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.c
26 * by moriyoshi koizumi <moriyoshi@php.net> on 4 dec 2002.
27 *
28 */
29
30 #ifdef HAVE_CONFIG_H
31 #include "config.h"
32 #endif
33
34 #include "mbfilter.h"
35 #include "mbfilter_utf7.h"
36
37 static int mbfl_filt_ident_utf7(int c, mbfl_identify_filter *filter);
38
39 static const unsigned char mbfl_base64_table[] = {
40 /* 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', */
41 0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x4b,0x4c,0x4d,
42 /* 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', */
43 0x4e,0x4f,0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5a,
44 /* 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', */
45 0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,
46 /* 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', */
47 0x6e,0x6f,0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,
48 /* '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/', '\0' */
49 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0x2b,0x2f,0x00
50 };
51
52 static const char *mbfl_encoding_utf7_aliases[] = {"utf7", NULL};
53
54 const mbfl_encoding mbfl_encoding_utf7 = {
55 mbfl_no_encoding_utf7,
56 "UTF-7",
57 "UTF-7",
58 (const char *(*)[])&mbfl_encoding_utf7_aliases,
59 NULL,
60 MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE | MBFL_ENCTYPE_GL_UNSAFE,
61 &vtbl_utf7_wchar,
62 &vtbl_wchar_utf7
63 };
64
65 const struct mbfl_identify_vtbl vtbl_identify_utf7 = {
66 mbfl_no_encoding_utf7,
67 mbfl_filt_ident_common_ctor,
68 mbfl_filt_ident_common_dtor,
69 mbfl_filt_ident_utf7
70 };
71
72 const struct mbfl_convert_vtbl vtbl_utf7_wchar = {
73 mbfl_no_encoding_utf7,
74 mbfl_no_encoding_wchar,
75 mbfl_filt_conv_common_ctor,
76 mbfl_filt_conv_common_dtor,
77 mbfl_filt_conv_utf7_wchar,
78 mbfl_filt_conv_common_flush
79 };
80
81 const struct mbfl_convert_vtbl vtbl_wchar_utf7 = {
82 mbfl_no_encoding_wchar,
83 mbfl_no_encoding_utf7,
84 mbfl_filt_conv_common_ctor,
85 mbfl_filt_conv_common_dtor,
86 mbfl_filt_conv_wchar_utf7,
87 mbfl_filt_conv_wchar_utf7_flush
88 };
89
90
91 #define CK(statement) do { if ((statement) < 0) return (-1); } while (0)
92
93 /*
94 * UTF-7 => wchar
95 */
mbfl_filt_conv_utf7_wchar(int c,mbfl_convert_filter * filter)96 int mbfl_filt_conv_utf7_wchar(int c, mbfl_convert_filter *filter)
97 {
98 int s, n;
99
100 n = -1;
101 if (filter->status != 0) { /* Modified Base64 */
102 if (c >= 0x41 && c <= 0x5a) { /* A - Z */
103 n = c - 65;
104 } else if (c >= 0x61 && c <= 0x7a) { /* a - z */
105 n = c - 71;
106 } else if (c >= 0x30 && c <= 0x39) { /* 0 - 9 */
107 n = c + 4;
108 } else if (c == 0x2b) { /* '+' */
109 n = 62;
110 } else if (c == 0x2f) { /* '/' */
111 n = 63;
112 }
113 if (n < 0 || n > 63) {
114 if (c == 0x2d) {
115 if (filter->status == 1) { /* "+-" -> "+" */
116 CK((*filter->output_function)(0x2b, filter->data));
117 }
118 } else if (c >= 0 && c < 0x80) { /* ASCII exclude '-' */
119 CK((*filter->output_function)(c, filter->data));
120 } else { /* illegal character */
121 s = c & MBFL_WCSGROUP_MASK;
122 s |= MBFL_WCSGROUP_THROUGH;
123 CK((*filter->output_function)(s, filter->data));
124 }
125 filter->cache = 0;
126 filter->status = 0;
127 return c;
