1748 | if (bidi_cache_idx == 0) | |
1749 | bidi_copy_it (&prev_bidi_it, bidi_it); | |
1750 | ||
1751 | old_level = bidi_it->resolved_level; | |
1752 | new_level = bidi_level_of_next_char (bidi_it); | |
1753 | if (bidi_it->ch == BIDI_EOB) | |
1754 | return; | |
1755 | ||
1756 | /* Reordering of resolved levels (clause L2) is implemented by | |
1757 | jumping to the other edge of the level and flipping direction of | |
1758 | scanning the buffer whenever we find a level change. */ | |
1759 | if (new_level != old_level) | |
1760 | { | |
1761 | int ascending = new_level > old_level; | |
1762 | int level_to_search = ascending ? old_level + 1 : old_level; | |
1763 | int incr = ascending ? 1 : -1; | |
1764 | int expected_next_level = old_level + incr; | |
1765 | ||
1766 | /* If we don't have anything cached yet, we need to cache the | |
1767 | previous character we've seen, since we'll need it to record | |
1768 | where to jump when the last non-base level is exhausted. */ | |
1769 | if (bidi_cache_idx == 0) | |
1770 | bidi_cache_iterator_state (&prev_bidi_it, 1); | |
1771 | /* Jump (or walk) to the other edge of this level. */ | |
1772 | bidi_find_other_level_edge (bidi_it, level_to_search, !ascending); | |
1773 | /* Switch scan direction and peek at the next character in the | |
1774 | new direction. */ | |
1775 | bidi_it->scan_dir = -bidi_it->scan_dir; | |
1776 | ||
1777 | /* The following loop handles the case where the resolved level | |
1778 | jumps by more than one. This is typical for numbers inside a | |
1779 | run of text with left-to-right embedding direction, but can | |
1780 | also happen in other situations. In those cases the decision | |
1781 | where to continue after a level change, and in what direction, | |
1782 | is tricky. For example, given a text like below: | |
1783 | ||
1784 | abcdefgh | |
1785 | 11336622 | |
1786 | ||
1787 | (where the numbers below the text show the resolved levels), | |
1788 | the result of reordering according to UAX#9 should be this: | |
1789 | ||
1790 | efdcghba | |
1791 | ||
1792 | This is implemented by the loop below which flips direction | |
1793 | and jumps to the other edge of the level each time it finds | |
1794 | the new level not to be the expected one. The expected level | |
1795 | is always one more or one less than the previous one. */ | |
1796 | next_level = bidi_peek_at_next_level (bidi_it); | |
1797 | while (next_level != expected_next_level) | |
1798 | { | |
1799 | expected_next_level += incr; | |
1800 | level_to_search += incr; | |
1801 | bidi_find_other_level_edge (bidi_it, level_to_search, !ascending); | |
1802 | bidi_it->scan_dir = -bidi_it->scan_dir; | |
1803 | next_level = bidi_peek_at_next_level (bidi_it); | |
1804 | } | |
1805 | ||
1806 | /* Finally, deliver the next character in the new direction. */ | |
1807 | next_level = bidi_level_of_next_char (bidi_it); | |
1808 | } | |
1809 | ||
1810 | if (bidi_it->scan_dir == 1 && bidi_cache_idx) | |
1811 | { | |
1812 | /* If we are at paragraph's base embedding level and beyond the | |
