1 /*
2 * "streamable kanji code filter and converter"
3 * Copyright (c) 1998-2002 HappySize, Inc. All rights reserved.
4 *
5 * LICENSE NOTICES
6 *
7 * This file is part of "streamable kanji code filter and converter",
8 * which is distributed under the terms of GNU Lesser General Public
9 * License (version 2) as published by the Free Software Foundation.
10 *
11 * This software 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 Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with "streamable kanji code filter and converter";
18 * if not, write to the Free Software Foundation, Inc., 59 Temple Place,
19 * Suite 330, Boston, MA 02111-1307 USA
20 *
21 * The author of this file:
22 *
23 */
24 /*
25 * the source code included in this files was separated from mbfilter_cp936.c
26 * by rui hirokawa <hirokawa@php.net> on 11 Aug 2011.
27 *
28 */
29
30 #include "mbfilter.h"
31 #include "mbfilter_gb18030.h"
32
33 #include "unicode_table_cp936.h"
34 #include "unicode_table_gb18030.h"
35
36 static int mbfl_filt_conv_gb18030_wchar_flush(mbfl_convert_filter *filter);
37 static size_t mb_gb18030_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
38 static void mb_wchar_to_gb18030(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
39
40 static const char *mbfl_encoding_gb18030_aliases[] = {"gb-18030", "gb-18030-2000", NULL};
41
42 const mbfl_encoding mbfl_encoding_gb18030 = {
43 mbfl_no_encoding_gb18030,
44 "GB18030",
45 "GB18030",
46 mbfl_encoding_gb18030_aliases,
47 NULL,
48 MBFL_ENCTYPE_GL_UNSAFE,
49 &vtbl_gb18030_wchar,
50 &vtbl_wchar_gb18030,
51 mb_gb18030_to_wchar,
52 mb_wchar_to_gb18030,
53 NULL
54 };
55
56 const struct mbfl_convert_vtbl vtbl_gb18030_wchar = {
57 mbfl_no_encoding_gb18030,
58 mbfl_no_encoding_wchar,
59 mbfl_filt_conv_common_ctor,
60 NULL,
61 mbfl_filt_conv_gb18030_wchar,
62 mbfl_filt_conv_gb18030_wchar_flush,
63 NULL,
64 };
65
66 const struct mbfl_convert_vtbl vtbl_wchar_gb18030 = {
67 mbfl_no_encoding_wchar,
68 mbfl_no_encoding_gb18030,
69 mbfl_filt_conv_common_ctor,
70 NULL,
71 mbfl_filt_conv_wchar_gb18030,
72 mbfl_filt_conv_common_flush,
73 NULL,
74 };
75
76 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
77
78 /* `tbl` contains inclusive ranges, each represented by a pair of unsigned shorts */
mbfl_bisec_srch(int w,const unsigned short * tbl,int n)79 int mbfl_bisec_srch(int w, const unsigned short *tbl, int n)
80 {
81 int l = 0, r = n-1;
82 while (l <= r) {
83 int probe = (l + r) >> 1;
84 unsigned short lo = tbl[2 * probe], hi = tbl[(2 * probe) + 1];
85 if (w < lo) {
86 r = probe - 1;
87 } else if (w > hi) {
88 l = probe + 1;
89 } else {
90 return probe;
91 }
92 }
93 return -1;
94 }
95
96 /* `tbl` contains single values, not ranges */
mbfl_bisec_srch2(int w,const unsigned short tbl[],int n)97 int mbfl_bisec_srch2(int w, const unsigned short tbl[], int n)
98 {
99 int l = 0, r = n-1;