128 }
129 }
130
131 switch (filter->status) {
132 /* directly encoded characters */
133 case 0:
134 if (c == 0x2b) { /* '+' shift character */
135 filter->status = 1;
136 } else if (c >= 0 && c < 0x80) { /* ASCII */
137 CK((*filter->output_function)(c, filter->data));
138 } else { /* illegal character */
139 s = c & MBFL_WCSGROUP_MASK;
140 s |= MBFL_WCSGROUP_THROUGH;
141 CK((*filter->output_function)(s, filter->data));
142 }
143 break;
144
145 /* decode Modified Base64 */
146 case 1:
147 case 2:
148 filter->cache |= n << 10;
149 filter->status = 3;
150 break;
151 case 3:
152 filter->cache |= n << 4;
153 filter->status = 4;
154 break;
155 case 4:
156 s = ((n >> 2) & 0xf) | (filter->cache & 0xffff);
157 n = (n & 0x3) << 14;
158 filter->status = 5;
159 if (s >= 0xd800 && s < 0xdc00) {
160 s = (((s & 0x3ff) << 16) + 0x400000) | n;
161 filter->cache = s;
162 } else if (s >= 0xdc00 && s < 0xe000) {
163 s &= 0x3ff;
164 s |= (filter->cache & 0xfff0000) >> 6;
165 filter->cache = n;
166 if (s >= MBFL_WCSPLANE_SUPMIN && s < MBFL_WCSPLANE_SUPMAX) {
167 CK((*filter->output_function)(s, filter->data));
168 } else { /* illegal character */
169 s &= MBFL_WCSGROUP_MASK;
170 s |= MBFL_WCSGROUP_THROUGH;
171 CK((*filter->output_function)(s, filter->data));
172 }
173 } else {
174 filter->cache = n;
175 CK((*filter->output_function)(s, filter->data));
176 }
177 break;
178
179 case 5:
180 filter->cache |= n << 8;
181 filter->status = 6;
182 break;
183 case 6:
184 filter->cache |= n << 2;
185 filter->status = 7;
186 break;
187 case 7:
188 s = ((n >> 4) & 0x3) | (filter->cache & 0xffff);
189 n = (n & 0xf) << 12;
190 filter->status = 8;
191 if (s >= 0xd800 && s < 0xdc00) {
192 s = (((s & 0x3ff) << 16) + 0x400000) | n;
193 filter->cache = s;
194 } else if (s >= 0xdc00 && s < 0xe000) {
195 s &= 0x3ff;
196 s |= (filter->cache & 0xfff0000) >> 6;
197 filter->cache = n;
198 if (s >= MBFL_WCSPLANE_SUPMIN && s < MBFL_WCSPLANE_SUPMAX) {
199 CK((*filter->output_function)(s, filter->data));
200 } else { /* illegal character */
201 s &= MBFL_WCSGROUP_MASK;
202 s |= MBFL_WCSGROUP_THROUGH;
203 CK((*filter->output_function)(s, filter->data));
204 }
205 } else {
206 filter->cache = n;
207 CK((*filter->output_function)(s, filter->data));
208 }
209 break;
210
211 case 8:
212 filter->cache |= n << 6;
213 filter->status = 9;
214 break;
215 case 9:
216 s = n | (filter->cache & 0xffff);
217 filter->status = 2;
218 if (s >= 0xd800 && s < 0xdc00) {
219 s = (((s & 0x3ff) << 16) + 0x400000);
220 filter->cache = s;
221 } else if (s >= 0xdc00 && s < 0xe000) {
222 s &= 0x3ff;
223 s |= (filter->cache & 0xfff0000) >> 6;
224 filter->cache = 0;
225 if (s >= MBFL_WCSPLANE_SUPMIN && s < MBFL_WCSPLANE_SUPMAX) {
226 CK((*filter->output_function)(s, filter->data));
227 } else { /* illegal character */
228 s &= MBFL_WCSGROUP_MASK;
229 s |= MBFL_WCSGROUP_THROUGH;
230 CK((*filter->output_function)(s, filter->data));
231 }
232 } else {
233 filter->cache = 0;
234 CK((*filter->output_function)(s, filter->data));
235 }
236 break;
237
238 default:
239 filter->status = 0;
240 break;
241 }
242
243 return c;
244 }
245
246 /*
247 * wchar => UTF-7
248 */
mbfl_filt_conv_wchar_utf7(int c,mbfl_convert_filter * filter)249 int mbfl_filt_conv_wchar_utf7(int c, mbfl_convert_filter *filter)