1813 | last cached position, the cache's job is done and we can | |
1814 | discard it. */ | |
1815 | if (bidi_it->resolved_level == bidi_it->level_stack[0].level | |
1816 | && bidi_it->charpos > bidi_cache[bidi_cache_idx - 1].charpos) | |
1817 | bidi_cache_reset (); | |
1818 | /* But as long as we are caching during forward scan, we must | |
1819 | cache each state, or else the cache integrity will be | |
1820 | compromised: it assumes cached states correspond to buffer | |
1821 | positions 1:1. */ | |
1822 | else | |
1823 | bidi_cache_iterator_state (bidi_it, 1); | |
1824 | } | |
1825 | } | |
1826 | ||
1827 | /* This is meant to be called from within the debugger, whenever you | |
1828 | wish to examine the cache contents. */ | |
1829 | void | |
1830 | bidi_dump_cached_states (void) | |
1831 | { | |
1832 | int i; | |
1833 | int ndigits = 1; | |
1834 | ||
1835 | if (bidi_cache_idx == 0) | |
1836 | { | |
1837 | fprintf (stderr, "The cache is empty.\n"); | |
1838 | return; | |
1839 | } | |
1840 | fprintf (stderr, "Total of %d state%s in cache:\n", | |
1841 | bidi_cache_idx, bidi_cache_idx == 1 ? "" : "s"); | |
1842 | ||
1843 | for (i = bidi_cache[bidi_cache_idx - 1].charpos; i > 0; i /= 10) | |
1844 | ndigits++; | |
1845 | fputs ("ch ", stderr); | |
1846 | for (i = 0; i < bidi_cache_idx; i++) | |
1847 | fprintf (stderr, "%*c", ndigits, bidi_cache[i].ch); | |
1848 | fputs ("\n", stderr); | |
1849 | fputs ("lvl ", stderr); | |
1850 | for (i = 0; i < bidi_cache_idx; i++) | |
1851 | fprintf (stderr, "%*d", ndigits, bidi_cache[i].resolved_level); | |
1852 | fputs ("\n", stderr); | |
1853 | fputs ("pos ", stderr); | |
1854 | for (i = 0; i < bidi_cache_idx; i++) | |
1855 | fprintf (stderr, "%*d", ndigits, bidi_cache[i].charpos); | |
1856 | fputs ("\n", stderr); | |
1857 | } | |
1858 | ||
1859 | #ifdef TEST_STANDALONE | |
1860 | ||
1861 | #include <sys/stat.h> | |
1862 | #include <signal.h> | |
1863 | ||
1864 | static char display_line[80]; | |
1865 | static int simulate_display; | |
1866 | static int incr = 1; | |
1867 | ||
1868 | void | |
1869 | signal_catcher (int sig) | |
1870 | { | |
1871 | if (simulate_display) | |
1872 | puts (display_line); | |
1873 | else | |
1874 | { | |
1875 | puts ("<"); | |
1876 | fflush (stdout); | |
1877 | } | |
1878 | signal (sig, SIG_DFL); | |
1879 | raise (sig); | |
1880 | } | |
1881 | ||
1882 | void | |
1883 | put_line (char *p) | |
1884 | { | |
1885 | if (simulate_display) | |
1886 | { | |
1887 | if (incr == -1) | |
1888 | { | |
1889 | if (p >= display_line) | |
1890 | memset (display_line, ' ', p - display_line + 1); | |
1891 | } | |
1892 | else | |
1893 | *p = '\0'; | |
1894 | fputs (display_line, stdout); | |
1895 | } | |
1896 | fflush (stdout); | |
1897 | } | |
1898 | ||
1899 | char * | |
1900 | init_display_direction (bidi_dir_t default_dir, int base_level) | |
1901 | { | |
1902 | char *p; | |
1903 | ||
1904 | /* To which display margin should we flush the lines? */ | |
1905 | switch (default_dir) | |
1906 | { | |
1907 | case NEUTRAL_DIR: | |
1908 | if ((base_level & 1) == 0) | |
1909 | { | |