100 while (l <= r) {
101 int probe = (l + r) >> 1;
102 unsigned short val = tbl[probe];
103 if (w < val) {
104 r = probe - 1;
105 } else if (w > val) {
106 l = probe + 1;
107 } else {
108 return probe;
109 }
110 }
111 return -1;
112 }
113
mbfl_filt_conv_gb18030_wchar(int c,mbfl_convert_filter * filter)114 int mbfl_filt_conv_gb18030_wchar(int c, mbfl_convert_filter *filter)
115 {
116 int k;
117 int c1, c2, c3, w = -1;
118
119 switch (filter->status) {
120 case 0:
121 if (c >= 0 && c < 0x80) { /* latin */
122 CK((*filter->output_function)(c, filter->data));
123 } else if (c > 0x80 && c < 0xff) { /* dbcs/qbcs lead byte */
124 filter->status = 1;
125 filter->cache = c;
126 } else {
127 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
128 }
129 break;
130
131 case 1: /* dbcs/qbcs second byte */
132 c1 = filter->cache;
133 filter->status = 0;
134
135 if (c1 >= 0x81 && c1 <= 0x84 && c >= 0x30 && c <= 0x39) {
136 /* 4 byte range: Unicode BMP */
137 filter->status = 2;
138 filter->cache = (c1 << 8) | c;
139 return 0;
140 } else if (c1 >= 0x90 && c1 <= 0xe3 && c >= 0x30 && c <= 0x39) {
141 /* 4 byte range: Unicode 16 planes */
142 filter->status = 2;
143 filter->cache = (c1 << 8) | c;
144 return 0;
145 } else if (((c1 >= 0xaa && c1 <= 0xaf) || (c1 >= 0xf8 && c1 <= 0xfe)) && (c >= 0xa1 && c <= 0xfe)) {
146 /* UDA part 1,2: U+E000-U+E4C5 */
147 w = 94*(c1 >= 0xf8 ? c1 - 0xf2 : c1 - 0xaa) + (c - 0xa1) + 0xe000;
148 CK((*filter->output_function)(w, filter->data));
149 } else if (c1 >= 0xa1 && c1 <= 0xa7 && c >= 0x40 && c < 0xa1 && c != 0x7f) {
150 /* UDA part3 : U+E4C6-U+E765*/
151 w = 96*(c1 - 0xa1) + c - (c >= 0x80 ? 0x41 : 0x40) + 0xe4c6;
152 CK((*filter->output_function)(w, filter->data));
153 }
154
155 c2 = (c1 << 8) | c;
156
157 if (w <= 0 &&
158 ((c2 >= 0xa2ab && c2 <= 0xa9f0 + (0xe80f-0xe801)) ||
159 (c2 >= 0xd7fa && c2 <= 0xd7fa + (0xe814-0xe810)) ||
160 (c2 >= 0xfe50 && c2 <= 0xfe80 + (0xe864-0xe844)))) {
161 for (k = 0; k < mbfl_gb18030_pua_tbl_max; k++) {
162 if (c2 >= mbfl_gb18030_pua_tbl[k][2] && c2 <= mbfl_gb18030_pua_tbl[k][2] + mbfl_gb18030_pua_tbl[k][1] - mbfl_gb18030_pua_tbl[k][0]) {
163 w = c2 - mbfl_gb18030_pua_tbl[k][2] + mbfl_gb18030_pua_tbl[k][0];
164 CK((*filter->output_function)(w, filter->data));
165 break;
166 }
167 }
168 }
169
170 if (w <= 0) {
171 if ((c1 >= 0xa1 && c1 <= 0xa9 && c >= 0xa1 && c <= 0xfe) ||
172 (c1 >= 0xb0 && c1 <= 0xf7 && c >= 0xa1 && c <= 0xfe) ||
173 (c1 >= 0x81 && c1 <= 0xa0 && c >= 0x40 && c <= 0xfe && c != 0x7f) ||
174 (c1 >= 0xaa && c1 <= 0xfe && c >= 0x40 && c <= 0xa0 && c != 0x7f) ||
175 (c1 >= 0xa8 && c1 <= 0xa9 && c >= 0x40 && c <= 0xa0 && c != 0x7f)) {
176 w = (c1 - 0x81)*192 + c - 0x40;
177 ZEND_ASSERT(w < cp936_ucs_table_size);
178 CK((*filter->output_function)(cp936_ucs_table[w], filter->data));
179 } else {
180 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