250 {
251 int s, n;
252
253 n = 0;
254 if (c >= 0 && c < 0x80) { /* ASCII */
255 if (c >= 0x41 && c <= 0x5a) { /* A - Z */
256 n = 1;
257 } else if (c >= 0x61 && c <= 0x7a) { /* a - z */
258 n = 1;
259 } else if (c >= 0x30 && c <= 0x39) { /* 0 - 9 */
260 n = 1;
261 } else if (c == '\0') { /* '\0' */
262 n = 1;
263 } else if (c == 0x2f) { /* '/' */
264 n = 1;
265 } else if (c == 0x2d) { /* '-' */
266 n = 1;
267 } else if (c == 0x20) { /* SPACE */
268 n = 2;
269 } else if (c == 0x09) { /* HTAB */
270 n = 2;
271 } else if (c == 0x0d) { /* CR */
272 n = 2;
273 } else if (c == 0x0a) { /* LF */
274 n = 2;
275 } else if (c == 0x27) { /* "'" */
276 n = 2;
277 } else if (c == 0x28) { /* '(' */
278 n = 2;
279 } else if (c == 0x29) { /* ')' */
280 n = 2;
281 } else if (c == 0x2c) { /* ',' */
282 n = 2;
283 } else if (c == 0x2e) { /* '.' */
284 n = 2;
285 } else if (c == 0x3a) { /* ':' */
286 n = 2;
287 } else if (c == 0x3f) { /* '?' */
288 n = 2;
289 }
290 } else if (c >= 0 && c < MBFL_WCSPLANE_UCS2MAX) {
291 ;
292 } else if (c >= MBFL_WCSPLANE_SUPMIN && c < MBFL_WCSPLANE_SUPMAX) {
293 s = ((c >> 10) - 0x40) | 0xd800;
294 CK((*filter->filter_function)(s, filter));
295 s = (c & 0x3ff) | 0xdc00;
296 CK((*filter->filter_function)(s, filter));
297 return c;
298 } else {
299 if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
300 CK(mbfl_filt_conv_illegal_output(c, filter));
301 }
302 return c;
303 }
304
305 switch (filter->status) {
306 case 0:
307 if (n != 0) { /* directly encode characters */
308 CK((*filter->output_function)(c, filter->data));
309 } else { /* Modified Base64 */
310 CK((*filter->output_function)(0x2b, filter->data)); /* '+' */
311 filter->status++;
312 filter->cache = c;
313 }
314 break;
315
316 /* encode Modified Base64 */
317 case 1:
318 s = filter->cache;
319 CK((*filter->output_function)(mbfl_base64_table[(s >> 10) & 0x3f], filter->data));
320 CK((*filter->output_function)(mbfl_base64_table[(s >> 4) & 0x3f], filter->data));
321 if (n != 0) {
322 CK((*filter->output_function)(mbfl_base64_table[(s << 2) & 0x3c], filter->data));
323 if (n == 1) {
324 CK((*filter->output_function)(0x2d, filter->data)); /* '-' */
325 }
326 CK((*filter->output_function)(c, filter->data));
327 filter->status = 0;
328 } else {
329 filter->status++;
330 filter->cache = ((s & 0xf) << 16) | c;
331 }
332 break;
333
334 case 2:
335 s = filter->cache;
336 CK((*filter->output_function)(mbfl_base64_table[(s >> 14) & 0x3f], filter->data));
337 CK((*filter->output_function)(mbfl_base64_table[(s >> 8) & 0x3f], filter->data));
338 CK((*filter->output_function)(mbfl_base64_table[(s >> 2) & 0x3f], filter->data));
339 if (n != 0) {
340 CK((*filter->output_function)(mbfl_base64_table[(s << 4) & 0x30], filter->data));
341 if (n == 1) {
342 CK((*filter->output_function)(0x2d, filter->data)); /* '-' */
343 }
344 CK((*filter->output_function)(c, filter->data));
345 filter->status = 0;
346 } else {
347 filter->status++;
348 filter->cache = ((s & 0x3) << 16) | c;
349 }
350 break;
351
352 case 3:
353 s = filter->cache;
354 CK((*filter->output_function)(mbfl_base64_table[(s >> 12) & 0x3f], filter->data));