1910 | p = display_line; | |
1911 | incr = 1; | |
1912 | } | |
1913 | else | |
1914 | { | |
1915 | p = display_line + sizeof (display_line) - 1; | |
1916 | *p-- = '\0'; | |
1917 | incr = -1; | |
1918 | } | |
1919 | break; | |
1920 | case L2R: | |
1921 | p = display_line; | |
1922 | incr = 1; | |
1923 | break; | |
1924 | case R2L: | |
1925 | p = display_line + sizeof (display_line) - 1; | |
1926 | *p-- = '\0'; | |
1927 | incr = -1; | |
1928 | break; | |
1929 | default: | |
1930 | abort (); | |
1931 | } | |
1932 | ||
1933 | return p; | |
1934 | } | |
1935 | ||
1936 | static char * | |
1937 | continuation_line (char *p, int need) | |
1938 | { | |
1939 | if (incr == -1) | |
1940 | { | |
1941 | if (p < display_line + need) | |
1942 | { | |
1943 | *p-- = '/'; | |
1944 | put_line (p); | |
1945 | putc ('\n', stdout); | |
1946 | memset (display_line, '>', sizeof(display_line) - 1); | |
1947 | p = display_line + sizeof (display_line) - 1; | |
1948 | *p-- = '\0'; | |
1949 | } | |
1950 | } | |
1951 | else | |
1952 | { | |
1953 | if (p > display_line + sizeof(display_line) - need - 2) | |
1954 | { | |
1955 | *p++ = '\\'; | |
1956 | put_line (p); | |
1957 | putc ('\n', stdout); | |
1958 | memset (display_line, '<', sizeof(display_line) - 1); | |
1959 | p = display_line; | |
1960 | } | |
1961 | } | |
1962 | ||
1963 | return p; | |
1964 | } | |
1965 | ||
1966 | int main (int argc, char *argv[]) | |
1967 | { | |
1968 | bidi_dir_t default_dir = NEUTRAL_DIR; | |
1969 | char lots_of_equals[] = "\n===============================================================================\n"; | |
1970 | ||
1971 | ||
1972 | if (argc > 1 && argv[1][0] == '-') | |
1973 | { | |
1974 | switch (argv[1][1]) | |
1975 | { | |
1976 | case 'R': | |
1977 | default_dir = R2L; | |
1978 | simulate_display = 1; | |
1979 | ++argv; | |
1980 | break; | |
1981 | case 'L': | |
1982 | default_dir = L2R; | |
1983 | simulate_display = 1; | |
1984 | ++argv; | |
1985 | break; | |
1986 | case 'N': | |
1987 | simulate_display = 1; | |
1988 | ++argv; | |
1989 | break; | |
1990 | default: | |
1991 | break; | |
1992 | } | |
1993 | bidi_overriding_paragraph_direction = default_dir; | |
1994 | } | |
1995 | ||
1996 | for (argv++; *argv; argv++) | |
1997 | { | |
1998 | FILE *in = fopen (*argv, "rb"); | |
1999 | struct stat stat_buf; | |
2000 | struct bidi_it iterator; | |
2001 | size_t i; | |
2002 | char *p = display_line; | |
2003 | int base_level; | |
2004 | unsigned char *s, *d, *s_end; | |
2005 | ||
2006 | if (!in || stat (*argv, &stat_buf)) | |
2007 | { | |
2008 | perror (*argv); | |
2009 | continue; | |
2010 | } | |
2011 | ||
2012 | if (stat_buf.st_size > input_buf_size) | |
2013 | { | |
2014 | input_buf = realloc (input_buf, stat_buf.st_size + 1); | |
2015 | if (!input_buf) | |
2016 | { | |
2017 | perror ("realloc input buffer"); | |
2018 | continue; | |
2019 | } | |
2020 | input_buf_size = stat_buf.st_size; | |
2021 | } | |
2022 | if (fread (input_buf, 1, stat_buf.st_size, in) != stat_buf.st_size) | |
2023 | { | |
2024 | perror ("reading input"); | |
2025 | continue; | |