181 }
182 }
183 break;
184
185 case 2: /* qbcs third byte */
186 c1 = (filter->cache >> 8) & 0xff;
187 c2 = filter->cache & 0xff;
188 filter->status = filter->cache = 0;
189 if (((c1 >= 0x81 && c1 <= 0x84) || (c1 >= 0x90 && c1 <= 0xe3)) && c2 >= 0x30 && c2 <= 0x39 && c >= 0x81 && c <= 0xfe) {
190 filter->cache = (c1 << 16) | (c2 << 8) | c;
191 filter->status = 3;
192 } else {
193 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
194 }
195 break;
196
197 case 3: /* qbcs fourth byte */
198 c1 = (filter->cache >> 16) & 0xff;
199 c2 = (filter->cache >> 8) & 0xff;
200 c3 = filter->cache & 0xff;
201 filter->status = filter->cache = 0;
202 if (((c1 >= 0x81 && c1 <= 0x84) || (c1 >= 0x90 && c1 <= 0xe3)) && c2 >= 0x30 && c2 <= 0x39 && c3 >= 0x81 && c3 <= 0xfe && c >= 0x30 && c <= 0x39) {
203 if (c1 >= 0x90 && c1 <= 0xe3) {
204 w = ((((c1 - 0x90)*10 + (c2 - 0x30))*126 + (c3 - 0x81)))*10 + (c - 0x30) + 0x10000;
205 if (w > 0x10FFFF) {
206 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
207 return 0;
208 }
209 } else { /* Unicode BMP */
210 w = (((c1 - 0x81)*10 + (c2 - 0x30))*126 + (c3 - 0x81))*10 + (c - 0x30);
211 if (w >= 0 && w <= 39419) {
212 k = mbfl_bisec_srch(w, mbfl_gb2uni_tbl, mbfl_gb_uni_max);
213 w += mbfl_gb_uni_ofst[k];
214 } else {
215 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
216 return 0;
217 }
218 }
219 CK((*filter->output_function)(w, filter->data));
220 } else {
221 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
222 }
223 break;
224
225 EMPTY_SWITCH_DEFAULT_CASE();
226 }
227
228 return 0;
229 }
230
mbfl_filt_conv_gb18030_wchar_flush(mbfl_convert_filter * filter)231 static int mbfl_filt_conv_gb18030_wchar_flush(mbfl_convert_filter *filter)
232 {
233 if (filter->status) {
234 /* multi-byte character was truncated */
235 filter->status = 0;
236 CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
237 }
238
239 if (filter->flush_function) {
240 (*filter->flush_function)(filter->data);
241 }
242
243 return 0;
244 }
245
mbfl_filt_conv_wchar_gb18030(int c,mbfl_convert_filter * filter)246 int mbfl_filt_conv_wchar_gb18030(int c, mbfl_convert_filter *filter)
247 {
248 int k, k1, k2;
249 int c1, s = 0, s1 = 0;
250
251 if (c >= ucs_a1_cp936_table_min && c < ucs_a1_cp936_table_max) {
252 if (c == 0x01f9) {
253 s = 0xa8bf;
254 } else {
255 s = ucs_a1_cp936_table[c - ucs_a1_cp936_table_min];
256 }
257 } else if (c >= ucs_a2_cp936_table_min && c < ucs_a2_cp936_table_max) {
258 if (c == 0x20ac) { /* euro-sign */
259 s = 0xa2e3;
260 } else {
261 s = ucs_a2_cp936_table[c - ucs_a2_cp936_table_min];
262 }
263 } else if (c >= ucs_a3_cp936_table_min && c < ucs_a3_cp936_table_max) {
264 s = ucs_a3_cp936_table[c - ucs_a3_cp936_table_min];
265 } else if (c >= ucs_i_cp936_table_min && c < ucs_i_cp936_table_max) {
266 s = ucs_i_cp936_table[c - ucs_i_cp936_table_min];