355 CK((*filter->output_function)(mbfl_base64_table[(s >> 6) & 0x3f], filter->data));
356 CK((*filter->output_function)(mbfl_base64_table[s & 0x3f], filter->data));
357 if (n != 0) {
358 if (n == 1) {
359 CK((*filter->output_function)(0x2d, filter->data)); /* '-' */
360 }
361 CK((*filter->output_function)(c, filter->data));
362 filter->status = 0;
363 } else {
364 filter->status = 1;
365 filter->cache = c;
366 }
367 break;
368
369 default:
370 filter->status = 0;
371 break;
372 }
373
374 return c;
375
376 }
377
mbfl_filt_conv_wchar_utf7_flush(mbfl_convert_filter * filter)378 int mbfl_filt_conv_wchar_utf7_flush(mbfl_convert_filter *filter)
379 {
380 int status, cache;
381
382 status = filter->status;
383 cache = filter->cache;
384 filter->status = 0;
385 filter->cache = 0;
386 /* flush fragments */
387 switch (status) {
388 case 1:
389 CK((*filter->output_function)(mbfl_base64_table[(cache >> 10) & 0x3f], filter->data));
390 CK((*filter->output_function)(mbfl_base64_table[(cache >> 4) & 0x3f], filter->data));
391 CK((*filter->output_function)(mbfl_base64_table[(cache << 2) & 0x3c], filter->data));
392 CK((*filter->output_function)(0x2d, filter->data)); /* '-' */
393 break;
394
395 case 2:
396 CK((*filter->output_function)(mbfl_base64_table[(cache >> 14) & 0x3f], filter->data));
397 CK((*filter->output_function)(mbfl_base64_table[(cache >> 8) & 0x3f], filter->data));
398 CK((*filter->output_function)(mbfl_base64_table[(cache >> 2) & 0x3f], filter->data));
399 CK((*filter->output_function)(mbfl_base64_table[(cache << 4) & 0x30], filter->data));
400 CK((*filter->output_function)(0x2d, filter->data)); /* '-' */
401 break;
402
403 case 3:
404 CK((*filter->output_function)(mbfl_base64_table[(cache >> 12) & 0x3f], filter->data));
405 CK((*filter->output_function)(mbfl_base64_table[(cache >> 6) & 0x3f], filter->data));
406 CK((*filter->output_function)(mbfl_base64_table[cache & 0x3f], filter->data));
407 CK((*filter->output_function)(0x2d, filter->data)); /* '-' */
408 break;
409 }
410
411 if (filter->flush_function != NULL) {
412 (*filter->flush_function)(filter->data);
413 }
414
415 return 0;
416 }
417
mbfl_filt_ident_utf7(int c,mbfl_identify_filter * filter)418 static int mbfl_filt_ident_utf7(int c, mbfl_identify_filter *filter)
419 {
420 int n;
421
422 switch (filter->status) {
423 /* directly encoded characters */
424 case 0:
425 if (c == 0x2b) { /* '+' shift character */
426 filter->status++;
427 } else if (c == 0x5c || c == 0x7e || c < 0 || c > 0x7f) { /* illegal character */
428 filter->flag = 1; /* bad */
429 }
430 break;
431
432 /* Modified Base64 */
433 case 1:
434 case 2:
435 n = 0;
436 if (c >= 0x41 && c <= 0x5a) { /* A - Z */
437 n = 1;
438 } else if (c >= 0x61 && c <= 0x7a) { /* a - z */
439 n = 1;
440 } else if (c >= 0x30 && c <= 0x39) { /* 0 - 9 */
441 n = 1;
442 } else if (c == 0x2b) { /* '+' */
443 n = 1;
444 } else if (c == 0x2f) { /* '/' */
445 n = 1;
446 }
447 if (n <= 0) {
448 if (filter->status == 1 && c != 0x2d) {
449 filter->flag = 1; /* bad */
450 } else if (c < 0 || c > 0x7f) {
451 filter->flag = 1; /* bad */
452 }
453 filter->status = 0;
454 } else {
455 filter->status = 2;
456 }
457 break;
458
459 default:
460 filter->status = 0;
461 break;
462 }
463
464 return c;
465 }
466