2026 | } | |
2027 | input_buf[stat_buf.st_size] = '\0'; | |
2028 | for (d = s = input_buf, s_end = s + stat_buf.st_size - 1; *s; s++) | |
2029 | { | |
2030 | if (*s != '\r' || s >= s_end || s[1] != '\n') | |
2031 | *d++ = *s; | |
2032 | } | |
2033 | stat_buf.st_size = d - input_buf; | |
2034 | input_buf[stat_buf.st_size] = '\0'; | |
2035 | ||
2036 | /* Done with administrivia, now for some real work... */ | |
2037 | signal (SIGABRT, signal_catcher); | |
2038 | signal (SIGINT, signal_catcher); | |
2039 | bidi_init_it (-1, default_dir, &iterator); | |
2040 | if (simulate_display) | |
2041 | { | |
2042 | p = init_display_direction (default_dir, | |
2043 | iterator.level_stack[0].level); | |
2044 | } | |
2045 | ||
2046 | memset (display_line, incr == -1 ? '>' : '<', sizeof (display_line) - 1); | |
2047 | display_line[sizeof (display_line) - 1] = '\0'; | |
2048 | base_level = iterator.level_stack[0].level; | |
2049 | ||
2050 | for (i = 0; i <= stat_buf.st_size; i++) | |
2051 | { | |
2052 | int c; | |
2053 | ||
2054 | bidi_get_next_char_visually (&iterator); | |
2055 | c = iterator.ch; | |
2056 | ||
2057 | if (c == '\n' || c == BIDI_EOB) | |
2058 | { | |
2059 | if (simulate_display) | |
2060 | { | |
2061 | put_line (p); | |
2062 | /* FIXME: if -R or -L, need to init paragraph here. */ | |
2063 | } | |
2064 | if (c == BIDI_EOB) | |
2065 | break; | |
2066 | putc (c, stdout); | |
2067 | } | |
2068 | else if (c >= LRE_CHAR && c <= LRM_CHAR) | |
2069 | { | |
2070 | if (simulate_display) | |
2071 | { | |
2072 | p = continuation_line (p, 5); | |
2073 | if (incr == -1) | |
2074 | { | |
2075 | memcpy (p - 4, bidi_name[c], 5); | |
2076 | p -= 5; | |
2077 | } | |
2078 | else | |
2079 | { | |
2080 | memcpy (p, bidi_name[c], 5); | |
2081 | p += 5; | |
2082 | } | |
2083 | } | |
2084 | else | |
2085 | fputs (bidi_name[c], stdout); | |
2086 | } | |
2087 | else if (c < ' ') | |
2088 | { | |
2089 | if (simulate_display) | |
2090 | { | |
2091 | p = continuation_line (p, 2); | |
2092 | if (incr == -1) | |
2093 | { | |
2094 | *p-- = '@' + c; | |
2095 | *p-- = '^'; | |
2096 | } | |
2097 | else | |
2098 | { | |
2099 | *p++ = '^'; | |
2100 | *p++ = '@' + c; | |
2101 | } | |
2102 | } | |
2103 | else | |
2104 | printf ("^%c", (c | 0x40)); | |
2105 | } | |
2106 | else | |
2107 | { | |
2108 | int c1 = (iterator.type == STRONG_R) ? bidi_mirror_char (c) : c; | |
2109 | ||
2110 | if (simulate_display) | |
2111 | { | |
2112 | p = continuation_line (p, 1); | |
2113 | *p = c1; | |
2114 | p += incr; | |
2115 | } | |
2116 | else | |
2117 | putc (c1, stdout); | |
2118 | } | |
2119 | ||
2120 | if (iterator.ch == '\n') | |
2121 | { | |
2122 | if (base_level != iterator.level_stack[0].level) | |
2123 | base_level = iterator.level_stack[0].level; | |
2124 | p = init_display_direction (default_dir, base_level); | |
2125 | memset (display_line, incr == -1 ? '>' : '<', | |
2126 | sizeof (display_line) - 1); | |
2127 | } | |
2128 | } | |
2129 | fputs (lots_of_equals, stdout); | |
2130 | fclose (in); | |
2131 | } | |
2132 | return 0; | |
2133 | } | |
2134 | #endif |