267 } else if (c >= ucs_ci_cp936_table_min && c < ucs_ci_cp936_table_max) {
268 /* U+F900-FA2F CJK Compatibility Ideographs */
269 if (c == 0xf92c) {
270 s = 0xfd9c;
271 } else if (c == 0xf979) {
272 s = 0xfd9d;
273 } else if (c == 0xf995) {
274 s = 0xfd9e;
275 } else if (c == 0xf9e7) {
276 s = 0xfd9f;
277 } else if (c == 0xf9f1) {
278 s = 0xfda0;
279 } else if (c >= 0xfa0c && c <= 0xfa29) {
280 s = ucs_ci_s_cp936_table[c - 0xfa0c];
281 }
282 } else if (c >= ucs_cf_cp936_table_min && c < ucs_cf_cp936_table_max) {
283 /* FE30h CJK Compatibility Forms */
284 s = ucs_cf_cp936_table[c - ucs_cf_cp936_table_min];
285 } else if (c >= ucs_sfv_cp936_table_min && c < ucs_sfv_cp936_table_max) {
286 /* U+FE50-FE6F Small Form Variants */
287 s = ucs_sfv_cp936_table[c - ucs_sfv_cp936_table_min];
288 } else if (c >= ucs_hff_cp936_table_min && c < ucs_hff_cp936_table_max) {
289 /* U+FF00-FFFF HW/FW Forms */
290 if (c == 0xff04) {
291 s = 0xa1e7;
292 } else if (c == 0xff5e) {
293 s = 0xa1ab;
294 } else if (c >= 0xff01 && c <= 0xff5d) {
295 s = c - 0xff01 + 0xa3a1;
296 } else if (c >= 0xffe0 && c <= 0xffe5) {
297 s = ucs_hff_s_cp936_table[c-0xffe0];
298 }
299 }
300
301 /* While GB18030 and CP936 are very similar, some mappings are different between these encodings;
302 * do a binary search in a table of differing codepoints to see if we have one */
303 if (s <= 0 && c >= mbfl_gb18030_c_tbl_key[0] && c <= mbfl_gb18030_c_tbl_key[mbfl_gb18030_c_tbl_max-1]) {
304 k1 = mbfl_bisec_srch2(c, mbfl_gb18030_c_tbl_key, mbfl_gb18030_c_tbl_max);
305 if (k1 >= 0) {
306 s = mbfl_gb18030_c_tbl_val[k1];
307 }
308 }
309
310 if (c >= 0xe000 && c <= 0xe864) { /* PUA */
311 if (c < 0xe766) {
312 if (c < 0xe4c6) {
313 c1 = c - 0xe000;
314 s = (c1 % 94) + 0xa1;
315 c1 /= 94;
316 s |= (c1 < 0x06 ? c1 + 0xaa : c1 + 0xf2) << 8;
317 } else {
318 c1 = c - 0xe4c6;
319 s = ((c1 / 96) + 0xa1) << 8;
320 c1 %= 96;
321 s |= c1 + (c1 >= 0x3f ? 0x41 : 0x40);
322 }
323 } else {
324 /* U+E766..U+E864 */
325 k1 = 0;
326 k2 = mbfl_gb18030_pua_tbl_max;
327 while (k1 < k2) {
328 k = (k1 + k2) >> 1;
329 if (c < mbfl_gb18030_pua_tbl[k][0]) {
330 k2 = k;
331 } else if (c > mbfl_gb18030_pua_tbl[k][1]) {
332 k1 = k + 1;
333 } else {
334 s = c - mbfl_gb18030_pua_tbl[k][0] + mbfl_gb18030_pua_tbl[k][2];
335 break;
336 }
337 }
338 }
339 }
340
341 /* If we have not yet found a suitable mapping for this codepoint, it requires a 4-byte code */
342 if (s <= 0 && c >= 0x0080 && c <= 0xffff) {
343 /* BMP */
344 s = mbfl_bisec_srch(c, mbfl_uni2gb_tbl, mbfl_gb_uni_max);
345 if (s >= 0) {
346 c1 = c - mbfl_gb_uni_ofst[s];
347 s = (c1 % 10) + 0x30;
348 c1 /= 10;
349 s |= ((c1 % 126) + 0x81) << 8;
350 c1 /= 126;
351 s |= ((c1 % 10) + 0x30) << 16;
352 c1 /= 10;
353 s1 = c1 + 0x81;
354 }
355 } else if (c >= 0x10000 && c <= 0x10ffff) {
356 /* Code set 3: Unicode U+10000..U+10FFFF */
357 c1 = c - 0x10000;
358 s = (c1 % 10) + 0x30;
359 c1 /= 10;
360 s |= ((c1 % 126) + 0x81) << 8;
361 c1 /= 126;
362 s |= ((c1 % 10) + 0x30) << 16;
363 c1 /= 10;
364 s1 = c1 + 0x90;
365 }
366
367 if (c == 0) {
368 s = 0;
369 } else if (s == 0) {
370 s = -1;
371 }
372
373 if (s >= 0) {
374 if (s <= 0x80) { /* latin */
375 CK((*filter->output_function)(s, filter->data));
376 } else if (s1 > 0) { /* qbcs */
377 CK((*filter->output_function)(s1 & 0xff, filter->data));
378 CK((*filter->output_function)((s >> 16) & 0xff, filter->data));
379 CK((*filter->output_function)((s >> 8) & 0xff, filter->data));
380 CK((*filter->output_function)(s & 0xff, filter->data));
381 } else { /* dbcs */
382 CK((*filter->output_function)((s >> 8) & 0xff, filter->data));
383 CK((*filter->output_function)(s & 0xff, filter->data));
384 }
385 } else {
386 CK(mbfl_filt_conv_illegal_output(c, filter));
387 }
388
389 return 0;
390 }
391
mb_gb18030_to_wchar(unsigned char ** in,size_t * in_len,uint32_t * buf,size_t bufsize,unsigned int * state)392 static size_t mb_gb18030_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
393 {
394 unsigned char *p = *in, *e = p + *in_len;
395 uint32_t *out = buf, *limit = buf + bufsize;
396
397 while (p < e && out < limit) {
398 unsigned char c = *p++;
399
400 if (c < 0x80) {
401 *out++ = c;
402 } else if (c > 0x80 && c < 0xFF && p < e) {
403 unsigned char c2 = *p++;
404 unsigned int s = (c << 8) | c2;
405
406 if (((c >= 0x81 && c <= 0x84) || (c >= 0x90 && c <= 0xE3)) && c2 >= 0x30 && c2 <= 0x39) {
407 if (p >= e) {
408 *out++ = MBFL_BAD_INPUT;
409 break;
410 }
411 unsigned char c3 = *p++;
412
413 if (c3 >= 0x81 && c3 <= 0xFE && p < e) {
414 unsigned char c4 = *p++;
415
416 if (c4 >= 0x30 && c4 <= 0x39) {
417 if (c >= 0x90 && c <= 0xE3) {
418 unsigned int w = ((((c - 0x90)*10 + (c2 - 0x30))*126 + (c3 - 0x81)))*10 + (c4 - 0x30) + 0x10000;
419 *out++ = (w > 0x10FFFF) ? MBFL_BAD_INPUT : w;
420 } else {
421 /* Unicode BMP */
422 unsigned int w = (((c - 0x81)*10 + (c2 - 0x30))*126 + (c3 - 0x81))*10 + (c4 - 0x30);
423 if (w <= 39419) {
424 *out++ = w + mbfl_gb_uni_ofst[mbfl_bisec_srch(w, mbfl_gb2uni_tbl, mbfl_gb_uni_max)];
425 } else {
426 *out++ = MBFL_BAD_INPUT;
427 }
428 }
429 } else {
430 *out++ = MBFL_BAD_INPUT;
431 }
432 } else {
433 *out++ = MBFL_BAD_INPUT;
434 }
435 } else if (((c >= 0xAA && c <= 0xAF) || (c >= 0xF8 && c <= 0xFE)) && (c2 >= 0xA1 && c2 <= 0xFE)) {
436 /* UDA part 1, 2: U+E000-U+E4C5 */
437 *out++ = 94*(c >= 0xF8 ? c - 0xF2 : c - 0xAA) + (c2 - 0xA1) + 0xE000;
438 } else if (c >= 0xA1 && c <= 0xA7 && c2 >= 0x40 && c2 < 0xA1 && c2 != 0x7F) {
439 /* UDA part 3: U+E4C6-U+E765 */
440 *out++ = 96*(c - 0xA1) + c2 - (c2 >= 0x80 ? 0x41 : 0x40) + 0xE4C6;
441 } else {
442 if ((s >= 0xA2AB && s <= 0xA9FE) || (s >= 0xD7FA && s <= 0xD7FE) || (s >= 0xFE50 && s <= 0xFEA0)) {
443 for (int i = 0; i < mbfl_gb18030_pua_tbl_max; i++) {
444 if (s >= mbfl_gb18030_pua_tbl[i][2] && s <= mbfl_gb18030_pua_tbl[i][2] + mbfl_gb18030_pua_tbl[i][1] - mbfl_gb18030_pua_tbl[i][0]) {
445 *out++ = s - mbfl_gb18030_pua_tbl[i][2] + mbfl_gb18030_pua_tbl[i][0];
446 goto next_iteration;
447 }
448 }
449 }
450
451 if ((c >= 0xA1 && c <= 0xA9 && c2 >= 0xA1 && c2 <= 0xFE) ||
452 (c >= 0xB0 && c <= 0xf7 && c2 >= 0xa1 && c2 <= 0xfe) ||
453 (c >= 0x81 && c <= 0xa0 && c2 >= 0x40 && c2 <= 0xfe && c2 != 0x7f) ||
454 (c >= 0xAA && c <= 0xfe && c2 >= 0x40 && c2 <= 0xa0 && c2 != 0x7f) ||
455 (c >= 0xA8 && c <= 0xa9 && c2 >= 0x40 && c2 <= 0xa0 && c2 != 0x7F)) {
456 unsigned int w = (c - 0x81)*192 + c2 - 0x40;
457 ZEND_ASSERT(w < cp936_ucs_table_size);
458 *out++ = cp936_ucs_table[w];
459 } else {
460 *out++ = MBFL_BAD_INPUT;
461 }
462 }
463 } else {
464 *out++ = MBFL_BAD_INPUT;
465 }
466 next_iteration: ;
467 }
468
469 *in_len = e - p;
470 *in = p;
471 return out - buf;
472 }
473
mb_wchar_to_gb18030(uint32_t * in,size_t len,mb_convert_buf * buf,bool end)474 static void mb_wchar_to_gb18030(uint32_t *in, size_t len, mb_convert_buf *buf, bool end)
475 {
476 unsigned char *out, *limit;
477 MB_CONVERT_BUF_LOAD(buf, out, limit);
478 MB_CONVERT_BUF_ENSURE(buf, out, limit, len);
479
480 while (len--) {
481 uint32_t w = *in++;
482 unsigned int s = 0;
483
484 if (w == 0) {
485 out = mb_convert_buf_add(out, 0);
486 continue;
487 } else if (w >= ucs_a1_cp936_table_min && w < ucs_a1_cp936_table_max) {
488 if (w == 0x1F9) {
489 s = 0xA8Bf;
490 } else {
491 s = ucs_a1_cp936_table[w - ucs_a1_cp936_table_min];
492 }
493 } else if (w >= ucs_a2_cp936_table_min && w < ucs_a2_cp936_table_max) {
494 if (w == 0x20AC) { /* Euro sign */
495 s = 0xA2E3;
496 } else {
497 s = ucs_a2_cp936_table[w - ucs_a2_cp936_table_min];
498 }
499 } else if (w >= ucs_a3_cp936_table_min && w < ucs_a3_cp936_table_max) {
500 s = ucs_a3_cp936_table[w - ucs_a3_cp936_table_min];
501 } else if (w >= ucs_i_cp936_table_min && w < ucs_i_cp936_table_max) {
502 s = ucs_i_cp936_table[w - ucs_i_cp936_table_min];
503 } else if (w >= ucs_ci_cp936_table_min && w < ucs_ci_cp936_table_max) {
504 /* U+F900-U+FA2F CJK Compatibility Ideographs */
505 if (w == 0xF92C) {
506 s = 0xFD9C;
507 } else if (w == 0xF979) {
508 s = 0xFD9D;
509 } else if (w == 0xF995) {
510 s = 0xFD9E;
511 } else if (w == 0xF9E7) {
512 s = 0xFD9F;
513 } else if (w == 0xF9F1) {
514 s = 0xFDA0;
515 } else if (w >= 0xFA0C && w <= 0xFA29) {
516 s = ucs_ci_s_cp936_table[w - 0xFA0C];
517 }
518 } else if (w >= ucs_cf_cp936_table_min && w < ucs_cf_cp936_table_max) {
519 /* CJK Compatibility Forms */
520 s = ucs_cf_cp936_table[w - ucs_cf_cp936_table_min];
521 } else if (w >= ucs_sfv_cp936_table_min && w < ucs_sfv_cp936_table_max) {
522 /* U+FE50-U+FE6F Small Form Variants */
523 s = ucs_sfv_cp936_table[w - ucs_sfv_cp936_table_min];
524 } else if (w >= ucs_hff_cp936_table_min && w < ucs_hff_cp936_table_max) {
525 /* U+FF00-U+FFFF HW/FW Forms */
526 if (w == 0xFF04) {
527 s = 0xA1E7;
528 } else if (w == 0xFF5E) {
529 s = 0xA1AB;
530 } else if (w >= 0xFF01 && w <= 0xFF5D) {
531 s = w - 0xFF01 + 0xA3A1;
532 } else if (w >= 0xFFE0 && w <= 0xFFE5) {
533 s = ucs_hff_s_cp936_table[w - 0xFFE0];
534 }
535 } else if (w >= 0xE000 && w <= 0xE864) {
536 /* PUA */
537 if (w < 0xE766) {
538 if (w < 0xE4C6) {
539 unsigned int c1 = w - 0xE000;
540 s = (c1 % 94) + 0xA1;
541 c1 /= 94;
542 s |= (c1 + (c1 < 0x06 ? 0xAA : 0xF2)) << 8;
543 } else {
544 unsigned int c1 = w - 0xE4C6;
545 s = ((c1 / 96) + 0xA1) << 8;
546 c1 %= 96;
547 s |= c1 + (c1 >= 0x3F ? 0x41 : 0x40);
548 }
549 } else {
550 /* U+E766-U+E864 */
551 unsigned int k1 = 0, k2 = mbfl_gb18030_pua_tbl_max;
552 while (k1 < k2) {
553 unsigned int k = (k1 + k2) >> 1;
554 if (w < mbfl_gb18030_pua_tbl[k][0]) {
555 k2 = k;
556 } else if (w > mbfl_gb18030_pua_tbl[k][1]) {
557 k1 = k + 1;
558 } else {
559 s = w - mbfl_gb18030_pua_tbl[k][0] + mbfl_gb18030_pua_tbl[k][2];
560 break;
561 }
562 }
563 }
564 }
565
566 /* While GB18030 and CP936 are very similar, some mappings are different between these encodings;
567 * do a binary search in a table of differing codepoints to see if we have one */
568 if (!s && w >= mbfl_gb18030_c_tbl_key[0] && w <= mbfl_gb18030_c_tbl_key[mbfl_gb18030_c_tbl_max-1]) {
569 int i = mbfl_bisec_srch2(w, mbfl_gb18030_c_tbl_key, mbfl_gb18030_c_tbl_max);
570 if (i >= 0) {
571 s = mbfl_gb18030_c_tbl_val[i];
572 }
573 }
574
575 /* If we have not yet found a suitable mapping for this codepoint, it requires a 4-byte code */
576 if (!s && w >= 0x80 && w <= 0xFFFF) {
577 /* BMP */
578 int i = mbfl_bisec_srch(w, mbfl_uni2gb_tbl, mbfl_gb_uni_max);
579 if (i >= 0) {
580 unsigned int c1 = w - mbfl_gb_uni_ofst[i];
581 s = (c1 % 10) + 0x30;
582 c1 /= 10;
583 s |= ((c1 % 126) + 0x81) << 8;
584 c1 /= 126;
585 s |= ((c1 % 10) + 0x30) << 16;
586 c1 /= 10;
587 s |= (c1 + 0x81) << 24;
588 }
589 } else if (w >= 0x10000 && w <= 0x10FFFF) {
590 /* Code set 3: Unicode U+10000-U+10FFFF */
591 unsigned int c1 = w - 0x10000;
592 s = (c1 % 10) + 0x30;
593 c1 /= 10;
594 s |= ((c1 % 126) + 0x81) << 8;
595 c1 /= 126;
596 s |= ((c1 % 10) + 0x30) << 16;
597 c1 /= 10;
598 s |= (c1 + 0x90) << 24;
599 }
600
601 if (!s) {
602 MB_CONVERT_ERROR(buf, out, limit, w, mb_wchar_to_gb18030);
603 MB_CONVERT_BUF_ENSURE(buf, out, limit, len);
604 } else if (s < 0x80) {
605 out = mb_convert_buf_add(out, s);
606 } else if (s > 0xFFFFFF) {
607 MB_CONVERT_BUF_ENSURE(buf, out, limit, len + 4);
608 out = mb_convert_buf_add4(out, (s >> 24) & 0xFF, (s >> 16) & 0xFF, (s >> 8) & 0xFF, s & 0xFF);
609 } else {
610 MB_CONVERT_BUF_ENSURE(buf, out, limit, len + 2);
611 out = mb_convert_buf_add2(out, (s >> 8) & 0xFF, s & 0xFF);
612 }
613 }
614
615 MB_CONVERT_BUF_STORE(buf, out, limit);
616 }
617
